AGC kinases and MAB4/MEL proteins maintain PIN polarity by limiting lateral diffusion in plant cells

Abstract

Polar subcellular localization of the PIN exporters of the phytohormone auxin is a key determinant of directional, intercellular auxin transport and thus a central topic of both plant cell and developmental biology. Arabidopsis mutants lacking PID, a kinase that phosphorylates PINs, or the MAB4/MEL proteins of unknown molecular function display PIN polarity defects and phenocopy pin mutants, but mechanistic insights into how these factors convey PIN polarity are missing. Here, by combining protein biochemistry with quantitative live-cell imaging, we demonstrate that PINs, MAB4/MELs, and AGC kinases interact in the same complex at the plasma membrane. MAB4/MELs are recruited to the plasma membrane by the PINs and in concert with the AGC kinases maintain PIN polarity through limiting lateral diffusion-based escape of PINs from the polar domain. The PIN-MAB4/MEL-PID protein complex has self-reinforcing properties thanks to positive feedback between AGC kinase-mediated PIN phosphorylation and MAB4/MEL recruitment. We thus uncover the molecular mechanism by which AGC kinases and MAB4/MEL proteins regulate PIN localization and plant development.

Keywords: Arabidopsis; cell polarity; lateral diffusion; plant development; polar auxin transport; positive feedback; protein phosphorylation.

Graphical abstract

Figure 1

MEL1 is recruited to the PM by PINs in planta (A) Expression pattern and subcellular localization of MEL1::MEL1-GFP and PIN2::MEL1-mCherry in the eir1-1 (pin2) background and after a backcross with the WT (Col-0). CO, cortex; EP, epidermis. The images are representative of 8 and 9 roots from 3 independent experiments (MEL1::MEL1-GFP), or 9 and 6 roots from 2 independent experiments (PIN2::MEL1-mCherry), respectively. The images are rotated 90° counterclockwise relative to the direction of growth. (B) PIN2::MEL1-mCherry translational reporter localized to ectopic membrane aggregations in epidermal cells of the pin2 mutant, instead of the apical PM as in the WT (compare to A). Introducing PIN2::PIN2-GFP into PIN2::MEL1-mCherry/pin2 restored WT-like apical PM localization of MEL1-mCherry, whereas the basally localized PIN1-GFP2 expressed from the PIN2 promoter caused MEL1-mCherry to localize basally. Images are representative of 18, 15, and 20 roots, respectively, from 4 independent experiments. Scale bars, 10 μm. See also Figure S1.

Figure 2

PINs recruit MEL1 to the PM in a phosphorylation-enhanced manner (A) Overexpression of PID led to a basal-to-apical switch of MEL1::MEL1-GFP localization in the cortex (co) cells, without affecting its apical localization in the epidermis (ep). Arrowheads indicate predominant MEL1-GFP localization in the cortex. The images are representative of 32 (Col-0) and 24 (35S::PID) roots analyzed in 2 independent experiments. (B) MEL1::MEL1-GFP in epidermal cells localized to the apical PM in Col-0 roots, whereas it displayed a range of localization defects in the pid wag1 wag2 background, including lateral, apolar, basal, and cytoplasmic localizations. Arrowheads indicate predominant MEL1-GFP PM localization; the asterisk indicates predominantly cytoplasmic localization. The images are representative of 4 independently transformed T1 plants per genotype. (C) BFA treatment (50 μM, 1 h) had no effect on apically localized MEL1-mCherry in the epidermis in the PIN2::PIN2-GFP background, as reported previously for MEL1::MEL1-GFP in the WT. However, basally localized MEL1-mCherry in the epidermis of the PIN2::PIN1-GFP2 background was largely dissociated from the PM upon BFA treatment (compare to Figure 1B). (D) Quantification of (C). The graph shows the ratios of PM/cytoplasm signal intensities. n indicates the number of cells from 4 different roots. The experiment was repeated independently twice with comparable results. (E) PIN2-Venus expressed from the PIN2 promoter restored WT-like apical PM localization of MEL1-mCherry in the pin2 mutant similar to PIN2-GFP, whereas the non-phosphorylatable PIN2^SA-Venus largely failed to do so (compare to Figure 1B). Images are representative of 16 roots per genotype analyzed in 3 independent experiments. (F) Scatterplot representation of PIN2-Venus and MEL1-mCherry colocalization in the images shown in (E). (G) Quantification of (E) and (F). The data from all experiments were pooled; each R value represents >20 cells from one root. Scale bars, 10 μm. See also Figure S2.

Figure 3

PINs, MAB4/MELs, and PID/WAGs physically interact with each other (A) In vitro pull-down of HIS-PIN2HL and/or HIS-PID with GST (negative control; left three lanes) or GST-MAB4 (middle three lanes). The input of HIS-tagged protein is shown in the right two lanes. The blot is representative of three independent experiments. The corresponding full western blot and Coomassie stain images are shown in Figures S3H and S3I. (B) Quantification of (A) and two independent additional experiments. Band intensities corrected for background intensity are shown. (C) In vitro pull-down of HIS-PID with GST-MEL1, GST-MAB4, and GST only, or non-induced GST-MEL1 lysate (−) as controls. Biologically independent lysates were used for the two GST-MEL1 lanes; the blot is representative of two technical replicates. (D) Quantification of (C) and one additional experiment. The GST-MEL1 group contains 4 data points, as two independent lysates were used in each experiment. Band intensities corrected for background intensity are shown. (E) In vitro pull-down of HIS-PIN2HL with GST-PID, GST-WAG1, and GST-WAG2, and GST only as control. The corresponding Coomassie stain is shown in Figure S3H. (F) Quantification of (E) and two independent additional experiments. Band intensities corrected for background intensity are shown. (G) In vivo FLIM-FRET imaging of PIN2::PIN2-GFP in the absence or presence of PIN2::MEL1-mCherry. Scale bar, 10 μm. (H) Quantitative analysis of (G). n indicates the total number of roots from 3 independent experiments. See also Figure S3.

Figure 4

MAB4/MELs interact with D6PK and promote PIN1 phosphorylation (A) In vitro pull-down of HIS-D6PK with GST- (negative control; 6^th lane) or GST-MEL1 (last lane). The input is shown in the left three lanes. The anti-HIS western blot and the Coomassie stain showing the loading with GST- are representative of three independent experiments. Note that the loading with GST-MEL1 can be seen on the anti-HIS western blot, as the anti-HIS antibody most likely recognizes the 5HIS stretch in the MEL1 sequence (aa 72–76). (B) Quantification of (A) and two independent additional experiments. Relative HIS-D6PK band intensities are shown. (C) In situ immunolocalization of PIN1 (green) and PIN1 phosphorylated at the S1 residue (magenta) in wild-type (Col-0) or mel1234 mutant root stele cells. Scale bars, 10 μm. (D) Quantitative analysis of (C). The boxplot shows the ratio of the PIN1-S1P/PIN1 signals at the PMs. n indicates the number of cells from five different roots. From 6 biological replicates in 3 independent experiments (3 with the S1P and 3 with the S4P antibody, which behaved identically under all conditions tested thus far³⁵), 4 showed comparable results, 1 showed no significant difference between the genotypes, and 1 showed an opposite trend. See also Figure S4.

Figure 5

MAB4/MEL proteins and PID/WAG kinases reduce PIN lateral diffusion (A) FRAP dynamics of PIN2-Venus in Col-0 and pid wag1 wag2 in root epidermis cells. (B) Quantitative analysis of (A). The experiment was repeated independently twice with comparable results. (C) FRAP dynamics of PIN2-Venus (WT) and PIN2^SA-Venus (SA) in root epidermis cells. The WT images are the same as the mock control in Figure S5C. (D) Quantitative analysis of (C). The WT control is the same as the mock control in Figure S5E. The experiment was repeated independently twice with comparable results. (E) FRAP dynamics of PIN2-GFP in the WT (Col-0) and mel1234 mutant root epidermis cells. (F) Quantitative analysis of (E). The experiment was repeated independently 3 times with comparable results. The violin plots (B, D, and F) show median values and probability density of the data after background subtraction and correction to photobleaching caused by iterative imaging. n refers to the number of cells from three different roots. Scale bars, 10 μm. See also Figures S5 and S6.

Figure 6

MEL1-mCherry diffuses fast compared to PIN2-GFP (A) FRAP dynamics of PIN2-GFP and MEL1-mCherry in the same root epidermis cells. Scale bar, 10 μm. (B) Quantitative analysis of (A). n indicates the number of cells from 3 different roots. The experiment was repeated independently twice with comparable results. The violin plots show median values and probability density of the data after background subtraction and correction to photobleaching caused by iterative imaging. n refers to the number of cells from three different roots.

Figure 7

Proposed model of PID-MAB4/MEL positive feedback loop mediating PIN polarity maintenance through limiting lateral diffusion Left: PID can interact with and phosphorylate the PIN hydrophilic loop (PINHL; P indicates a phospho-residue). However, without MAB4/MELs, there is still increased lateral diffusion of PINs. Middle: without PID, unphosphorylated PINHLs do attract MAB4/MELs but at a much lower efficiency, leading to more lateral diffusion. Right: when all three are present, the interaction of PID with the PINHL and the subsequent phosphorylation attract MAB4/MELs that act as scaffolds to form PIN/PID/MAB4 complexes, increasing PINHL phosphorylation and limiting lateral diffusion by PIN complex formation.