Functional anatomy of the Arabidopsis cytokinesis-specific syntaxin KNOLLE

Authors

  • Sonja Touihri,

    1. Zentrum für Molekularbiologie der Pflanzen, Entwicklungsgenetik, University of Tübingen, Auf der Morgenstelle 3, 72076 Tübingen, Germany
    Search for more papers by this author
  • Christian Knöll,

    1. Zentrum für Molekularbiologie der Pflanzen, Entwicklungsgenetik, University of Tübingen, Auf der Morgenstelle 3, 72076 Tübingen, Germany
    Search for more papers by this author
  • York-Dieter Stierhof,

    1. Zentrum für Molekularbiologie der Pflanzen, Mikroskopie, University of Tübingen, Auf der Morgenstelle 5, 72076 Tübingen, Germany
    Search for more papers by this author
  • Isabel Müller,

    1. Zentrum für Molekularbiologie der Pflanzen, Entwicklungsgenetik, University of Tübingen, Auf der Morgenstelle 3, 72076 Tübingen, Germany
    Search for more papers by this author
    • Present address: Regierung von Oberbayern, Technischer Umweltschutz – Gentechnik, Maximilianstraße 39, 80538 München, Germany.

  • Ulrike Mayer,

    1. Zentrum für Molekularbiologie der Pflanzen, Entwicklungsgenetik, University of Tübingen, Auf der Morgenstelle 3, 72076 Tübingen, Germany
    2. Zentrum für Molekularbiologie der Pflanzen, Mikroskopie, University of Tübingen, Auf der Morgenstelle 5, 72076 Tübingen, Germany
    Search for more papers by this author
  • Gerd Jürgens

    Corresponding author
    1. Zentrum für Molekularbiologie der Pflanzen, Entwicklungsgenetik, University of Tübingen, Auf der Morgenstelle 3, 72076 Tübingen, Germany
    Search for more papers by this author

(fax +49 7071 295797; e-mail gerd.juergens@zmbp.uni-tuebingen.de).

Summary

In plant cytokinesis, Golgi/trans-Golgi network-derived vesicles are targeted to the plane of cell division where they fuse with one another to form the partitioning membrane (cell plate). This membrane fusion requires a specialised syntaxin (Qa-SNARE), named KNOLLE in Arabidopsis. KNOLLE is only made during the M-phase of the cell cycle, targeted to the plane of cell division and degraded in the vacuole at the end of cytokinesis. To identify the parts of KNOLLE required for proper targeting and function in membrane fusion, we generated chimeric syntaxins comprising complementary fragments from KNOLLE and MVB-localized PEP12 (SYP21). Surprisingly, targeting of the chimeric protein was not specified by the C-terminal membrane anchor. Rather the N-terminal region including helix Ha and the adjacent linker to helix Hb appeared to played a critical role. However, deletion of this N-terminal fragment from KNOLLE (KNΔ1–82) had the same effect as its presence in the chimeric protein (KN1–82-PEP1264–279), suggesting that targeting to the plane of cell division occurs by default, i.e. when no sorting signal would target the syntaxin to a specific endomembrane compartment. Once the full-length syntaxin accumulated at the plane of division, phenotypic rescue of the knolle mutant only required the SNARE domain plus the adjacent linker connecting helix Hc to the SNARE domain from KNOLLE. Our results suggest that targeting of syntaxin to the plane of cell division occurs without active sorting, whereas syntaxin-mediated membrane fusion requires sequence-specific features.

Introduction

Syntaxins (Qa-SNARE proteins) mediate membrane fusion by forming trans-SNARE complexes through interaction with an R-SNARE protein (vesicle-associated membrane protein, VAMP) on the transport vesicle and other Q-SNARE protein(s) on the target membrane (Jahn and Scheller, 2006; Lipka et al., 2007). In Arabidopsis there is a large family of syntaxins (SYP, syntaxin of plants), which has been subdivided into five major groups including SYP1 and SYP2 (Sanderfoot et al., 2000; Lipka et al., 2007). The SYP1 group comprises plasma membrane-localised syntaxins as well as the cytokinesis-specific syntaxin KNOLLE (SYP111). Within the SYP1 group, KNOLLE represents a distinct member that has only been identified in flowering plant species including both dicots and monocots (Figure S1 in Supporting Information). KNOLLE is exclusively expressed during the M-phase and accumulates at the plane of cell division. KNOLLE is required for homotypic fusion of membrane vesicles that generates the partitioning membrane during cytokinesis, as knolle mutants accumulate a band of unfused vesicles in the plane of cell division (Lukowitz et al., 1996; Lauber et al., 1997). Finally, KNOLLE protein is trafficked to the vacuole for degradation at the end of cytokinesis. In contrast, SYP2 syntaxins localise to the pre-vacuolar compartment (PVC)/multivesicular bodies (MVBs)/late endosomes (PEP12/SYP21) (da Silva Conceição et al., 1997) and/or the vacuolar membrane (VAM3/SYP22), and appear to be functionally redundant in vacuolar trafficking (Uemura et al., 2010). Thus, the two groups of syntaxins are trafficked to different subcellular compartments, which would explain the inability of PEP12 to substitute for KNOLLE in cytokinesis even when expressed from the KNOLLE promoter (Müller et al., 2003).

Syntaxins display a conserved domain organisation (Jahn and Scheller, 2006; Lipka et al., 2007). The very N-terminus comprises an unfolded stretch of up to 40 amino-acid residues followed by three helical domains, Ha, Hb and Hc, which are separated from one another by short linkers and perform regulatory functions (Margittai et al., 2003; Van Komen et al., 2006). A prominent linker separates helix Hc from the adjacent SNARE domain that interacts with cognate SNARE domains from other proteins to form the four-helical bundle of SNARE complexes (Antonin et al., 2002; Bracher and Weissenhorn, 2004). Approximately 20 hydrophobic amino-acid residues adjacent to the SNARE domain constitute the C-terminal membrane anchor (tail anchor) by which syntaxins, like other tail-anchored proteins, are inserted into the endoplasmic reticulum (ER) membrane via the GET pathway (Schuldiner et al., 2008; Borgese and Fasana, 2011).

To identify the domains required for proper targeting and function of KNOLLE, we generated transgenic plants expressing chimeric syntaxins in which fragments of KNOLLE were replaced by homologous fragments of PEP12. Our results indicate that a specific region in the N-terminal half of syntaxins is critical for targeting, whereas the C-terminal membrane anchor does not confer specificity of targeting. Furthermore, both the SNARE domain and the adjacent linker connecting helix Hc to the SNARE domain are essential for proper functioning of KNOLLE.

Results

Targeting of syntaxin to the plane of cell division

The cytokinesis-specific syntaxin KNOLLE is only expressed in dividing cells, accumulating at the plane of cell division, whereas PEP12 (SYP21) is located at the PVC/MVBs in both dividing and non-dividing cells even when expressed from the KNOLLE promoter (Müller et al., 2003; Figure 1, endogenous KNOLLE (#1), transgenic PEP12 (#2)). To identify protein regions that might serve as targeting signals, we swapped homologous segments between the two syntaxins and analysed the subcellular localisation of the chimeric proteins as well as their ability to rescue the knolle-X37 mutant (Lukowitz et al., 1996). All chimeras were Myc-tagged and their expression levels were analysed (Figure S2). Their subcellular protein localisation was detected with anti-Myc antibody, whereas the endogenous KNOLLE protein was detected with specific antiserum. However, the KNOLLE antiserum also detected chimeric proteins to various degrees. The C-terminal membrane anchor (tail anchor) has been reported to mediate the subcellular localisation of syntaxin 1/HPC (Suga et al., 2003). Surprisingly, KNOLLE with the tail anchor of PEP12 in place of its own was not targeted to the PVC/MVBs but instead co-localised with endogenous KNOLLE at the cell plate and moreover rescued the knolle mutant (Figure 1, transgene #3; Table 1). This result indicates that the tail anchor does not confer specificity of localisation or function.

Figure 1.

 The N-terminal half of KNOLLE and PEP12 mediates correct targeting.
(a) Schematic representation of predicted domain organisation of syntaxins KNOLLE and PEP12 (top) and their chimeras comprising complementary fragments of KNOLLE (grey) and PEP12 (blue). Relevant amino-acid positions for KNOLLE and PEP12 are indicated above and below the schematics, respectively. Transgenes (numbered on the left) encode N-terminally Myc-tagged proteins and are expressed from the KNOLLE promoter. For each transgene, the subcellular protein localisation and its ability to rescue knolle are indicated on the right. Ha, Hb, Hc, helices; SNARE, SNARE domain; TA, tail anchor. CP, cell plate; PVC, pre-vacuolar compartment.
(b) Immunofluorescence images of the subcellular localisation of chimeric proteins (Myc, red) in mitotic and non-mitotic cells of seedling roots. Numbers on the left indicate transgenes depicted in (a). Endogenous KNOLLE (KN, green) and DNA (blue, overlay) are also shown. Scale bars, 5 μm.

Table 1.   Rescue analysis of knolle mutant seedlings in transgenic lines
Construct numberT2 seedlingsa 
Nb (n lines)PPT-res.c (%)kn/kn (%)d PCR genotype
  1. aFrom kn-X37 heterozygous lines expressing the indicated construct (#; see Figures).

  2. bN, total number of seedlings analysed; n lines, number of independent transgenic kn-X37 heterozygous lines.

  3. cPercentage of seedlings displaying phosphinotricine (PPT; 15 mg l−1) resistance associated with the transgene; approximately 75% indicates the presence of one copy.

  4. dkn/kn (%), percentage of kn-X37 homozygous mutant seedlings; PCR genotype of pooled kn seedlings, presence or absence of transgene indicating no rescue or rescue, respectively.

31135 (7 lines)773Rescue
No transgene
11958 (5 lines)7319No rescue
Transgene
121227 (6 lines)7118No rescue
Transgene
14971 (4 lines)763Rescue
No transgene
15910 (4 lines)774Rescue
No transgene
161069 (5 lines)744Rescue
No transgene
171724 (8 lines)7317No rescue
Transgene
191156 (8 lines)734Rescue
No transgene
201203 (4 lines)754Rescue
No transgene
21941 (5 lines)7317No rescue
Transgene
221671 (5 lines)7219No rescue
Transgene
231521 (4 lines)7220No rescue
Transgene

The next step was to generate reciprocal half–half chimeras. Interestingly, localisation of these chimeras depended on the origin of the N-terminal half such that KNOLLE(N)-PEP12(C) located at the cell plate whereas the reciprocal chimera localised to the PVC or vacuole (Figure 1, transgenes #4 and #5). This result was confirmed in subsequent experiments in which the N-terminus and all three helices Ha, Hb and Hc, or only the first two helices, Ha and Hb, were from KNOLLE (Figure 2, transgenes #7, #9). Moreover, the reciprocal chimeras were targeted to the PVC or vacuole (Figure 2, transgenes #6 and #8), which was confirmed by ultrastructural localisation using immunogold labeling (Figure 3a,b). The chimeras that ended up at the vacuole were associated with the cytosolic face of the tonoplast membrane, which would explain their stability. These results reveal alternative locations of overlapping sets of reciprocal chimeras and thus support the notion that a targeting signal resides in the N-terminal 120 amino-acid residues of KNOLLE and/or the N-terminal 93 amino-acid residues of PEP12.

Figure 2.

 N-terminal 120 amino-acid residues of KNOLLE or PEP12 are sufficient for proper targeting.
(a) Schematic representation of predicted domain organisation of syntaxins KNOLLE and PEP12 (top) and their chimeras comprising complementary fragments of KNOLLE (grey) and PEP12 (blue). Relevant amino-acid positions for KNOLLE and PEP12 are indicated above and below the schematics, respectively. Transgenes (numbered on the left) encode N-terminally Myc-tagged proteins and are expressed from the KNOLLE promoter. For each transgene, the subcellular protein localisation and its ability to rescue knolle are indicated on the right. Ha, Hb, Hc, helices; SNARE, SNARE domain; TA, tail anchor. CP, cell plate; PVC, pre-vacuolar compartment.
(b) Immunofluorescence images of the subcellular localisation of chimeric proteins (Myc, red) in mitotic and non-mitotic cells of seedling roots. Numbers on the left indicate transgenes depicted in (a). Endogenous KNOLLE (KN, green) and DNA (blue, overlay) are also shown. Scale bars, 5 μm.

Figure 3.

 Ultrastructural localisation of immunogold-labelled chimeric syntaxins.
(a, b) Chimeric proteins with two (#6, a) or three (#8, b; see Figure 2a) N-terminal helices from PEP12 localised to the tonoplast.
(c–e) Chimeric protein with N-terminal helix Ha and linker to helix Hb from PEP12 (#10; see Figure 4a) was detected at the tonoplast (c), the Golgi stacks and trans-Golgi network (TGN) (d, e) but not at the cell plate (e).
(f, g) Chimeric protein with N-terminal helix Ha from PEP12 (#12; see Figure 4a) localised to Golgi and TGN (f, g) but not to the cell plate (g). CP, cell plate; CW, cell wall; G, Golgi stack; T, trans-Golgi network; V, vacuole. Scale bars: (a–c) 500 nm; (d–g) 250 nm.

The fungal toxin brefeldin A (BFA) inhibits sensitive ARF guanine–nucleotide exchange factors (ARF-GEFs) and thus blocks the corresponding trafficking pathways (Peyroche et al., 1999; Geldner et al., 2003). Treatment of Arabidopsis seedling roots with BFA resulted in prominent aggregates of endosomal membranes called BFA compartments (Geldner et al., 2001, 2003). Unlike KNOLLE, PEP12 did not accumulate in BFA compartments (Figure S3; Reichardt et al., 2007). It is interesting to note that the chimeras that behaved like KNOLLE also shared its response to BFA treatment and accumulated in BFA compartments, whereas the reciprocal class of chimeras behaved like PEP12 and did not localise to BFA compartments (Figure S3).

Trafficking to the plane of cell division by default

To identify a region critical for targeting to the plane of cell division, we exchanged shorter N-terminal fragments between KNOLLE and PEP12 (Figure 4). A chimera that contained only the N-terminal 82 amino-acid residues of KNOLLE, including the first helix Ha and the small adjacent linker to helix Hb, accumulated at the plane of cell division whereas the reciprocal chimera did not and rather stayed at the Golgi/trans-Golgi network (TGN) or was passed on to the PVC/vacuole (Figures 3c–e and 4, transgenes #10 and #11; Figure S4b). The subcellular localisation of chimera #10 at the Golgi/TGN was confirmed by double-labelling with the TGN marker ARF1 (Figure S5b). The ultrastructural analysis confirmed the localisation of chimera #10 to Golgi, TGN or vacuole (Figure 3c–e). In contrast, shortening of the KNOLLE-derived N-terminus by six amino-acid residues resulted in vacuolar trafficking of the chimeric protein whereas the reciprocal chimera accumulated in the Golgi/TGN, as also indicated by co-localisation with ARF1 in BFA compartments (Figure 4, transgenes #12, #13; Figures S4 and S5). Again, the ultrastructural analysis confirmed the localisation of the reciprocal chimera to Golgi or TGN (Figure 3f,g). These results suggested that the linker separating helices Ha and Hb might be part of the signal targeting to the plane of cell division.

Figure 4.

 The N-terminal region including the Ha–Hb linker is sufficient for proper targeting of KNOLLE.
(a) Schematic representation of predicted domain organisation of syntaxins KNOLLE and PEP12 (top) and their chimeras comprising complementary fragments of KNOLLE (grey) and PEP12 (blue). Relevant amino-acid positions for KNOLLE and PEP12 are indicated above and below the schematics, respectively. Transgenes (numbered on the left) encode N-terminally Myc-tagged proteins and are expressed from the KNOLLE promoter. For each transgene, the subcellular protein localisation and its ability to rescue knolle are indicated on the right. Ha, Hb, Hc, helices; SNARE, SNARE domain; TA, tail anchor. CP, cell plate; PVC, pre-vacuolar compartment; TGN, trans-Golgi network.
(b) Immunofluorescence images of the subcellular localisation of chimeric proteins (Myc, red) in mitotic and non-mitotic cells of seedling roots. Numbers on the left indicate transgenes depicted in (a). Endogenous KNOLLE (KN, green) and DNA (blue, overlay) are also shown. Scale bars, 5 μm.

To address the trafficking role of the N-terminal fragment up to the start of helix Hb, we swapped homologous subfragments between KNOLLE and PEP12 and we also deleted this N-terminal fragment containing the presumed targeting sequence (Figure 5). Interestingly, all chimeras derived from KNOLLE but containing either the very N-terminal 20 amino-acid residues from PEP12 or only the helix Ha or the helix Ha plus the linker to helix Hb from PEP12 were targeted to the plane of cell division, in contrast to chimeras #8 and #10 (Figure 5, transgenes #14–16). These results suggest that if there is a putative sorting signal of PEP12 that acts negatively on trafficking to the division plane it was incomplete in these chimeras. Alternatively, there might be redundant sorting signals for targeting to the cell plate, such that at least one remains intact in each chimera #14–16. Moreover, these chimeric proteins were also able to rescue the knolle mutant, suggesting that the N-terminal region does not contain KNOLLE-specific information (Table 1). The notion of a sorting signal that acts negatively on trafficking to the division plane implies that KNOLLE reaches the plane of cell division by default, i.e. in the absence of such a signal. To test this notion, we deleted the N-terminal region up to the start of helix Hb from the two syntaxins, KNOLLE and PEP12. Truncated KNOLLE accumulated at the plane of cell division and appeared to be more stable than full-length KNOLLE, staying in the plasma membrane of recently divided cells. In contrast, truncated PEP12 was detected in small patches near the cell plate (Figure 5, transgenes #17 and #18). Whereas truncated KNOLLE co-localised with full-length KNOLLE, truncated PEP12 did not do so and rather resembled the PVC localisation of PEP12 (Figure S6a). Additional staining signals might suggest that the truncated PEP12 protein was partially retained at earlier compartments along the pathway, possibly at the ER (Figure S6b). However, double labeling with the Golgi marker γCOP/SEC21 (Movafeghi et al., 1999) and the TGN marker ARF1 (Stierhof and El Kasmi, 2010) did not reveal co-localisation, in contrast to the co-localisation of ARF1 with truncated KNOLLE (Figure S6b). This surprisingly complex trafficking of truncated syntaxins and their chimeras with N-terminal fragments of mixed origin will be discussed below (see Discussion).

Figure 5.

 N-terminally truncated KNOLLE is targeted to the plane of cell division.
(a) Schematic representation of predicted domain organisation of syntaxins KNOLLE and PEP12 (top) and their chimeras comprising complementary fragments of KNOLLE (grey) and PEP12 (blue). Relevant amino-acid positions for KNOLLE and PEP12 are indicated above and below the schematics, respectively. Transgenes (numbered on the left) encode N-terminally Myc-tagged proteins and are expressed from the KNOLLE promoter. For each transgene, the subcellular protein localisation and its ability to rescue knolle are indicated on the right. Ha, Hb, Hc, helices; SNARE, SNARE domain; TA, tail anchor. CP, cell plate; ER, endoplasmic reticulum; PM, plasma membrane.
(b) Immunofluorescence images of the subcellular localisation of chimeric proteins (Myc, red) in mitotic and non-mitotic cells of seedling roots. Numbers on the left indicate transgenes depicted in (a). Endogenous KNOLLE (KN, green) and DNA (blue, overlay) are also shown. Scale bars, 5 μm.

KNOLLE regions necessary for syntaxin function in cytokinesis

The function of syntaxin in cytokinesis requires high-level expression during M-phase, targeting to the plane of cell division and specific activity during membrane fusion (Müller et al., 2003). One protein region known to play an important role is the SNARE domain, which interacts with cognate SNARE-domain proteins to form specific complexes that mediate membrane fusion (Sanderfoot et al., 2001; Uemura et al., 2004; Lipka et al., 2007). Indeed, KNOLLE syntaxin bearing the SNARE domain of PEP12 in place of its own was properly targeted to the plane of cell division but failed to rescue the knolle mutant (Figure 6, transgene #21; Table 1). To identify additional domains important for KNOLLE function, we replaced the helices Hb, Hc and the interjacent linker with the homologous region of PEP12, and we also generated a related chimera in which the swapped region included the linker between helix Ha and helix Hb as well (Figure 6a, transgenes #19 and #20). Surprisingly, these two chimeras not only localised to the plane of cell division but they were also able to rescue the knolle mutant, indicating that KNOLLE function does not depend on the sequence of the exchanged protein region (Figure 6b, transgenes #19 and #20; Table 1). In contrast, chimeric proteins that in addition had the linker between helix Hc and the SNARE domain provided by PEP12 localised properly but were unable to rescue the knolle mutant (Figure 6, transgenes #22 and #23; Table 1). Thus, both the SNARE domain and the adjacent linker connecting helix Hc to the SNARE domain are necessary for KNOLLE function in membrane fusion during cytokinesis.

Figure 6.

 Syntaxin regions requiring the KNOLLE sequence for function in cytokinesis.
(a) Schematic representation of predicted domain organisation of syntaxins KNOLLE and PEP12 (top) and their chimeras comprising complementary fragments of KNOLLE (grey) and PEP12 (blue). Relevant amino-acid positions for KNOLLE and PEP12 are indicated above and below the schematics, respectively. Transgenes (numbered on the left) encode N-terminally Myc-tagged proteins and are expressed from the KNOLLE promoter. For each transgene, the subcellular protein localisation and its ability to rescue knolle are indicated on the right. Ha, Hb, Hc, helices; SNARE, SNARE domain; TA, tail anchor. CP, cell plate; PM, plasma membrane.
(b) Immunofluorescence images of the subcellular localisation of chimeric proteins (Myc, red) in mitotic and non-mitotic cells of seedling roots. Numbers on the left indicate transgenes depicted in (a). Endogenous KNOLLE (KN, green) and DNA (blue, overlay) are also shown. Scale bars, 5 μm.

Discussion

Targeting to the plane of cell division

Many plasma membrane-localised proteins are known to accumulate at the plane of cell division during cytokinesis, suggesting that trafficking to the division plane is mechanistically related to trafficking to the plasma membrane in interphase (Reichardt et al., 2007). This includes cell wall-modifying enzymes such as endoxyloglucan transferase that are secreted from the cell in interphase and are targeted to the cell plate in dividing cells (Yokoyama and Nishitani, 2001). This has led to the notion that trafficking to the plasma membrane or the plane of cell division occurs by default (Jürgens and Pacher, 2003). There is good evidence for the idea of a default pathway to the plasma membrane, since a secretory form of soluble GFP, which is translocated across the ER membrane because of its N-terminal signal peptide, is secreted from the cell (Batoko et al., 2000). However, a fairly large number of plasma membrane-localised proteins are endocytosed during cytokinesis and then targeted from endosomes to the plane of cell division rather than being recycled to the plasma membrane (Reichardt et al., 2007). There are a few known exceptions to this rule, KNOLLE being one of them. KNOLLE is specifically synthesised during M-phase, targeted to the plane of cell division and degraded in the vacuole at the end of cytokinesis (Völker et al., 2001; Reichardt et al., 2007). Our analysis of KNOLLE-PEP12 chimeras revealed that the C-terminal membrane anchor (tail anchor) is not critical for sorting between the plane of cell division and the PVC. This is unlike the situation of proteins with a transmembrane domain, which traffic to the plasma membrane only if their transmembrane domain exceeds a certain length (Brandizzi et al., 2002). In contrast, KNOLLE-PEP12 chimeric proteins were targeted to the plane of cell division if the N-terminal region including helix Ha and the linker to helix Hb were from KNOLLE but trafficked to the PVC or vacuole if this critical region was from PEP12, regardless of the length of the swapped sequence and the origin of the remainder of the chimera. The simplest interpretation of these results would be that the sorting signal resides there and might promote vacuolar trafficking and/or prevent trafficking to the cell plate. Alternatively, KNOLLE might have multiple sorting signals in different regions of the protein but the vacuolar sorting signal of PEP12, which resides in the N-terminal region, is epistatic to them all. This rather unlikely interpretation is formally equivalent to the much simpler assumption that PEP12 has a positive sorting signal whereas KNOLLE has none. Experimentally, the two interpretations cannot be distinguished. For example, both interpretations would be compatible with the observed trafficking to the cell plate of KNOLLE proteins that have small N-terminal subfragments replaced with those of PEP12.

Deleting this critical region still resulted in the accumulation of KNOLLE (transgene #17) at the plane of cell division, whereas PEP12 (transgene #18) trafficking to the PVC/MVB appeared compromised. Although some truncated PEP12 was localised in a pattern akin to that of full-length PEP12, others were seemingly retained along the pathway, including possible retention in the ER. That some truncated PEP12 appears to localise like full-length PEP12 was unexpected, considering the observation that KNOLLE-PEP12 (transgene #5) reaches the plane of cell division, which suggests that there are no positive sorting signals to the vacuole in truncated PEP12 or KNOLLE sorting to the division plane is epistatic to those vacuolar signals. However, the reciprocal chimera PEP12-KNOLLE (transgene #4) takes the vacuolar route, suggesting the opposite epistatic relationship, if any. Another possible explanation for the vacuolar trafficking of some truncated PEP12 might be that such protein might tend to fold abnormally and thus be subject to vacuolar degradation.

The significance of this reduced trafficking efficiency of PEP12 is not clear at present. Conceivably, sorting of PEP12 might already occur at the exit from the ER whereas KNOLLE seems to ‘go with the flow’. Alternatively, KNOLLE and PEP12 might use different ER exit signals that are located in different regions of the two syntaxins but are both recognised by the COPII subunit SEC24. For example, a di-acidic peptide consisting of five amino acids (MELAD) residing in the linker between Hb and Hc appears to mediate export of Golgi-localised SYP31 from the ER (Chatre et al., 2009). In yeast, however, Sec24p has distinct binding sites for the SYP31 orthologue Sed5p and other SNAREs, recognising a different signal residing in the linker between Hc and the SNARE domain of Sed5p (Miller et al., 2003, 2005). Interestingly, chimeras with the N-terminal region of mixed KNOLLE–PEP12 origin were largely retained at the Golgi/TGN (if not trafficked to the plane of cell division), suggesting that sorting to the vacuolar pathway was incomplete or defective. All these observations support the notion that syntaxin trafficking to the plane of cell division occurs by default, i.e. if there is no signal that mediates sorting to some alternative endomembrane destination, as analysed here for PEP12. How default traffic to the plane of cell division is brought about mechanistically is not clear at present. Because both newly synthesised and endocytosed membrane proteins accumulate there, it seems reasonable to presume that the phragmoplast microtubules provide some guidance to the targeting process. Indeed, KNOLLE is dispersed in the dividing cell rather than accumulating at the plane of division when the formation of microtubules is blocked in mutant embryos lacking tubulin-folding cofactors (Mayer et al., 1999). However, no kinesin motor protein has been identified that would move the cytokinetic vesicles to the plane of cell division.

In general, not much is known about the targeting of syntaxins to their site of action in any system (Salaün et al., 2004). Interestingly, the plasma membrane-localised syntaxin 3 is prevented by its SNARE domain from entering the outer segment area of photoreceptor cells, which appears to be the default targeting area of the plasma membrane (Baker et al., 2008). In epithelial cells, syntaxin 3 is targeted to the apical domain of the plasma membrane, which appears to require a four amino-acid motif, FMDE, in helix Ha but not the SNARE domain (Sharma et al., 2006). Syntaxin 4 is targeted to the complementary baso-lateral area of the plasma membrane of epithelial cells, which appears to require a small motif N-terminal to helix Ha (Torres et al., 2011). Thus, different parts of syntaxin mediate sorting in positive or negative ways, and there seems to be no unifying principle.

Which parts of KNOLLE are required for action of syntaxin at the division plane?

When KNOLLE-related SYP1 syntaxins were tested for their ability to substitute for KNOLLE when expressed from the KNOLLE promoter, the most closely related syntaxin SYP112 rescued the knolle mutant whereas PEN1 (SYP121) was unable to do so (Müller et al., 2003). Here, we swapped domains between the SYP1 syntaxin KNOLLE and the SYP2 syntaxin PEP12 in order to identify relevant features for syntaxin function in cytokinesis. Interestingly, both the very N-terminus and the very C-terminus of KNOLLE could be exchanged with those of PEP12 without compromising the knolle-rescuing ability of the chimeric protein. This is interesting because other syntaxins have been shown to require a specific N-peptide motif for interacting with cognate SM protein (Südhof and Rothman, 2009). In contrast, deletion of the N-terminal region up to the start of helix Hb rendered the truncated KNOLLE protein unable to rescue the knolle mutant, although trafficking to the plane of cell division was not affected. Thus, the N-terminus might contain a sequence-unspecific element essential for syntaxin action at the plane of cell division. However, we cannot exclude the possibility that N-terminally truncated KNOLLE protein might be structurally altered in such a way that its activity is impaired without compromising its trafficking. Furthermore, a major part of the N-terminal helical region, including the linker preceding helix Hb, both helices Hb and Hc and the interjacent linker when derived from PEP12, did not interfere with syntaxin function of the chimeric protein in cytokinesis. In contrast, the SNARE domain and the linker preceding this domain had to be of KNOLLE origin for the chimeric syntaxin to rescue the knolle mutant. The sequence-specific requirement for the SNARE domain from KNOLLE can be easily related to its interaction with cognate partners to form SNARE complexes that mediate membrane fusion (McNew et al., 2000). Very recently, PEN1 (SYP121) was relieved of its inability to rescue the knolle mutant by replacing its SNARE domain with that of KNOLLE (Reichardt et al., 2011). In contrast, the requirement for the linker domain was unexpected and its role in syntaxin function during cytokinesis is at present unknown. It remains to be determined whether the sequence-specific requirement of the linker is unique to KNOLLE and what its functional significance might be.

Experimental Procedures

Secondary structure prediction software

For the prediction of the secondary structure the following software was used: Predict protein (http://www.Predictprotein.org), psipred (http://bioinf.cs.ucl.ac.uk/psipred); memsat3 (http://bioinf.cs.ucl.ac.uk/psipred) and swiss-model (http://swissmodel.expasy.org/). Additionally we also used previously published secondary structure predictions (Lukowitz et al., 1996).

Cloning strategy and plant transformation

The swapping constructs were made by overlap-extension PCR (Higuchi et al., 1988). All swapping constructs were based on the coding sequence of KNOLLE and PEP12.

For all PCR reactions, Pfu polymerase from Fermentas (http://www.fermentas.com/en/home) was used. The final PCR products were cloned into the MCS of a KNOLLE expression cassette (Müller et al., 2003) within the binary vector pGreen-II-BASTA (Hellens et al., 2000), using the respective restriction sites (XbaI/AvrII or XbaI/EcoRI). All constructs were N-terminally Myc-tagged. They were checked by restriction digest and confirmed by sequencing (GATC GmbH, http://www.gatc-biotech.com/). Sequence analysis was carried out with Vector NTI (alignment-program; Informax, http://www.invitrogen.com/site/us/en/home/LINNEA-Online-Guides/LINNEA-Communities/Vector-NTI-Community/vector-nti-software.html). Restriction enzymes were purchased from MBI Fermentas, synthetic oligonucleotides from ARK (Sigma, http://www.sigmaaldrich.com/). The floral-dip method was used for Agrobacterium tumefaciens-mediated plant transformation (Clough and Bent, 1998). Oligonucleotide primers used for PCR are listed in Table S1.

Plant selection and growth

Arabidopsis thaliana plants were grown as described previously (Mayer et al., 1991). All swapping constructs were introduced into plants heterozygous for the kn mutation X37 (Ler/Nd) (Lukowitz et  al., 1996). T1 plants were grown on soil and selected for transformants by spraying twice with a 1:1000 Basta solution (183 g l−1 glufosinate; Bayer CropScience, http://www.bayercropscience.com/) (Völker et al., 2001). Basta-resistant plants were tested by PCR for the presence of the transgene. The primers used were specific for the Myc-tag and the KN 3′ untranslated region (UTR). Seedlings used for microscopy were grown on 2.15 g l−1 Murashige and Skoog (MS) medium (½-MS medium) + 1% sucrose + 8 g l−1 agar with or without phosphinothricin (15 μg ml−1). The plates were vertically oriented in growth chambers at 23°C, under cycles of 16-h light and 8-h darkness for 4–5 days.

Drug treatment

Four- to 5-day-old seedlings were incubated with inhibitors at room temperature (22°C) for the indicated times. The incubation was done in 1 ml of liquid ½-MS medium containing 50 μm brefeldin A (BFA) [stock solution 50 mm BFA in DMSO:ethanol (1:1), Sigma-Aldrich]. Drug treatment was stopped by fixation with 4% paraformaldehyde in microtubule-stabilising buffer (MTSB) [50 mm piperazine-1,4-bis(2-ethanesulfonic acid) (PIPES), 5 mm EGTA, 5 mm MgSO4, pH 7.0, adjust pH with KOH].

Western blot and immunofluorescence

Preparation of protein extracts, western blot and whole-mount immunofluorescence were performed as described (Lauber et al., 1997). We used the following antibodies and dilutions: as primary antibodies, mouse anti-Myc monoclonal antibody 9E10 (1:1000 for western blot, 1:600 for immunofluorescence; Santa Cruz Biotechnology, http://www.scbt.com/), mouse anti-tubulin antibody (1:6000 for western blot, Sigma-Aldrich), rabbit anti-KNOLLE serum (1:3000 for western blot, 1:2000 for immunofluorescence; Lauber et al., 1997), rabbit anti-ARF1 serum (1:1000; Pimpl et al., 2000), rabbit anti-Sec21/γCop serum (1:1000; Movafeghi et al., 1999); as secondary antibodies, fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit (1:600; Dianova, http://www.dianova.com/), Alexa488-conjugated goat anti-rabbit (1:600; Invitrogen, http://www.invitrogen.com/) and Cy3-conjugated goat anti-mouse (1:600; Dianova). 4′,6-Diamidino-2-phenylindole (DAPI) staining was performed as described in Völker et al. (2001). The immunostained samples were analysed using a confocal laser scanning microscope (CLSM; Leica, http://www.leica.com/). All images were acquired with a 63× water-immersion objective and processed with Adobe Photoshop CS3 and Adobe Illustrator CS3 (http://www.adobe.com/).

Immuno-electron microscopy

Root tips were fixed with 4% formaldehyde in microtubule-stabilizing buffer (MTSB; 30–60 min), followed by treatment with 8% formaldehyde in MTSB for another 60–90 min. Fixed samples were embedded in 10% gelatin, infiltrated with a mixture of polyvinylpyrrolidone and sucrose (Tokuyasu, 1989), and were then frozen in liquid nitrogen. Thawed ultrathin cryosections were labelled with rabbit anti-GFP antibody (1:500; Torrey Pines Scientific, http://torreypinesscientific.com/) or mouse monoclonal anti-Myc antibody 9E10 (approximately 5 μg ml−1) diluted in blocking buffer. Bound primary antibodies were detected with goat anti-rabbit or anti-mouse IgG coupled to Nanogold (1:80 in blocking buffer; Nanoprobes, http://www.nanoprobes.com/). After silver enhancement (HQ silver, 8.5 min; Nanoprobes), labelled sections were stained with uranyl acetate and embedded in methyl cellulose.

Acknowledgements

We thank Ulrike Hiller and Dagmar Ripper for technical assistance, Karin Schumacher and Peter Pimpl for kindly providing experimental materials and Cornelia Krause, Sandra Richter, Hauke Beckmann, Misoon Park and Manoj Singh for critical reading of the manuscript. This work was supported by grant Ju 179/15-1 from the Deutsche Forschungsgemeinschaft to GJ.

Ancillary