BLOS1, a putative BLOC-1 subunit, interacts with SNX1 and modulates root growth in Arabidopsis

Vesicle trafficking performs conserved functions in eukaryotes. It is involved in communication among different cell compartments and the exchange of information between cells and their environment. Endosomal transport plays a central role in vesicle trafficking as it connects anterograde transport from the Golgi apparatus to the endosomes and the lysosomes, endocytic transport from the plasma membrane to the endosomes and retrograde transport from the endosomes to the Golgi (Bonifacino and Rojas, 2006; Li et al., 2007). Cargo is sorted from the early endosomes to the cell surface, recycling endosomes or late endosomes and lysosomes (Gruenberg, 2001; Bonifacino and Glick, 2004). Components destined for the lysosomes are sorted to the late endosomes, or multivesicular bodies, which contain whorls of membranes and vesicles. The late endosomes morph into the lysosomes with a highly regulated fusion machinery (Raposo et al., 2007). The lysosomes serve as the major degradative compartments in eukaryotic cells (Kornfeld and Mellman, 1989). In addition, the secretory lysosomes or some specialized lysosome-related organelles play important functions in secretion via exocytosis (Raposo et al., 2007).

In contrast to mammalian cells, plant vesicle trafficking is less well understood. The identification of genes encoding components of vesicle trafficking in plant genomes suggests the conserved role of vesicle trafficking in plants. Endosomal transport plays a key role in many processes in plants including embryo differentiation (Goh et al., 2007; Silady et al., 2008), cell fate specification (Shen et al., 2003; Tian et al., 2007), cell wall remodeling (Baluska et al., 2002), gravitropism (Silady et al., 2004; Abas et al., 2006) and polar auxin transport (Gälweiler et al., 1998; Steinmann et al., 1999; Geldner et al., 2001; Jallais et al., 2006; Jallais et al., 2007; Jallais et al., 2008; Dhonukshe et al., 2008). Endosomal trafficking is essential for polar transport of the phytohormone auxin and its influx (AUX) and efflux (PIN) carriers (Paciorek et al., 2005; Müller et al., 2007). Targeting of PIN1 protein to the basal plasma membrane requires PIN1 recycling from the plasma membrane to the endosomes mediated by GNOM, an endosomal GDP and GTP exchange factor for ARF GTPases (Geldner et al., 2003). By contrast, PIN2 transport requires the SNX1-positive endosomes. SNX1 is a newly described plant-sorting nexin that is localized to the prevacuolar compartment (PVC) in plants (Jaillais et al., 2006). Besides SNX1, another component of retromer, VPS29, is proposed to function in endosomal recycling of PINs. In the vps29 mutant, intracellular accumulation of PIN1 and, to a lesser degree, of PIN2 and other membrane proteins was observed. VPS29 colocalizes with SNX1 and is required for maintaining the SNX1 endosomal morphology but not the GNOM endosomes. A current model is that PINs are internalized first through the GNOM endosomes and then routed to the SNX1 endosomes for recycling (Jaillais et al., 2007). Nevertheless, the destination of the endocytosed PIN1 and PIN2 remains to be clarified.

It has been demonstrated that plants have evolved different vacuoles to adapt their metabolism. Vacuoles, including the protein storage vacuole (PSV) and the lytic vacuole (LV), are important structures in vesicle trafficking. The lytic vacuoles share lysosome-like properties but also function as the protein storage vacuoles for protein accumulation (Frigerio et al., 2008). LVs and PSVs have different pH in their lumen, as highlighted by staining with pH- or acidic protease-sensitive dyes. The general consensus stemming from extensive tonoplast intrinsic protein (TIP) isoform analysis is that α-TIP is enriched in PSVs, γ-TIP in LVs and γ-TIP in storage vacuoles (Jauh et al., 1998; Jauh et al., 1999). However, the transport machinery of cargoes from the endosomes to LVs or PSVs is lacking. Meanwhile, the physiological role of LVs in the maintenance of the steady-state levels of the endocytosed proteins remains to be elucidated.

The BLOC-1 complex consists of at least eight subunits in both human and mice, including dysbindin, snapin, muted, pallidin, cappuccino, BLOS1, BLOS2 and BLOS3. It is involved in vesicle trafficking from the endosomes to the lysosomes in higher eukaryotes (Li et al., 2007; Raposo et al., 2007; Truschel et al., 2009). Human Hermansky-Pudlak syndrome (HPS) is caused by the mutations of genes encoding dysbindin or BLOS3 (Li et al., 2003; Morgan et al., 2006). HPS is characterized by the defects in the biogenesis of lysosome-related organelles due to impaired cargo transport from the endosomes to the lysosomes (Li et al., 2007; Raposo et al., 2007). Whether there exists a homologous BLOC-1 complex in plants is unknown. Sequence comparisons reveal putative homologs of BLOS1 (also known as AtGCN5L1) and BLOS2 in Arabidopsis. Whether these homologs play similar roles in endosomal transport in Arabidopsis requires investigation. We here report the first evidence of a putative BLOC-1 in Arabidopsis and its possible involvement in the transport of PIN1 and PIN2, mediated by SNX1, to control root development.

Multiple protein sequences from different species were aligned using the ClustalW2 program (http://www.ebi.ac.uk/Tools/clustalw2/) and GeneDoc (http://www.nrbsc.org/downloads/) (supplementary material Fig. S1). The percentage of identity between the Arabidopsis and human proteins was calculated using the BLASTP algorithm (http://blast.ncbi.nlm.nih.gov/). The calculated sequence identities to human homologs for BLOS1 (At2g30330) and BLOS2 (At5g49550) were 36% and 29%, respectively. Interaction between BLOS2 and BLOS1 was revealed by yeast two-hybrid assays (Fig. 1A) and split-YFP assays (Fig. 1B). Next, our work is mainly focused on the functional studies of BLOS1.

Tissue-specific expression of BLOS1 was tested by real-time PCR with RNA extracted from adult plants. BLOS1 expression was found in all major tested organs including the root, stem, leaf, flower and silique, with highest expression in the leaf (Fig. 1C). Western blotting by a polyclonal anti-BLOS1 antibody was consistent with these findings (Fig. 1D).

Tissue-specific expression of BLOS1 promoter activity was further analyzed by histochemical localization of GUS activity. In 5-day-old germinating seedlings, GUS expression was detected in the vascular tissues (Fig. 2A), shoot apical meristem (Fig. 2B) and primary root (Fig. 2C). At 9 days, GUS expression was clearly detectable in the adventitious root (Fig. 2D) and the lateral root primordia (Fig. 2E). GUS activity in up to 4-week-old transgenic BLOS1 pro::GUS seedlings showed expression in the flowers, anthers and silique (Fig. 2F–H).

To study the subcellular localization of BLOS1, a protoplast system was adopted. Sixteen hours after cotransformation with different organellar markers, confocal images showed that BLOS1 partially colocalized with endosome markers ARA7 (Fig. 3A) and VAMP727 (Fig. 3B), PVC and vacuole markers PEP72 (Fig. 3C) and VAM3 (Fig. 3D). These data indicate that BLOS1 is probably localized at the endosomal compartments.

Histochemical GUS staining suggested that BLOS1 might play a role in root development. As there is no T-DNA exonal insertion line available, an XVE-inducible RNAi construct was made to specifically knockdown BLOS1. BLOS1 full-length cDNA was subcloned into an XVE vector to produce sense and antisense strands. Western blotting showed that the protein level of BLOS1 in RNAi line L4 was apparently lower than that in the wild type after estradiol inducement (Fig. 4A), which is consistent with the lower level of BLOS1 transcripts by northern blotting after estradiol inducement (data not shown). Our work focused on root phenotype analysis in transgenic line L4, although phenotypes in other tissues might exist. The knockdown line L4 had longer primary roots and more lateral roots (Fig. 4B,C).

To explore the function of BLOS1, full-length BLOS1 cDNA was used as a bait to screen an Arabidopsis cDNA library. SNX1 (At5g06140) was identified as a novel interactor (Fig. 5A) via the GCN5L domain of BLOS1 (Fig. 5B). Likewise, BLOS2 interacted with SNX1 in the yeast two-hybrid assay (Fig. 5C), further supporting that BLOS1 and BLOS2 might function in the same complex. In addition, BLOS1 and SNX1 colocalized in protoplast (Fig. 5D). To further confirm this interaction, we found that BLOS1 interacted with SNX1 in onion and protoplast by split-YFP assays (Fig. 5E,F).

Considering a possible role for BLOS1 in endosomal trafficking, we hypothesized that BLOS1 recruits PINs sorted by SNX1 and transports a portion of PINs to the vacuoles for degradation. Immunocytochemical analysis of PIN1 and PIN2 in BLOS1 RNAi knockdown lines showed that the protein levels of PIN1 and PIN2 on the plasma membrane increased after inducement (Fig. 6C,D) compared with wild type (Col-0) after inducement (Fig. 6A,B). However, the polar distributions of both PIN1 and PIN2 were not apparently altered. To rule out the possibility of transcriptional upregulation of PIN1 and PIN2 by BLOS1, we performed real-time PCR and the results showed no significant increase of PIN1 or PIN2 transcripts in the BLOS1 RNAi-L4 after inducement by estradiol (Fig. 6E).

To further study whether the accumulation of PIN1 and PIN2 on the plasma membrane is due to a defect in the endocytosis of these PINs, the recruitment to the endosomes after tracing for 15 minutes in L4 was examined. Recruitment was not apparently impaired as controlled by the effects of brefeldin A (BFA) treatment in FM4-64 endocytic assays (Fig. 7), suggesting that the endocytosis of PIN proteins from the plasma membrane to the endosomes might not be affected in these lines.

Although the cellular function of the endosomal transport complex BLOC-1 in mammals requires further definition (Li et al., 2007), we here provide the first report of its possible involvement in endosomal trafficking in plants. BLOS1 and BLOS2 in Arabidopsis are probably two homologous subunits of BLOC-1, as revealed by sequence comparison, biochemical features and cellular function in endosomal transport. Downregulation of BLOS1 causes longer primary and lateral roots, suggesting that the homologous BLOC-1 complex might be involved in root development by negatively regulating the stability of PIN1 and PIN2 through the interaction with SNX1.

Thus far, at least eight subunits have been identified in the mammalian BLOC-1 complex, which is involved in cargo-specific sorting from the endosomes to the lysosomes (Li et al., 2007; Setty et al., 2007; Truschel et al., 2009). Sequence comparison reveals only two putative homologs (BLOS1 and BLOS2) of BLOC-1 subunits in Arabidopsis. Whether the composition of BLOC-1 is the same as its mammalian counterpart remains to be characterized biochemically. As shown in this study, BLOS1 and BLOS2 interact directly, suggesting a conservative role of BLOC-1 in protein trafficking from the endosome to the vacuole in plants.

Cargo sorting from the trans-Golgi network (TGN) or from the plasma membrane (PM) to the lysosomes might share similar machinery. It is evident that BLOC-1 directs cargoes from the TGN or from the PM to the lysosome-related organelles or to the lysosomes for degradation (Setty et al., 2007; Iizuka et al., 2007; Truschel et al., 2009). In Arabidopsis, it is known that PIN proteins rapidly and constitutively cycle between the PM and the endosomes (Geldner et al., 2001). Internalization of PIN2 is clathrin-mediated (Dhonukshe et al., 2007) and it is then transferred to the sorting endosomes via SNX1 (Jaillais et al., 2006). However, the destination of the internalized PIN proteins remains uncertain. One possibility is the temporary storage in specific vesicles for fast recycling to the PM. Another possible fate is the proteasomal degradation of ubiquitinated PINs (Abas et al., 2006). Third, there is evidence that PINs could be targeted for vacuolar degradation in the absence of light (Laxmi et al., 2008), which might be mediated by the ESCRT complex (Spitzer et al., 2009). Nevertheless, the molecular aspects of this route are primitive. Currently, there are no direct interactions reported among the PINs, SNX1 and ESCRT. Here, we show for the first time physical interactions between SNX1 and BLOS1/2, which might provide insights into the complexity of PIN protein homeostasis.

It is probable that a portion of PIN proteins could be targeted for vacuolar degradation mediated by SNX1 and BLOS1/2, suggesting a conservative role of BLOC-1 in mammals and plants for vacuolar targeting. First, both BLOS1 and BLOS2 interact with SNX1, which probably occurs within the endosomal compartments. Second, SNX1 is reported to mediate PIN2 transport (Jaillais et al., 2006). Third, knockdown of BLOS1 leads to an increase of PIN1 and PIN2, which is probably caused by impaired vacuolar degradation as discussed below.

The augmentation of PIN1 and PIN2 could be a result of: (1) transcriptional upregulation of PIN1 and PIN2 by BLOS1; (2) impaired endocytosis of PIN1 and PIN2; and (3) delayed proteosomal or lysosomal protein degradation to facilitate membrane reinsertion of PIN1 and PIN2. As we showed in this study, BLOS1 is mostly expressed in the cytosol and the transcripts of PIN1 and PIN2 are not significantly increased in the BLOS1 RNAi lines (Fig. 6E); therefore, it is unlikely that BLOS1 acts as a transcription factor to regulate the expression of PIN1 and PIN2. In the FM4-64 traced assay, the endocytosis from the PM to the endosomes appears to be unaffected (Fig. 7). We have not ruled out the possibility that BLOS1 might affect the proteosomal degradation of ubiquinated PIN1 and PIN2. Based on the function of its mammalian counterpart, it is more likely that the impaired lysosomal targeting of PIN1 and PIN2 caused by the deficiency of BLOS1 might facilitate these PIN proteins to take an alternative route to the PM through membrane reinsertion. In agreement with this phenomenon, it has been reported that the melanosomal protein TYRP1 (Setty et al., 2007), the dopamine receptor D2 (Ji et al., 2009) or the NMDA receptor NR2A (Tang et al., 2009) are increased in the PM through a similar mechanism when BLOC-1 is deficient.

The timing and location of auxin accumulation determines spatiotemporal facets of developmental reprogramming. Localized increase of auxin in a single pericycle cell causes the initiation and formation of lateral roots (Dubrovsky et al., 2008). Moreover, the stabilization of PIN2 affects its abundance and distribution and leads to defects in auxin distribution and root development (Abas et al., 2006). Thus, the regulation of the homeostasis of PIN1 and PIN2 by vesicle trafficking is important to plant growth. The accumulation of PIN1 and PIN2 in the BLOS1 RNAi lines is probably caused by impaired degradation through the endosomal-vacuole pathway, which might, in turn, lead to an increase of auxin transport that initiates the development of longer primary roots and the formation of more lateral roots.

It must be re-emphasized that the root phenotypes of BLOS1 knockdown lines are opposite to the snx1 lines (Jaillais et al., 2006), although we have shown the physical interaction of BLOS1/2 with SNX1. One possible explanation is that SNX1 is responsible for the sorting of endocytosed PIN1 and PIN2 where they might be recycled to the PM via VPS29 or retromer (Jaillais et al., 2007). The recycled PIN1 and PIN2 are reduced in the snx1 lines (Kleine-Vehn et al., 2008) and thus show the delayed root development. The opposite root phenotypes in snx1 and in BLOS1 RNAi lines agree with the opposite changes in PIN1 and PIN2 stability. BLOS1/2 or BLOC-1 might act in a different pathway through the interaction with SNX1 for vacuolar degradation of PIN1 and PIN2. It is possible that this proportion of PIN1 and PIN2 could be marked with mono-ubiquitin and recognized by ESCRT (Spitzer et al., 2009). To support this, our preliminary data shows that BLOC-1 interacted with ESCRT-1 in a yeast two-hybrid assay (data not shown). In the mammalian cell, it is suggested that BLOC-1 acts downstream of ESCRT (Truschel et al., 2009). Thus, we proposed a model for a dual role of SNX1, one for fast recycling of PIN1 and PIN2 via retromer (Jaillais et al., 2007) and another for vacuolar targeting of PIN1 and PIN2 via BLOC-1 (Fig. 8). Similarly, in mammalian cells, the role of SNX1 has been reported to be involved in both the retrograde endosomal transport via retromer (Gullapalli et al., 2004) and the lysosomal degradative pathway of cell surface receptors such as EGFR (Zhong et al., 2002) and PAR1 (Gullapalli et al., 2006). Our findings suggest that the vacuolar degradative pathway is important in the homeostasis of PIN1 and PIN2 to regulate the plant growth and development.

Plants were grown under a 16:8-hour light:dark cycle at 22°C on solid Murashige and Skoog (MS) agar [half-strength MS salts, 3% (w/v) sucrose and 1% (w/v) agar]. For root analysis, seedlings were grown vertically on half MS medium in Petri dishes for the indicated period of time. For growing the inducible RNAi lines, experiments were carried out by using both wild-type (WT, Col-0) and transgenic plants treated with 10 μM beta-estradiol, whereas dimethyl sulfoxide (DMSO, a solvent for estradiol) was used as treatment for the control lines.

The yeast two-hybrid screen was performed by following the user manual of the MATCHMAKER 3 Yeast Two-Hybrid System (Clontech, Mountain View, CA) with minor modifications. Full-length BLOS1 cDNA (At2g30330) was amplified by PCR using the primers 5′-CCATATGATGAATACGCCGATGTCACTCTC-3′ and 5′-CCTGCAGACATGATGGAATCCTACTCAGGC-3′. The PCR product was cloned into the NdeI and PstI sites of pGBKT7 (binding domain, BD) and pGADT7 (activation domain, AD) in-frame. Yeast cells were first transformed with pGBKT7-BLOS1 and sequentially transformed with the Arabidopsis cDNA library constructed in the GAL4 AD vector pGAD-GH (Xie et al., 1999). The transformed cells were plated on dropout selection medium lacking Trp, Leu, His and adenine (Ade) and incubated at 30°C for 3–8 days.

To further confirm the screening results by paired yeast two-hybrid assays, full-length BLOS2 (At5g49550) and SNX1 (At5g06140) cDNAs were amplified with primer pairs P_BLOS2 5′-AACATATGATGGCGGATTCACGAGATGATCTA-3′, 5′-GGGAATTCTCAGTCTGTAACAACAGAATCATTAG-3′ and P_SNX1 5′-AACATATGATGGAGAGCACGGAGCAGCGAGGAA-3′, 5′-CCGAATTCTTAGACAGAATAAGAAGCTTCAAGTTTTGG-3′. The PCR products were respectively cloned into pGBKT7 and pGADT7 vectors in frame.

To create inducible dsRNAi transgenic lines, primers 5′-GGCTCGAGACTAGTATGAATACGCCGATG-3′ and 5′-CCGAATTCAAGCTTCCATCGTCTTCATCC-3′ were used to amplify a 400 bp BLOS1 cDNA PCR product. The remaining procedures were as described previously (Guo et al., 2003).

Two peptides LEKEKSEEESETLE and DDNRRSSLQLREKTERSRKEAI selected from BLOS1 were synthesized to generate rabbit polyclonal antibodies. The specificity of purified BLOS1 antibody from peptide DDNRRSSLQL-REKTERSRKEAI was examined in protoplast protein extracts. Expression levels of BLOS1 in different tissues were analyzed by purified BLOS1 antibody (1:4000) and protein loadings were revealed by Ponceau S staining or by tubulin. The procedures of western blotting were followed as previously described (Müller et al., 1998).

Real-time RT-PCR analysis (performed in MJ Bio-Rad PTC-200) was conducted by following the ‘Guide to Performing Relative Quantitation of Gene Expression Using Real-Time Quantitative PCR Protocol’ (Applied Biosystems, Foster City, CA). Primer pairs 5′-ACCGAGAGATCGAGAAAGGAAGCA-3′ and 5′-CGAATTTCGCCACCGTAATGGCAA-3′ were used to detect BLOS1 mRNA. Primer pairs for detecting PIN1 and PIN2 mRNA were as follows: PIN1-F: 5′-ACTCTCAGATGCAGGTCTAGGCAT-3′; PIN1-R: 5′-ACAGTGCTAAGAATGTCAGGATGCAC-3′; PIN2-F: 5′-CGTCCATCCTGATATTCTCAGCAC-3′; PIN2-R: 5′-CATACACCTAAGCCTGACCTGGAA-3′. Control primers for actin were 5′-GGAACTGGAATGGTGAAGGCTG-3′ and 5′-CGATTGGATACTTCAGAGTGAGGA-3′.

Total RNA from 2-week-old seedlings were isolated by the Plant RNeasy Prep Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. RNA northern blot analysis was performed (20 μg of total RNA per lane) with a ³²P-radiolabeled full-length BLOS1 probe. The procedure was carried out as described (Sun et al., 2003).

The open reading frame sequences of BLOS1, BLOS2 and SNX1 were amplified by PCR with gateway-specific primer pairs Pgw_BLOS1 5′-AAAAAGCAGGCTGGATGAATACGCCGATGTCACTC-3′, 5′-AGAAAGCTGGGTATTGTTGATCTTCATGAATGTTTCG-3′, Pgw_BLOS2 5′-AAAAAGCAGGCTGGATGGCGGATTCACGAGATGA-3′, 5′-AGAAAGCTGGGTATCAGTCTGTAACAACAGAATCATT-3′, Pgw_SNX1 5′-AAAAAGCAGGCTGGATGGAGAGCACGGAGCAGCCGA-3′, 5′-AGAAAGCTGGGTATTAGACAGAATAAGAAGCTTCAAG-3′ and gateway-control primer pairs tB1 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCT3′ and tB2 5′-GGGGACCACTTTGTACAAGAAAGCTGGGT-3′. The resulting PCR products were cloned into plasmids pUC-SPYCE-gw and pUC-DYNn-gw to give rise to pUC-SPYCE-gw-BLOS1, pUC-DYNn-gw-BLOS2 and pUC-DYNn-gw-SNX1. pUC-DYNn-gw contains N-terminal YFP modified from pUC-SPYNE-gw. To generate the gateway constructs, we followed the gateway cloning protocol (Invitrogen, Carlsbad, CA).

For the promoter-GUS assay, a 2187 bp genomic fragment upstream of the BLOS1 TAA stop codon, together with a 1148 bp genomic fragment downstream of the BLOS1 TAA stop codon, were PCR-amplified with the primer pairs P₂₁₈₇ 5′-GGCTGCAGGGTGATCCATGTTAAAGA-3′, 5′-CCGGATCCTTGTTGATCTTCATGAATG-3′ and P₁₁₄₈ 5′-CCGAGCTCGAGCAAGTTATTGTGATTC-3′, 5′-GCCGGAATTCGAGAAATAAAACTTGGC-3′, respectively. The 2187 bp (PstI/BamHI) and 1148 bp (SacI/EcoRI) fragments were cloned into the pBI101 vector in frame to produce the P_BLOS1-GUS fusion construct. The construct was introduced into Arabidopsis wild-type plants by Agrobacterium-mediated vacuum infiltration (Bechtold and Pelletier, 1998). After selection on MS agar plates containing 50 mg/ml kanamycin, the resulting transgenic plants were stained in GUS staining buffer (0.1% Triton X-100, 5 mM Fe²⁺CN, 5 mM Fe³⁺CN and 1 mg/ml X-Gluc; Sigma-Aldrich, St Louis, MO) in 50 mM sodium phosphate buffer, pH 7.0) and mounted in 50% glycerol.

A 783 bp DsRed fragment was amplified from pDsRed2N1 (Clontech) by primer pairs 5′-AGTGAACCGTCAGATCCGCTA-3′ and 5′-CCTCTAGAGTGGTATGGC TGAT-3′. After digestion with KpnI and XbaI, the resulting PCR product was cloned into pWM101. The BLOS1 and BLOS2 cDNAs were amplified from wild-type Columbia (Col-0) with the primers 5′-CGAGAGCTCATGAATACGCCGATGTCAC-3′, 5′-TGCGGTACCTGTTGATCTTCATGAATGTT-3′ and 5′-GAGAGCTCATGGCGGATTCACGAGATGATCTAGCTG-3′, 5′-GAGGTACCCTGTAACAACAGAATCATTAGAGCGACGC-3′, respectively. The PCR products were cloned into the SacI/KpnI sites of the CaMV35S:DsRed2N1:pWM101vector. SNX1 was amplified by the primers 5′-TGTGAGCTCATGGAGAGCACGGAGCAGCC-3′ and 5′-ACCGGTACCACAGAATAAGAAGCTTCAAGT-3′. The PCR products were cloned into the SacI/KpnI sites of the CaMV35S:GFP:pWM101 vector. The resulting constructs were introduced into Arabidopsis protoplast as described (Hwang and Sheen, 2001).

Immunocytochemical analysis of PIN1 and PIN2 was performed on whole-mount roots of 5-day-old seedlings. Mouse anti-PIN1 (Gälweiler et al., 1998) and guinea pig anti-PIN2 (Müller et al., 1998) were diluted to 1:50 and 1:100, respectively. For the secondary antibodies, Alexa Fluor 488 goat anti-mouse IgG and Alexa Fluor 555 goat anti-guinea pig IgG (Invitrogen) were diluted to 1:500. The detailed procedures were previously described (Friml et al., 2002a; Friml et al., 2002b). Imaging was performed using a Zeiss LSM 510 Meta inverted confocal microscope and accompanying software.

For FM4-64 treatments, seedlings were incubated for 15 minutes at room temperature in half-liquid medium containing FM4-64 (5 μM) and rinsed three times in half-liquid medium. BFA (25 μM in DMSO) treatments were performed in half-liquid medium for 1 hour before observation.

Quantitative data in paired groups was analyzed using the Student's t-test. Qualitative data was compared using the χ² test.

Sequence data from this article can be found in the EMBL and GenBank data libraries under the following accession numbers: BLOS1 (At2g30330, NM_128586), BLOS2 (At5g49550, NM_124332) and SNX1 (At5g06140, NM_120696). The accession numbers of the homologs of BLOS1 and BLOS2 are listed in Fig. S1 in the supplementary material.

This work was supported in part by grants from the National Natural Science Foundation of China (30730049) and from the National Basic Research Program of China (2007CB947200). We thank Fugang Ren and Bingshan Wang for their technical support in BiFC assays. We are thankful to Richard T. Swank for proofreading this manuscript.