Syntaxin 1C, a soluble form of syntaxin, attenuates membrane recycling by destabilizing microtubules

Soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) machinery is a central regulator of specific interactions between vesicular v-SNARE (e.g. VAMP; also known as synaptobrevin) and target t-SNARE [e.g. SNAP25 and syntaxin (STX)] (Rothman, 1994). The STX family consists of 19 members, most of which associate with specific membrane compartments by virtue of a hydrophobic trans-membrane domain (TMD) in their C-terminal region and regulate membrane transport between intracellular compartments and target organelles (Teng et al., 2001). For example, STX1A is involved in the docking and/or fusion of synaptic vesicles to the plasma membrane (PM) at active zones in neurons (Bennett et al., 1992; Inoue et al., 1992). Several syntaxins have soluble variants lacking a TMD (e.g. STX1BΔTMD, STX1C, 2D, 3D, 16C), which are generated by alternative splicing (Ibaraki et al., 1995; Jagadish et al., 1997; Pereira et al., 2008; Quinones et al., 1999; Simonsen et al., 1998). However, the functions and the mechanism of action of these soluble syntaxins have not yet been determined.

To better understand the action of soluble syntaxins, we focused on STX1C. STX1C, an alternative splice variant of the stx1A gene, is deleted hemizygously in patients with the neurodevelopmental disorder Williams syndrome, which shows characteristic cognitive profiles, such as hyperactivity, poor attention span, good memory, remarkably spared linguistic abilities and visual spatial deficits (Nakayama et al., 1997; Nakayama et al., 1998). STX1C inhibits intracellular glucose transport by suppressing the translocation of glucose transporter-1 (GLUT1) to the PM in glial cells (Nakayama et al., 2004). STX1C possesses a large part of the N-terminal domain of STX1A, whereas the novel C-terminal domain of 34 residues is generated by the insertion of a 91-bp splicing region. This insertion results in a loss of the TMD and the functional SNARE domain, the helical structure necessary for association with v-SNARE, resulting in cytoplasmic expression (Jagadish et al., 1997; Jahn and Sudhof, 1999; Nakayama et al., 2003). It is generally believed that membrane-bound syntaxins, such as STX1A, regulate membrane trafficking through the SNARE domain (Jahn and Sudhof, 1999). However, the mechanism by which the soluble syntaxin STX1C affects GLUT1 translocation remains unknown.

GLUT1, which is a representative member of the high-affinity facilitative glucose transporter (GLUT) family including GLUT3 and GLUT4, is responsible for the entry of glucose into cells and for maintaining cell metabolism and homeostasis throughout the periphery and the brain (Olson and Pessin, 1996). Recently, it has been demonstrated that GLUT translocation to the PM is dependent on the cytoskeletal systems including microtubules (MTs) and actin filaments (Dransfeld et al., 2001; Singh et al., 1998; Tong et al., 2001; Wang et al., 1998). For example, drugs that influence the function of tubulin polymerization, such as vinblastine and paclitaxel, alter glucose uptake activity by GLUT1 in glioma cells (Singh et al., 1998). It has also been reported that the GLUT4 vesicular transport depends on cortical actin remodeling (Dransfeld et al., 2001; Tong et al., 2001; Wang et al., 1998) that is regulated by a direct link between SNARE proteins and the MT network (Pooley et al., 2008). Furthermore, MT disruption by nocodazole inhibits the motility of the vesicle-transport (VT)-containing PM protein reduced folate carrier (RFC), as well as final translocation of RFC to the PM in epithelial cells (Marchant et al., 2002). Interestingly, the STX1C N-terminal region possesses a tubulin-binding domain (TBD) that is found in the MAP2 and tau protein family (Lewis et al., 1988) (hereafter referred to as MAP2/tau). In MAP2/tau, the TBD regulates MT dynamics together with regulatory proteins, including MT plus-end-binding proteins and soluble factors, such as cytoplasmic linker proteins (CLIPs) and stathmin (Howard and Hyman, 2009). This leads us to postulate that the TBD of STX1C also affects cytoskeletal transport systems.

In this study we used an lung epithelial cell line, FRSK, to examine the function of the N-terminal TBD of STX1C to elucidate how this soluble syntaxin suppresses GLUT1 translocation. Our data indicate that STX1C delays the recycling phase of GLUT1 to the PM. Thus, we report the first characterization of the activity of a soluble syntaxin and establish that STX1C alters the cellular distribution of GLUT1 through suppression of MT stability and VT motility through its TBD. This suggests a functional role of STX1C in the suppression of GLUT1 translocation and intracellular glucose transport.

To investigate how the TBD of STX1C is involved in MT polymerization in vivo, we first examined the association between STX1C and tubulin by using wild-type (WT) and mutated peptides of the predicted TBD that correspond to residues 89–106 of STX1C. As shown in Fig. 1A, given that half of the residues (i.e. 89–98) of STX1C show high homology with the TBD of MAP2/tau, we attempted to replace the first four residues of the peptide; valine, arginine, serine and lysine (VRSK), with alanines (AAAA) or glycines (GGGG). In competition assays the WT peptide significantly inhibited the association between recombinant His-tagged STX1C and tubulin at 0.5 and 5 μM compared with 0 μM of WT competitor, whereas the mutated peptides (AAAA and GGGG) did not out-compete the association between recombinant His-tagged STX1C and tubulin (Fig. 1B). This indicates that the VRSK sequence in TBD is crucial for the interaction between STX1C and tubulin.

Thus, we prepared several DsRed–STX1C expression constructs with WT or mutated TBDs (pDsRed–STX1C-WT, and STX1C-AAAA and STX1C-GGGG, respectively) for transfection experiments. Next, for the in vivo MT-polymerization assay, we selected several types of cell that did not endogenously express STX1C or STX1A, both of which are alternative splice variants of the stx1A gene with a common N-terminus. We did this in order to avoid the potentially complicating effects on dynamics caused by endogenous wild-type STX1C isoforms, and to more directly identify loss of function on dynamics by the STX1C TBD mutants as shown in a previous study on the effect of Tau mutants on dynamics (Bunker et al., 2006). We cloned cell lines stably expressing GFP–α-tubulin and chose a lung epithelial cell line, FRSK, which has thin cell periphery suitable for in vivo imaging (Fig. 1C).

We previously reported that STX1C suppresses glucose transport through the inhibition of GLUT1 translocation to the PM in glioblastoma cells (Nakayama et al., 2004). However, we did not know whether exogenous STX1C and the TBD affected glucose uptake activity or GLUT1 translocation in epithelial cells. Therefore, to understand the effect of STX1C and the TBD on glucose uptake activity, we exogenously introduced DsRed vector (Vec), STX1C-WT, STX1C-AAAA or STX1C-GGGG to at least three different transformant cell lines for each construct. Western blotting analysis with the 14D8 antibody, which is reactive to common N-termini of STX1C and STX1A, showed that the expression of each introduced construct in the various transformants was almost the same (Fig. 1C). We also found that GLUT1 was the most abundant of the high affinity glucose transporters in FRSK cells and that the amount of GLUT1 protein was similar among the transformants (Fig. 1C), indicating that overexpression of the exogenous STX1C constructs did not affect GLUT1 expression in FRSK cells.

Next, we investigated glucose uptake by GLUT1 using 2-deoxy-D-glucose (2-DG), a non-metabolic glucose analog. As shown in Fig. 1D a 2-DG kinetic assay was carried out using 0.1–100 mM 2-DG, to examine the Michaelis constant (K_m) and the maximum velocity of the reaction (V_max). The V_max value was significantly reduced in the transformant cells expressing STX1C-WT without changing the K_m value, compared with the FRSK parent cells (Pt), cells expressing Vec and TBD-mutated constructs (AAAA, GGGG; Table 1). This result was also confirmed by the Eadie–Hofstee plot (Fig. 1D, inset). This reduced V_max value was not caused by a change in basal glucose transport by the sodium-dependent glucose transporter (SGLT), because 2-DG uptake through SGLT, measured under sodium-free conditions, accounted for only ~10% of the total 2-DG uptake and was not altered by overexpression of exogenous constructs in epithelial cells (data not shown) or glioblastoma cells, as reported previously (Nakayama et al., 2004). These results suggest that the suppression of glucose uptake by STX1C-WT might be due to a decrease in the number of GLUT1 molecules on the PM, but not to a change in the glucose transport rate by a single GLUT1 molecule.

Finally, to interpret results of the kinetic assay, we examined the effect of STX1C TBD on GLUT1 translocation in each of the transformant cells. Using immunofluorescence microscopy, we found that exogenous STX1C-WT was expressed in the cytosol, and suppressed GLUT1 translocation to the PM (Fig. 1E). Furthermore, to investigate the effect of STX1C TBD mutants on GLUT1 translocation, mutated constructs were transfected into FRSK cells. In contrast to STX1C-WT, expression of mock vector or TBD-mutated constructs (STX1C-AAAA and -GGGG), did not suppress GLUT1 translocation (Fig. 1E). Similar results were also obtained in cells transiently transfected with each construct (data not shown). These results suggest that the TBD domain of STX1C is also responsible for the suppression GLUT1 translocation.

To investigate the mechanism of suppression, cloned FRSK cells stably expressing GFP–α-tubulin were transfected with STX1C constructs and the dynamics of MT polymerization was visualized using time-lapse microscopy. Fig. 2A shows a representative fluorescence time-lapse image of living FRSK cells stably expressing GFP–α-tubulin. As shown in Fig. 2B, we captured time-lapse images of pioneer MTs in the flat peripheral region of these cells. Most MT plus-ends were oriented toward the leading edge and exhibit non-equilibrium polymerization behavior, referred to as dynamic instability, with stochastic switching between phases of growth and shortening, consistent with previous studies in endothelial and fibroblast cells (Desai and Mitchison, 1997; Wittmann and Waterman-Storer, 2001). Fig. 2B also shows typical dynamic MT transition between the three phases (i.e. growth, shortening and attenuation) as defined in the Materials and Methods (see also supplementary material Movie 1). Typical MT life-history plots for cells transfected with mock vector (Vec), STX1C-WT, STX1C-AAAA or STX1C-GGGG are presented in Fig. 2C (see also supplementary material Movies 2–5). From these individual MTs, we determined the dynamic instability of MTs, as described in the Materials and Methods.

We first assessed the ability of each construct to influence the velocity and distance that dynamic MTs exhibit in the growth or shortening phase. As shown in Fig. 3A, STX1C-WT only significantly (P<0.01) reduced the velocity of MT in the shortening phase, whereas the velocity during the growth phase remained unchanged relative to MTs in Vec, STX1C-AAAA or STX1C-GGGG. By contrast, the ability to influence MT dynamics was not significantly different between the four constructs (Fig. 3B). STX1C-WT increased the attenuation time of MTs, compared with Vec, STX1C-AAAA or STX1C-GGGG; the two TBD mutants (AAAA and GGGG) having no effect (Fig. 3C).

We then calculated the fraction of time that dynamic MTs spent in each phase relative to the total time tracked. We found that STX1C-WT significantly (P<0.01) influenced the fraction of time that dynamic MTs spent in each phase relative to Vec, STX1C-AAAA and STX1C-GGGG (Table 2). For example, STX1C-WT significantly (P<0.01) increased the fraction of time MTs spent attenuated, by 25.3% (from 44.4 to 59.4%; Table 2), while reducing the fraction of time MTs spent growing and shortening by 42.3% (from 30.1 to 17.3%; Table 2) and 8.6% (from 25.5 to 23.3%; Table 2) respectively, relative to the Vec treatment (Table 2). By contrast, the two STX1C TBD mutants did not influence the phase distribution of the MTs relative to that of the Vec treatment.

To determine whether the different STX1C constructs affected MT dynamicity, the percentage of dynamic MTs versus total MTs in the cells was initially analyzed using the criteria described in the Materials and Methods. For the analysis of the density of total MTs per analyzed area, a region of the cell periphery was randomly chosen, and the number of MTs in the area was counted. STX1C-WT significantly (P<0.01) decreased the density of total MTs relative to MTs in Vec- or STX1C-AAAA-treated cells and the density of dynamic MTs relative to MTs in Vec-, STX1C-AAAA- and STX1C-GGGG-treated cells (Table 3). Additionally, STX1C-WT significantly (P<0.01) decreased the percentage of dynamic MTs versus total-MTs, by 24% (from 86.6 to 65.8%; Table 3) relative to MTs in Vec-treated cells (Table 3). By contrast, two of the STX1C TBD mutants did not influence the density of dynamic or total MTs or the percentage of dynamic MT relative to MTs in Vec-treated cells. In particular, the STX1C-AAAA mutant was significantly (P<0.01) compromised in all abilities to influence the density of dynamic or total-MTs and the percentage of dynamic MT, compared with STX1C-WT. A similar result was obtained from the analysis of the dynamicity, which is a measure of the visually detectable amount of dynamic instability occurring in a MT population. It was calculated as the total length grown and shortened divided by the time period observed through the process of dynamic instability. STX1C-WT decreased the dynamicity of MTs compared with those in the Vec control cells (from 21 to 18.8%, a decrease of 11%; Table 3), whereas the two TBD mutants had no effect (Table 3).

We also tried to perform time-lapse analysis by using U87MG glioblastoma cells expressing endogenous STX1C (supplementary material Fig. S1). The effect of STX1C TBD mutants on MT polymerization indexes in cells expressing endogenous STX1C was similar to those in cells not expressing endogenous STX1C (supplementary material Table S1), except for the effect on the velocity and distance in the growth phase, which might be due to other factors in cells expressing endogenous STX1C. The result suggests that the effect on MT polymerization indexes in FRSK cells not expressing endogenous STX1C is not the result of side-effects of other factors.

Thus, these data suggest that, because a similar effect was also obtained in U87MG-expressing endogenous STX1C, STX1C-WT plays a role as a negative regulator, leading to the suppression of MT stability through its TBD.

MT dynamics are related to the motility of vesicles containing PM protein (Marchant et al., 2002). Thus, we attempted to study the effect of STX1C on VT motility. It is generally accepted that GLUT1 is expressed and recycled constitutively (Olson and Pessin, 1996), however, the recycling process remains unclear. Thus, we investigated GLUT1 recycling using an anti-GLUT1 monoclonal IgG. Given that this antibody detects the ectodomain of GLUT1, depending on the kind of cell (Kinet et al., 2007; Mueckler et al., 1985), we first confirmed that the monoclonal antibody was useful as an antibody recognizing the extracellular domain of GLUT1 in FRSK cells (supplementary material Fig. S2). Next, according to an established method for examining membrane recycling (Kamiguchi et al., 1998), we investigated the fate of endocytosed GLUT1 using the GLUT1-ectodomain antibody (see the Materials and Methods). Fig. 4A illustrates that internalized-GLUT1 molecules were recycled back to PM after an extra 45 minutes in culture (Fig. 4Ag,h,i), whereas at time-points excepting 45 minutes, recycling of GLUT1 molecules was not detected (Fig. 4A).

To investigate VT motility in FRSK cells, we used the fluorescent probe FM1-43, which is useful for measuring membrane trafficking (Brumback et al., 2004). Using this fluorescent probe, we first studied the colocalization of FM1-43 vesicles and GLUT1 vesicles in FRSK cells. Double immunostaining using anti-GLUT1 IgG and FM1-43FX, a fixable analog of FM1-43, revealed that FRSK cells have a heterogeneous population of single or aggregated vesicles of ~0.5-3 μm diameter. This is consistent with the observed vesicle size in a variety of epithelial cells reported previously (Jerdeva et al., 2005; Sahlen et al., 2002). Double immunostaining revealed that 95.04±0.71% (33 cells) of FM1-43FX vesicles that colocalized with GLUT1 vesicles in the flat peripheral region were ≤1.5 μm diameter (Fig. 4B). Therefore, FM1-43 vesicles ≤1.5 μm in diameter in the FRSK cell periphery could be used to monitor GLUT1 vesicles. We also examined the colocalization of FM1-43FX vesicles and MTs in FRSK cells. Double immunostaining using FM1-43FX and anti-β-tubulin IgG revealed that 99.52±0.10% (86 cells) of vesicles ≤1.5 μm diameter in the flat peripheral region colocalized with MTs (Fig. 4C). To assess whether vesicles in the flat peripheral region moved with MTs, we further studied VT motility in living cells stably expressing GFP–α-tubulin using FM4-64, an alternative analog of FM1-43. Simultaneous time-lapse observations with GFP and FM4-64 using a DualView system revealed that vesicles in the flat peripheral region were actually moving along with MTs (Fig. 4D). These data indicate that GLUT1 vesicles ≤1.5 μm in diameter, move with MTs.

On the basis of these observations, we performed live microscopy studies on vesicles moving with MTs in living cells. Fig. 5A shows a representative FM1-43 time-lapse image of a living parent FRSK cell. The FM1-43 fluorescent signal was captured as time-sequential images (Fig. 5A, lower panel). Vesicles, visualized as punctate signals showed complex transport behavior, which included long-range bi-directional motions and short-range motions, such as rotation and oscillation, in FRSK cells (see also supplementary material Movie 6). To assess the exact motility of secretory vesicles moving with MTs, with the exception of vesicles rotating and oscillating, only FM1-43 vesicles anterogradely moving toward the PM in the flat peripheral region were analyzed until the vesicle moved in another direction, switched to short-range motions or disappeared (Materials and Methods). Time-lapse observations with FM dye also revealed that as the diameter increased to >2–3 μm, the vesicles in the FRSK cells gradually lost long-range motion. Thus, only FM1-43 vesicles of ~0.5–1.5 μm in the flat peripheral region were analyzed. To study the effects of STX1C and the TBD on VT motility, cells were transfected with Vec, STX1C-WT or STX1C-AAAA, and changes in movement distance of individual vesicles were measured over time by tracking the positions of directional movement ends before another directional movement or disappearance. Fig. 5B,C shows representative FM1-43 time-lapse images and vesicle movement plots, respectively, for cells transfected with a mock vector, STX1C-WT or STX1C-AAAA. A VT velocity analysis revealed that epithelial GLUT1 vesicles were transported with a maximum velocity of ~1.2 μm/second and an average velocity during a long-range run of ~0.3 μm/second, which is similar to the value of MT-dependent velocity of vesicles in epithelial and endothelial cells (Manneville et al., 2003; Marchant et al., 2002). A detailed velocity analysis revealed that STX1C-WT significantly (P<0.01) suppressed the average velocity of MT-dependent VT by ~60% compared with Vec and STX1C-AAAA (Fig. 5C, upper panel; Table 4), whereas there was no difference in the distance of directional vesicle movement between constructs (Fig. 5C, lower panel; Table 4). These data indicate that the time parameter in the equation (velocity=distance/time) is enlarged by STX1C-WT, resulting in slow MT-dependent VT, and suggest that the TBD of STX1C also affects the transport velocity of MT-dependent secretory vesicles.

Next, we investigated the effect of STX1C on the recycling profile of GLUT1 after its endocytosis. As in Fig. 4B, GLUT1 recycling periods were examined but with an extra culture period of 15–75 minutes after incubation with GLUT1 IgG, and masking with unconjugated secondary antibody. Fig. 6A–E shows images at 30, 45, 60, 75 and 90 minutes time points after incubation with GLUT1 IgG and extra culture of cells transfected with Vec, STX1C-WT or STX1C-AAAA. At the 45-minute time point, which is the recycling time of GLUT1 to the PM (as shown in Fig. 4A), positive labeling with Alexa-Fluor-488-conjugated secondary antibody in cells transfected with Vec, STX1C-WT or STX1C-AAAA was detected in 50.6, 21.0 and 49.4%, respectively, of at least 174 cells examined (Fig. 6B; Table 5). STX1C-WT suppressed the cell-recycling rate of GLUT1 to the PM by 58% compared with Vec and STX1C-AAAA. By contrast, STX1C-WT increased the rate of cell recycling of GLUT1 to the PM at the 60 minutes, 75 minutes and 90 minutes compared with Vec, and at the 60 minutes and 90 minutes compared with STX1C-AAAA (Fig. 6C–E; Table 5). This is probably the result of GLUT1 vesicles not arriving at the PM at 45 minutes in STX1C-WT-treated cells, but instead being delayed and arriving at the PM at 60 and 75 minutes. At 15 and 30 minutes, GLUT1 recycling to the PM was only observed at extremely low levels in cells with each construct (Fig. 6A; Table 5). These observations demonstrated that STX1C-WT decreased the rate of cell recycling of GLUT1 to the PM at the 45 minutes time point and increased it thereafter, compared with Vec and STX1C-AAAA (Fig. 6A–E; Table 5).

We further examined whether this effect on the recycling profile was due to an alteration of the GLUT1 internalization process by STX1C. Fig. 6F shows immunofluorescence images of endocytosed GLUT1 using a GLUT1 ectodomain antibody (see the Materials and Methods). The average density (number per μm²) of GLUT1 vesicles in the peripheral region was calculated in at least 81 cells. An internalization assay of GLUT1 demonstrated that STX1C-WT did not significantly (P>0.05) alter the density of endocytosed GLUT1 vesicles compared with Vec or STX1C-AAAA (Fig. 6G).

Conversely, during microscopic observation, we noted that the signal of GLUT1 recycled to PM at 45 minutes appeared to be weaker in cells transfected with STX1C-WT than the other constructs. Thus, we next examined whether the observed decrease in GLUT1 recycled to the PM by STX1C at this stage was because less GLUT1 was internalized, given the possibility that the amount of GLUT1 on the PM in first internalization step had already been suppressed (Fig. 1D–E). To quantify GLUT1 in the process of internalization and recycling we conducted flow cytometry (FCM) analysis using the GLUT1 ectodomain antibody (see the Materials and Methods). Trypsin treatment did not affect the GLUT1 recycling profile in FRSK cells (supplementary material Fig. S3). Therefore, using FCM analysis in permeabilized cells, we quantitatively evaluated the endocytotic (Fig. 7A) and recycling (Fig. 7B) processes using Qdot705-conjugated secondary antibody, because this fluorescent analog showed a better signal-to-noise ratio than Alexa Fluor 488 or Alexa Fluor 680R-PE (data not shown). The FCM analysis demonstrated that there was a reduction in the most frequent signal intensity of Qdot705 bound to the endocytosed antibody–GLUT1 complex in cells stably expressing STX1C-WT, but not those expressing Vec or STX1C-AAAA (Fig. 7A, lower panel). Furthermore, STX1C-WT decreased the average mean fluorescence intensity (MFI) of Qdot705 bound to endocytosed-GLUT1 by 14% or 15% compared with Vec or STX1C-AAAA, respectively (Table 6), as shown by 11 independent experiments. This result indicates a reduction of internalized GLUT1 molecules, which is consistent with GLUT1 suppression on the PM by STX1C (Fig. 1D–E) and a reduction of endocytosed GLUT1 molecules per one internalized vesicle from Fig. 6G. Similarly, STX1C-WT reduced the most frequent signal intensity of Qdot705 bound to GLUT1 that was recycled back to the PM (Fig. 7B, lower panel) by 30 or 15% of that seen with Vec or STX1C-AAAA, respectively, as demonstrated in nine independent experiments (Table 6). Furthermore, the ratio of recycled GLUT1 to endocytosed GLUT1 decreased in cells with STX1C-WT, which did not occur with Vec and STX1C-AAAA (Table 6). These quantitative observations indicate that the reduction of recycled GLUT1 at 45 minutes caused by STX1C-WT reflected a decrease in the quantities of endocytosed GLUT1, and that STX1C-WT also decreased recycling of GLUT1 at this time.

We demonstrated that STX1C with a wild-type TBD suppressed the growth phase instead of increasing the pause phase in MT polymerization, as well as reducing the MT dynamicity and the density of total and dynamic MTs, suggesting that STX1C acts as a negative factor for MT stability. Additionally, the TBD of STX1C was found to be responsible for the suppression of VT velocity of the GLUT1 secretory vesicle and delay in the recycling phase of the GLUT1 vesicle, but did not affect the endocytotic process of GLUT1. These observations suggest that suppression of the MT dynamics and VT velocity of GLUT1 secretory vesicles delays the recycling phase of the GLUT1 vesicle, resulting in increased intracellular accumulation of GLUT1 vesicles through alteration of the cellular distribution of GLUT1. Additionally, these phenomena were not affected by the inserted portion unique to STX1C (amino acids 226–260), because the TBD mutant with the unique portion (AAAA) did not inhibit glucose transport or MT-dynamics. The unique portion with repetitive proline residues causes loss of the functional helical structure and the TMD, but does not affect MT dynamics.

We demonstrated that STX1C with a wild-type TBD associated directly with tubulin and reduced the percentage of growth time, decreasing the velocity and the density of total and dynamic MTs, the percentage of dynamic MTs and the dynamicity, while increasing the time and the percentage of time spent in the attenuated state of MT polymerization. This is consistent with previous in vitro studies using artificial STX1AΔTM, which has almost the same N-terminal sequence as that of STX1C, with the exception of the last half of the SNARE domain; artificial STX1AΔTM protein associated with tubulin and suppressed in vitro MT polymerization activity in a reassembly assay (Fujiwara et al., 1997; Itoh et al., 1999). These characteristics appear to be partially compatible with the dynamics of tubulin polymerization in MAP/tau, which showed a marked increase in the percentage of time MTs spend in the attenuated state as well as the percentage of non-dynamic MTs in the cells (Bunker et al., 2004). However, given that STX1C decreased not only dynamic MT density but also total MT density, it appears that MAP/tau acts as a MT stabilizing factor (Howard and Hyman, 2009), whereas STX1C acts as a negative factor for MT stability. Additionally, STX1C did not result in binding along the length of MTs, as did MAP/tau, or in accumulation at the plus-end of MTs, as did MT plus-end tracking proteins (Howard and Hyman, 2009). Additionally, it appeared to be cytoplasmic (Nakayama et al., 2003). It has been reported that MT density is reduced in cells with increased stathmin, which is an MT-instability factor distributed throughout the cytoplasm, and that it reduces available tubulin subunits by directly binding to the tubulin subunit, but not to MTs (Jourdain et al., 1997; Ringhoff and Cassimeris, 2009). STX1C might be a sequestering protein, in a manner similar to that of stathmin.

We also demonstrated that alterations in MT instability, through the effect of the STX1C TBD, affected the motility of vesicles containing GLUT1 protein as well as their translocation to the PM, consistent with a previous report (Marchant et al., 2002). Although this effect on VT motility might directly reflect the state of individual MTs, such as destabilization and disruption, it is also possible that alterations in the association between vesicles and MTs could affect VT motility. The STX family of molecules, which now consists of 19 members, regulates every process of membrane transport between intracellular compartments and delivery to the target organelles (Teng et al., 2001). GLUT1 vesicles are known to interact with cytoskeletal proteins, such as myosin VI, α-actinin-1 and the kinesin-1 family member 1B (KIF-1B) through GLUT1CBP (also known as TIP2 and GIPC1), resulting in targeting of GLUT1 to specific subcellular sites, either by tethering the transporter to cytoskeletal motor proteins or by anchoring it to the actin cytoskeleton (Bunn et al., 1999; Reed et al., 2005), whereas syntaxin has not been shown to participate in GLUT1-vesicle transport.

Recently, it has been reported that the translocation of GLUT4, a member of the facilitative GLUT family which includes GLUT1, depends on syntaxins and the cytoskeletal system (Dransfeld et al., 2001; Perera et al., 2003; Pooley et al., 2008; Tong et al., 2001; Wang et al., 1998). For example, STX6, 8 and 12 colocalize on the GLUT4 vesicle, whereas STX6 regulates transport of the GLUT4 vesicle in a membrane-trafficking step that sequesters GLUT4 away from traffic destined for the PM (Perera et al., 2003). Centromere protein F (CENPF), which provides a link between proteins of the SNARE system, including STX4, associates directly the MT network, resulting in the regulation of GLUT4 vesicular transport (Pooley et al., 2008). In neurons, syntabulin, a motor-protein-linked factor associated with both STX1A and KIF5B has also been implicated in the anterograde axonal transport of STX1A-containing vesicles on the MT network (Cai et al., 2007; Su et al., 2004). Although FRSK cells do not express STX1A, it is possible that other PM syntaxins colocalize on GLUT1 vesicles, as in the case of GLUT4. For example, a ubiquitously distributed motor-protein-linked factor for STX2 and STX3, such as PM syntaxin, might be a candidate GLUT1 colocalizing protein and a competitor of STX1C, because the amino acid sequence of the TBD and the SNARE motif was particularly highly conserved between them (Teng et al., 2001). Alternatively, it has been reported that STX2 and STX3 bind to Munc18b, which is located along MTs and affects granule exocytosis associated with the MT network in mast cells (Martin-Verdeaux et al., 2003). It is also possible that STX1C plays a role as a binding competitor with Munc18b and STX2 or STX3 in the MT network, because STX1C has the ability to bind to Munc18b (Jagadish et al., 1997). Further studies of motor-protein-linked factors and the function of Munc18b with STX1C within the MT network could help provide an understanding of the regulatory mechanism of STX1C in GLUT1-vesicle transport.

Taken together, our results suggest a STX1C-TBD-dependent mechanism for the regulation of GLUT1 translocation and GLUT1-vesicle transport, providing important insights into the physiological function and the mechanism of soluble syntaxins, including STX1C.

For the expression of DsRed monomer fusion protein, the STX1C open reading frame (AB086954M) was subcloned in-frame into the pDsRed-C1 expression vector (Clontech) for time-lapse image analysis and into the pPro Ex-1 vector (Invitrogen) for the in vitro binding assay. Mutated STX1C expression vectors were constructed from the pDsRed–STX1C expression plasmid using the KOD-Plus Mutagenesis kit (Toyobo, Osaka, Japan). All the constructs were verified by sequencing.

The rat epithelial cell line, FRSK, was purchased from the Health Science Research Resources Bank (HSRRB; Osaka, Japan). Cells were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium (Invitrogen), supplemented with 10% (v/v) FCS (Sigma), penicillin (100 μg/ml) and streptomycin (100 μg/ml). For time-lapse analysis, cells were plated on poly-D-lysine-coated 35-mm glass bottom micro-well dishes and cultured in HEPES-buffered Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 10% FCS at 37°C in 5% CO₂. For transfection, cells were treated with a mixture of plasmid DNA and LipofectAMINE-2000 reagent (Invitrogen). For stable transformants of GFP–α-tubulin or DsRed derivatives, cells were selected with G418 (Sigma) and cloned.

In vitro binding studies were carried out according to our previous method with slight modifications (Fujiwara et al., 1997; Itoh et al., 1999). Briefly, the purified tubulin fraction was incubated with 6His-STX1C-linked Ni⁺ resin in the presence of various concentrations of synthetic peptides (Biosynthesis). After washing with the binding buffer [10 mM HEPES (pH 7.6), 150 mM KCl, 0.1% protease inhibitors cocktail, 1% phosphatase inhibitor cocktail], the bound material was eluted with Laemmli buffer containing 10 mM EDTA. The blotted membrane were probed with a rabbit anti-STX1C polyclonal antibody (Nakayama et al., 2003) or a mouse anti-β-tubulin monoclonal antibody (DM1A; Sigma).

For GLUT immunoblotting, a total cell membrane preparation was made as described previously (Nakayama et al., 2004). Immunoblotted signals were detected using the ECL system (Amersham) and were quantified using LAS-3000 (Fujifilm).

Glucose transport was assayed by measuring the uptake of [³H]2-deoxy-D-glucose (2-DG), essentially as described previously (Nakayama et al., 2004). The 2-DG uptake was linear between 0 and 30 minutes of incubation.

After FRSK parent cells were transfected with DsRed constructs, cells were incubated with 5 nM FM1-43 or FM4-64 (both from Molecular Probes) for 1 hour at 37°C. All experiments were done at 37°C. After incubation, cells were washed twice with DMEM and incubated with HEPES-buffered DMEM containing 10% FCS for 30 minutes at 37°C. After incubation, cells were observed using a time-lapse video microscopy system.

Analysis of microtubule (MT) dynamics and vesicle-transport (VT) motility was carried out according to previous studies with slight modifications (Bunker et al., 2006; Szodorai et al., 2009). MT dynamics and VT experiments were done using FRSK cells stably expressing a GFP–α-tubulin and the FRSK parent cell line, respectively. Cells were grown on 35-mm culture dishes and observed in a control chamber equipped with a microscope maintained at 37±0.5°C by an MI-IBC heat regulator (Olympus). For MT and vesicle dynamics experiments, cell images were obtained at 2.13- to 3.15-second intervals on an IX71 microscope (Olympus) and an ORCA-ER cooled CCD camera (Hamamatsu). EGFP and FM1-43 were excited through a 485DF15 filter (Omega Opticals, Brattleboro, VT) and their emissions were collected through 540/50BP filters (Chroma). DsRed and FM4-64 were excited through a 575DF25 filter (Omega Opticals) and their emissions were collected through a 624/40BP filter (Semrock, Lake Forest, IL). MT ends and transport vesicles at the lamellar edge of interphase cells were imaged using the AquaCosmos image analysis software (Hamamatsu).

To evaluate the motility of MT-dependent secretory vesicles, with the exception of short-range motions such as vesicle rotation and oscillation, only FM1-43 vesicles anterogradely moving towards the PM in the flat peripheral region were analyzed until the vesicle was seen to move in another direction, switch to short-range motion, or disappear. At least 38 randomly selected MTs from more than 12 independent cells or at least 37 randomly selected transport vesicles from more than four independent cells were observed. The distance traveled from a point of origin (from a pause point to the next pause point) and the velocity of individual MT ends and transport vesicles were determined using the distance measurement command on the AquaCosmos image analysis software. These values were transferred to a Microsoft Excel spreadsheet and used to analyze the dynamics of individual MTs and vesicles. Changes in MT length greater than 0.5 μm were designated as growth or shortening events. Periods in which length changes were less than 0.5 μm were designated as phases of attenuated MT dynamics (pause). The fraction of time spent was calculated as the time that MTs spent in three phases (growth, shortening, pausing) relative to the total time tracked. The density of dynamic MTs and total MTs was calculated as the number of MTs divided by the total area measured. We operationally defined dynamic MTs as those grown or shortened without a pause phase of more than a quarter of the lifespan (measured in minutes) of an individual MT. The percentage of dynamic MTs was calculated as the number of dynamic MTs divided by the total number of MTs. Dynamicity was calculated as the time spent growing and shortening divided by the total time measured.

FRSK cells stably expressing GFP–α-tubulin were loaded with 5 nM FM4-64 for 1 hour at 37°C and then washed extensively in DMEM. GFP–α-tubulin and FM4-64 were simultaneously excited (492DF18 filter; Omega Opticals). Their emissions were detected with a DualView setup (Optical Insights) with appropriate filters (dichroic filter 565cdxr, emission filter 670DF40) and projected onto the two halves of the CCD-EM chip (Hamamatsu). Image sequences were acquired in the time-lapse mode at 2.5- to 3-second intervals on an IX81 microscope (Olympus) using a 100× Olympus objective lens (numerical aperture, 1.4) and analyzed using the MetaMorph imaging software (Molecular Devices).

Translocation of GLUT1 in FRSK cells was observed as described previously (Nakayama et al., 2004). Double staining of FM1-43FX and β-tubulin was carried out according to a previous report (Tojima et al., 2007). Images were obtained using a confocal scanning laser microscope, FV-1000 (Olympus) that was equipped with a triple band-pass filter set.

Internalized and recycled GLUT1 in FRSK cells was visualized as described previously (Kamiguchi et al., 1998). For the internalization assay, cells were incubated with mouse monoclonal anti-GLUT1 (40 μg/ml) (R&D Systems) for 30 minutes at 37°C to allow for endocytosis of the IgG bound to GLUT1 on the PM. After rinsing at 4°C, the cells were fixed with 4% formaldehyde for 30 minutes. Because this fixation protocol did not permeabilize the cells, subsequent incubation with unlabeled anti-mouse IgG (400 μg/ml; Jackson Laboratory) for 30 minutes at 37°C specifically blocked the cell-surface IgG. Then, the cells were fixed again with 4% formaldehyde for 10 minutes to immobilize the unlabeled secondary antibody. After washing, the cells were permeabilized and blocked with 0.1% Triton X-100 and 10% horse serum in PBS for 30 minutes. Internalized GLUT1 was visualized by incubating the cells with Alexa-Fluor-488-conjugated anti-mouse IgG (1:100) for 30 minutes at 20°C. For analysis of endocytosed GLUT1 vesicles, the density of vesicles was calculated by counting positive signals existing in at least 10 rectangle regions in a transfected cell. The average density was calculated in at least 81 cells.

For double staining with endocytic vesicles, cells were cultured on a gridded slide dish. Mouse monoclonal anti-GLUT1 (40 μg/ml) along with 10 μM FM1-43FX (Invitrogen) were taken up by living cells for 30 minutes at 37°C, followed by destaining with DMEM containing 1 mM ADVASEP-7 (Biotium, Hayward, CA) for 10 minutes at 37°C. After cells were fixed with 4% formaldehyde, FM1-43FX images were obtained. GLUT1 IgGs on the surface of the cells were blocked with unlabeled anti-mouse IgG (400 μg/ml) and permeabilized with a 0.1% Triton X-100. After blocking with 10% goat serum, cells were visualized with an Alexa-Fluor-594-conjugated anti-mouse IgG. GLUT1 signals were then obtained.

For the recycling assay, cells were incubated with mouse monoclonal anti-GLUT1 (40 μg/ml) for 30 minutes at 37°C to allow for endocytosis of the GLUT1–IgG complex. The cells were cooled to 4°C and incubated with unlabeled anti-mouse IgG (400 μg/ml) for 30 minutes at 4°C to block the cell-surface IgG. After extensive washes at 4°C, the cells were incubated at 37°C for various periods in prewarmed 10% FCS–DMEM. This incubation allowed the cells to recover and proceed with the trafficking of endocytosed GLUT1 that had been tagged with the anti-GLUT1 IgG. The cells were then fixed with 4% formaldehyde, and recycled GLUT1 on the cell surface was detected by visualizing the anti-GLUT1 IgG that had not been blocked with the unconjugated secondary antibody. This was done by incubating the unpermeabilized cells with Alexa-Fluor-488-conjugated anti-mouse IgG (1:100) for 30 minutes at 20°C. Images were obtained using an ORCA-ER camera. In internalization and recycling experiments, Alexa-Fluor-488-conjugated secondary antibody did not recognize newly synthesized GLUT1 transported into the PM from the soma because it was not bound to the IgG. We operationally defined the cell recycled GLUT1 on the PM as that showing continuous GLUT1 signals on more than 12.5% of the cell circumference.

FCM analysis of internalized and recycled GLUT1 was carried out according to a previous study with slight modifications (Reddy et al., 2001). FRSK cells stably expressing plasmids encoding DsRed derivatives were used for FCM analysis. After the cells were trypsinized, floating cells were neutralized with excess FCS and incubated with mouse monoclonal anti-GLUT1 (40 μg/ml) for 30 minutes at 37°C to allow for endocytosis of the GLUT1–IgG complex. According to the immunocytochemical procedures for internalized and recycled GLUT1, the GLUT1 signal was detected with Qdot705-conjugated anti-mouse IgG antibody. Soaking, washing and fixation were conducted with TBS buffer. For confirmation of the GLUT1 signal, a portion of the prepared cells was visualized with Alexa-Fluor-488-conjugated anti-mouse IgG antibody and examined microscopically. Cells labeled with Qdot705 were analyzed using a FACSAria II (BD Bioscience). DsRed was excited with a 532-nm laser line (diode-pumped solid state laser) and its emission was collected through a 610/20 BP filter. Qdot705 was excited with a UV laser at 355 nm, and the fluorescence signal was collected through 670LP filter. The data were analyzed using FlowJo software (Tree Star; www.freestar.com).

The data are expressed as means ± s.e.m. and were analyzed statistically using Microsoft Excel and Prism software (GraphPad Software). P-values of less than 0.05 were considered to be significant.

We express our gratitude to M. Sanada (Kyorin University) for experimental assistance. We are greatly indebted to T. Tojima, H. Akiyama, R. Itofusa and K. Ohtawa (RIKEN-BSI) for advice on the measurement of MT dynamics and VT motility, measurement with the DualView system, histochemical experiments and FCM analysis, respectively.