A novel di-acidic motif facilitates ER export of the syntaxin SYP31.

It is generally accepted that ER protein export is largely influenced by the transmembrane domain (TMD). The situation is unclear for membrane-anchored proteins such as SNAREs, which are anchored to the membrane by their TMD at the C-terminus. For example, in plants, Sec22 and SYP31 (a yeast Sed5 homologue) have a 17 aa TMD but different locations (ER/Golgi and Golgi), indicating that TMD length alone is not sufficient to explain their targeting. To establish the identity of factors that influence SNARE targeting, mutagenesis and live cell imaging experiments were performed on SYP31. It was found that deletion of the entire N-terminus domain of SYP31 blocked the protein in the ER. Several deletion mutants of different parts of this N-terminus domain indicated that a region between the SNARE helices Hb and Hc is required for Golgi targeting. In this region, replacement of the aa sequence MELAD by GAGAG or MALAG retained the protein in the ER, suggesting that MELAD may function as a di-acidic ER export motif EXXD. This suggestion was further verified by replacing the established di-acidic ER export motif DLE of a type II Golgi protein AtCASP and a membrane-anchored type I chimaera, TMcCCASP, by MELAD or GAGAG. The MELAD motif allowed the proteins to reach the Golgi, whereas the motif GAGAG was found to be insufficient to facilitate ER protein export. Our analyses indicate that we have identified a novel and transplantable di-acidic motif that facilitates ER export of SYP31 and may function for type I and type II proteins in plants.

Introduction

SNAREs (soluble N-ethyl-maleimide sensitive factor attachment receptor proteins) are critical components of the molecular machinery that facilitates vesicular transport in the secretory pathway of eukaryotic cells (Hong, 2005; Lipka ; Moreau ). These proteins favour membrane fusion of apposing lipid bilayers, and contribute to the targeting of membrane components through the secretory pathway. SNAREs are engaged in several processes in plant biology (cell growth and development, autophagy, gravitropism, stress responses, and resistance to pathogens) that are related directly or indirectly to their fusogenic properties (Pratelli ; Surpin and Raikhel, 2004).

Most SNAREs and SYP31 are type IV proteins anchored to the membrane via a C-terminal (C-ter) transmembrane domain (TMD) anchor (Borgese ; Burri and Lithgow, 2004; Chou and Shen, 2007); only a few SNAREs are either isoprenylated or bi-palmitoylated (e.g. SNAP-25) to be anchored to the membrane (Hong, 2005). SNAREs are also characterised by (i) a basic domain close to the membrane, (ii) one, or rarely two SNARE motifs (coiled-coil domains), which are responsible for SNARE pairing and membrane fusion (Hu ), and (iii) a regulatory N-terminal (N-ter) domain. Although the central role of SNAREs in membrane fusion has been largely investigated, only a few studies have been devoted to the question of how these proteins are targeted to specific membranes. For Ykt6, a C-ter isoprenylated SNARE, targeting is dependent on the N-ter longin domain and not driven by the isoprenylation itself (Hasegawa ). It has also been proposed that the longin domain may interact with a targeting component on the membrane (Hasegawa ).

In the case of type IV SNARE proteins (C-ter TMD anchor), it is clear that the length of their TMD is not sufficient to explain their location. Although TMD influences the localization of type I membrane proteins depending on their length (Brandizzi ), this criterion seems not to be sufficient to explain most of the locations of the ER/Golgi SNAREs which have a different topology from type I with the N-ter facing the cytosol. For type I membrane protein, a 17 amino acid (aa) TMD caused protein retention in the ER; whereas a 20 aa and a 23 aa TMDs caused protein re-distribution to the Golgi and the plasma membrane, respectively (Brandizzi ). However, in yeast, Bet1, Gos1, Sft1, and Sed5 that have a 15 aa TMD are located in the Golgi whereas Sec22 and Slt1, which are located in the ER, have a 19–20 aa TMD (Burri and Lithgow, 2004). In Arabidopsis, Sec22 and SYP31 (homologue to Sed5) have both a 17 aa TMD and different subcellular localizations, Sec22 in the ER/Golgi and SYP31 in the Golgi (Uemura ; Chatre ). Although for type II TMD proteins, such as plant glycosyltransferases, some correlation was found between TMD length and ER/intraGolgi distribution (Saint-Jore Dupas et al., 2006), the correct targeting and subcompartmentation of these proteins in the early secretory pathway was also dependent on other signals present in their cytoplasmic tail (Saint-Jore Dupas et al., 2006). Hanton showed that di-acidic DXE motifs can be functional in type I, type II, and multispanning membrane proteins. Yuasa found that a di-basic motif is involved in the ER export of the type II protein prolyl 4-hydroxylase. Recently, Schoberer have determined that single basic amino acids of the cytoplasmic tail of N-glycan-processing enzymes are responsible for their ER export to the Golgi. In addition, the entire cytosolic N-ter longin domains of the vacuolar VAMP7 SNAREs appeared to be critical for their targeting (Uemura ). Finally, it has also been shown that the longin domain of Sec22 can be critical for ER export in yeast (Liu ). Several ER export motifs have been shown to overcome the influence of the TMD on protein targeting (Hanton ; Yuasa ; Matheson ). However, the influence of such peptides on SNARE trafficking has yet to be demonstrated.

SYP31 is a type IV syntaxin that is required for anterograde trafficking between the ER and the Golgi (Bubeck ) and is localized in the Golgi (Chatre ; Bubeck ). In this work, it was investigated whether the N-ter cytosolic domain of SYP31 (beyond the SNARE domain) is required for ER export of this syntaxin to the Golgi. Deletions and mutations were used to screen the entire cytosolic N-ter domain sequence of SYP31 for its role in ER export to the Golgi. Deletion mutants in the N-ter domain indicated that a region between the helices Hb and Hc was required for Golgi targeting. In this region, it was found that the pentapeptide MELAD contains a novel di-acidic motif EXXD. Mutagenesis analyses showed that such a motif is transplantable and that it can functionally replace the known di-acidic motif DLE of the Golgi protein AtCASP (CASP).

Materials and methods

Cloning of mutant forms of SYP31 and CASP

The primers and fragments used for overlapping PCR cloning of the various mutant forms of SYP31 are described in Supplementary Table S1 at JXB online. PCR fragments were subcloned in the binary vector pVKH18En6 bearing the coding region of fluorescent proteins using the cloning sites XbaI and SalI for C-ter fusions) or BamHI and SacI (N-ter fusions). Plasmid amplification and selection were performed in the E. coli DH5α strain. Positive constructs were sequenced and transformed into Agrobacterium tumefaciens (strain GV301) for subsequent analyses in tobacco leaf epidermal cells.

For the SYP31 mutant form MALAG, overlapping PCR were realized with the forward 5′-ACCCTTCAGAACATGGCGCTTGCTGGTGGGAACTATTCA-3′ and the reverse 5′-TGAATAGTTCCCACCAGCAAGCGCCATGTTCTGAAGGGT-3′ primers (see Supplementary Table S1 at JXB online).

Constructs CASP, CASPDXE1, TMcCCASP, and TMcCCASPDXE1 have already been described by Hanton . Constructs CASPmeladg, CASPgagagg, TMcCCASPmeladg, and TMcCCASPgagagg were generated with specific primers to replace the sequences DLE and GLG by the sequences MELADG and GAGAGG.

Plant material and transient expression systems

Four week-old tobacco (N. tabacum cv. Xanthi) greenhouse plants grown at 22–24 °C were used for A. tumefaciens-mediated transient expression (Batoko ). A. tumefaciens carrying the constructs in the transforming binary vectors were cultured at 28 °C to stationary phase (approximately 24 h), washed, and resuspended in infiltration medium [MES 50 mM pH 5.6, glucose 0.5% (w/v), Na3PO4 2 mM, acetosyringone (Aldrich) 100 μM from 200 mM stock in dimethyl sulphoxide]. The bacterial suspension was inoculated using a 1 ml syringe without a needle by gentle pressure through a small puncture on the abaxial epidermal surface (Brandizzi , b). Transformed plants were then incubated under normal growth conditions for 2 d at 22–24 °C.

Confocal microscopy and fluorescence quantification

Transformed leaves were analysed 24 h, 48 h, or 72 h after infection of the lower epidermis. Images were captured with the following confocal laser scanning microscopes: upright Leica TCS SP2 with a ×63 oil immersion objective (IFR 103, INRA-Bordeaux), and an upright Zeiss LSM510 META with a ×63 water immersion objective (Department of Biology, University of Saskatchewan). The emitted fluorescence of co-expressed -GFP and -YFP constructs was captured by alternately switching the 488 nm and 514 nm excitation lines of an argon ion laser using the multitrack facility of the microscopes. Imaging settings were as described by Brandizzi , b), and Leica and Adobe PHOTOSHOP 7.0 software were used for post-acquisition image processing. Non-saturating imaging settings using low laser output were used to capture images destined for quantification, as described by Brandizzi . Quantification of fluorescence of the different constructs in the ER and the Golgi was performed by single-imaging frame collection from cells expressing each construct, using identical laser output levels and imaging conditions, as in Hanton . Measurements of fluorescence levels were made within a two μm2 circle using ImageJ 1.34-s software in the post-acquisition analysis. Sixty cells were analysed for each fluorescent construct. Ten Golgi and ten ER zones were used per cell, so that a total of 600 Golgi and ER zones were individually quantified. Statistical analyses of the fluorescence means were made with the Student's t test, and the data presented in the Results section have P values <0.01.

Results

The N-ter domain is required for SYP31 targeting to the Golgi

To determine the sequences required for ER export of SYP31 to the Golgi, mutant constructs of SYP31 were spliced to the coding sequence of YFP for subcellular localization analyses (Fig. 1). Coupling YFP to the C-ter of SYP31 transforms SYP31 in a type II membrane protein; this results in exposure of YFP in the ER lumen without disruption of Golgi targeting (Chatre ; see Supplementary Fig. S1 at JXB online), as also observed for several other SNAREs (Zeng ; Bubeck ).

Fig. 1.

SYP31 mutant constructs coupled to YFP. Drawings summarizing the different mutant forms of SYP31 made by deletions in the N-ter domain or mutations in the MELAD motif, and tested for Golgi targeting. Ha, Hb, and Hc represent the three α-helices of the N-ter domain. Snare corresponds to the coiled-coil snare domain for snare interactions. TM is the transmembrane domain. M represents the region between the helices Hb and Hc where the MELAD sequence is present.

It has been reported that the SNARE regions of the rat Bet1 and yeast Sec22 are involved in correct protein targeting (Joglekar ; Liu ). Therefore, a YFP fusion of a SNARE domain deletion of SYP31 was analysed first. Live cell imaging analyses of cells expressing this construct showed that the protein was exported to the Golgi (see Supplementary Fig. S2 at JXB online). This result suggests that the SNARE domain of SYP31 is not required for ER export. We therefore focused on the reminder of the N-ter cytosolic part of SYP31. For this, several deletion and missense mutant forms of a SYP31-YFP construct (summarized in Fig. 1) were produced. Sequence alignments of the different mutant forms of the cytosolic N-ter of SYP31-YFP are shown in Supplementary Fig. S3 at JXB online. The entire cytosolic portion of the protein was first deleted (–Nter mutant). In contrast to the wild-type SYP31–YFP construct, which is targeted to the Golgi apparatus (see Supplementary Fig. S1 at JXB online; Fig. 2A), the N-ter deleted construct (–Nter) was localized mainly in the ER (Fig. 2B). As a consequence it was hypothesized that a portion of the cytosolic N-ter domain contained critical domains for ER to Golgi transport of SYP31.

Fig. 2.

Subcellular localization of different SYP31–YFP mutant constructs expressed in tobacco leaf epidermal cells. (A) Golgi location of SYP31–YFP. (B) The mutant protein without its N-ter domain (–Nter) is retained in the ER. (C) The mutant protein Ha is also retained in the ER. (D, E, F,) The mutant proteins MHc, HaMHc, and +M-Hc are targeted to the Golgi. (G, H, I) The mutant proteins –M-Hc, GAGAG, and MALAG are retained in the ER, and a few aggregates are also observed. Bars=10 μm.

To establish which region in the N-ter domain could be involved in SYP31 Golgi targeting, a portion of the N-ter extremity of SYP31 was removed, corresponding to amino acids 1–29 and 73–157, leaving intact the Ha helix [Ha mutant (Fig. 1; see Supplementary Fig. S3 at JXB online). The Ha mutant construct labelled the ER network (Fig. 2C), suggesting that one or both of the deleted regions to either end of the Ha helix in the N-ter domain of SYP31, but not the Ha helix itself, was involved in targeting SYP31 to the Golgi.

A construct starting with the M region, which spans between helices Hb and Hc , and contains the MELAD sequence (Fig. 1; MHc, grey box), was targeted to the Golgi (Fig. 2D). These results suggest that the N-ter extremity of SYP31 is not required for Golgi targeting and that the presence of the M region and/or the Hc helix was sufficient for correct localization of the construct to the Golgi. This was confirmed by colocalization experiments with the ER and Golgi marker Erd2-GFP (Boevink ) as shown in Supplementary Fig. S4A–C at JXB online.

Addition of the Ha helix to the MHc contruct (HaMHc) did not alter Golgi targeting (Fig. 2E). Similarly, the presence or absence of the helix Hb did not influence the Golgi localization of the chimaeras (data not shown).

To determine whether the M region and/or the Hc helix was responsible for proper protein targeting to the Golgi, the –M-Hc and +M-Hc constructs (Fig. 1; see Supplementary Fig. S3 at JXB online) were used. Figure 2F shows that the +M-Hc mutant construct was correctly targeted to the Golgi, as shown by colocalization with the Golgi marker Erd2GFP (Boevink ) (see Supplementary Fig. S4D–F at JXB online). By contrast, the –M-Hc form failed to reach the Golgi efficiently and it was mostly distributed in the ER (Fig. 2G; see Supplementary Fig. S4G–I at JXB online). These data strongly suggested that the M region but not the Hc helix was implicated in SYP31 export to the Golgi.

Mutation of the EXXD motif decreases SYP31 ER export

Analysis of the M region sequence revealed the presence of an aa stretch, MELAD (Fig. 1; see Supplementary Fig. S3 at JXB online), which appeared to contain a di-acidic-like motif, EXXD. To determine whether this MELAD motif is involved in ER export of SYP31 to the Golgi, this sequence was replaced by GAGAG (Fig. 1; see Supplementary Fig. S3 at JXB online). It was hypothesized that disruption of the MELAD sequence would affect ER export of SYP31. Consistent with our hypothesis, the mutated protein was highly retained in the ER (Fig. 2H; see Supplementary Fig. S4J–O at JXB online).

To exclude the possibility that ER retention could be due to saturation of the export machinery in conditions of protein over-expression, the subcellular localization of SYP31 and its GAGAG (SYP31GAGAG) mutant version was followed in a time-course experiment. A Golgi location of SYP31 was found from 24 h and 48 h of expression (see Supplementary Fig. S5A, B at JXB online), whereas the ER location for the GAGAG mutant form was observed throughout this time period (see Supplementary Fig. S5C, D at JXB online). Analyses were not performed at 72 h because these constructs lead to tissue damage.

Retention of the SYP31GAGAG in the ER was quantified following established protocols (Hanton ). The pixel intensity was quantified within regions of interest of similar areas (see Materials and methods). For this, fluorescence intensities of 600 Golgi and ER regions were quantified in cells expressing either wild-type SYP31–YFP or the SYP31GAGAG mutant. In order to be able to compare and validate the quantifications, cells with similar levels of fluorescence were used. This was established by comparing the average pixel intensity of the fluorescence associated with membranes. Therefore any difference in subcellular location cannot be attributed to a putative different saturation of the secretory pathway. The results were expressed as the ER/Golgi fluorescence intensity ratios. As expected with the wild-type protein construct SYP31, no fluorescence was found in the ER (Fig. 3). By contrast, the fluorescence ratio of ER/Golgi reached a value of about 1 for the SYP31GAGAG mutant (Fig. 3), confirming our qualitative observation (Fig. 2H) that the ER export of this mutant is defective.

Fig. 3.

Fluorescence intensity quantification of the localization of the wt SYP31, and the GAGAG and MALAG mutant constructs expressed in tobacco leaf epidermal cells. The results are expressed as the ratio of fluorescence intensity in the ER membranes to that in the Golgi bodies. Measurements of fluorescence levels were made within a two μm2 circle using ImageJ 1.34-s software in post-acquisition analysis. Sixty cells were analysed for each fluorescent construct. Ten Golgi and ten ER zones were used per cell, so that a total of 600 Golgi and ER zones were individually quantified. The data presented correspond to means ±SD.

To pinpoint whether the putative di-acidic motif EXXD in the MELAD sequence was critical for SYP31 ER export, a mutant form of SYP31 was produced in which the MELAD sequence was replaced by MALAG (SYP31MALAG). As seen for the SYP31GAGAG, the MALAG mutant construct also labelled the ER network (Fig. 2I), and the quantification of the fluorescence, in the same conditions as described above, in both the ER network and the Golgi bodies indicated that the SYP31MALAG was retained to the same extent as the SYP31GAGAG mutant in the ER (Fig. 3), strongly suggesting that the EXXD sequence corresponds to a functional export motif that contributes to targeting of SYP31 to the Golgi.

The EXXD motif can functionally replace canonical di-acidic export motifs

To test the activity of the EXXD motif further, the type II Golgi matrix protein CASP was used (Hanton ; Renna ); disruption of the di-acidic signal DXE in CASP leads to its retention in the ER. Therefore, the CASP DXE sequence was replaced with the entire MELADG sequence. The resultant construct, CASPmeladg, was expressed in tobacco leaf epidermal cells and its distribution was compared with that of wild-type CASP. The wild-type CASP construct was efficiently transported to the Golgi (Fig. 4) as shown in colocalization analyses with Erd2YFP (see Supplementary Fig. S6A–C at JXB online; Hanton ; Renna ). As compared to the wild-type construct, CASPmeladg was mainly trafficked to the Golgi (Fig. 4; see Supplementary Fig. S6G–I at JXB online), showing that the EXXD sequence could substantially function to facilitate ER export of the CASP mutant. To determine whether the neighbouring aa could influence the activity of the EXXD sequence in the CASP mutant, the MELADG motif was mutated to GAGAGG (CASPgagagg). It was found that the CASPgagagg mutant was mostly retained in the ER (Fig. 4; see Supplementary Fig. S6J–L at JXB online) similarly to the ER export incompetent CASPDXE1 (Hanton ; Fig. 4; Supplementary Fig. S6D–F at JXB online). These data support the suggestion that the activity of the EXXD motif is influenced by the aa environment when transplanted into a protein that is different from SYP31.

Fig. 4.

The di-acidic MELAD motif is functional in the type II Golgi matrix protein CASP. (A) Presentation of the different constructs of the type II Golgi matrix protein CASP that were used coupled to GFP. CASP, the AtCASP protein in the type II orientation with the TMD close to the C-ter; CASPDXE1, the mutant version of AtCASP with the first DXE (DLE) motif replaced by a non-functional GLG motif (Hanton ); CASPmeladg, a mutant version of AtCASP with the DLE motif replaced by the MELADG motif; CASPgagagg, a mutant version of AtCASP with the DLE motif replaced by the GAGAGG motif. (B) Fluorescence intensity quantification of the localization of the different constructs expressed in tobacco leaf epidermal cells. The ratios of fluorescence in the ER membranes to that of the Golgi bodies are reported. Six-hundred Golgi and ER zones were individually quantified as indicated in Fig. 3, and the data presented correspond to means ±SD.

In order to test whether this motif could also be functional in the reverse orientation, it was transplanted in a type I protein. For this, an artificial type I protein was used, TMcCCASP that contains an EXD motif exposed in the cytosol (Hanton ). As observed earlier, the TMcCCASP construct was efficiently transported to the Golgi (Hanton ; Fig. 5; see Supplementary Fig. S7A at JXB online). The TMcCCASP construct bearing the MELADG sequence (TMcCCASPmeladg) was also transported to the Golgi, but again with a lower efficiency (Fig. 5; see Supplementary Fig. S7C at JXB online). Then the replacement of the DLE and MELADG sequences by the mutated ones GLG (TMcCCASPDXE1) and GAGAGG (TMcCCASPgagagg) were compared. It was found that the mutated protein construct TMcCCASPDXE1 (Fig. 5; see Supplementary Fig. S7B at JXB online) and TMcCCASPgagagg (Fig. 5; see Supplementary Fig. S7D at JXB online) were retained in the ER. It was also established that overexpression of the different constructs of CASP (type II proteins such as the YFP-coupled versions of SYP31) and TMcCCASP (type I version) did not saturate the secretory pathway up to 72 h (see Supplementary Figs S8, S9 at JXB online).

Fig. 5.

The di-acidic MELAD motif is functional in a type I version of CASP. (A) Presentation of the different constructs of the type I version of AtCASP that were used coupled to GFP. TMcCCASP, the AtCASP protein in the type I orientation with the TMD close to the N-ter and the DLE motif in a reverse orientation (Hanton ); TMcCCASPDXE1, the mutant version of TMcCCASP with the DLE motif replaced by a non-functional GLG motif (Hanton ); TMcCCASPmeladg, a mutant version of TMcCCASP with the DLE motif replaced by the MELADG motif; TMcCCASPgagagg, a mutant version of TMcCCASP with the DLE motif replaced by the GAGAGG motif. As compared to mutants in Fig. 4, the g was added and conserved according to the aa present in the sequence of SYP31. (B) Fluorescence quantification of the localization of the different constructs expressed in tobacco leaf epidermal cells. The ratios of fluorescence in the Golgi bodies to that in the ER membranes are reported. Six-hundred Golgi and ER zones were individually quantified as indicated in Fig. 3, and the data presented correspond to means ±SD.

These results therefore strongly support the conclusion that the sequence MELADG and the included EXXD di-acidic motif are able to replace the DXE motif in both the type II CASP and the type I TMcCCASP proteins.

Discussion

The mechanisms for the transport of transmembrane proteins in plant cells have received much attention in the last few years, both for the characterization of the protein machineries involved and the sequence motifs required for their traffic (Hanton ; Matheson ; Lipka ; Moreau ; Robinson ; Bassham and Blatt, 2008; Langhans ; Nielsen ; Rojo and Denecke, 2008).

Our objective was to determine putative ER export signals in the transmembrane SNARE protein SYP31, since TMD length alone was not compatible with Golgi targeting. The presence of an ER export signal either in the SNARE domain itself or in the cytoplasmic section of SYP31 is hypothesized. It has been shown that the SNARE motif of the rat Bet1 is required for its intracellular targeting (Joglekar ). Similarly, a 10 aa stretch in the SNARE domain of Sec22 has been reported to be involved in efficient ER export of this v-SNARE in yeast (Liu ). Deletion of the SNARE domain did not result in ER retention of the constructs, ruling out a role for the SNARE domain in ER export to the Golgi, as has also been found for the animal homologue syntaxin 5 (Joglekar ).

We found that deletion of the M domain comprising amino acids 108–128 in the N-ter cytosolic section of SYP31 resulted in a redistribution of the construct to the ER. Within the M domain, the presence of a putative di-acidic motif (M)ELAD(G) of the type EXXD was identified. This motif was able to replace the DXE motif in a putative Golgi matrix protein and, as determined for the DXE motif (Hanton ), it was able to function in both orientations. Although the EXXD motif seemed weaker than the DXE motif in the type I and type II versions of CASP (Figs 4, 5), it remained functional, overcoming the 17 aa TMD present in CASP that would be expected to retain the protein in the ER (Brandizzi ). Consequently, even though the EXXD motif seems weaker than the DXE in driving export out of the ER, the signal is hierarchically stronger than TMD length.

Interestingly, deletion of the entire N-ter cytosolic section up to the SNARE domain did not lead to a total retention/block of SYP31 in the ER (see the few dots and/or putative clustered Golgi stacks in Fig. 2B). This is reminiscent of what was observed for GONST1 (Hanton ), suggesting that, although the di-acidic motif can clearly contribute to the export of SYP31 out of the ER to the Golgi, other attributes may influence protein targeting. In adddition, the decrease of SYP31 transport to the Golgi, as quantified in this study, may have been underestimated for two reasons: (i) overexpression of the constructs may have favoured their diffusion through the Golgi (Cosson ), and (ii) ER export sites are closely associated with the Golgi at the resolution of confocal microscopy (daSilva ), and the puncta quantified as Golgi may have been, in fact, a mixture of Golgi bodies and ER export sites, therefore contributing towards an underestimation of the capacity of the di-acidic motif EXXD to drive SYP31 out of the ER. In spite of these methological limitations, our data clearly establish the function of the di-acidic motif EXXD in SYP31 for export of the protein out of the ER.

That a di-acidic motif contributes to ER export and Golgi targeting of SYP31 may imply an interaction of this motif with the COPII machinery (Votsmeier and Gallwitz, 2001). Recently, Hanton have shown that the recruitment of the Sec24 COPII coat protein to ER export sites is influenced by the expression of membrane cargo proteins, and that this recruitment can also be signal-dependent since TMcCCASP (with a functional DXE sequence), but not the non-functional mutant form TMcCCASPDXE1, induced the recruitment of Sec24. Recently, Sieben demonstrated in vivo by a FRET approach that the interaction between the K+ channel KAT1 protein and Sec24 required a di-acidic ER export motif. In addition, the work conducted by Schoberer has suggested that single basic amino acids of the cytoplasmic tail of N-glycan processing enzymes may interact with the COPII machinery for their ER export. It is therefore possible to envisage an interaction between SYP31 and Sec24 for its export out of the ER to reach the Golgi. A very close EXXD motif [(M)ELAD(L)] has recently been found in the SREBP escort protein Scap, and to be required for interaction with Sec23/Sec24 in animal cells (Sun ). A specific interaction of Sed5 (yeast homologue of SYP31) with Sec24 has also been demonstrated in yeast (Peng ), but whether this interaction was involved in the transport of Sed5 itself to the Golgi, in the selection of other cargo proteins, or in the targeting of COPII vesicles to the Golgi was not clear. The latter assumption is highly speculative since Cao reported that the docking of ER-derived vesicles in yeast was independent of SNARE proteins, and it has been reported that Sed5 is not required on the donor vesicle for Golgi fusion (Miller ).

At least three independent cargo-binding sites have been demonstrated for the yeast Sec24 (Mossessova ; Miller ). It has been shown that the ‘A-site’ and the ‘B-site’ of Sec24 can interact with the Sed5 YNNSNPF and LXXLE motifs, respectively. The ‘B-site’ of Sec24 can also interact with the DXE motif (Mossessova ). By comparing the sequences of Sed5 and SYP31, the sequences SSNPF (aa 192–196), and LPPL (aa 204–207)/MEMSLL (aa 231-236) in SYP31 are suggested as candidate motifs for interaction with the ‘A-site’ and the ‘B-site’ of Sec24, respectively. By analogy with yeast, it is hypothesized that the ‘B-site’ of the plant Sec24 may interact with both the LPPL (aa 204–207)/MEMSLL (aa 231–236) sequences and the EXXD di-acidic motif of SYP31. It is also envisaged that the ‘B-site’ of Sec24 would have access to the di-acidic EXXD motif of SYP31 in its ER conformation, and that the ‘B-site’ of Sec24 would have access to the LPPL (aa 204–207)/MEMSLL (aa 231–236) sequences of SYP31 in its Golgi conformation (SNARE complex). As a consequence, a defined site in Sec24 may compete for different target motifs in SYP31 according to the subcellular location, the function, and environment of the target protein concerned.

Supplementary data

Supplementary data are available at JXB online.

A. Primers used for cloning of the different mutant forms of SYP31. B. PCR fragments used for cloning each mutant of SYP31.

The Golgi localization of SYP31-YFP through its co-location with Erd2-GFP.

The Golgi localization of a SYP31-YFP mutant protein deleted of its SNARE domain, which excludes any role of this domain in Golgi targeting.

The sequences of the different domains of SYP31 from the N-terminus up to the beginning of the SNARE domain.

The co-expression of Erd2-GFP with various mutant constructs of SYP31-YFP expressed in tobacco leaf epidermal cells.

The localization, after 24 h and 48 h of expression in tobacco leaf epidermal cells, of SYP31 in the Golgi bodies and the GAGAG mutant version in the ER.

The co-expression of Erd2-YFP with various mutant constructs of the type II Golgi matrix protein CASP expressed in tobacco leaf epidermal cells.

The co-expression of the various mutant constructs of the type I version of AtCASP (TMcCCASP-GFP) in tobacco leaf epidermal cells.

The time-course (from 24 to 72 h) of the expression in tobacco leaf epidermal cells of CASPmeladg and CASPgagagg.

The time-course (from 24–72 h) of the expression in tobacco leaf epidermal cells of TMcCCASPmeladg and TMcCCASPgagagg.