Syntaxins and interacting SNARE proteins enable membrane fusion in diverse trafficking pathways. The Arabidopsis SYP1 family of plasma membrane-localized syntaxins comprises nine members, of which KNOLLE and PEN1 play specific roles in cytokinesis and innate immunity, respectively. To identify mechanisms conferring specificity of action, we examined one member of each subfamily—KNOLLE/SYP111, PEN1/SYP121 and SYP132—in regard to subcellular localization, dynamic behavior and complementation of knolle and pen1 mutants when expressed from the same promoters. Our results suggest that cytokinesis-specific syntaxin requires high-level accumulation during cell-plate formation, which necessitates de novo synthesis rather than endocytosis of pre-made protein from the plasma membrane. In contrast, syntaxin in innate immunity does not need upregulation of expression but instead requires pathogen-induced and endocytosis-dependent retargeting to the infection site. This feature of PEN1 is not afforded by SYP132. Additionally, PEN1 could not substitute for KNOLLE because of SNARE domain differences, as revealed by protein chimeras. In contrast, SYP132 was able to rescue knolle as did KNOLLE-SYP132 chimeras. Unlike KNOLLE and PEN1, which appear to have evolved to perform specialized functions, SYP132 stably localized at the plasma membrane and thus might play a role in constitutive membrane fusion.
SNARE proteins constitute a family of membrane-anchored proteins that play key roles in membrane fusion events of intracellular trafficking pathways by forming SNARE complexes that dock membranes to be fused. Their main characteristic feature is an evolutionarily conserved domain of 60–70 amino acids arranged in heptad repeats, which has been designated the SNARE domain (1). Based on the conserved amino-acid residue at the center of the SNARE domain, SNARE proteins have been classified into R- (arginine) and Q- (glutamine) SNAREs. The Q-SNARE family is further divided into four subfamilies (Qa-, Qb-, Qc- and Qb,c-SNAREs) based on differences in the structure of the SNARE domain (2). Each SNARE complex is formed by association of four interacting SNARE domains, one each from VAMP/R-SNARE on the donor membrane and three from Q-SNAREs on the acceptor membrane: one from syntaxin/Qa-SNARE and two from either SNAP25/Qb,c-SNARE or one each from two t-SNARE light chains/Qb- and Qc-SNAREs (1).
Syntaxins/Qa-SNAREs are conserved among all eukaryotic organisms, a fact that emphasizes a universal role for syntaxins in membrane trafficking (3). Compared to other organisms like yeast and mammals, which have two and four genes encoding plasma membrane-localized syntaxins, respectively, plant genomes harbor an increased number of genes for syntaxins involved in the late secretory pathway (4–6). The Arabidopsis genome encodes 18 putative syntaxins representing 5 different Syntaxin of Plant (SYP) families, of which the 9 members of the SYP1 family have been localized to the plasma membrane (5–7). Although, in principle, redundancy would explain the occurrence of the numerous SYP1 syntaxins in Arabidopsis, there is also evidence for functional diversification.
SYP1 syntaxins show different spatio-temporal expression profiles (8). For instance, SYP132 is expressed ubiquitously in all tissues throughout plant development, whereas SYP124, SYP125 and SYP131 are only expressed in pollen, and SYP123 appears to be exclusively expressed in root hair cells during root development (8). The SYP1 family also includes the functionally well-characterized syntaxins KNOLLE/SYP111 (9,10) and PEN1/SYP121/SYR1 (11–14). KNOLLE is a specialized SYP1 syntaxin of flowering plants that seems to be exclusively required for cytokinesis and has no ortholog in lower plants or non-plant organisms, suggesting that other SYP1 syntaxins played a comparable role in plant cytokinesis before KNOLLE evolved (6). KNOLLE gene expression is confined to late G2 and M phases of the cell cycle, which is mediated by mitosis-specific activator (MSA) promoter elements that bind R1R2R3-Myb transcription factors (9,15). In addition, KNOLLE protein only accumulates during mitosis, localizing to the forming cell plate that eventually separates the daughter cells, and is degraded immediately after completion of cytokinesis (10,16). Like KNOLLE, the plasma membrane-localized syntaxin PEN1 also arose late in plant evolution (6). PEN1 is involved in non-host penetration resistance against the powdery mildew Blumeria graminis f. sp. hordei (B. g. hordei) and mediates vesicle fusion at B. g. hordei—Arabidopsis interaction sites (12,13). PEN1/SYP121 also known as SYR1 was originally identified in tobacco for its involvement in potassium and chloride channel response to the plant hormone ABA in guard cells (11). SYR1/SYP121 has been shown to affect KAT1 potassium channel activity, KAT1 trafficking to the plasma membrane, and directly interacts with the channel subunit KC1 via an FxRF motif (14,17,18). PEN1/SYP121 has also been subjected to structure–function analysis. However, that study did not address the problem of syntaxin specificity (19). A third member of the SYP1 family, SYP132, is ubiquitously expressed in Arabidopsis(5) and has been related to the evolutionarily most ancient branch of SYP1 proteins (6). So far no function for SYP132 has been reported in Arabidopsis. The tobacco ortholog NbSYP132 contributes to resistance against bacterial pathogens, mediating secretion of pathogenesis-related protein 1 (20), whereas Medicago MtSYP132 has been localized to the plasma membrane surrounding Rhizobium infection threads and to the symbiosome membrane (21).
We have previously analyzed mechanisms of KNOLLE specificity in cytokinesis and identified the mitosis-specific expression of KNOLLE as a major determinant (22). The closest paralog of KNOLLE named SYP112 was functionally equivalent to KNOLLE when expressed from the KNOLLE promoter. In contrast, PEN1 was not able to substitute for KNOLLE in cytokinesis when expressed like KNOLLE. The results suggested functional divergence of the two SYP1 proteins but did not offer any mechanistic explanation. To analyze mechanisms that ensure the specific biological activity of KNOLLE and PEN1, we expressed these specialized SYP1 syntaxins and SYP132 from the same set of promoters in stably transformed Arabidopsis plants. The transgenically made proteins were analyzed for their ability to substitute for KNOLLE or PEN1 in the respective mutant as well as their subcellular localization and dynamics. We also tested chimeric proteins that were generated by exchanging the SNARE domain between KNOLLE and PEN1 or SYP132. We show that KNOLLE is completely functional during cytokinesis when carrying the SNARE motif of SYP132 but fails to fulfill proper cell-plate formation when carrying the SNARE motif of PEN1. Our results suggest that functional specificity of the specialized SYP1 syntaxins KNOLLE and PEN1 diverged from the presumably ancient SYP132. KNOLLE function appears to require high-level expression during the mitosis immediately preceding cytokinesis, whereas PEN1 function displays striking subcellular protein dynamics that seems to enable rapid retargeting to infection sites via endocytosis.
Unlike KNOLLE, PEN1 and SYP132 syntaxins are stable proteins
In dividing cells of the Arabidopsis seedling root, KNOLLE accumulates at the trans-Golgi network (TGN) in early mitosis, localizes to the cell plate during cytokinesis and takes the degradation route via multivesicular bodies (MVBs) to the lytic vacuole shortly after completion of the newly made plasma membrane (10,16,23). To analyze the subcellular localization and protein behavior of PEN1 and SYP132 in comparison to KNOLLE, we generated transgenic lines stably expressing Myc-tagged PEN1 (Myc-PEN1) or SYP132 (Myc-SYP132) under control of the KNOLLE cis-regulatory sequences (22). Colocalization analyses indicated that both Myc-PEN1 and Myc-SYP132 accumulated at KNOLLE-positive compartments: at the TGN in early mitosis and at the cell plate during cytokinesis, although the SYP132 signal was weaker than the PEN1 signal at the cell plate (Figure 1A–F; 22). After formation of the cell plate, neither Myc-PEN1 nor Myc-SYP132 colocalized at KNOLLE-labeled MVBs (Figure 1G–L). Both Myc-PEN1 and Myc-SYP132 proteins were more stable than endogenous KNOLLE as well as transgenically made Myc-KNOLLE and still detectable in interphase cells, when the KNOLLE promoter is not active (Figure 1N,O; compare with Figure 1M). Intriguingly, PEN1 accumulated much more strongly at the cell plate than at the plasma membrane during cytokinesis, whereas SYP132 accumulated evenly at both the cell plate and the plasma membrane (Figure S1). Thus, both PEN1 and SYP132 localize at the division plane in dividing cells and at the plasma membrane in interphase, indicating that they do not take the KNOLLE degradation pathway after cell-plate formation.
SYP132 but not PEN1 can substitute for KNOLLE function
Previously we have shown that PEN1 cannot substitute for KNOLLE function; however, the reason for this has not been clarified so far (22). To characterize this inability in more detail we performed a comparative analysis of the knolle complementation competence of PEN1, SYP132 and KNOLLE when expressed from the KNOLLE promoter. Progeny from five KN:Myc-KNOLLE, nine KN:Myc-PEN1 and five KN:Myc-SYP132 transgenic lines that were also knolle heterozygous were phenotypically analyzed. All five KN:Myc-KNOLLE transgenic lines completely rescued the knolle mutant phenotype. Two out of five KN:Myc-SYP132 transgenic lines rescued the knolle mutant phenotype completely, which was not the case for any of the nine KN:Myc-PEN1 transgenic lines. Examples of high-level and low-level expression lines for each transgene are shown in Figure 2 and Table S1. We observed phenotypic variation between different transgenic lines ranging from severe knolle seedlings (no rescue) to normal seedlings (complete rescue) (Figure 2A/1). Partially rescued seedlings initially developed like wild type but were later arrested in growth and eventually died (Figure 2A/2,3). To address why different Myc-PEN1 and Myc-SYP132 transgenic lines rescued knolle mutants to different degrees, we analyzed their expression levels (Figure 2B,C). Whereas PEN1 transgenic lines only rescued knolle partially at both low and high levels of expression, SYP132 transgenic lines rescued knolle partially or completely, depending on the level of expression (Figure 2A,B). Surprisingly, the protein levels of the low expression lines were still much higher than the Myc-KNOLLE protein level required for rescuing the knolle mutant completely (Figure 2B). As PEN1 and SYP132 are more stable than KNOLLE, the protein level we detected by immuno-blotting analysis does not represent their expression levels in mitotic cells. KNOLLE mRNA and protein are only present in dividing cells (9,10) and thus, the transcript level of each SYP1 transgene should represent the respective de novo synthesized protein amount in dividing cells. Therefore, we compared levels of transcript accumulation between the different SYP1 transgenes. Intriguingly, although all Myc-KNOLLE transgenic lines rescued the knolle mutant, some lines displayed a lower level of transcript accumulation than did Myc-PEN1 or Myc-SYP132 transgenic lines (Figure 2C). Thus, the rescue ability appeared not to depend on mere expression quantity but rather might reflect a qualitative difference between the SYP1 syntaxins in regard to their ability to promote cytokinesis.
SYP132 cannot replace PEN1 in pathogen penetration resistance
PEN1 was shown to play a role in non-host resistance to fungal pathogens by contributing to localized cell wall deposition (formation of papillae), which prevents barley powdery mildew Blumeria graminis f. sp. hordei (B. g. hordei) spores from invading Arabidopsis leaves (12). Although pen1 mutants display no obvious phenotype in the absence of pathogen attack, papilla formation upon attempted fungal ingress is delayed and penetration resistance significantly reduced (12,24). To analyze whether SYP1 syntaxins are able to substitute for PEN1 in penetration resistance, transgenic plant lines constitutively expressing SYP1 syntaxins from the UBIQUITIN 10 (UBQ10) promoter (25) were generated and these plants were crossed with pen1-1 mutant plants (12). pen1-1 mutant plants expressing UBQ10:RFP-PEN1 or UBQ10:RFP-SYP132 were inoculated with powdery mildew B. g. hordei. B. g. hordei–Arabidopsis interaction sites were visualized by established coomassie blue and callose staining protocols that allow quantification of fungal invasion rates (Figure 3C; 12). Only about 20% of the fungal penetration attempts were successful in wild-type control plants, whereas the success rate increased to about 90% in the pen1-1 mutant (Figure 3C). This was also reflected in significantly different levels of fluorescent epidermal cells showing hypersensitive-like cell death response (Figure 3D,E) because of activation of post-invasion defense mechanisms (26). UBQ10:RFP-PEN1 fully restored wild-type invasion resistance, whereas UBQ10:RFP-SYP132 complemented pen1-1 only partially, allowing an intermediate 60% of fungal sporelings to invade epidermal leaf cells (Figure 3C). We did not recover any UBQ10:RFP-KNOLLE transgenic plants that expressed KNOLLE at detectable levels. To analyze how this functional difference between the two SYP1 syntaxins PEN1 and SYP132 in pathogen resistance comes about, we investigated their behavior at the subcellular level.
Quantitative live-cell imaging revealed that RFP-PEN1 expressed from the UBQ10 promoter strongly accumulated at the Arabidopsis–B. g. hordei interaction sites 16 h after inoculation (Figure 4A–C). At sites of attempted fungal penetration, RFP-PEN1 signal intensity was about 4.5 times higher than at other plasma membrane regions of the same cell (Figures 4C and S2), indicative of the recently described active translocation to and accumulation in a pathogen-induced plasma-membrane microdomain (24,27). In contrast, RFP-SYP132 expressed from the UBQ10 promoter showed only about 1.4 times higher signal intensity at plant–pathogen interaction sites than elsewhere at the same plasma membrane (Figures 4D–F and S2). In summary, the poor ability of UBQ10:RFP-SYP132 to prevent pathogen entry correlated well with its comparatively inefficient recruitment to fungal invasion sites rather than with any conceivable activity differences to PEN1 at the penetration site.
PEN1 but not SYP132 constitutively cycles between the plasma membrane and endosomes
To address why PEN1 and SYP132 differed in their ability to substitute for knocked-out SYP1 family members, we analyzed their subcellular localization and dynamics in seedling root cells. Both Myc-PEN1 and Myc-SYP132 expressed from the KNOLLE promoter localize at the plasma membrane and at the cell plate. Additionally, Myc-PEN1 stained some endomembrane compartments in interphase cells, unlike SYP132 (Figure S3). To identify these compartments we did double-labeling experiments with specific subcellular markers (Figure 5A–H). Myc-PEN1-positive puncta were distinct from the Golgi stacks that were labeled by the γCOP subunit of the coat protein I (COPI) complex mediating retrograde transport from the cis-Golgi to the endoplasmic reticulum (ER) (Figure 5A; 28). Additionally, there was also no colocalization with the trans-Golgi labeled by the yellow fluorescent protein (YFP)-tagged rat sialyl transferase, N-ST-YFP (Figure 5E; 29). In contrast, the vesicle formation-initiating GTPase ARF1 that localizes to the Golgi/TGN/early endosome (16,23) labeled some Myc-PEN1-positive punctate structures (Figure 5B–D). ARF1 was localized mainly to the TGN and also to the Golgi stacks by immunogold labeling (23). Thus, we would expect some ARF1-positive compartments not to be labeled by endocytosed RFP-PEN1. In addition, live-cell imaging revealed complete colocalization of GFP-PEN1-positive puncta with the endocytic tracer FM4-64 within 10 min of incubation (Figure 5F–H; 12). Thus, in interphase cells, PEN1 localizes to the plasma membrane and to the TGN, which functionally corresponds to early endosomes in plants (30,31).
The fungal toxin brefeldin A (BFA) reversibly inhibits vesicle trafficking by blocking the activity of sensitive ARF guanine-nucleotide exchange factors (ARF-GEFs) (32,33). BFA treatment of Arabidopsis seedling roots traps cycling plasma-membrane proteins in endosomal BFA compartments by inhibiting ARF-GEFs required for recycling (31,32). In contrast, secretory traffic from the ER to the plasma membrane is not inhibited by BFA in Arabidopsis so that newly synthesized proteins are not trapped in BFA compartments (16,31,34). When KN:Myc-SYP1 transgenic root tips were treated with BFA, all three transgenically made SYP1 proteins accumulated in large BFA compartments in mitotic cells (Figure S4A–C). In non-dividing cells, however, Myc-PEN1 localized to BFA compartments, whereas Myc-SYP132 still labeled the plasma membrane (Figure S4B,C) and KNOLLE protein was not detected because of its specific degradation shortly after completion of cytokinesis (Figure S4A). Because the transgenically made SYP1 proteins were expressed from the KNOLLE promoter, and therefore not synthesized during interphase, only labeled SYP1 protein that was endocytosed from the plasma membrane accumulated in BFA compartments. These results indicate that PEN1, but not SYP132, cycles constitutively between the plasma membrane and endosomal compartment(s) in interphase cells.
In mitotic cells, both PEN1 and SYP132 could be detected in BFA compartments. To determine whether PEN1 and SYP132 are newly synthesized or endocytosed before accumulating in BFA compartments during mitosis, we blocked the secretory pathway by BFA treatment of gnl1 mutant seedlings expressing KN:Myc-PEN1 and KN:Myc-SYP132 transgenes. GNL1 is a BFA-resistant ARF-GEF that mediates retrograde transport from the Golgi stacks to the ER (34,35). Treating gnl1 mutant seedlings with BFA leads to the inhibition of ER–Golgi traffic, as shown by the ER accumulation of KNOLLE and the Golgi marker N-ST-YFP as well as the fusion of Golgi stacks with the ER (16,35). Without BFA treatment, both Myc-PEN1 and Myc-SYP132 accumulated at the cell plate in mitotic cells of gnl1 mutant seedlings (Figure 5I,M). After BFA treatment, however, the newly synthesized Myc-SYP1 syntaxins were trapped in the ER and, thus, did not reach the plane of cell division (Figure 5J–L,N–P, open arrowheads). Intriguingly, Myc-PEN1 also accumulated in BFA compartments, whereas Myc-SYP132 did not but rather labeled the plasma membrane (Figure 5K; closed arrowhead, compare with Figure 5O). This indicates that PEN1 is both secreted and endocytosed during mitosis, whereas SYP132 is not endocytosed and only the de novo synthesized protein is transported through the secretory pathway to the cell plate. It should be noted that genomic GFP fusions of SYP132 also label the cell plate in dividing cells, in addition to labeling the plasma membrane in all cells (8). Considering our results, this observation suggests that the endogenous promoter of SYP132 is also active during M phase such that newly synthesized SYP132 is targeted to the cell plate.
Although BFA is commonly believed to be a highly specific inhibitor of membrane trafficking with known molecular targets, there is always a remote possibility of non-specific side effects (36). To examine in a different way whether SYP132 is indeed not endocytosed during cytokinesis, we expressed SYP132 exclusively during interphase and investigated its ability to reach the cell plate. PEN1 was used as a control and both SYP1 proteins were expressed from the cis-regulatory elements of the HISTONE 4 (H4) gene because H4 mRNA appears in S-phase and is completely degraded before the onset of mitosis (37,38). Both RFP-PEN1 and RFP-SYP132 were present at the plasma membrane in interphase cells as they were when expressed from the KNOLLE promoter (Figure S4D–F,H–J, compare with Fig. 1H,I). Additionally, RFP-PEN1 but not RFP-SYP132 colocalized with ARF1-labeled endosomes and accumulated in BFA compartments upon BFA treatment (Figure S4G,K). In mitotic cells, we observed RFP-PEN1 fluorescence at the division plane, colocalizing with KNOLLE at the cell plate (Figure 5Q–S). In contrast, RFP-SYP132 did not label the developing cell plate at all, indicating that the protein is not internalized during mitosis (Figure 5T–V). Taken together, our results demonstrate that SYP1 syntaxins behave differently at the subcellular level: PEN1 cycles constitutively between the plasma membrane and endosomes, whereas SYP132, once delivered, remains statically at the plasma membrane. This difference in localization dynamics might also explain two observations reported above. First, PEN1 but not SYP132 accumulates more strongly at the cell plate than at the plasma membrane in dividing cells (Figure S1). In the case of cycling PEN1, both de novo synthesized and endocytosed proteins are targeted to the cell plate, whereas in the case of SYP132, only de novo synthesized protein accumulates at the cell plate. Second, SYP132 when expressed from the UBQ10 promoter was not able to completely rescue the pen1 mutant during pathogen attack because only newly synthesized protein was targeted to the infection site. To perform a more rigorous test of this idea, we crossed KN:Myc-SYP1 transgenic plants with pen1-1 mutant plants and analyzed their homozygous pen1-1 mutant progeny that also expressed the transgene. In accordance with the exclusive activity of the KNOLLE promoter in proliferating tissues (10), all transgenically made SYP1 proteins were detected in flowers (Figure 3A). However, Myc-PEN1 and Myc-SYP132 also accumulated in leaves, again indicating that these proteins were more stable than KNOLLE (Figure 3B). We tested their ability to restrict pathogen entry in pen1-1 mutants (Figure 3C–H; 12). Myc-PEN1 completely rescued pen1-1, reducing fungal entry rates to the wild-type level (Figure 3C,F). In contrast, Myc-SYP132 entirely failed to rescue pen1-1 (Figure 3C,G). In addition, Myc-SYP132 expressed in the wild-type background did not interfere with endogenous PEN1 function (Figure 3C,H). Thus, when expressed from the KNOLLE promoter, SYP132 cannot take over PEN1 function in pathogen resistance. Thus, the ability of syntaxin to rescue pen1 requires dynamic localization behavior. However, this cannot explain the ability of SYP132 to rescue knolle in cytokinesis. As SYP132 but not PEN1 can substitute for KNOLLE function (Figure 2), differences in amino-acid sequence rather than protein abundance at the cell plate account for their different rescuing ability.
Significance of the SNARE domain for KNOLLE function
Although PEN1 localized at the cell plate, it did not substitute for KNOLLE during cell-plate formation. A possible reason for its failure might be an insufficient interaction with other SNARE partners of the cytokinesis SNARE complex. The SNARE domain was identified as the main determinant of specificity between interacting SNARE proteins (39–43). Although the SNARE domain of SYP1 syntaxins is very conserved by amino-acid sequence, we cannot rule out that subtle differences account for the efficiency of interaction. To test if the SNARE domain contributes to functional specificity of SYP1 syntaxins, we generated chimeric proteins by swapping the SNARE domains between KNOLLE and PEN1 or SYP132, and expressed these chimeric proteins from the KNOLLE promoter (Figure 6A). Both KNOLLE carrying the SNARE domain of PEN1 (KNOLLE-PEN1SND) and PEN1 carrying the SNARE domain of KNOLLE (PEN1-KNOLLESND) localized at the cell plate and the plasma membrane in dividing cells (Figure 6B,C). In interphase cells both chimeric proteins were detected at the plasma membrane, indicating that they were stable and not degraded after cytokinesis. The same was observed for the KNOLLE-SYP132 chimeras as both ‘KNOLLE-SYP132SND’ and ‘SYP132-KNOLLESND’ localized at the cell plate as well as the plasma membrane in dividing cells and at the plasma membrane in non-dividing cells, further demonstrating high protein stability (Figure 6D,E). Thus, replacing the SNARE domain of KNOLLE with that of PEN1 or SYP132 leads to increased protein stability, suggesting that the SNARE domain of KNOLLE in conjunction with the remainder of the KNOLLE protein promotes its rapid turnover. All chimeric constructs were tested for their ability to rescue the knolle mutant. Remarkably, all proteins harboring the SNARE domain of KNOLLE were able to completely rescue the knolle mutant phenotype (Table S1). In addition, chimeric KNOLLE protein harboring the SNARE domain of SYP132 was also able to rescue the knolle mutant phenotype, consistent with the rescue ability of SYP132 (Table S1). On the contrary, chimeric KNOLLE protein carrying the SNARE domain of PEN1 did not rescue the knolle phenotype, and thus behaved like the wild-type PEN1 protein (Table S1). These results suggest that the SNARE domain contributes to syntaxin function in cytokinesis.
knolle mutant rescue by PEN1-KNSND is limited to expression during mitosis
Replacing the SNARE domain of PEN1 with that of KNOLLE rendered PEN1 competent to rescue the knolle mutant when expressed during mitosis. Considering that PEN1 is endocytosed and reaches the plane of cell division during cytokinesis, we addressed the possibility that chimeric PEN1-KNOLLESND protein can rescue the knolle mutant if expressed before mitosis. To this end we generated plants expressing RFP-PEN1-KNOLLESNDor the reciprocal construct RFP-KNOLLE-PEN1SND under the control of the S-phase-specific HISTONE 4 promoter. The chimeric proteins accumulated at the plasma membrane in non-dividing cells and at the cell plate during mitosis (Figure 6F,G). H4:RFP-KNOLLE-PEN1SND was not able to rescue the knolle mutant phenotype as might have been expected because KN:RFP-KNOLLE-PEN1SND also did not rescue knolle. Surprisingly, H4:RFP-PEN1-KNOLLESND also did not rescue the knolle mutant phenotype in contrast to KN:RFP-PEN1-KNOLLESND (Table S1), although expression levels of the RFP-PEN1-KNOLLESND protein were comparable for either promoter (Figure S5). These results clearly demonstrate that KNOLLE function depends on (i) high-level de novo protein synthesis during mitosis and (ii) its specific SNARE domain.
Our study addressed functional divergence versus redundancy among members of the SYP1 syntaxin family, focusing on one representative from each of the three subgroups, KNOLLE/SYP111, PEN1/SYP121 and SYP132. KNOLLE and PEN1 have their specific biological roles in cytokinesis and non-host pathogen defense, respectively (9,12). To identify mechanisms defining specificity of protein function, we expressed SYP1 proteins from a specific set of promoters, which eliminated differences in gene expression conferred by the endogenous promoters (8), and we swapped protein domains.
KNOLLE plays a unique role in somatic cytokinesis, which cannot easily be substituted for by other syntaxins (22). Its closest paralog SYP112 is essentially functionally equivalent, including rapid degradation at the end of cytokinesis, but lacks the strong expression of KNOLLE during mitosis preceding cytokinesis (22). The same study revealed the inability of PEN1 to substitute for KNOLLE when expressed from the KNOLLE cis-regulatory sequences but did not identify a plausible molecular mechanism for this failure. As PEN1 when expressed like KNOLLE accumulated at the cell plate there seemed to be some functional difference between the two proteins, although the levels of protein accumulation had not been compared. Our present study revealed that PEN1 was indeed expressed during M phase at least as strongly as KNOLLE but failed to rescue the knolle mutant. By contrast, SYP132 when expressed like KNOLLE rescued the knolle mutant completely, indicating a clear functional difference between PEN1 on one hand and KNOLLE and SYP132 on the other. SNARE domain swaps between KNOLLE and PEN1 or SYP132 yielded chimeric proteins of which PEN1 protein with the SNARE domain of KNOLLE was able to rescue the knolle mutant completely. Thus, the SNARE domain appears to be a critical determinant of KNOLLE protein functional specificity. This is a surprising result, considering the earlier observation that both KNOLLE and PEN1 interact with the same Qb,c-SNARE SNAP33 (13,44). Although the same interacting Qb,c-SNARE is involved in cytokinesis and in non-host pathogen defense, the R-SNARE might be different between the two SNARE complexes. Alternatively, the rate of assembly or disassembly of the two SNARE complexes might be different.
KNOLLE expression outside mitosis appears to be detrimental, although 35S promoter-driven expression resulted in KNOLLE accumulation near the apical plasma membrane in growing root hairs (45). However, using the S-phase-specific H4 promoter, KNOLLE accumulation was severely impaired, whereas both PEN1 and SYP132 were expressed and accumulated at the plasma membrane. As PEN1-KNOLLESND protein was able to rescue the knolle mutant completely when expressed from the KNOLLE promoter, i.e. during M phase immediately preceding cytokinesis, and this chimeric protein was stable and not deleterious, we expressed the same protein from the S-phase-specific H4 promoter. Surprisingly, early expression of the PEN1-KNOLLESND followed by its transient storage at the plasma membrane and subsequent endocytosis during cytokinesis enabled its accumulation in the plane of cell division but did not rescue the knolle mutant, although the protein level was comparable to that of the same protein made from the KNOLLE promoter. This strongly suggests that the cytokinesis-specific syntaxin needs to be synthesized immediately before cell-plate formation, whereas the same protein when endocytosed is ineffective. This is in line with the observation that BFA-induced inhibition of ER–Golgi traffic in gnl1 mutant seedlings impairs cytokinesis (16). There is no obvious reason for this difference in efficacy between newly made and endocytosed syntaxin in cytokinesis, especially because the TGN acts as a sorting station that directs both secretory and endocytosed proteins to the plane of cell division in cytokinesis (31).
The comparative analysis of PEN1 and SYP132 revealed an important difference in the dynamic behavior of the two proteins. Whereas PEN1 cycled continually between the plasma membrane and endosomal compartments, SYP132 appeared to associate stably with the plasma membrane. This was observed in the root cells in which ER–Golgi traffic was inhibited by BFA treatment of gnl1 mutant seedlings: when expressed from the KNOLLE promoter, PEN1 was trapped in the ER but still accumulated in endosomal BFA compartments, whereas SYP132 was only detected in the ER and at the plasma membrane. The same difference was observed when the two proteins were expressed from the S-phase-specific H4 promoter: only PEN1 accumulated at the plane of division (cell plate) during cytokinesis, whereas SYP132 stayed at the plasma membrane. Thus, PEN1 appears to be a highly dynamic protein, whereas SYP132 once made appears to be firmly anchored at the plasma membrane.
PEN1 plays an important role in Arabidopsis non-host resistance to fungal pathogens (12,24). Endogenous PEN1 appears to be moderately up-regulated in response to pathogen attack and accumulates rapidly at the site of infection. We used two different promoters to analyze the relevance of syntaxin retargeting for mounting a successful defense against non-adapted powdery mildew fungi. The UBQ10 promoter is constitutively active and thus provides, during pathogen attack, both newly synthesized syntaxin and syntaxin made earlier and then stored at the plasma membrane. In contrast, the KNOLLE promoter is only active in proliferating cells but not in mature leaf cells. Thus, in the latter case, only syntaxin made earlier and then stored at the plasma membrane is available during pathogen attack. Our data suggest that PEN1 is highly dynamic such that endocytic retargeting of plasma membrane-localized syntaxin is sufficient and no newly made syntaxin is necessary for mounting a successful defense during pathogen attack. In contrast, SYP132 only partially rescued the compromised pathogen defense of pen1 leaves when expressed from the UBQ10 promoter but had no effect when expressed from the KNOLLE promoter. Thus, even newly synthesized SYP132 might not be efficiently targeted to the site of infection. One possible explanation for this difference in subcellular dynamics between PEN1 and SYP132 might be that the strong accumulation of syntaxin at the infection site does not result from directional secretion but rather requires endosomal retargeting. Similar observations were made in polar targeting of PIN proteins in Arabidopsis(46,47).
Compared to the two specialized SYP1 syntaxins KNOLLE and PEN1, SYP132 might represent a rather general syntaxin function at the plasma membrane, possibly involved in constitutive fusion of secretory vesicles. No knockout mutants of SYP132 are known. However, SYP132 appears to be broadly if not ubiquitously expressed during development (8). Furthermore, sequence comparisons with SYP1 syntaxins from lower plants suggest that SYP132 might play an ancient role in secretory traffic to the plasma membrane (6).
In an evolutionary scenario that associates SYP132 with plasma-membrane syntaxins of primitive land plants, KNOLLE and PEN1 appear to have evolved divergently to serve their respective highly specialized function. KNOLLE has adopted an exclusive role in membrane fusion during cytokinesis, which for yet unknown reasons requires high-level expression immediately before cytokinesis. In addition to, and possibly as a consequence of, the dramatic change in gene regulation leading to high-level protein accumulation, KNOLLE protein has become highly unstable, being targeted to the vacuole for degradation at the end of cytokinesis. Interestingly, during KNOLLE evolution, there seems to have been no substantial functional change from the presumably ancient SNARE domain of SYP132, unlike the SNARE domain of PEN1. In contrast, PEN1 displays dramatic subcellular protein dynamics, as evidenced by its continual cycling in non-infected cells and the retargeting to the plane of cell division during cytokinesis. This dynamic behavior also enables PEN1 to act in plant innate immunity, facilitating its rapid relocation from the plasma membrane to fungal infection sites via endocytosis and retargeting. It remains to be determined how the attacked plant cell reorganizes its membrane trafficking to fend off fungal intruders.
Plant growth, transformation and selection
Arabidopsis thaliana plants were grown on half-strength Murashige and Skoog (MS) medium (+1% sucrose for microscopy) or on soil at 18–23°C, with cycles of 16 h light and 8 h dark. Landsberg/Niederzenz (Ler/Nd) plants heterozygous for the knolle mutation X37-2 (9) or Columbia (Col) plants homozygous for the pen1-1 mutation (12) were transformed with Agrobacterium tumefaciens, using the floral-dip method (48). T1 plants from bulk-harvested seeds were selected for transformants either grown on soil by spraying twice with a 1:1000 dilution of Basta® (45) (183 g/L Glufosinate-ammonium, Bayer) or grown on half-strength MS medium containing 50 µm kanamycin. BASTA-resistant plants were genotyped for knolle X37-2 (22) and kanamycin-resistant plants for pen1-1(12) as described.
Transgenic constructs for subcellular localization and for knolle rescue were expressed from the mitosis-specific KNOLLE promoter. KN and PEN1 coding sequences were cloned into the KNOLLE cassette as described (22). SYP132 CDS was amplified from a flower and silique cDNA library of the Landsberg ecotype (49) by polymerase chain reaction (PCR) according to standard procedures using Taq DNA Polymerase (Peq Lab Biotechnologie GmbH). SYP132 CDS was directionally cloned into the KNOLLE cassette in the pBluescript vector via restriction sites SmaI and EcoRI (22). The KN:Myc-SYP132 insert was introduced into the pBAR-B vector via the restriction sites HpaI and SpeI.
Constructs for SNARE domain swaps were generated by primer extension PCR according to standard procedures using Taq DNA Polymerase (Peq Lab Biotechnologie GmbH). Constructs were cloned into the multiple cloning site (MCS) of the pGreenIIB containing the KNOLLE cassette carrying an N-terminal RFP-tag.
A HISTONE 4 (H4) expression cassette was generated by amplifying 543 bp upstream and 177 bp downstream of the H4 gene. The 5′ and 3′ sequences were directionally cloned into the binary pGreenIIB-vector using the restriction sites SacI/XbaI and EcoRI/KpnI, respectively, surrounding an MCS containing XbaI, SmaI and EcoRI. An N-terminal RFP-tag was introduced by XbaI/SmaI. SYP1 coding sequences were introduced via SmaI/EcoRI sites.
The UBQ10 promoter (25) constructs for pen1-1 rescue were generated by directionally cloning Arabidopsis syntaxin coding sequences into the binary pGreenIIB-vector downstream of the UBQ10 promoter via the restriction sites SmaI and SpeI.
The constructs were checked by restriction digest and sequencing using the ABI PRISM Big Dye Terminator Cycle Sequencing Kit and the ABI-Sequencer 310 (Applied Biosystems) or using GATC (Konstanz) service before transformation into A. tumefaciens strain GV3101. Standard protocols were used for molecular biology (50). Restriction enzymes were purchased from MBI Fermentas and synthetic oligonucleotides from ARK (Sigma-Aldrich).
For RT-PCR analysis, total RNA was isolated from 100 mg Arabidopsis seedlings, using the ‘Trizol-method’(51) or the RNAeasy plant mini kit (Qiagen). After removal of contaminating DNA (DNase I, Fermentas), first strand cDNA was synthesized with Superscript II RNaseH-Reverse Transcriptase (Invitrogen), using the dT-anchor-random II primer. As a control, we used ACTIN2. All primer sequences are listed in Table S2.
Western blot analysis
Preparation of protein extracts and Western blots was performed as described (10). For protein extraction we used one inflorescence, five rosette leaves or 50 mg seedlings. Rabbit anti-KNOLLE antiserum was used at 1:5000 dilution (10), mouse anti-α-tubulin monoclonal antibody at 1:4000 (Sigma-Aldrich), rat anti-RFP monoclonal antibody at 1:1500 (chromotek), sheep anti-rabbit IgPOD polyclonal antibody at 1:1000 (Boehringer), goat anti-mouse IgPOD polyclonal antibody at 1:10 000 (Boehringer), goat anti-rat IgPOD at 1:1000 (Sigma-Aldrich) and mouse anti-Myc-POD monoclonal antibody at 1:1000 (Roche). Independent T1 knolle heterozygous lines for each transgene (KNOLLE, PEN1 and SYP132) were investigated for expression level. For phenotypic and transcript level analyses, we used the strongest and weakest expression line of each transgenic construct.
Inhibitor treatment and FM 4-64 staining
Three- to five-day-old seedlings were incubated in 1 mL of liquid medium (half-strength MS medium) containing 50 µm BFA. FM 4-64 dissolved in water was used at 4 µm final concentration. Seedlings were incubated with inhibitors and dye at room temperature for the indicated times followed by fixation with 4% paraformaldehyde in microtuble stabilizing buffer (MTSB) (50 mm PIPES, 5 mm EGTA, 5 mm MgSO4, adjust pH with KOH). The following stock solutions were used: 50 mm BFA (Sigma-Aldrich) in dimethyl sulphoxide (DMSO):ethanol (1:1), and 2 mm FM 4-64 (Molecular Probes) in water. Control treatments were performed with equal amounts of the respective solvents.
Antibody staining and confocal laser-scanning microscopy
Whole-mount immunofluorescence was performed as described (10). Antibodies and dilutions were as follows: rabbit anti-KNOLLE antiserum (1:2000) (10), mouse anti-Myc monoclonal antibody 9E10 (1:600; Santa Cruz Biotechnology), rabbit anti-SEC21/γCOP polyclonal antibody (1:1000) (16), rabbit anti-ARF1 polyclonal antibody (1:5000) (52), fluorescein isothiocyanate (FITC)-conjugated secondary goat anti-rabbit antibody (1:600, Dianova), Cy3-conjugated secondary goat anti-mouse antibody (1:600, Dianova). DAPI (4′,6-diamidino-2-phenylindole) staining was performed as described (45). Immunofluorescence and live-cell microscopy were done with a Leica TCS-SP2/SP5 confocal laser-scanning microscope. All confocal laser-scanning microscopy (CLSM) images were obtained using the Leica Confocal software and a 63× water-immersion objective. Images were processed using Adobe Photoshop CS3.
B. g. hordei inoculation and quantification of pen1 mutant phenotype
Four-week-old Arabidopsis plants were inoculated with powdery mildew Blumeria graminis hordei from barley. Live-cell imaging of single leaves was performed at 12–16 h after inoculation. Penetration rescue analyses were performed at 72 h after inoculation.
Individual B. g. hordei–Arabidopsis interaction sites were characterized microscopically for failed and successful invasion (efficient papilla formation versus haustorium formation and hypersensitive-response-like cell death) using aniline blue and coomassie blue as recently described (12). The experiment was repeated three times, and 100 interaction sites per genotype were scored each time.
We thank Stefan Driessen, Ulrike Hiller and Alexandra Matei for technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft through an AFGN grant to G. J.