Arabidopsis thaliana (Columbia ecotype) was used as the WT plant in this study. Arabidopsis plants were transformed using Agrobacterium tumefaciens strain C58C1 (pMP90) (Bechtold and Pelletier, 1998). All seeds were vernalized at 4°C for 3 days and germinated at 22°C under a 16-h-light/8-h-dark photoperiod. Seeds were grown on the medium containing 4.3 g/L Murashige and Skoog (MS) nutrient mix (Sigma–Aldrich), 1% sucrose, 0.5 g/L MES (pH 5.7 with KOH), and 0.8% agarose. Transformed plants were selected on the hygromycin-containing (30 μg/mL) medium.

The binary vector pCAMBIA 1300-NOS with modified cloning sites (Lee et al., 2010) was used for transgene construction. ProPINs:PINs:GFP, ProPINs:M3PINs:GFP, and ProPINs:3m1PINs:GFP constructs were generated by replacing the ProE7 fragment of ProE7:PINs:GFP, ProE7:M3PINs:GFP, and ProE7:3m1PINs:GFP (Sasayama et al., 2013) with the promoter region of PINs (ProPINs). ProPINs were amplified by PCR using the Arabidopsis genomic DNA as template and the primers listed in Supplementary Table S1.

For heterologous expression of PIN-HL proteins in Escherichia coli (E. coli) for the in vitro phosphorylation assay, the PIN-HL part was obtained from Arabidopsis cDNA by using the primers in Supplementary Table S1 and cloned into the pGEX-4T-1 vector (GE Healthcare, Inchon, Korea) for fusion with glutathione S-transferase (GST) at the N-terminus. The protein kinase PID used in the kinase assay was as described in Ganguly et al. (2012a).

All constructs were confirmed by nucleotide sequencing, and at least five independent transgenic lines for each construct were analyzed.

Phyllotatic analysis was conducted by numbering the 3-week-old vegetative leaves from early to late developmental stages, and the angles between leaves were measured by the Leica Application suite (v.2.8.1). The seed number per silique was measured under the Leica MZFLIII dissecting microscope. For the root gravitropism assay, 3-day-old seedlings grown vertically in the light were turned 90° for the indicated period of time (2–10 h) in the dark condition before observing root bending.

Observation of fluorescent reporters was conducted as described previously (Ganguly et al., 2010; Ganguly et al., 2012a). To observe cytological effects of brefeldin A (BFA), seedlings were pretreated with cycloheximide (50 μM) for 30 min and followed by BFA (25 μM) treatment for 30 min. BFA was dissolved in DMSO (0.05% final), and control treatments included 0.05% DMSO. Quantification of PM-localized PIN:GFP signals was performed using the histogram function of Adobe Photoshop CS6 (Adobe Systems).

To observe gravitropic changes of PIN7:GFP localization in root columella cells, 4-day-old seedlings were transferred to a slide glass containing half-strength MS medium and grown for 24 h vertically in the dark condition. Gravity stimulation was applied for the indicated time period by turning the slide 90°, subsequently the changes of PIN7:GFP signal were analyzed under a confocal microscope. To visualize cell boundaries, FM4-64 (2 μM) was applied for 3 min.

Expression of GST-fused kinase (PID) and substrates (WT and M3-mutated PIN1-HL) in E. coli (BL21DE3) cells was induced by 1 mM isopropyl β-D-1-thiogalactopyranoside for 3 h at 28°C, and the cells were harvested by centrifugation at 3900 g for 10 min. The harvested cells were lysed with B-PER bacterial protein extraction reagent (Thermo Scientific, Seoul, Korea) with the Halt protease inhibitor complex (Santa Cruz), and the GST-fused proteins were purified using the GST-bound agarose resin (Elpis biotech).

For in vitro phosphorylation assays, l μg of kinase and substrate proteins, respectively, were mixed in 30 μl kinase buffer (25 mM Tris–HCl pH 7.5, 1 mM DTT, 5 mM MgCl₂, 100 uM sodium orthovanadate, 30 mM β-glycerophosphate) and 1 μCi [γ-32P] ATP for 1 h at 30°C and, the reaction was halted by adding 5 μl of SDS-PAGE loading buffer and boiling for 10 min. The reaction samples were then separated in 10% acrylamide gels and stained with Coomassie Brilliant Blue. Autoradiography was performed using BAS imaging plate (BAS 2040, Fujifilm) and Bio-imaging Analyzer (BAS-2500, Fujifilm). The intensity of the phosphorylation bands was estimated by ImageJ software¹ (the National Institutes of Health, Bethesda, MD, USA). To estimate the phosphorylated band intensity, the background intensity value of the same lane was subtracted from the intensity value of the phosphorylated PIN band.

In the root, PIN1 is mainly localized at the basal PM of vascular cells and the basal and inner lateral sides of pericycle and endodermal cells (Gälweiler et al., 1998). In order to study the effect of M3 and 3m1 phosphorylation sites on subcellular localization of PIN1, WT PIN1 and mutated PIN1 at the M3 or 3m1 region (M3PIN1 or 3m1PIN1; Supplementary Figure S1) of the HL domain were expressed under its own promoter (ProPIN1) in the pin1 mutant background.

In vascular cells, WT PIN1, as expected, predominantly localized to the basal PM, whereas the basal localization of M3PIN1 was disturbed so as for the mutant PIN1 to be more localized to lateral sides than WT PIN1 (Figure 1A). Moreover, M3PIN1 tended to be more internalized than WT PIN1. Statistical analysis obviously revealed the defects of M3PIN1 and 3m1PIN1 in basal localization by showing considerably decreased ratio (~40%) of apical/basal to lateral localization (Figures 1B,C). In endodermal cells, while WT PIN1 localized predominantly in basal and inner lateral sides, mutation of M3 or 3m1 disturbed this polarity by delivering these mutant PIN1 proteins to the outer lateral PM (Figures 1D–F). These results indicate that the proper subcellular PIN1 polarity requires the M3 phosphorylation residues.

When roots were treated with the recycling inhibitor BFA, both WT PIN1 and M3PIN1 formed internal BFA compartments (Figure 1G), suggesting that, though M3PIN1 tends to internalize, some M3PIN1 proteins still are able to target to the PM and properly recycle like WT PIN1. This result in root meristematic vasculature cells is consistent with our previous observation that M3PIN1 and M3PIN7 are partially internalized and accumulated in BFA compartments in the root hair cell (Sasayama et al., 2013). However, M3PIN3 was almost completely internalized so as not to form BFA compartments in the root hair cell (Ganguly et al., 2012a). On the other hand, all 3m1 mutants of PIN1, PIN3, and PIN7 formed BFA compartments in the root hair cell as they were partially internalized (Ganguly et al., 2012a; Sasayama et al., 2013). These results, together with the developmental phenotypic effects as will be mentioned, suggest that the phosphorylatable residues in the M3 site function redundantly.

In a previous study, ectopic cell division was observed in the M3PIN3-expressing root pericycle (Ganguly et al., 2012a). M3PIN1 and 3m1PIN1 also caused considerable increase of ectopic cell division in the pericycle (Figures 1H,I). It is likely that the disruption of PIN subcellular polarity generally causes ectopic cell division probably by interfering local auxin concentration gradients.

The pin1 mutant shows structural abnormalities in inflorescence axes, flowers, and leaves (Okada et al., 1991). To investigate the biological roles of M3 and 3m1 phosphorylation sites, the phenotypes of pin1 complemented with M3PIN1 and 3m1PIN1 were analyzed.

The loss of PIN1 caused aberrant cotyledon morphogenesis such as single, multiple, or fused cotyledons (Figure 2A), most likely by affecting auxin-gradient formation during embryogenesis. For the phenotypic analysis of cotyledons, the segregated F1 progeny seedlings of the self-fertilized heterozygous pin1 mutant background were observed because homozygous M3PIN1-complemented pin1 mutants barely produced seeds for next generation. While complementation with WT PIN1 greatly restored the normal cotyledon phenotype, complementation with M3PIN1 or 3m1PIN1 only partially rescued the pin1 cotyledon defects (Figure 2B). Interestingly, the phenotype ratio of ‘fused cotyledon’ was rather increased by M3 or 3m1 mutation in comparison with the complete loss of PIN1 (pin1) (Figures 2A,B). Furthermore, some severe defects such as reduced or no cotyledons were observed by the complementation with both M3PIN1 and 3m1PIN1, indicating that these PIN1 mutants, by impairing its native polarity, lead to alteration of local auxin gradients and subsequent morphogenetic changes during embryogenesis.

Because PIN1 is associated with leaf initiation and phyllotaxis (Okada et al., 1991; Reinhardt et al., 2003; Prasad et al., 2011), we examined rosette leaf phyllotaxis of the complemented lines with WT PIN1, M3PIN1 or 3m1PIN1. WT plants showed a typical phyllotactic distribution pattern with the peak at 130–140° of leaf angle (Figures 3A,B). This distribution pattern was dissipated in the homozygous pin1 mutant by considerably deviating its peak from the typical angle of 130–140°. Complementation of homozygous pin1 with WT PIN1 restored the peak at 130–140°. However, M3PIN1- or 3m1PIN1-complementation failed to restore the WT phyllotactic peak, indicating that the M3 phosphorylation site is necessary for PIN1’s function in phyllotaxis.

The homozygous pin1 mutant forms a pin-like inflorescence stem lacking lateral shoot organs (Okada et al., 1991; Figures 4A,C). M3PIN1-complemented plants grew a pin-like primary inflorescence stem that, however, was soon aborted (Figure 4C). This early abortion of the primary stem by M3PIN1 led to growth of secondary inflorescence stems (Figure 4A). In contrast, 3m1PIN1-complemented plants kept growing the primary stem, although the size was smaller than that of WT plants (Figure 4B). The M3PIN1 plants, though growing secondary stems, eventually produced aborted shoot apexes in these secondary stems (Figure 4D). The homozygous pin1 mutant did not abort the primary stem early but maintained its growth for a while (Okada et al., 1991; Figure 4A).

In addition to their effects on inflorescence stem development, M3 or 3m1 mutation also affected silique development so as to produce aberrant or no silique (Figures 4D–F). The WT plants generally produced over 1.5 cm siliques. However, the number of siliques over 1 cm was decreased considerably by 3m1 and greatly by M3 mutation (Figure 4E). Consistently, the seed number per silique also decreased by these mutations to 11% (M3) and 67% (3m1) of the WT level (40.9 seeds/silique) (Figures 4F,G).

These results together suggest more than that the M3 phosphorylation site is necessary for normal biological functions of PIN1. M3PIN1- and 3m1PIN1-complemented transformants revealed new phenotypes that are not shown in the pin1 mutant, such as generation of fused cotyledons and early abortion of the primary inflorescence stem. As we described earlier for the root tissues (Figure 1), the partial loss of polarity or altered intracellular trafficking of M3PIN1 and 3m1PIN1 could cause spatial changes of auxin gradients, which subsequently leads to developmental defects.

In the root columella cell, upon changes of gravity vector, PIN3 and PIN7 relocalize so as to redistribute auxin for gravitropic bending of the root (Friml et al., 2002b; Kleine-Vehn et al., 2010). Under a constant gravity vector, these PINs symmetrically distribute in the columella cell PM. However, changes of the gravity vector cause endocytosis of those PINs and selective recycling of them toward the PM side facing gravity direction (bottom), resulting in higher auxin flow and inhibition of cell growth at the bottom side of the root.

Here, we tested whether the M3 phosphorylation site is implicated in PIN7’s function during root gravitropism. Similarly as in the PIN1 study, WT or M3-mutated PIN7 with the GFP tag under the PIN7 promoter was complemented into the homozygous pin7 mutant. We first observed the intracellular trafficking behavior of WT PIN7 and M3PIN7 upon gravity vector change. WT PIN7 accumulated into endocytotic vesicles within 15 min after gravity stimulus of the root whereas M3PIN7 did not show obvious accumulation in the endocytotic vesicle upon the gravity stimulus (Figure 5A). Before gravity stimulus, WT PIN7 and M3PIN7 were symmetrically localized at the columella cell PM (Figure 5A). After a 40-min gravity stimulus, WT PIN7 asymmetrically redistributed to the PM side facing to the new gravity direction (Figures 5B,C). However, M3PIN7 did not change its symmetrical distribution after the same gravity stimulus (Figures 5B,C). For a statistical analysis, we took the ratio of PIN7 signal intensity between lateral (L) and basal (B) sides of the columella cell (L/B, Figures 5B,C) as a polarity parameter. While WT PIN7 significantly increased its lateral localization from 0.83 to 1.46 of L/B after the 40-min gravity stimulus, M3PIN7 maintained almost the same ratio of L/B (~1.0) regardless of the gravity stimulus, suggesting that phosphorylation of the M3 site is essential for gravity-stimulated transcytosis and relocalization of PIN7 in the root columella cell.

Next we examined whether the defects of M3PIN7 in subcellular relocalization affect the gravitropic response of the M3PIN7 plant root. The pin7 mutant plants exhibited root gravitropic defects as reported previously (Kleine-Vehn et al., 2010; Figures 5D,E). While complementation with WT PIN7 rescued root gravitropism of pin7, complementation with M3PIN7 failed to rescue (Figures 5D,E). In the gravitropic kinetics analysis of the root, the M3PIN7-complemented mutant showed relatively decreased gravitropic responses compared to WT and WT PIN7-complemented plants (Figures 5D,E). These results suggest that the M3 phosphorylation residues are necessary for PIN7-mediated root gravitropism most likely by modulating the transcytotic relocalization of PIN7 after gravity stimuli.

As conducted for PIN1 and PIN7, we analyzed the homozygous pin2 mutant plants complemented with WT PIN2, M3PIN2, or 3m1PIN2 with the GFP tag under the PIN2 promoter. WT PIN2 polarly localized to the upper side of the root epidermal cell PM as previously reported (Müller et al., 1998; Figure 6A). However, M3PIN2 showed almost no expression and 3m1PIN2 showed weak expression (Figure 6A), which is consistent with our previous result showing that M3PIN2 and 3m1PIN2, when expressed under a root hair-specific promoter (ProE7), did not express in the root hair cell (Sasayama et al., 2013). Accordingly to this results, the pin2 mutant plant complemented with M3PIN2 or 3m1PIN2 failed to restore root gravitropism whereas the complementation with WT PIN2 restored it (Figure 6B). Intriguingly, the treatment with wortmannin, an inhibitor for vacuolar trafficking, was reported to restore both M3- and 3m1PIN2 protein signals in the cytoplasm but not in the PM of the root hair cell, suggesting that the M3 phosphorylation site is implicated in regulation of vacuolar targeting of PIN2 (Sasayama et al., 2013). These results suggest that a similar phosphorylation motif of PIN-HL can play different roles depending on PIN species.

Although the phosphorylatable Ser/Thr residues in long PINs are mostly conserved, the flanking residues vary particularly in PIN2 (Supplementary Figure S1). For the M3 region around the first two Ser residues, while TVRK(/R)SNAS is the consensus sequence in other long PINs, VVK(/R)RSXAS is the consensus of PIN2 orthologs. For the third Ser, PIN2 includes NKS rather than RRS. Finally for the region around forth Thr and fifth Ser, the PIN2 consensus is TPRAS whereas other long PINs have TPRPS. Although it has remained to be characterized, these syntactic variations around the M3 phosphorylatable residues might lead PIN2 to have different trafficking behavior most likely by different reader proteins against these different syntaxes around the phosphorylation residues.

In order to know whether the 5 Ser/Thr residues of the M3 site are involved in phosphorylation of PIN1, we performed in vitro phosphorylation assays with heterologously expressed PID kinase and M3- or 3m1-mutated PIN1-HL domain as kinase substrate and [γ-³²P] ATP as a phosphate source. PID showed autophosphorylation (Figure 7). Compared with the WT PIN1-HL phosphorylation level, PID-mediated phosphorylation of M3-mutated PIN1-HL was considerably decreased (Figure 7). Conversely, 3m1 mutation marginally decreased the phosphorylation level (Figure 7). In our previous study with PIN3, M3 mutation showed a greater effect than 3m1 mutation on PID-mediated phosphorylation of PIN3-HL (Ganguly et al., 2012a). These results suggest that 5 Ser/Thr residues in the M3 site of long PINs are redundant targets of PID. Because the M3 mutation could not completely inhibit PID-mediated phosphorylation of the long PINs, PID is likely to phosphorylate other residues of the HL domain. The Ser residues of 3 repeats of TPRXS have been reported as the PID’s phosphorylation targets outside the M3 site (Dhonukshe et al., 2010; Huang et al., 2010). Therefore, this study consistently shows that the M3 site of long PIN-HL is a partial target of PID.

Although the native function of M3 phosphorylation site was studied with PIN3 (Ganguly et al., 2012a) and the role of M3 site of PIN1, 2, and 7 in subcellular trafficking was shown in the root hair model system (Sasayama et al., 2013), the native function of M3 site of PIN1, 2, and 7 in their own expression domains had not been characterized. Our current study, by analyzing the native biological roles of M3 phosphorylation site of PIN1, 2, and 7, demonstrates that the M3 phosphorylation site is generally required for proper trafficking and intracellular polarity, native biological functions, and PID-mediated phosphorylation of long PINs.