Polar-localized NPH3-like proteins regulate polarity and endocytosis of PIN-FORMED auxin efflux carriers

The phytohormone auxin is transported from its site of biosynthesis by an intercellular transport system, which is called polar auxin transport. Polar auxin transport establishes asymmetrical auxin distribution in organs and tissues. This process occurs at various developmental stages, such as apical-basal axis formation during embryogenesis, aerial organ formation, root development, vascular patterning, gravitropism, and phototropism (Rolland-Lagen, 2008; Vanneste and Friml, 2009). The auxin efflux carriers of the PIN-FORMED (PIN) family are crucial components of the polar auxin transport (Petrásek et al., 2006). PIN proteins show polar localization at the plasma membrane that correlates with and determines the direction of intercellular auxin flow (Wisniewska et al., 2006).

Several factors that are important for polar PIN delivery have been identified. These include the fact that polar PIN targeting requires a balanced sterol composition in the plasma membrane (Willemsen et al., 2003; Men et al., 2008). In addition, constitutive subcellular dynamics is important for PIN polarity. PIN proteins constitutively undergo clathrin-dependent endocytosis, GNOM-dependent recycling and retromer-dependent vacuolar targeting (Steinmann et al., 1999; Geldner et al., 2003; Jaillais et al., 2006; Dhonukshe et al., 2007; Dhonukshe et al., 2008; Jaillais et al., 2007; Kleine-Vehn et al., 2008). Even though all the mentioned factors influencing PIN polarity, knowledge about the molecular components specifically controlling this process is limited. Phosphorylation of PIN proteins (Zhang et al., 2009; Huang et al., 2010) by a Ser/Thr kinase PINOID (PID) is known to be crucial for apical PIN delivery, while protein phosphatase 2A functions antagonistically (Benjamins et al., 2001; Christensen et al., 2000; Friml et al., 2004; Michniewicz et al., 2007). The MACCHI-BOU 4/ENHANCER OF PINOID/NAKED PINS IN YUC MUTANTS 1 (MAB4/ENP/NPY1) gene was genetically identified as a factor involved in organ formation together with PID (Treml et al., 2005; Furutani et al., 2007; Cheng et al., 2007). The gene encodes a NON-PHOTOTROPIC HYPOCOTYL 3 (NPH3)-like protein of unknown function. In cotyledon development, the mab4 mutation reduces PIN1 abundance in the plasma membrane and when combined with the pid mutation it causes complete reversal of PIN1 polarity, indicating that MAB4 regulates polar auxin transport in organogenesis through the control of PIN1 localization together with PID.

In the Arabidopsis genome, four genes, MAB4/ENP/NPY1-LIKE1 (MEL1), MEL2, MEL3 and MEL4 have a noticeably higher level of homology to MAB4. Recently, these genes have been reported as NPY5, NPY3, NPY4 and NPY2, and shown to function redundantly with MAB4 in organ formation (Cheng et al., 2008). Here, we show, using genetic analysis, the functional redundancy between MAB4 and MEL genes not only in organ formation but also in root gravitropism, thus demonstrating more extensive functions of MAB4 subfamily genes. In multiple mutants of the MAB4 subfamily members, the abundance and polar localization of PIN proteins were severely reduced, leading to defective expression of DR5rev::GFP, an auxin responsive marker (Friml et al., 2003). Pharmacological analyses showed that mel mutations affected PIN2 internalization from the plasma membrane, but did not affect intracellular PIN2 trafficking. Furthermore, all MAB4 subfamily proteins are localized at the cell periphery with polarity in plants. The MAB4 polarity was almost identical to PIN polarity. These results suggest that the MAB4 subfamily proteins specifically regulate the retention of PIN proteins in the plasma membrane with polarity in auxin-related morphogenesis.

Arabidopsis thaliana ecotype Columbia (Col) was used as wild type. The following mutant alleles were used: mab4-1 (Col), mab4-2 (Col) (Furutani et al., 2007) and pin1-201 (Col) (Furutani et al., 2004). mel1-1 (GABI_027H10), mel1-2 (SAIL_792_G03), mel2-1 (SALK_072281), mel2-2 (SALK_119048), mel3-1 (SALK_142094), mel3-2 (SALK_058416), mel4-1 (SALK_023554) and mel4-2 (SALK_046452) were obtained from the ABRC (Arabidopsis Biological Resource Center) and NASC (Nottingham Arabidopsis Stock Centre) (Alonso et al., 2003; Sessions et al., 2002; Rosso et al., 2003) (see Fig. S1 in the supplementary material). pin2-201 (SAIL_177_A12) carries a T-DNA insertion at the 4th exon and is supposed to be a null allele. This allele was obtained from the Syngenta Arabidopsis Insertion Library (SAIL) and was backcrossed five times to Col prior to root gravitropism assay (Sessions et al., 2002). Plants were grown on soil as previously reported (Fukaki et al., 1996), and siliques were collected for analyses of reporter assay and embryo immunolocalization. For analyses of root gravitropism assay and reporter assay, seeds were surface sterilized and germinated on Murashige and Skoog plates, as previously described (Furutani et al., 2004).

To construct the plasmid promoter (pro)MEL:MEL-GFP, MEL1-4 promoters, the 2.9 kb, 2.9 kb, 2.9 kb and 2.7 kb 5′ sequence of respective genes, were inserted into the binary destination vector pGWB4, which contains a Gateway conversion cassette in front of the GFP-coding region (GWB vectors were kindly provided by Dr Tsuyoshi Nakagawa, Shimane University, Japan) (Nakagawa et al., 2007). Full-length cDNA of MEL1, MEL2, MEL3 and MEL4 were inserted into the pGWB4 including each promoter. For 35S:MAB4-GFP, the MAB4-coding region was inserted into the pGWB5 between the cauliflower mosaic virus 35S promoter and the GFP-coding region. The resulting plasmids were introduced into Agrobacterium tumefaciens strain MP90. proMEL1:MEL1-GFP, proMEL3:MEL3-GFP and proMEL4:MEL4-GFP vectors were transformed into the mel1-1 mel2-1 mel3-1 mel4-1 quadruple plants by the floral dip method (Clough and Bent, 1998). proMEL2:MEL2-GFP was transformed into heterozygous mab4-2 and homozygous mel2-1 plants. Transformants were selected on germination medium containing 30 μg/ml kanamycin. Homozygous lines were identified in the T3 generation, and T3 or T4 homozygous lines were used for reporter analysis.

For histological analysis, roots tips were stained with 50 μg/ml of propidium iodide (Sigma-Aldrich), and fluorescence was imaged by confocal laser-scanning microscopy (FV1000; Olympus). For confocal microscopy, dissected embryos were mounted in 7% glucose and root tips were mounted in MS liquid medium [1/2 MS salt mixture, 1% glucose and 0.05% MES (pH 5.8)]. Whole-mount immunofluorescence was performed manually using a protocol described previously (Sauer et al., 2006). Antibodies were diluted as follows: 1:500 for rabbit anti-MAB4, 1:2000 for rabbit anti-PIN2, 1:1000 for mouse anti-GFP (nacalai tesque; used in Fig. 5), 1:500 for rabbit anti-GFP (Molecular Probes; used in Fig. 6 and Fig. S12 in the supplementary material), 1:500 for Alexa488- and Alexa546-conjugated anti-mouse and rabbit secondary antibodies (Invitrogen; used in Fig. 5), and 1:500 and 1:600 for FITC- and CY3-conjugated anti-mouse and anti-rabbit secondary antibodies (Dianova; used in Fig 6 and Fig. S12 in the supplementary material), respectively. Starch granules were visualized as described previously (Willemsen et al., 1998). FRAP analysis was performed as described previously (Grebe et al., 2003; Men et al., 2008).

Drugs were exogenously applied by incubation of 5-day-old seedlings in MS liquid medium supplemented with FM4-64 [15 mM stock in dimethylsulfoxide (DMSO)] (7.5 μM), BFA (10 mM stock in DMSO) (25/50 μM), cycloheximide (CHX) (10 mM stock in ethanol) (50 μM), wortmannin (10 mM stock in DMSO) (15 μM), Tyrphostin A23 (TyrA23) (30 mM stock in DMSO) (15/30/45/60 μM), Tyrphostin A51 (TyrA51) (30 mM stock in DMSO) (45 μM) or 1-naphthalene acetic acid (NAA) (10 mM stock in DMSO) (10 μM). Control treatments contained an equivalent amount of solvent (DMSO or ethanol). For FM4-64 and BFA treatment experiments, seedlings were rinsed twice in MS liquid medium after 1 hour of treatment. For BFA washout experiments, 5-day-old seedlings were treated with 50 μM BFA for 1 hour and then washed with the MS liquid medium for 2 hours. For wortmannin treatment, seedlings were washed twice in the MS liquid medium after 2 hours of treatment. When we performed CHX-pre-treatment (45 minutes) and co-incubation with CHX and BFA or wortmannin, we obtained the same results as those without CHX pre-treatment and co-incubation. For TyrA23 or TyrA51 treatment, plants were pretreated with TyA23 or TyA51 for 30 minutes, followed by co-incubation with BFA (25 μM) and TyA23 or TyA51.

The Arabidopsis genome encodes 31 NPH3-like proteins, most of which have not been well characterized (Kimura and Kagawa, 2006). MAB4 is a member of a small subfamily consisting of five proteins of unknown function. We focus on the four family members MEL1-4, which display higher homology to MAB4 than NPH3 or RPT2 (ROOT PHOTOTROPIC RESPONSE 2) (Motchoulski and Liscum, 1999; Sakai et al., 2000) (see Fig. S1A in the supplementary material). MEL1 (At4g37590), MEL2 (At5g67440), MEL3 (At2g14820) and MEL4 (At2g23050) genes have been also recently reported by Cheng et al., who named them NPY5, NPY3, NPY2 and NPY4, respectively (Cheng et al., 2008). We identified T-DNA insertion alleles, called mel1-1, mel1-2, mel2-1, mel2-2, mel3-1, mel3-2, mel4-1 and mel4-2 (see Fig. S1B,C in the supplementary material). However, we could not identify any phenotypes at various developmental stages in each single mutant background, suggesting that MAB4 subfamily genes might function redundantly. To investigate the role of MEL genes, we combined mel mutations with the mab4 mutation. mab4 single mutants display mild defects in organ formation including cotyledons and floral organs (Furutani et al., 2007). Both mel1 and mel2 mutations enhanced the mab4 mutant phenotypes, as previously reported (Cheng et al., 2008). The mel1-1 or mel2-1 mutation caused severe defects in cotyledon and floral organ development in the mab4-2 background (data not shown). Subsequently, mab4-2 mel1-1 mel2-1 triple mutants displayed severe pin-like inflorescences (data not shown). These results indicate that MEL1 and MEL2 regulate organ formation redundantly with MAB4.

Next, we constructed multiple mutants of mel mutants. All combinations of double mutants failed to exhibit obvious phenotypes (data not shown). Among all combination of triple mutants, mel1-1 mel3-1 mel4-1 and mel2-1 mel3-1 mel4-1 triple mutants displayed mild defects of root gravitropism (data not shown). Furthermore, the mel1-1 mel2-1 mel3-1 mel4-1 quadruple mutants exhibited severer defects in root gravitropism. The wild-type roots grew in the direction of gravity (Fig. 1A,C), while the quadruple mutant roots lengthened substantially in a random direction, but responded slightly to gravity (Fig. 1B,C). However, the mutant roots had well-organized cell layers and amyroplasts in columella cells, which function as statoliths in root gravitropism, as seen in wild-type roots (Fig. 1D,E). We also obtained the same results from the other combination of mutant alleles, mel1-2, mel2-2, mel3-2 and mel4-2, although these quadruple mutants displayed milder defects in root gravitropism than did mel1-1 mel2-1 mel3-1 mel4-1 (data not shown). These results indicate that MEL1, MEL2, MEL3 and MEL4 gene products mediate the gravity-controlled orientation of the primary root growth redundantly.

The phenotypes of mab4 and mel multiple mutants are similar to those of pin mutants, namely the pin-like inflorescences in pin1 mutants and the defective root gravitropism in pin2 mutants (Fig. 1C) (Okada et al., 1991; Müller et al., 1998; Luschnig et al., 1998). Previously, it was reported that MAB4/ENP is important for PIN1 localization during cotyledon development (Treml et al., 2005; Furutani et al., 2007). To test a role for MAB4 subfamily genes in the control of PIN localization, we analyzed PIN1 localization in cotyledon primordia of mab4-2 mel1-1 mel2-1 triple mutants, and PIN1 and PIN2 localization in mel1-1 mel2-1 mel3-1 mel4-1 quadruple mutants. As mab4-2 mel1-1 mel2-1 triple mutants were sterile, we analyzed PIN1-GFP expression in embryos obtained from siliques of mab4-2^+/−mel1-1^−/− mel2-1^−/− plants. In mab4-2 embryos, the abundance of PIN1-GFP in the plasma membrane was reduced from the heart stage, as previously reported (Furutani et al., 2007). The reduction of PIN1-GFP expression was restricted to the protodermal cell layer of cotyledon primordia in the mab4-2 mutant background. By contrast, in embryos of the triple mutants, the severe reduction in the abundance of PIN1-GFP in the plasma membrane was found not only in the protodermal cell layer but also in provascular tissues of cotyledon primordia from the heart stage, although PIN1-GFP was normally expressed in the provasculature of hypocotyle and radicle (Fig. 2A-D). These results indicate that MEL1 and MEL2 regulate PIN1 localization in cotyledon development mainly in the provascular tissue in the same way that MAB4 does in the protodermal cell layer.

Next, we analyzed the localization of PIN1-GFP and PIN2-GFP in agravitropic roots of mel1-1 mel2-1 mel3-1 mel4-1 quadruple mutants. PIN1-GFP was normally expressed in the stele, pericycle and endodermis of the quadruple roots (Fig. 2E,F). However, PIN1-GFP polarity was disordered in the endodermal cell layer of the quadruple mutants. In the wild type, PIN1-GFP was mainly localized in the basal and inner lateral side of the plasma membrane in endodermal cells (Fig. 2E). Notably, in the mutants, the GFP signal was additionally detectable in the outer lateral side of the plasma membrane (Fig. 2F). To investigate the disruption of PIN1-GFP polarity in more detail, we measured the ratio of GFP intensity at the apical-basal (A-B) side and outer lateral (OL) side of the plasma membrane (see Fig. S2 in the supplementary material). In the mel1-1 mel2-1 mel3-1 mel4-1 quadruple mutants, the ratio of the A-B density to the OL density was severely lower than that in wild-type roots (Fig. 2N). Next, we analyzed the PIN2-GFP localization in the mel1-1 mel2-1 mel3-1 mel4-1 roots. In the epidermis and cortex, the abundance of polar localized PIN2-GFP in the plasma membrane was severely reduced, although PIN transcripts were not affected (Fig. 2G; see Fig. S3 in the supplementary material). In addition, we observed an increase in PIN2-GFP signal in the outer lateral side of the plasma membrane in the epidermis of the quadruple mutants compared with wild type (Fig. 2H,I). Subsequently, we measured the ratio of PIN2-GFP intensity between the A-B and OL sides of the plasma membrane in wild type and quadruple mutants. The intensity ratio of A-B side to OL side strongly decreased in the quadruple mutants (Fig. 2O). To investigate whether mel mutations specifically affect PIN localization, we analyzed the localization of the non-PIN protein EGFP-LTI6a, which is localized all over the plasma membrane, in wild type and quadruple mutants (Capel et al., 1997; Navarre and Goffeau, 2000; Cutler et al., 2000). We could not detect any differences in EGFP-LTI6a expression between wild type and the quadruple mutants (Fig. 2J-M). When we also calculated the ratio of GFP intensity of the A-B side of the plasma membrane to that of the OL side, there was no difference between them (Fig. 2P). These results indicate that MEL1, MEL2, MEL3 and MEL4 specifically regulate the polarized localization of PIN1 and PIN2 proteins in the root tip.

Next, we investigated the distribution of the auxin response in the various combinations of mab4 and mel mutations. To achieve this, we analyzed the expression pattern of the auxin responsive marker, DR5rev::GFP in mab4-2 mel1-1 mel2-1 embryos and mel1-1 mel2-1 mel3-1 mel4-1 roots. As reported previously (Friml et al., 2003), in heart-stage embryos, DR5rev::GFP was expressed in the tips of two cotyledon primordia and radicle (Fig. 3A,B), sometimes the expression was detectable in only one primordium. Later, additional signal of DR5rev::GFP was found in the provascular tissues (Fig. 3C,D). By contrast, in the mab4-2 mel1-1 mel2-1 embryos, expression of DR5rev::GFP often disappeared in the apical region of the embryo, although DR5rev::GFP was widely expressed in the basal region of the embryo (Fig. 3E,F). At the torpedo stage, the GFP signal in the provascular tissue was restricted to the tops of cotyledon primordia (Fig. 3G,H). We also found an aberrant expression pattern of DR5rev::GFP in mel1 mel2 mel3 mel4 roots. Expansion of DR5rev::GFP expression was observed in the lateral root caps, but not in wild-type roots (Fig. 3I,J). These results indicate that the establishment of auxin maxima was disturbed in mab4-2 mel1-1 mel2-1 embryos and mel1-1 mel2-1 mel3-1 mel4-1 roots correlating with the disordered polar PIN localization.

To investigate how mel mutations affect polar localization of PIN proteins, we analyzed the lateral mobility of functional PIN2-GFP at the plasma membrane in mel1 mel2 mel3 mel4 mutants by fluorescence recovery after photo-bleaching (FRAP) experiments. We bleached PIN2-GFP fluorescence from the plasma membrane and monitored FRAP in roots treated with the protein biosynthesis inhibitor CHX and energy inhibitors, which abolish energy-dependent traffic. FRAP kinetics monitoring lateral diffusion did not obviously differ between wild-type and mel1-1 mel2-1 mel3-1 mel4-1 mutant cells in which fluorescence had been bleached at the plasma membrane (see Fig. S4A-C in the supplementary material). Therefore, our findings show that mel mutations do not appear to affect the rate of PIN2-GFP lateral diffusion when membrane trafficking is blocked.

To address another possibility that mel mutations affect PIN trafficking, PIN2-GFP trafficking was analyzed in the mutants. PIN proteins continuously undergo endocytosis and recycling between the plasma membrane and endosomes. Furthermore, PIN proteins are sorted to the lytic vacuolar compartments via a pre-vacuolar compartment (PVC). Treatment with the vesicle trafficking inhibitors BFA and wortmannin, affects PIN trafficking at the respective points of drug action. BFA treatment inhibits PIN recycling and leads to PIN accumulation in aggregated endosomes, so called BFA compartments. Wortmannin affects localization of vacuolar sorting receptors and alters the PVC identity, leading to PIN2 relocalization to wortmannin-induced compartments. When mel1-1 mel2-1 mel3-1 mel4-1 or wild-type roots were treated with BFA or wortmannin, PIN2-GFP was internalized into the BFA compartment or wortmannin-induced compartments (see Fig. S5A,B,E,F in the supplementary material). To investigate the sensitivity to the drug in the quadruple mutants in more detail, we treated roots with a range of BFA concentrations. However, we could not detect any differences in sensitivity between wild type and the quadruple mutants (data not shown). The effects of BFA treatment on vesicle trafficking are reversible. When BFA was washed from the medium, the BFA compartments disappeared and PIN2-GFP reverted to the plasma membrane (see Fig. S5C in the supplementary material). In mel quadruple mutants, BFA washout led to reversion to normal recycling of PIN2-GFP to the plasma membrane in a similar manner to wild-type roots (see Fig. S5D in the supplementary material). These results indicate that mel mutations do not affect the recycling of PIN2 proteins back to the plasma membrane. Taken together, the results suggest that intracellular trafficking of PIN2-GFP is unaffected in mel multiple mutants. To investigate PIN internalization mediated by clathrin-dependent endocytosis in the mutants, we treated PIN2-GFP expressing plants with TyrA23, an inhibitor of recruitment of endocytic cargo into the clathrin-mediated pathway, in the presence of BFA (Ortiz-Zapater et al., 2006; Dhonukshe et al., 2007). In the wild-type root, only the endocytic tracer FM4-64 accumulated in BFA compartments, whereas PIN2-GFP was retained at the plasma membrane (Fig. 4A). Similar results were also obtained for the plasma membrane protein, EGFP-LTI6a. TyrA23, but not a close structural analog, TyrA51, inhibited the BFA-induced internalization of EGFP-LTI6a (Fig. 4C; see Fig. S6 in the supplementary material). By contrast, in mel1-1 mel2-1 mel3-1 mel4-1 roots treated with TyrA23 and BFA, PIN2-GFP accumulated in BFA compartments where it colocalized with FM4-64 (Fig. 4B), although EGFP-LTI6a was retained in the plasma membrane (Fig. 4D). To examine the sensitivity to TyrA23, treatment was performed over a wide range of concentrations. TyrA23 blocked PIN2 internalization in a concentration-dependent manner, but the mutants were less sensitive than wild type (see Fig. S7 in the supplementary material). In addition, auxin treatment was recently reported to block PIN internalization from the plasma membrane. We exogenously treated PIN2-GFP-expressing seedlings with auxin in the presence of BFA and examined the effects on PIN internalization in the mel multiple mutants. PIN2-GFP was kept in the plasma membrane of wild-type cells, whereas PIN2-GFP accumulated in the intracellular compartments in the mutant cells (Fig. 4E,F). These results indicate that mel mutations specifically modulate clathrin-dependent PIN2 internalization from the plasma membrane.

To examine the organ- and tissue-specific expression patterns of the MAB4 and MEL genes in plants, promoter fragments of MAB4, MEL1, MEL2, MEL3 and MEL4 were fused to the β-glucuronidase (GUS) reporter gene and introduced into Arabidopsis wild-type plants. Furthermore, we performed in situ hybridization using specific anti-sense probes for each gene. We obtained the same results with regard to expression pattern using these two methods (see Figs S8, S9, S10 in the supplementary material). The signal from proMAB4:GUS, proMEL1:GUS and proMEL2:GUS was detected in organ primordia such as cotyledons, leaves and floral organs (see Figs S8, S10 in the supplementary material). In the radicle and root, the promoter of the MAB4 and all MEL genes was strongly active (see Fig. S8 in the supplementary material). Interestingly, we also found activity of the MEL1 and MEL2 promoters in stomatal lineage cells of cotyledons and true leaves (see Fig. S8F,I in the supplementary material). These results indicate that the expression patterns of MAB4, MEL1 and MEL2 mark active proliferating organs, including shoot meristem, young leaf primordia and floral organ primordia, whereas in the root tip MAB4, MEL1, MEL2, MEL3 and MEL4 were expressed in an overlapping unique manner. These expression patterns are consistent with the results of genetic analyses.

To investigate the functional domain of MAB4 and MEL proteins in plant cells, we performed immunolocalization analysis for MAB4 using a MAB4-specific antibody and expressed functional MEL-GFP fusion proteins under the control of their own promoters. To confirm the specificity of the MAB4 antibody, we performed immunolocalization analysis towards wild-type and mab4 embryos, and compared the signal between them. In the wild-type embryos, the signals were detected in the domain where the MAB4 mRNA is detectable, whereas signals were not detectable in the mab4 embryos (see Fig. S11 in the supplementary material). The functionality of the corresponding MEL-GFP fusion proteins was confirmed by complementation of the mutant phenotype. In wild-type embryos, MAB4 was localized in the periphery of protodermal cells with polarity at the early heart stage (Fig. 5A). The MAB4 polarity was converged to the tips of cotyledon primordia (Fig. 5A). It has previously been reported that PIN1 polarity was also converged toward the tips of cotyledon primordia in the protodermal cell layer (Benková et al., 2003). To investigate the spatial and temporal relationship between MAB4 and PIN1 localizations during embryogenesis, we performed double-immunostaining analysis using MAB4- and GFP-specific antibodies in the PIN1-GFP-expressing embryos. At the globular and heart stage, MAB4 was colocalized with PIN1-GFP in the protodermal cells (Fig. 5B-D). These results indicate that MAB4 was localized peripherally with polarity in the outer cells, and that MAB4 polarity was absolutely identical to PIN1 polarity. Next, to examine the subcellular localization of MEL proteins, we analyzed the localization of the functional MEL-GFP in the embryo. MEL1-GFP was localized in the upper side of the plasma membrane towards the tips of cotyledon primordia in the protodermal cells, and in the basal side of provascular cells and radicle tip cells (Fig. 5E-G). In the proMEL2:MEL2-GFP-expressing plants, MEL2-GFP signals were detected in the identical domains to MEL1-GFP signals (data not shown). MEL3-GFP was basally localized in the plasma membrane in the QC and columella initial cells, and weakly in the cytosol (Fig. 5H-J). MEL4-GFP was basally localized in the provascular cells of the basal region and radicle tips (Fig. 5K-N). Furthermore, we analyzed the localization of MEL proteins in the root expressing functional MEL-GFP proteins. In the stele, pericycle, endodermis and cortex, MEL1-GFP was basally localized, whereas in the epidermis it was in the upper side of the plasma membrane (Fig. 6A,B). In the QC and columella cells, MEL1-GFP was localized weakly in the cytosol, sometimes close to the basal side of the plasma membrane (see Fig. S12A in the supplementary material). The localization pattern of MEL2-GFP and MEL4-GFP was almost identical to that of MEL1-GFP in the root (data not shown; Fig. 6E,F). In the epidermis, the signal of MEL3-GFP was detected in the upper side of the plasma membrane (Fig. 6C,D). In the QC and columella initial cells, MEL3-GFP was weakly localized in the cytosol, sometimes close to the basal side of the plasma membrane (see Fig. S12B in the supplementary material). These results indicate that MEL proteins are mainly localized in the cell periphery with polarity; however, in specific cells such as QC and columella initial cells, MEL proteins tend to diffuse into the cytosol sometimes close to the plasma membrane. In addition, in the domain where expression of MEL genes is overlapping, the subcellular distribution pattern of their proteins was completely identical.

PIN proteins are internalized from the plasma membrane and then recycled to the plasma membrane or targeted to the vacuole through endosomes. To investigate whether polarized MEL proteins are also internalized from the plasma membrane and traffic between the intracellular compartments, we performed a pharmacological analysis using BFA as an inhibitor of vesicle trafficking. Interestingly, when we treated proMEL1:MEL1-GFP-expressing plants with BFA, we could not find any change in the MEL1-GFP signal close to the plasma membrane, although the fluorescence of FM4-64 presented in the BFA compartments (Fig. 6G-I). Under the same conditions, PIN2-GFP accumulated in the BFA compartments and co-stained with FM4-64 in the epidermis of the root (Fig. 6J-L). In addition, we performed double-immunostaining analysis using PIN2 and GFP specific antibodies in BFA-treated proMEL3:MEL3-GFP roots. The GFP signal was localized to the cell periphery, while PIN2 was localized not only in the plasma membrane but also in the BFA compartments (Fig. 6M-P). These results indicate that MEL proteins are not internalized from the plasma membrane and that their localization is confined to the cell periphery, whereas PIN2 proteins are internalized.

To examine whether polar localization of the MAB4 protein depends on the PIN proteins, we analyzed MAB4 localization in embryos developing in pin1-201/+ siliques. The mutant embryos were confirmed by cotyledon defects. The abundance of MAB4 in the plasma membrane was significantly reduced in pin1 embryos (Fig. 7A,B). If the observed effects of the pin1 mutation on the MAB4 localization were caused by changes in auxin distribution, the exogenous application of auxin could also affect the localization. When 35S:MAB4-GFP seedlings were treated with synthetic auxin NAA, the signal of MAB4-GFP in the plasma membrane became weak and its polarity was slightly disturbed (Fig. 7C-F). These results indicate that correct auxin distribution, possibly established by PIN proteins, is important in the MAB4 localizations in the plasma membrane.

Here, we show functional redundancy between MAB4 subfamily genes in auxin-related morphogenesis, not only in organ formation but also root gravitropism. In their multiple mutants, PIN abundance in the plasma membrane was severely reduced with weakened polarity. Pharmacological analysis demonstrated that the mutations affected PIN internalization from the plasma membrane, but did not affect intracellular PIN trafficking. In addition, all MAB4 subfamily proteins are localized at the cell periphery, with polarity almost identical to PIN polarity. Our data suggest that the MAB4 subfamily proteins specifically regulate polarity and endocytosis of PIN proteins in the plasma membrane.

MAB4, besides functionally redundant genes MEL1/NPY5 and MEL2/NPY3, has been identified as a factor that regulates organ formation through the control of polar auxin transport (Furutani et al., 2007; Treml et al., 2005; Cheng et al., 2008). Besides organ formation, our results show a crucial and redundant role for MEL genes in root gravitropism. mel1 mel2 mel3 mel4 quadruple mutants displayed defects in root gravitropism, as did pin2 mutants and plants treated with an inhibitor of polar auxin transport, indicating that MEL genes control polar auxin transport in root gravitropism as well as during organ formation (Fig. 1). In addition, NPH3-like proteins RPT2 and NPH3 have been reported to function in the phototropism of hypocotyl and root, where polar auxin transport is also involved (Motchoulski and Liscum, 1999; Sakai et al., 2000). These data suggest that the NPH3 family generally might control polar auxin transport in plant development and its modulation in response to the environment or endogenous signals. However, the currently understood functions of NPH3 family members are not the full story in various aspects of polar auxin transport regulated development. We could not find obvious vascular development phenotypes regulated by polar auxin transport in any combinations of mel mutations, even though some MEL genes are expressed in vascular tissues. This raises the possibility of additional redundancy at this developmental stage. The genes At3g26490, At1g67900 and At5g47800 are candidate redundant factors that belong to the subclass that includes the MAB4 and MEL genes. The prospective analyses of these genes and delineation of their relationship to MAB4 family genes would provide insight into all functions of the NPH3 family genes.

MAB4 was reported to control polar auxin transport through the control of PIN1 localization during cotyledon development (Treml et al., 2005; Furutani et al., 2007). Our results show that mutations of the MEL genes also specifically affect PIN localization, but not general plasma membrane protein localization (Fig. 2). This is seen in the polar localization of the MAB4 subfamily proteins in the cell periphery, which is almost identical to that of PIN proteins (Figs 5, 6). These observations lead to the suggestion that the MAB4 subfamily genes specifically control the localization of the corresponding PIN proteins in their expression domains at the various developmental stages.

PIN proteins in the plasma membrane are constitutively internalized to the endosomes as well as to other plasma membrane proteins. Some of them are recycled back to the plasma membrane depending on the activity of GNOM, and others are targeted to the lytic vacuole in a retromer-dependent manner and degraded there. Our pharmacological analysis and FRAP experiment in the mutant background suggest that mel mutations affect PIN2 internalization from the plasma membrane, but neither the lateral mobility nor intracellular vesicle trafficking of PIN2 proteins (Fig. 4; see Figs S4, S5 in the supplementary material). In addition, the insensitivity of the MAB4 subfamily proteins to BFA treatment indicates that the functional site is the plasma membrane and not the intracellular compartments (Fig. 6). Taken together, our results suggest that the MAB4 subfamily specifically regulates endocytosis of PIN proteins to maintain its polar localization. This begs the question as to the mechanism of action of the MAB4 subfamily in retention of PIN localization. It is possible that the MAB4 subfamily blocks the internalization of PIN proteins from the plasma membrane. Mutations of MAB4 subfamily genes could enhance PIN2 internalization leading to the reduction of polarized PIN2 abundance and lateral relocation by recycling. This scenario is consistent with previous reports that PIN1 internalization from lateral and apical membranes established basal polarity and interference with PIN1 endocytosis prevents PIN1 polarization (Dhonukshe et al., 2008). PIN proteins are retained at the membrane where the MAB4 subfamily proteins are localized, whereas PIN proteins are internalized where the MAB4 subfamily proteins are not. Recently, PID and related AGC kinases were reported to trigger endocytosis-dependent apical PIN recycling through direct phosphorylation of PIN proteins (Michniewicz et al., 2007; Dhonukshe et al., 2010). PID is localized all over the plasma membrane in a non-polarized fashion, whereas MAB4 localization is polarized. These data suggest that PID subfamily promotes PIN internalization in the absence of MAB4 subfamily, while MAB4 subfamily accumulates PIN proteins in the plasma membrane, maybe through the control the activity of PID. Furthermore, a recent study demonstrated that AUXIN-BINDING PROTEIN 1 (ABP1) recruits clathrin to the plasma membrane to promote endocytosis and that auxin binding to ABP1 interferes with this action (Robert et al., 2010). Considering that mel mutations affected PIN localization specifically, the MAB4 subfamily might confine the ABP1-mediated signaling to PIN endocytosis.

In cells displaying PIN polarization in the plasma membrane, the MAB4 subfamily proteins certainly exhibit identical polar localization patterns. However, in specific cells, such as columella and QC cells, the signals from MAB4 family proteins were slightly diffused in the cytosol (see Fig. S12 in the supplementary material), whereas PIN proteins are localized all over the plasma membrane (Friml et al., 2002a; Friml et al., 2002b). This situation is very similar to PIN apolarization in mel multiple mutants. It is, thus, conceivable that the MAB4 polarization in the cell periphery strongly correlates with PIN polarity. Cells displaying inconsistency between MAB4 and PIN localization patterns accumulate more auxin than do neighboring cells, as seen by the fact that DR5rev::GFP is highly expressed in columella and QC cells. In addition, we show defective PIN localization and auxin distribution in the mutants of the MAB4 subfamily genes and vice versa (Figs 2, 3, 7). These strongly suggest the existence of a feedback regulation in the establishment of the MAB4 and PIN polarity. Which came first: MAB4 polarity or PIN polarity? At this moment, it is a chicken and egg situation. Further detailed analyses will provide us with insights into the molecular mechanisms that control cell polarity in the polar auxin transport system.

We thank Ben Scheres for providing us with PIN1-GFP expressing plants, and ABRC and NASC for providing materials. We also thank Asami Mori for excellent technical assistance. This work was partly supported by a Ministry of Education, Culture, Sports, Science and Technology, through Grant in Aid for Scientific Research on Priority Areas (14036222) to M.T., by Grant-in-Aid for Young Scientists (20770034), and by Global COE Program in NAIST (Frontier Biosciences: strategies for survival and adaptation in a changing global environment), MEXT, Japan to M.F.