In receptor-mediated transport pathways in mammalian cells, clathrin-coated vesicle (CCV) µ-adaptins are the main binding partners for the tyrosine sorting/internalization motif (YXXØ). We have analyzed the function of the µA-adaptin, one of the five µ-adaptins from Arabidopsis thaliana, by pull-down assays and plasmon resonance measurements using its receptor-binding domain (RBD) fused to a histidine tag. We show that this adaptin is able to bind the consensus tyrosine motif YXXØ from the pea vacuolar sorting receptor (VSR)-PS1, as well as from the mammalian trans-Golgi network (TGN)38 protein. Moreover, the tyrosine residue was revealed to be crucial for binding of the complete cytoplasmic tail of VSR-PS1 to the plant µA-adaptin. The trans-Golgi localization of the µA-adaptin strongly suggests its involvement in Golgi- to vacuole-trafficking events.
Vesicle formation and vesicle targeting/fusion are processes essential for intracellular protein transport (Kirchhausen, 2000a). Proteins carried by vesicles may be transported either as bulk cargo or selectively through the agency of a transmembrane receptor. Clathrin-coated vesicles (CCVs) are present in all eukaryotic cells and are the vehicles of receptor-mediated post-Golgi and endocytic protein transport (Marsh and McMahon, 1999; Robinson and Bonifacino, 2001). In both cases, the extracytoplasmic (lumenal) domains of the receptors possess motifs, which are responsible for binding the transport cargo molecules, while the cytoplasmic tails of the receptors interact with adaptor (AP) complexes. Recruitment of the latter from the cytosol is followed in a highly coordinated fashion by the assembly of clathrin triskelions, providing the vesicle with a coat and culminating in vesicle budding (Hirst and Robinson, 1998; Kirchhausen, 2000a; Owen et al., 1999).
Early in vitro binding studies on extracts from mammalian cells pointed to the AP-2 complex, a component of the CCV coat, as being the likely receptor-binding partner (Pearse, 1988; Sosa et al., 1993). Yeast two-hybrid studies subsequently identified the µ1- and µ2-adaptins as interacting with a tyrosine motif YXXØ in the receptor cytoplasmic tails (Ohno et al., 1995), an observation later confirmed directly by plasmon resonance (Boll et al., 1996). The YXXØ motif, where Ø is a bulky hydrophobic residue and X is any amino acid, has now been well characterized (Marks et al., 1997), and is present in the cytoplasmic tail of many integral proteins of the plasma membrane (PM) of mammalian cells, as well as the trans-Golgi network (TGN), e.g. TGN38 (Stephens and Banting, 1998; Stephens et al., 1997).
There can be no doubt that the β1- and β2-adaptins function as clathrin heavy chain (CHC)-binding partners (Owen et al., 2000; Shih et al., 1995). On the other hand, the µ1–µ3-adaptins have been unequivocally established as being the main binding partners for the cytoplasmic tails of various transmembrane receptors (Heilker et al., 1999). Nevertheless, the large subunits, α-, β1- and β2-adaptins have also been reported to bind to epidermal growth factor (EGF)- and asiaglycoprotein (ASGP)-receptors (Beltzer and Spiess, 1991; Rapoport et al., 1998; Sorkin and Carpenter, 1993). All µ-adaptins have in common a bipartite structure with their aminoterminal one-third responsible for the interaction with β-adaptins, whereas the carboxyterminal two-thirds represent the receptor-binding domain (RBD; Aguilar et al., 1997). Recently, the crystal structure of the RBD of µ2-adaptin from mouse complexed with the internalization/sorting signal peptide of TGN38 has been resolved at 2.7 Å (Owen and Evans, 1998). The amino acids crucial for binding to the tyrosine sorting motif YXXØ are located within the β-sheet strands β1, β2, β15, and β16. These authors demonstrated that two hydrophobic pockets are required for binding to the YXXØ motif, one for the binding of the tyrosine residue and the other one for the binding of the hydrophobic (Ø) residue.
All members of the VSR family contain the putative tyrosine sorting motif YXXØ, the amino acids (Y)Tyrosine, (M)Methionine, (P)Proline, (L)Leucine (YMPL) within their cytoplasmic tails (Paris et al., 1997; Shimada et al., 1997). Although VSRs have no homology to the mammalian mannose-6-phosphate receptors (MPRs), a number of the proteins, which participate in vacuolar protein sorting, are nevertheless homologous to those involved in lysosomal protein transport, e.g. the target soluble N-ethylmaleimide-sensitive factor attachment protein receptors (t-SNAREs) AtVAMP3p and AtPEP12p (Bassham and Raikhel, 1999; Sanderfoot et al., 1999), as well as a dynamin-I-like protein (Jin et al., 2001). This suggests that a similar pathway to the lytic vacuole exists for hydrolytic enzymes in plants (Jiang and Rogers, 2003; Robinson et al., 2000). However, unequivocal proof for an interaction between plant CCV coat components and a VSR has yet to be established.
In this paper, we now provide conclusive evidence that a plant µ-adaptin is the binding partner for the YXXØ sorting motif in the cytoplasmic tail of a plant VSR. Furthermore, we show by gel filtration analysis subcellular fractionation and immunogold staining of cryosections that the ArabidopsisµA-adaptin is part of a putative plant AP complex, and that it is also localized to the trans-Golgi.
Sequence comparisons of ArabidopsisµA-adaptin
We have previously described the cDNA sequence of Arabidopsis thalianaµA-adaptin (At-µA-Ad) medium chain as sharing homologies not only with the µ-adaptin sequences of Dictyostelium discoideum and Caenorhabditis elegans but also with mouse µ1- and µ2-adaptins (Happel et al., 1997). In addition, Arabidopsis contains four further genomic plant µ-adaptin sequences as revealed by databank searching. While At-µA-Ad is located on chromosome 5, the highly related At-µB1-Ad and At-µB2-Ad (86% identity and 90% similarity) are both located on chromosome 1 (genes T16B5.13 and F8A5.29, respectively). At-µC-Ad (gene F22K18.250) is located on chromosome 4 and At-µD-Ad (gene F25P12.96) is located on chromosome 1.
All plant µ-adaptin sequences resemble their mammalian counterparts in size: µA-adaptin 438 amino acids (aa), µB1-adaptin 411 aa, µB2-adaptin 428 aa, µC-adaptin 451 aa, and µD-adaptin 445 aa, all with predicted molecular masses of around 50 kDa. Comparison of µA-adaptin sequence with the other plant µ-adaptin sequences reveals almost equal similarities (Table 1), with a slightly higher degree of similarity to µB1- and µB2-adaptin than to µC- and µD-adaptin sequences. When compared with mammalian sequences, the highest degree of similarity was observed to the mouse µ1- and µ2-adaptin sequences, while the lowest degree of similarity is to mammalian µ3A- and µ4-adaptins. Although µA-adaptin shows an almost equally high similarity to mammalian µ1- and µ2-adaptins, and cannot be correlated to either one on the basis of similarity alone, it is the only plant µ-adaptin with the highest similarity to mammalian µ2-adaptin (Table 2). As µ2-adaptin is reported to have the highest affinity and broadest specificity for YXXØ signals (Ohno et al., 1998), we decided to take µA-adaptin as an exemplary representative of the plant µ-adaptins to prove their general function. Comparison of the other plant µ-adaptins with the mammalian µ-adaptins reveals a significant similarity of both µB1- and µB2-adaptins to mouse µ1-adaptin and for µD-adaptin to human µ3A-adaptin, while µC-adaptin cannot be correlated unequivocally to any of the mammalian medium subunits (Table 2).
Table 1. Comparison of similarities between At-µA-Ad (GenBank Accession no. AF009631) and µ-adaptin sequences from Arabidopsis and mammals by pairwise blast alignment (Blosum62 matrix)
Table 2. Comparison of similarities between plant and mammalian µ-adaptin sequences using pairwise blast alignment (Blosum62 matrix)
Identities/similarities are given in percentage. The clones are identical to those in Table 1.
In their paper on the crystal structure analysis of the RBD of µ2-adaptin complexed with the tyrosine motifs of TGN38 and the EGF-receptor, Owen and Evans (1998) had taken a plant µ-adaptin sequence into comparison, which is identified here as the µA-adaptin sequence. As shown in Figure 1, we have demonstrated by alignment of the five plant µ-adaptin sequences that all of the amino acids crucial for the binding of the tyrosine motif are also highly conserved in the five plant µ-adaptin sequences. Almost all amino acid exchanges within the β-sheet strands between the plant µ-adaptins are conserved, thus leading to the expectation of an analogous function for all plant µ-adaptins.
Expression of the RBD of µA-adaptin
Based on the significant similarity between plant µA- and mammalian µ1- and µ2-adaptin sequences we expected the plant µA-adaptin to bind tyrosine motifs, which are located within the cytoplasmic tails of membrane proteins. To demonstrate an interaction between µA-adaptin and the YXXØ motif in pull-down assays, it was first necessary to express both proteins as fusion proteins. We made use of the fact that mammalian µ-adaptins have a bipartite structure, with the aminoterminal one-third spanning the β-adaptin-binding region (residues 1–145 in mouse µ1,2-adaptin), and the carboxyterminal two-thirds comprising the receptor-binding region (residues 147–423 of µ1-adaptin; residues 164–435 of µ2-adaptin). Separate expression of either region does not result in a loss of functionality (Aguilar et al., 1997). Based on this knowledge, we have fused the carboxyterminal two-thirds comprising the RBD of µA-adaptin (residues 144–438) to a histidine tag and expressed this fusion protein, which has a molecular mass of 38 kDa (Figure 2a,b).
In order to monitor the expression of the RBD, as well as to obtain an insight into the intracellular localization of µA-adaptin, two independent sets of peptide antibodies were raised (Figure 2c). To avoid a cross-reaction with the other plant µ-adaptins, epitopes were chosen in regions of the µA-adaptin sequence, which share extremely low similarities and no continuous stretch of identical residues with either of the other four adaptin sequences (Figure 1). The peptide sequence of the first epitope is located at the border between the aminoterminal one-third and the carboxyterminal two-thirds, while the peptide sequence of the second epitope is located within the RBD region of the µA-adaptin. Both antibodies recognize their epitope when coupled to BSA but do not recognize BSA alone, indicating that labeling is specific for both antibodies (Figure 2c). In addition, both antibodies cross-reacted with the histidine-tagged RBD fusion protein of µA-adaptin in immunoblots, which was additionally confirmed by labeling with the histidine antibody (Figure 2b).
µA-adaptin is able to bind to threefold repeats of tyrosine motifs (YXXØ)
A putative tyrosine motif in the carboxyterminal end of all plant VSRs, 606YMP609L in VSR-PS1, corresponds to the consensus motif YXXØ in mammalian receptors. However, before attempting to show that the plant µA-adaptin is a binding partner for plant sorting receptors, we first investigated its binding capacity toward the tyrosine motif of TGN38, a protein that cycles between the TGN and the PM in mammalian cells (Bos et al., 1993; Humphrey et al., 1993; Wong and Hong, 1993). The tyrosine motif (YQRL) in the cytoplasmic tail of TGN38 is known to function in both the endocytic and the lysosomal transport pathways, and is also well known for its binding to mammalian AP complexes and isolated µ-adaptins and exhibits the highest affinity observed so far, especially when used as a triple repeat (Ohno et al., 1995).
For heterologous binding experiments, the threefold repeat of the wild-type (wt) tyrosine motif of TGN38 (SDYQRL)3× was expressed as a glutathione-S-transferase (GST) fusion protein, immobilized on GSH–Sepharose and incubated with the histidine-tagged RBD of µA-adaptin. A mutated sorting signal of TGN38, in which the critical tyrosine is substituted for alanine (SDAQRL)3×, served as a negative control. Binding of µA-adaptin to GST-(SDYQRL)3× in this pull-down assay was detected by immunoblotting using a histidine antibody.
As shown in Figure 3(a), the plant µA-adaptin bound to the TGN38 motif (SDYQRL)3× with high specificity while almost no binding to the control was detectable. Furthermore, binding of µA-adaptin to the alanine-substitution mutant (Y/A) was reduced to 16% demonstrating that the critical tyrosine residue of TGN38 is also crucial for binding to plant µA-adaptin. These results provide strong evidence in support of the notion that the binding mechanism for tyrosine motifs is conserved between plant and mammalian µ-adaptins. The functional analogy between plant and mammalian µ-adaptins gains further support from our homologous binding experiments using plant µA-adaptin and the triple repeat of the plant VSR-PS1 (AQYMPL)3×.
As seen in Figure 3(b), the threefold repeat of the pea VSR-PS1 motif recognized the µA-adaptin, showing a weaker binding compared to the TGN38 motif, which is not surprising (see above). This interaction is nevertheless specific, as binding to the alanine substitution mutant (Y/A) was reduced by more than 60% and was not detectable in the control (GST). In additional experiments, the binding of the histidine-tagged RBD of µA-adaptin to the tyrosine motifs of TGN38 and VSR-PS1 was revealed to be saturable using 6–8 µg for TGN38 and 12 µg for VSR-PS1 (data not shown).
Interaction of VSR-PS1 cytoplasmic tail with µA-adaptin
Binding of the threefold repeats of the TGN38 and the VSR-PS1 sorting motifs has revealed that ArabidopsisµA-adaptin can function as a binding partner in both cases. To confirm this in a situation that more resembles the in vivo state, the tyrosine motif of VSR-PS1 was examined when located in the complete cytoplasmic tail sequence. In a first set of experiments, the full cytoplasmic tail sequence of VSR-PS1, as well as its alanine substitution mutant (Y/A), was fused to GST and used to assay its binding capacities to the histidine-tagged RBD of µA-adaptin. As shown in Figure 4(a), µA-adaptin binding to the complete VSR-PS1 cytoplasmic tail was clearly detectable well above over a low background binding to the control GST. However, we noticed significant binding of µA-adaptin to the mutated cytoplasmic tail, estimated as being 50% compared to binding of the wt VSR-PS1 cytoplasmic tail. To corroborate our findings through a complementary approach, we also measured µA-adaptin binding to the cytoplasmic tail of VSR-PS1 by surface plasmon resonance (SPR) measurements (BIAcore 3000).
The GST-tagged wt cytoplasmic tail of VSR-PS1 was captured with GST antibodies, which were immobilized on the sensor surface. Subsequently, the histidine-tagged RBD of µA-adaptin was passed over the sensor surface in order to monitor binding to the VSR-PS1 fusion protein. As shown in Figure 4(b), binding of the VSR-PS1 cytoplasmic tail to µA-adaptin was detectable and exhibited an equilibrium rate constant of 144 nm (see Table 3 for details). Simultaneously, µA-adaptin binding to two further controls in addition to GST was recorded, one comprising the alanine substitution (YL/AA) mutant of VSR-PS1, in which the putative sorting motif YMPL was altered and a control fusion protein in which the VSR-PS1 cytoplasmic tail amino acid residues were randomly mixed (GST-GSGYEHPIASERIKDVPYMQERISQDGRRQLM-YNQAHNM). The binding of the µA-adaptin RBD to these controls was reduced to 20% (690 nm for the alanine substitution mutant and 633 nm for the random mutant), showing that binding of µA-adaptin to the VSR-PS1 cytoplasmic tail is dependent upon the sorting motif. Furthermore, comparison of µA-adaptin binding to the wt VSR-PS1 cytoplasmic tail and the alanine substitution (YL/AA) mutant confirms the results obtained in the pull-down assay, namely demonstrating that the YMPL motif in the cytoplasmic tail of VSR-PS1 is necessary for binding of µA-adaptin and thus reconstitutes a classical tyrosine sorting motif.
Table 3. Kinetic values for the µA–adaptin interaction with the VSR-PS1 cytoplasmic tail
ka (m−1 sec−1)
Sensorgrams such as those shown in Figure 4 were used to calculate the rate constants for association, dissociation, and the equilibrium rate constant for the binding of µA-adaptin to the VSR-PS1 cytoplasmic tail. Note that the affinity of µA-adaptin for the wt cytoplasmic tail was around four times higher as compared to all other controls, showing the specificity of the µA-adaptin binding and the requirement of a functional tyrosine sorting motif.
9.6 × 103
5.6 × 10−3
2.8 × 104
4.1 × 10−3
6.0 × 103
3.8 × 10−3
5.7 × 103
3.9 × 10−3
The assembly status of µA-adaptin
A primary question is whether µA-adaptin is a subunit of a putative AP complex or exists as a single protein associated with clathrin as for example AP180/CALM (Ahle and Ungewickell, 1986; Dreyling et al., 1996). In order to detect the presence of µA-adaptin in a putative plant AP complex, we chose an approach that has been used extensively for the purification of mammalian AP complexes. Therefore, we isolated CCV from Arabidopsis cell suspension cultures and also from pig brain for comparison. After their removal from the vesicles, the coat proteins were subjected to gel filtration using a Superose 6 column. As shown in Figure 5, clathrin from both organisms is found in fractions corresponding to the molecular mass of triskelia (670 kDa). The commercially available monoclonal antibody 100-1 recognizes mammalian β1- and β2-adaptins (Figure 5a), but also at least one β-adaptin from Arabidopsis (Figure 5b). This antibody has already been used to detect plant β-adaptins (Holstein et al., 1994). Both the mammalian β- and µ-adaptins are clearly separated from the clathrin peak, thus confirming the successful separation of clathrin coat proteins into clathrin and AP complex peaks, the latter with an average molecular mass of about 280 kDa (Schröder and Ungewickell, 1991). The plant β-adaptin(s) elutes in fractions corresponding to the mammalian AP-complex peak, but more importantly, the plant µA-adaptin shows almost the same distribution as its mammalian counterpart. The mammalian µ-adaptin has a peak in fraction 30, which is shared by the µA-adaptin, which in addition has a second peak at around 66 kDa (fraction 34), corresponding to monomers. Thus, the gel filtration characteristics of the µA-adaptin are consistent with its presence in a multimeric AP complex. As all our antibody preparations proved to be unsuitable for immunoprecipitation, we were unfortunately not able to identify the other µA-adaptin-interacting subunits.
Intracellular localization of the µA-adaptin
Subcellular fractionation of Arabidopsis cell cultures was performed in order to gain an insight into the intracellular localization of µA-adaptin. Therefore, microsomal membranes were separated on a linear sucrose gradient (20–55% w/w) under isopycnic conditions (Figure 6). A clear separation of PM (39.5–46.5% w/w) from the bulk of Golgi membranes (27–32% w/w) was obtained. This is shown in immunoblots using standard marker antibodies, namely against the 100 kDa subunit of PM-ATPase (Villalba et al., 1991), and against reversibly glycosylated protein (RGP), a 40-kDa Golgi marker (Dhugga et al., 1997).
The distribution of the TGN, of a pre-vacuolar compartment (PVC) and also of an Arabidopsis VSR were also scrutinized in this gradient using a commercially available antibody against the TGN localized syntaxin TLG2a (Bassham et al., 2000), a polyclonal antibody directed against the PVC-localized syntaxin Syp21 (Sanderfoot et al., 1998) as well as a polyclonal antibody against VSR-At1 from Arabidopsis, respectively (Sohn et al., 2003). The distribution of the bulk of VSR membranes (28.5–37% w/w) is very similar to that of the PVC marker Syp21 (28.5–36% w/w) and also to that of the TGN marker TLG2a (30.5–38% w/w). The VSR antibody revealed a double band, indicating the possibility of two VSR isomers labeled by this antibody.
µA-adaptin (30.5–41.5% w/w) co-fractionates with the TGN, the VSR(s) and Syp21, and also partially overlaps with the Golgi. It does not correlate with the distribution of the PM marker.
Clathrin is distributed throughout the gradient. It is detected in PM-enriched fractions as well as in those fractions, which contain the PVC, the TGN and the Golgi apparatus. Such a distribution is in accordance with the sites of CCV formation in the cell (the PM, the TGN, and the recycling endosomes). It is also in agreement with the notion that clathrin functions together with VSR(s) and µA-adaptin at the TGN.
In order to confirm that the Golgi/TGN, and not the PM, is the location of the µA-adaptin, we have performed immunofluorescence studies on paraffin sections as well as immunogold electron microscopy on ultra-thin cryosections prepared from Arabidopsis root tips (Figure 7). As seen in Figure 7(a), a punctate labeling pattern was obtained suggestive of either the PVC, endosomes or the Golgi apparatus. In order to identify the structure labeled by the µA-adaptin antibody, we used the same peptide antibody on ultra-thin cryosections (Figure 7c,d). The gold label revealed the µA-adaptin to be restricted to Golgi stacks, especially at the periphery of their trans face. Label was absent from the PM (Figure 7c).
Features of Arabidopsisµ-adaptins
In mammalian cells, the µ1- and µ2-adaptins are the most prominent binding partners at the Golgi and PM for the YXXØ sorting/internalization motifs located in the cytoplasmic tails of a number of receptors and transmembrane proteins (Marks et al., 1997). While mammals have six µ-adaptins present in four AP complexes (Kirchhausen, 1999), yeast has three AP complexes with four µ-adaptin homologs (Yeung et al., 1999). A. thaliana has in total five µ-adaptin sequences, which in size resemble their mammalian and yeast counterparts.
All amino acids crucial for binding to the tyrosine and the bulky hydrophobic amino acid of the tyrosine motif are strictly conserved in plant µ-adaptins with the exception of the exchange of 423E in the µA-adaptin sequence and for that of the 438R residue in the µC-adaptin sequence. Interestingly, changes at these positions are also found in µ1–µ3-adaptins from various non-plant species. The variability at this position may reflect the preference for different amino acids in the Y + 3 position in various tyrosine motifs and may point to the existence of other tyrosine motifs in plant proteins.
Sequence comparison between µA-adaptin and mammalian µ-adaptins, reveals an almost equal similarity to both, mammalian µ1- and µ2-adaptins. This is in agreement with the fact that in the tyrosine motif of VSR-PS1, the Y + 2 position (proline) and the Y + 3 position (leucine) match the preferences of both µ1- and µ2-adaptins (Ohno et al., 1998). Moreover, µA-adaptin contains the consensus motif GYPQ within its aminoterminal domain, which is identical to that from mouse µ1- and µ2-adaptin. The mammalian motif is also able to interact with the dileucine signal of the invariant chain (Bremnes et al., 1998), implicating similar kinds of sorting motifs in plant transmembrane proteins as well.
Compared to µA-adaptin, µB1- and µB2-adaptins possess the most appropriate amino acid sequences that are required for the structure of the third pocket, which is responsible for binding to a hydrophobic amino acid at the Y − 3 position (Owen et al., 2001). A scrutiny of the cytoplasmic tail sequence of VSR-PS1 reveals that a hydrophobic amino acid is also at the Y − 3 position, as is also the case for P-selectin (Owen et al., 2001). It is therefore possible that a third hydrophobic pocket in plant µ-adaptins may play an additional role in binding to tyrosine motifs (see below).
Binding of µA-adaptin to YXXØ motifs
The tyrosine motif sequences of VSR-PS1 and six Arabidopsis VSR homologs are 100% identical (YMPL), and in one other Arabidopsis VSR homolog, there is only one amino acid exchange (YIPL). Furthermore, a comparison of the full-length sequences from VSR-PS1 (NP471; Paris et al., 1997) with AtELP (Ahmed et al., 1997) has revealed an overall identity at the amino acid level of 72% (Paris et al., 1997). Based on a blast search (data not shown), even higher overall similarities (82%) to other members of the Arabidopsis VSR family were observed, some of them described earlier as being localized at the PM (Laval et al., 1999). Therefore the tyrosine motifs and the cytoplasmic tails from both plants may be considered to be equal in terms of adaptin binding.
Earlier attempts to elucidate an interaction between a receptor and an AP complex from our and other laboratories have been based solely on heterologous binding experiments. While in one set of experiments, AtELP YXXØ-peptides were used for competition (Sanderfoot et al., 1998), in the other two cases, the receptor interacted with coat protein fractions, but in neither case could it be ruled out that proteins other than the AP complex subunits were responsible for the binding (Butler et al., 1997; Robinson et al., 1998b).
In our initial bead-binding experiments with the receptor binding domain of the µA-adaptin from Arabidopsis, we used the threefold repeat of the tyrosine motif of TGN38 and of the VSR-PS1 (AQYMPL) from pea, as the same strategy is an established procedure to verify AP complex binding to sorting signals in mammalian systems (e.g. Boge et al., 1998; Boll et al., 1996; Höning et al., 1996; Heilker et al., 1996; Ohno et al., 1995; Rodionov and Bakke, 1998). Furthermore, the µ-adaptin subunit alone or a µ-adaptin fragment comprising the RBD as a binding partner instead of complete AP complexes in binding experiments have been successfully used previously (Aguilar et al., 1997; Hofmann et al., 1999; Ohno et al., 1995; Rodionov and Bakke, 1998; Stephens et al., 1997). Our heterologous binding experiment using the threefold repeat of the TGN38 motif proves that the RBD of µA-adaptin is functionally conserved between plants and mammals, as µA-adaptin was capable of binding to the tyrosine motif and especially because the critical tyrosine residue was also crucial for binding. Thus, binding of the heterologous TGN38 motif by µA-adaptin demonstrates the conservation of plant µ-adaptin functions.
We were able to confirm these results in homologous binding studies by using the tyrosine motif of the VSR-PS1 receptor, not only as a threefold repeat in pull-down assays, but also as a single motif located within the cytoplasmic tail in plasmon resonance measurements. The equilibrium dissociation constant we measured for the VSR-PS1 cytoplasmic tail (KD = 144 nm) is within the range of values obtained in numerous studies for other AP-complex-binding studies (Heilker et al., 1999). The binding of the cytoplasmic tail of VSR-PS1 to ArabidopsisµA-adaptin was of high specificity for the tyrosine motif because all three controls show equilibrium constants, which are more than four times lower (584–690 nm). Here, the alanine mutant (YL/AA) exhibited the lowest affinity binding, suggesting that the interaction between µA-adaptin and the VSR-PS1 motif is based on a ‘two-pinned plug’ rather than on a ‘three-pinned plug’ (Owen et al., 2001). The cytoplasmic tails of VSR-PS1 and Arabidopsis VSR homologs contain another tyrosine motif (YMDS/A) upstream of the YMPL motif, which does not fit the consensus sequence. Furthermore, there was no other mammalian consensus sorting motif observed within the VSR-PS1 cytoplasmic tail, so that the YMPL sequence can be regarded as the only likely functional sorting motif.
In conclusion, the binding studies reported above represent a direct interaction between an AP complex subunit and a sorting receptor motif in plants. The strong binding of µA-adaptin to TGN38 is in perfect agreement with previous studies, which show that the tyrosine motif of TGN38 exhibits by far the highest affinity towards all mammalian µ-adaptins, especially when used as a triple repeat (Boll et al., 1996; Ohno et al., 1995). As both the position and the residues surrounding the critical tyrosine are important determinants of interaction (Ohno et al., 1996), the binding of multiple repeats of a single motif does not reflect the in vivo situation, and the binding values obtained cannot be compared to repeats from other motifs or even to single motifs located in their natural environment. Thus, it is not surprising that the TGN38 motif displays a stronger binding affinity towards µA-adaptin than does the plant motif itself. In our studies, we have used the pea VSR-PS1 tyrosine motif together with the ArabidopsisµA-adaptin. Although, they do not represent the correct in vivo combination, they served well to demonstrate the conserved features of receptor motif µ-adaptin interactions. Elucidation of the correct in vivo binding partners can only be achieved by comparative binding studies with the five plant µ-adaptins and seven Arabidopsis VSR homologs.
µA-adaptin is a subunit of the putative AP complex
Gel filtration chromatography is a common procedure to reveal that AP complex subunits are constituents of heterotetrameric complexes with a molecular mass of about 280 kDa (Dell'Angelica et al., 1997, 1999; Hirst et al., 1999; Schröder and Ungewickell, 1991; Simpson et al., 1997). As a subunit of a putative plant AP complex µA-adaptin was expected to be eluted in fractions corresponding to the Stoke's radius of AP complexes. Here, we were able to demonstrate that the gel filtration characteristics of clathrin coat proteins from mammals and plants are identical, namely their separation into a clathrin and an AP complex peak. Both, the mammalian µ-adaptin and the plant µA-adaptin were found in fractions that correspond to their presence in AP complexes, as well as in fractions that correspond to their monomeric or partially assembled AP complex status (Jarousse and Kelly, 2000). While the mammalian β1- and β2-adaptins showed the same broad distribution as mammalian µ-adaptin, plant β-adaptin(s) was located in the fractions that corresponded to the position of mammalian AP complexes. As µA-adaptin and also plant β-adaptin(s) behave like their mammalian counterparts in gel filtration, we conclude that they are also subunits of a putative plant AP complex.
µA-adaptin localizes to the Golgi/TGN and not the PM
As previously mentioned, as µA-adaptin showed almost the same degree of similarity towards mammalian µ1- and µ2-adaptins, indeed with a slight preference for the latter, it might be expected to be localized to the PM rather than the Golgi/TGN.
Our immunolabeling data, however, convincingly demonstrate that the µA-adaptin is not present at the PM. They are also supported by subcellular fractionation experiments, which clearly show that µA-adaptin-containing fractions do not co-equilibrate with the PM.
In isopycnic sucrose density gradients, the distributions of µA-adaptin and VSR-At1 are almost identical, suggesting that they may be located on the same membrane. Indeed, based on the binding studies described above, this could be expected. According to Li et al. (2002), the bulk of VSR(s) is found at the PVC under steady state conditions. Moreover, the multi-vesiculate nature of the PVC has recently been established through the immunogold labeling of high-pressure-frozen specimens, and by the selective response of the PVC to the phosphatidyl-inositol-3 (PI3)-kinase inhibitor wortmannin (Tse et al., 2004). Our gradient results showing almost identical profiles for VSR-At1 and the PVC marker Syp21, also support this co-localization. On the other hand, the TGN marker TLG2a also shows the same distribution, and as indicated by immunogold labeling, the µA-adaptin is restricted to the trans-/TGN region of the Golgi apparatus. However, this apparent contradiction can be explained in the following way: VSRs cycle between the TGN, where they selectively bind cargo ligands, and the PVC, where the ligands dissociate. VSRs rapidly enter and leave the TGN, where they are collected into CCV with the help of µA-adaptin-containing AP complexes. The coat of the CCV dissociates before fusion of the VSR-containing vesicle with the PVC. If ligand dissociation is slow, VSRs will accumulate in the PVC. Thus, VSRs may be present in both TGN (small amounts) and PVC (the majority) membranes, but the µA-adaptin only becomes associated with TGN membranes. This scenario is based on the assumption that retrograde transport of VSRs occurs without the mediation of µA-adaptin containing AP complexes. Whereas in yeast, retrograde transport of the VSRs occurs via retromer-coated vesicles (Seaman and Williams, 2002; Seaman et al., 1998), this does not appear to be the case for mannosyl 6-phosphate receptors, which are recycled from the pre-lysosomal compartment in animal cells via AP1-bearing vesicles (Huang et al., 2001; Waguri et al., 2003). It is possible that in plants other medium adaptin subunits, such as the µB1/B2, which are closely related to the µ1-adaptin, serve as substitutes for this transport step.
Escherichia coli strains used in this study were DH5α and BL-21DE3 (Hanahan, 1985) and were grown in Luria-Bertani (LB) medium (Sambrook et al., 1989). Cell cultures of A. thaliana (At-7) were obtained from the Max Planck Institute in Cologne, Germany, and grown in the dark under constant rotation (90 r.p.m) and harvested 7 days after inoculation. A. thaliana var. Columbia (wt) was grown in liquid culture (Gamborgs medium; #G-5893, Sigma, Deisenhofen, Germany) for 10 days under constant rotation (90 r.p.m.) with a cycle of 16 h light and 8 h dark.
Fusion protein constructs
The cDNA sequences of the threefold repeats of the TGN38 tyrosine sorting motif (SDYQRL)3×, of the VSR-PS1 motif (AQYMPL)3×, as well as of their alanine substitutions (SDAQRL)3× or (AQAMPL)3× were synthesized as oligonucleotides corresponding to the plus and the complementary minus strands, respectively (Nucleic Acids Product Supply, Göttingen, Germany). In the following, the cDNA sequences of the respective plus strands are given. The cDNA sequence of (SDYQRL)3× is 5′-GATCCAGTGACTACCAACGTTTGAGTGACTACCAACGTTTGAGTGACTACCAACGTTTGTGA-3′. The (SDAQRL)3× cDNA sequence is 5′-GATCCAGTGACGCTCAACGTTTGAGTGACGCTCAACGTTTGAGTGACGCTCAACGTTTGTGA-3′. The (AQYMPL)3× cDNA sequence is 5′-GATCCGCACAATATATGCCCTTGGCACAATATATGCCCTTGGCACAATATATGCCCTTGTGA-3′. The (AQAMPL)3× cDNA sequence is 5′-GATCCGCACAAGCTATGCCCTTGGCACAAGCTATGCCCTTGGCACAAGCTATGCCCTTGTGA-3′. The cDNA sequence of the GST-VSR-PS1 wt cytoplasmic tail fusion protein is 5′-GGATCCGTGTATAAATATAGAATTAGGCAATACATGGATTCT GAAATCAGAGCAATCATGGCACAATATATGCCCTTGGACAGC CAAGAAGAAGGTCCAAATCACGTCAATCATCAAAGAGGTTGAATTC-3′ with a BamHI site in 5′- and an EcoR1 in 3′ added for insertion in frame with the GST-coding sequence into pGEX 2T (Amersham Pharmacia Biotech, Freiburg, Germany). The sequence of the randomly mixed complete cytoplasmic tail sequence of VSR-PS1 is 5′-GGATCCGGATATGAACACCCCATCGCATCTGAAAGAATTAA AGATGTCCCATATATGCAAGAAAGAATCTCTCAAGACGGTAG AAGGCAATTGATGTACAATCAAGCACATAATATGTGA-3′.
To anneal the complementary strands, 4 µg of each of the plus and minus strands were mixed, respectively, incubated for 10 min at 95°C in a water bath and cooled slowly over night (Sambrook et al., 1989). Fusion of each double-stranded oligonucleotide to the GST-tag occurred via its BamHI site at the 5′ end and its blunted 3′ end into the expression vector pGEX-4T-3 (Amersham Pharmacia Biotech). The vector was opened with BamHI and SmaI prior to fusion. The construct was transformed into the DH5α strain, and the plasmid DNA was isolated using a Midiprep kit (Qiagen, Hilden, Germany) and then subjected to commercially sequencing (MWG AG Biotech, Ebersberg, Germany).
The cDNA comprising the sequence of the RBD of µA-adaptin was cut out from λ ZipLox vector using the restriction enzymes BstEII (position 602 bp) and HindIII (position 1724 bp). The resulting 1122-bp fragment was then cloned into the expression vector pQE30 (Qiagen), which had been opened using SmaI and HindIII. Ligation finally resulted in the construct of histidine-tagged RBD of µA-adaptin ((His)6x-RBD-µA-adaptin).
Expression and purification of fusion proteins
Luria-Bertani (0.5 l) were inoculated in a 1 : 10 dilution from an overnight culture. Cells were grown until an OD600 of 0.7 was obtained. For induction isopropyl-beta-d-thiogalactopyranoside (IPTG) was added to a final concentration of 1 or 2 mm for (His)6x- and GST fusion proteins, respectively. Induction occurred at room temperature 3 h for GST fusion proteins and 2 h for (His)6x fusion proteins. Bacteria were obtained by centrifugation at 4000 g at 4°C for 15 min and frozen at −80°C after washing once with PBS.
Purification of GST fusion proteins was performed as described by Smith and Johnson (1988). Bacteria obtained from 0.5 l cultures were re-suspended in 20 ml of buffer A (50 mm Tris, pH 7.5, 100 mm KCl, 5 mm EDTA) plus protease inhibitors (2 mm leupeptin, 0.7 µm pepstatin, 2 g ml−1 aprotinin, 0.15 mm phenylmethylsulfonyl fluoride) and disrupted by sonication (six bursts of 30 sec with 90% power (Bandelin Sonoplus GM70). Centrifugation at 100 000 g for 20 min removed the cell debris. The supernatants were then incubated with 500 µl of GSH–Sepharose beads on a rotator at 4°C, sequentially washed with 30 ml of buffer A and with 10 ml of buffer B (50 mm Tris–HCl (pH 7.5), 1 mm KCl, 5 mm EDTA). Elution of GST fusion proteins was achieved at a final concentration of 5 mm glutathione.
For the purification of (His)6x-RBD of µA-adaptin bacteria obtained from 1-l culture were re-suspended in 20 ml of lysis buffer (10 mm Tris–HCl (pH 8.0), 50 mm NaH2PO4, 100 mm NaCl). The cells were disrupted by sonication as described above and then centrifuged for 20 min at 100 000 g. The supernatant was incubated with 750 µl Talon™ metal affinity resin (Clontech, Palo Alto, USA) for 1 h at 4°C on a rotator. The Talon™ was separated from the solution and washed sequentially with 20 ml of lysis buffer, 20 ml of washing buffer (20 mm Tris–HCl (pH 8.8), 100 mm NaCl), and 1 ml of washing buffer, including 50 mm Imidazol. An Imidazole concentration of 500 mm was used to obtain optimal elution.
For binding, both the GST fusion proteins and the (His)6x fusion proteins were changed into binding buffer (100 mm Tris–HCl (pH 7.5), 5 mm EDTA, 0.1% (v/v) Triton X-100; Rodionov and Bakke, 1998) using NAP-5 columns (Amersham Pharmacia Biotech). 100 µg packed glutathione–Sepharose beads were prepared according to the manufacturer's instructions (Amersham Pharmacia Biotech) and subsequently incubated with 50 µg of the respective GST fusion protein for 30 min at 4°C. The beads were washed three times with 500 µl of binding buffer and centrifuged at 510 g. For each binding experiment, 30 µl of the pre-incubated GSH beads were incubated with 1–16 µg of the (His)6x-µA-RBD construct and BSA in a final concentration of 1% (w/v), filled up to 100 µl final volume and incubated for 1.5 h at 4°C on a rotator. The beads were washed two times with 200 µl of binding buffer, and the final pellet was re-suspended in twofold sample buffer (Laemmli, 1970). The samples were boiled at 95°C for 1 min and subjected to SDS–PAGE. Each binding assay was independently performed three times with each of the three samples.
Association of the VSR-PS1 cytoplasmic tail with the RBD of µA-adaptin was analyzed in real-time by SPR using a BIAcore 3000 biosensor (BIAcore AB, Freiburg, Germany). In brief, an anti-GST monoclonal antibody (BIAcore AB) was immobilized on all four flow cells of a C1 sensor chip, which has a flat carboxymethylated surface at equal densities according to the manufacturers' instructions. Subsequently, the chip was used to capture the GST fusion proteins harboring the sorting motif sequences at a flow rate of 5 µl min−1 followed by the analysis of µA-adaptin binding. All interaction experiments were performed with binding buffer (100 mm Tris–HCl (pH 7.5), 5 mm EDTA, 0.1% (v/v) Triton X-100) at a flow rate of 20 µl min−1. Association for 1 min was followed by dissociation for 1 min during which the binding buffer was perfused. A short-pulse injection (15 sec) of 20 mm NaOH/0.5% SDS was used to regenerate the sensor chip surface after each experimental cycle. The GST-EnvCD-derived sensor chip remained stable and retained its specific binding capacity for all experimental cycles of association/dissociation and regeneration. The (His)6x-RBD of µA-adaptin was used at final concentrations ranging from 600 to 1000 µm. To exclude distortions because of injection and mixing, segments of the sensorgrams recorded 15–20 sec after switching from buffer flow to AP solution or 5–10 sec after switching back to running buffer were used for the calculations of association and dissociation rate, respectively.
Kinetic parameters and equilibrium dissociation constants were determined from sensorgrams recorded at different AP concentrations. The association constant ka, the dissociation constant kd, and the equilibrium constant KD = kd/ka were calculated using biacore kinetic evaluation software, assuming pseudo-first order kinetics A + B ⇔ AB. The model calculates the association rate constant ‘ka’ and the steady-state response level ‘Req’ by fitting data to the equation R = Req(1−e), where t is the time in seconds, Req is the steady-state response level, and C the molar concentration of APs in the injection solution. The steric interference factor n, which describes the valency of the interaction between µA and VSR-PS1 was set to 1. The dissociation rate constant, kd, was determined by fitting the data to the equation R = R0e, where R0 is the response level at the beginning of the dissociation phase. This model has recently been applied to describe the interactions of APs with cytosolic domains (Heilker et al., 1996; Höning et al., 1997) and is described in more detail elsewhere by Jonsson et al. (1991).
Gel electrophoresis, immunoblotting, and protein determination
SDS-gradient gels (10–19%) were prepared as previously described by Holstein et al. (1996). Proteins were blotted onto nitrocellulose (Towbin et al., 1979) and visualized with the Supersignal West Pico ECL kit (Pierce, Rockford, USA). Bands on the ECL film were quantified by using the BASYS gel-analyzing system (BioTec Fisher, Reiskirchen, Germany). Protein concentrations were determined according to Bradford (1976).
For immunoblot studies, the following antibodies were used: the affinity-purified polyclonal µA-adaptin peptide antibodies #9139 (20 ng µl−1) and #2080 (184 ng µl−1), a polyclonal antiserum against RGP (1 : 10.000 dilution; supplied by Dr K. Dhugga, Pioneer Hi-Bred International Johnston, IA, USA), a monoclonal PM-ATPase antibody (1 : 500 dilution; supplied by Dr W. Michalke, University of Freiburg, Germany), a polyclonal syntaxin of plants (Syp)21 antibody (1 : 1000, raised according to Sanderfoot et al., 1998), a monoclonal CHC antibody (1 : 500 dilution) and a monoclonal AP-50 antibody (1 : 250 dilution) both from BD, Tranduction Laboratories, Franklin Lakes (USA), the monoclonal 100-1 β1/β2-adaptin antibody (1 : 100 dilution; Sigma, Taufkirchen, Germany), the TLG2a antibody in a 1 : 1000 dilution from Rose Biotechnology Inc. (http://www.rosebiotech.com) and an (His)6x antibody (BioScience, Göttingen, Germany) in a 1 : 2000 dilution and a GST antibody (Amersham Pharmacia Biotech) at a 1 : 1000 dilution. Secondary antibodies against mouse or rabbit coupled to horseradish-peroxidase, respectively, were obtained from Sigma (Taufkirchen, Germany).
For immunofluorescence studies, the affinity-purified polyclonal µA-adaptin peptide antibody #2080 (184 ng µl−1) was used in a 1 : 10 dilution and the secondary antirabbit IgG coupled to Cy-5 in a final concentration of 1 : 200 (Jackson ImmunoResearch Laboratory Inc., West Grove, PA, USA).
For immunolabeling on cryosections, the primary antibody was the affinity-purified µA-adaptin #2080 antibody in a 1 : 50 dilution. The secondary goat antirabbit IgG coupled to 10-nm-diameter particles of gold was obtained from BioCell (Cardiff, UK) and diluted in 1 : 30 in PBS containing 1% (w/v) BSA.
Peptide coupling to BSA
Each peptide (4 µg µl−1) was coupled to H2 fraction of BSA (20 µg µl−1; Applichem, Germany) via 1% (v/v) glutaraldehyde (Sigma, Deisenhofen, Germany) in a final volume of 1 ml on a rotator at 4°C for 2 h. Subsequently samples were dialyzed for 16 h at 4°C against a buffer containing 0.1 m NaHCO3, 0.5 m NaCl (pH 8.3). A protein dot-blot immunoassay was made with 10 and 25 ng of either BSA alone or the conjugate. The sera of #2080 and #9139 were used in 1 : 500 dilutions each in PBS/5% (w/v) skim milk. The proteins were visualized with the Supersignal West Pico ECL kit (Pierce, Rockford, USA).
Twenty grams of Arabidopsis cell culture were homogenized in 10 ml TDKE-buffer (25 mm Tris–HCl (pH 7.0), 1 mm dithiothreitol, 10 mm KCl, 3 mm EDTA), including the final concentrations of the following antiproteases: 2 mm leupeptin, 0.7 µm pepstatin, 2 g ml−1 aprotinin, and 0.15 mm phenylmethylsulfonyl fluoride using a ‘Zauberstab’-blendor. After filtration through three layers of miracloth, the homogenate was pre-centrifuged for 15 min at 8000 g and the resulting supernatant for another 15 min at 18 000 g in a Heraeus swing-out rotor #2704.
The supernatant was then applied on a 5-ml 60% (w/w) sucrose cushion in TDKE buffer and centrifuged for 45 min at 100 000 g in a swing-out TST28.38 rotor. The interphase was removed and put on a linear 20–55% (w/w) sucrose gradient in TDKE buffer and spun for 20 h at 100 000 g in a swing-out TST28.38 rotor. Fractions of 1.5 ml were taken, and aliquots of each fraction were subjected to SDS–PAGE after TCA precipitation.
Gel filtration on superose 6/FPLC
Clathrin-coated vesicles were isolated from pig brain as described elsewhere by Lindner (1994) and from Arabidopsis cell culture cells. Fifty grams of frozen cell culture were homogenized using a mortar and a pestle and subsequently re-suspended in 50 ml of buffer A (0.1 m 2-morpholinoethansulfonic acid (MES) (pH 6.4), 0.5 mm MgCl2, and 1 mm EGTA, 1 mm EDTA + inhibitors, see above). Pre-centrifugation was performed as described above, and the supernatant was subsequently centrifuged (T12.50 rotor, 4°C, 120 000 g, 1.5 h) to obtain the microsomal pellet. The obtained pellet was re-suspended in 4 ml of buffer A + 300 mm sucrose. A ficoll/sucrose (12.5% Ficoll 400, 12.5% sucrose) solution was added in a 1 : 1 ratio, and the mixture was centrifuged in an AT-5 rotor (40 min, 43 000 g). The obtained supernatant was then fivefold diluted and centrifuged for 1 h at 110 000 g in T12.50 rotor to obtain a pellet enriched in CCV. This pellet was re-suspended in 1 ml of buffer A and 1 g Hydroxyapatite gel equilibrated before in buffer A (DNA-grade Biogel, Bio-Rad, Munich, Germany) was incubated in a batch procedure for 1 h at 4°C with slow motion and then washed two times with 5 mm potassium phosphate (PP) in buffer A (2 min, 4°C, 600 g). CCVs were removed from Hydroxyapatite gel by stepwise elution using 300, 400, and 500 mm PP in buffer A. CCV of each elution step were then subjected to centrifugation (1 h, T12.50 rotor at 110 000 g), and the CCV pellets were combined in a total volume of 2 ml buffer A.
To remove the coat proteins from mammalian or plant CCV, approximately 1 mg CCV were incubated in final 0.5 m Tris/buffer A for 30 min at 4°C. Centrifugation for 1 h at 120 000 g in a T12.50 rotor then separated the soluble coat proteins from the vesicle membranes. About 0.5 mg/0.5 ml coat proteins were applied on a Superose 6 column connected to a FPLC (Amersham Pharmacia Biotech). The flow rate was 0.3 ml min−1, and the fraction size was 0.5 ml. Each fraction was analyzed on SDS–PAGE. Calibration of Superose 6 column was performed using the Molecular weight marker kit (MW-GF-1000) from Sigma (Deisenhofen, Germany).
Immunofluorescence studies on paraffin sections of Arabidopsis root tip cells
Paraffin embedding of 5-mm-long tips of Arabidopsis roots from 10-day-old seedlings grown in liquid culture was performed as described elsewhere by Jauh et al. (1998). Immunofluorescence studies were performed as outlined by Paris et al. (1996) using the antibody as given above. Images were taken with the Bio-Rad Confocal laser scanning microscope MRC-10242PMT system and the laser sharp program versus 3.02, with a laser power of 1%. The recorded images were processed by adobe photoshop 7.0 software.
Cryosectioning and immunogold labeling
Root tips 5 mm long were excised and transferred into a primary fixative (1.5% (w/v) paraformaldehyde, 0.2% glutaraldehyde (v/v) in 0.1 m PP buffer (pH 7.4) for 12–16 h at 4°C, then infiltrated with 2.3 m sucrose in PP buffer over night at 4°C, and finally embedded in 10% (w/v) (37°C) gelatine solubilized in PP buffer. Gelatine cubes (0.8 mm3) were infiltrated in 2.3 m sucrose for 10 min at 4°C and then mounted onto specimen stubs, frozen in liquid nitrogen, and cut into ultra-thin sections at −80°C with a Leica UCT microtome (Leica, Bensheim, Germany). Frozen sections were picked up as described by Liou et al. (1996) and transferred to formvar-carbon-coated nickel grids. Immunogold labeling was performed as described by Pimpl et al. (2000), except that the staining procedure with uranyl acetate/oxalate was omitted.
We thank Ruth Pilot and Eva Besemfelder-Butz for their excellent technical assistance and are indebted to Drs Stefan Hillmer and Dirk Wenzel for support at the electron microscope and with the cryosectioning, respectively. We thank Dr Giselbert Hinz for helpful comments on the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft (TP A7 and TP A5, SFB 523) and by grants to N.P. and J.-M.N. from the Swiss National Science Foundation (Grant 31-46926.96).