VAMP8 is abundantly expressed in tissues with epithelial morphology
We and others previously identified VAMP8/endobrevin as a homolog of neuronal VAMP2 [17,18]. Northern blot analysis revealed that VAMP8 transcript is abundantly expressed in kidney and, to a lower level, in other tissues . This expression pattern suggested that VAMP8 may be important in a trafficking step specialized in polarized epithelial cells. In order to conduct a more detailed analysis of the expression and function of VAMP8, we generated affinity-purified rabbit polyclonal antibodies using the full-length VAMP8 protein lacking its carboxyl-terminal hydrophobic domain as an immunogen. These antibodies recognized a single band of 14 kDa in rat kidney postnuclear supernatant ( Fig. 1A). This band was eliminated by preincubating the antibodies with soluble recombinant VAMP8, but not with recombinant VAMP1 or VAMP4, indicating the specificity of the antibodies ( Fig. 1A). To analyze the protein expression pattern of VAMP8, affinity-purified polyclonal antibodies were used to detect VAMP8 protein in multiple tissues. In agreement with our northern blot results, we found VAMP8 abundantly expressed in kidney, as well as in lung and spleen with the highest expression level in lung ( Fig. 1B). In liver and testis, the VAMP8 protein is expressed at a somewhat lower level. In contrast, VAMP8 expression in brain and heart was significantly lower and only detectable upon longer exposure. In agreement with the observed tissue distribution, we also detected a single 14 kDa band in most cell lines examined, with its highest expression level in MDCK cells (data not shown). Thus, although the VAMP8 protein is present in most tissues and cell lines examined, VAMP8 seems to be abundantly expressed in tissues with epithelial morphology. This expression pattern suggests that VAMP8 is involved in a trafficking pathway common to most cell types, but likely serves a specialized function in epithelial cell types.
Figure 1. VAMP8 is abundantly expressed in tissues with epithelial morphology. A) Affinity-purified anti-VAMP8 antibodies preincubated with control buffer, recombinant VAMP1 or VAMP4 but not antibodies preincubated with recombinant VAMP8 recognize a 14 kDa protein in kidney postnuclear supernatant (PNS). B) PNSs from rat brain, heart, kidney, spleen, lung, liver, and testis (30 μg/lane) were analyzed by western blot using anti-VAMP8 polyclonal antibodies. A single band is detected at 14 kDa.
Download figure to PowerPoint
VAMP8 is apically targeted in nephric tubule epithelium and in MDCK cells
The abundant expression of VAMP8 in tissues and cell lines with epithelial morphology prompted us to further investigate its localization in polarized epithelial cell types. The renal epithelium represents an excellent example of a highly polarized epithelium crucial for ion homeostasis of the organism. To this end, we sectioned rat kidney and immunolabeled with affinity-purified rabbit antibodies to VAMP8 and antibodies to β-catenin, a basolateral membrane component of the cadherin–catenin adhesion complex . In these sections, VAMP8 immunoreactivity was restricted to the lumenal oriented, apical membranes of nephric tubular cells ( Fig. 2A). Cross-sections show that cuboidal epithelial cells of all tubules examined, regardless of the region of the nephron, are VAMP8 positive on their lumenal side. Interestingly, the surrounding kidney connective tissue and blood vessels seem to be devoid of VAMP8 staining ( Fig. 2A). As anticipated, the anti-β-catenin antibody selectively stained the lateral membranes of the tubule epithelium ( Fig. 2B,E). Merging both images clearly demonstrates the opposing subcellular localization of VAMP8 on the apical, and β-catenin on the basolateral membrane ( Fig. 2C).
Figure 2. VAMP8 is apically targeted in nephric tubule epithelium. Sections of rat kidney were double-labeled with rabbit polyclonal antibodies directed against VAMP8 (A) and syntaxin 13 (D) and with a monoclonal mouse antibody against β-catenin (B and E). Merged images are shown in (C and F). Note that anti-VAMP8 antibodies strongly labeled the lumenal/apical side of the ascending and descending nephric tubules. Scale bars, 15 μm.
Download figure to PowerPoint
When rat kidney sections were labeled for other ubiquitously expressed SNARE proteins, such as the endosomal syntaxins 7 (data not shown) and 13 ( Fig. 2D), we did not observe a strictly apical or basolateral staining pattern. Instead, the immunoreactivity for syntaxin 7 and 13 was distributed throughout the cytoplasm and concentrated in the perinuclear region, respectively. The apical localization of VAMP8 in nephric tubule epithelium suggests that this SNARE mediates an apical membrane transport event in polarized cells.
Next, we assessed the distribution of VAMP8 in filter-grown MDCK cells. VAMP8 in these polarized cells displays a diffuse staining pattern that is predominantly localized to the apical side of the cells ( Fig. 3A,D,G,J). Previous studies in non-polarized cells have shown that epitope-tagged, as well as endogenously expressed VAMP8 colocalizes with markers of the early endocytic compartment [17,18]. Double immunofluorescence of VAMP8 with the apical PM protein gp135  revealed a partial but not complete overlap ( Fig. 3B,C,E,F). VAMP8 is also apically localized in polarized CaCo-2 cells (data not shown). As seen in Fig. 3G, VAMP8 immunoreactivity was also associated with perinuclear membrane structures reminiscent of the staining patterns observed for endosomal markers such as TfR or syntaxin 13 (data not shown). In contrast, anti-E-cadherin antibodies stained lateral membranes, indicating that cell–cell contacts among the individual cells were fully established in these filter-grown cultures ( Fig. 3H,K). Little overlap was observed between the VAMP8 and E-cadherin staining ( Fig. 3I,L).
Figure 3. VAMP8 localization in MDCK cells. Polarized filter-grown MDCK cells were fixed, permeabilized and double-stained with rabbit anti-VAMP8 antibodies (A, D, G, and J) and mouse monoclonal antibodies to the apical PM protein gp135 (B and E) or to E-cadherin (H and K). Merged images are shown in (C, F, I, and L). Confocal images in the upper panels were acquired along the x-y axis of the cell monolayer. The x-z views, in the lower panels, were constructed by averaging sections over a line at each z position in 0.5 μm steps. Scale bars, 6 μm.
Download figure to PowerPoint
VAMP8 cycles apically in MDCK cells
To identify SNARE proteins which are involved in apical and basolateral cycling in MDCK cells, we established an in vitro assay that allowed us to monitor the cycling of VAMP and syntaxin molecules between intracellular compartments and the apical or basolateral PM. We first transiently transfected MDCK cells with different SNARE-GFP fusion constructs. All the fluorescent protein fusions had GFP at the carboxyl-terminus of the SNARE protein, which would result in the exposure of the GFP to the exterior of the cell if the SNARE reached the PM. Therefore, SNARE-GFP proteins that cycle through the apical or basolateral PM could bind anti-GFP antibody in the extracellular media. For all constructs, we confirmed that the localization of the expressed SNARE-GFP fusion did not differ from the localization of endogenous proteins. Transfected cells were plated onto Transwell tissue culture chambers at high density, and sufficient time was allowed for the formation of confluent monolayers of polarized cells. Anti-GFP antibody was then added to either the apical or basolateral side of the cells and allowed to be internalized.
Fig. 4 shows representative examples for apical or basolateral uptake of anti-GFP antibodies by VAMP8-GFP, rbet1-GFP, and VAMP3-GFP transfected cells. As outlined in Materials and Methods, we incubated cells for 1 h with anti-GFP antibody followed by a 1-h chase. Staining with secondary antibody revealed that anti-GFP antibody was taken up by VAMP8-GFP transfected cells from the apical but not from the basolateral side ( Fig. 4A–D). This uptake of antibody is specific, since untransfected cells did not internalize any antibody, and also the intracellular structures stained by the secondary antibody corresponded to SNARE-GFP fluorescence. In contrast, rbet1, a SNARE that mediates vesicular trafficking between the endoplasmic reticulum and the Golgi apparatus [22,23], did not appear to cycle through either the apical or basolateral PM ( Fig. 4E–H).
Figure 4. VAMP8 is involved in apical recycling. Filter-grown MDCK cells transfected with either VAMP8-GFP (A-D), rbet1-GFP (E-H), or VAMP3-GFP (I-L) were incubated with rabbit anti-GFP antibodies added to the apical (the two left columns) or basolateral (the two right columns) side of the Transwell chamber. After a 1-h chase, monolayers were fixed and internalized antibodies were visualized with Texas Red-conjugated anti-rabbit IgG. Scale bar, 15 μm.
Download figure to PowerPoint
VAMP3/cellubrevin localizes to recycling endosomes and has been previously implicated in TfR recycling in non-polarized cells [19,24]. Interestingly, a VAMP3-immunoglobulin fusion protein has been previously shown to recycle through the PM in an antibody-uptake experiment in non-polarized cells . In our recycling assay we found that VAMP3-GFP recycled through the apical PM, and to a lesser extent also through the basolateral PM, of polarized MDCK cells ( Fig. 4I–L). To rule out the possibility that basolateral uptake or recycling might be impaired in our experimental setup, we added fluorescently labeled human Tf to the basolateral side of MDCK cells stably expressing human TfR. In agreement with previous studies , Tf was efficiently internalized when added to the basolateral but not the apical side, indicating that access to the basolateral membrane was not markedly decreased in our culture conditions (data not shown).
To more quantitatively assess the rate of recycling, we cytofluorometrically determined the ratio between the fluorescence obtained for internalized anti-GFP antibody and the GFP fluorescence of our SNARE-GFP constructs. The average ratio obtained for apical uptake of GFP antibody from ten randomly picked cells transfected with VAMP8-GFP was 0.36±0.09, whereas the average basolateral uptake of VAMP8-GFP was 0.11±0.06 ( Fig. 5). When we omitted the anti-GFP antibody in control experiments, we obtained an antibody to GFP fluorescence ratio of 0.09±0.6, a value nearly identical to the number obtained for basolateral recycling of VAMP8-GFP. This background fluorescence is likely due to non-specific staining by the secondary antibody and/or by bleed-through from the GFP to the Texas Red detection channel. The ratios for apical and basolateral recycling of rbet1-GFP were 0.13±0.05 and 0.15±0.06, respectively ( Fig. 5). The cytofluorometric measurements for VAMP3-GFP revealed an apical ratio of 0.24±0.08 and a basolateral ratio of 0.19±0.02 ( Fig. 5). Although the value for apical recycling of VAMP3-GFP is lower than the apical VAMP8-GFP value, it is statistically above the apical rbet1 level as determined by a 2-tailed paired t-test (p=0.010). Likewise, basolateral uptake by VAMP3-GFP was also significantly above the basolateral levels for rbet1 (p=0.039) and VAMP8 (p=0.004).
Figure 5. Differential requirement of SNARE proteins in apical and basolateral recycling pathways in polarized MDCK cells. Filter-grown MDCK cells transfected with the indicated SNARE-GFP constructs were assayed for apical (Ap) or basolateral (Bl) recycling as described in Materials and Methods. The ratios between the fluorescence obtained for internalized anti-GFP antibody and the GFP fluorescence are given on the y-axis of each graph. Each ratio represents the mean (±standard deviation) of the values obtained from ten randomly picked SNARE-GFP expressing cells. In control experiments (Co), anti-GFP antibody was not added to the monolayers. The uptake of Oregon green 488-conjugated Tf is given in arbitrary units.
Download figure to PowerPoint
To extend our studies of SNARE proteins in apical and basolateral trafficking we tested the recycling of several other SNARE-GFP proteins in our anti-GFP antibody-uptake assay. Syntaxin 7 and syntaxin 13 have been shown to function in endosomal trafficking [27,28], whereas syntaxin 6 and VAMP4 are implicated in trans-Golgi network trafficking [29,30]. Apical as well as basolateral uptake of anti-GFP antibody was more heterogeneous for syntaxin 7-GFP than for all other constructs. Whereas eight out of ten syntaxin 7-GFP transfected cells displayed an apical uptake ratio of ∼0.1, the two remaining cells had an apical uptake ratio of ∼0.4. Likewise, three out of ten syntaxin 7-GFP transfected cells showed a basolateral uptake value of ∼0.3, whereas the remaining seven cells showed a basolateral uptake which was not markedly above background levels ( Fig. 5). This observed heterogeneity for syntaxin 7 recycling ratios does not correlate with expression levels and the source of the heterogeneity is not clear at this time. The apical and basolateral recycling values obtained for syntaxin 13-GFP, syntaxin 6-GFP, and VAMP4-GFP were not significantly above background levels, indicating that in polarized MDCK cells these SNARE proteins do not recycle through either PM domain at a rate detectable by our system ( Fig. 5). Taken together, we were able to observe clear differences between the rates at which various SNARE proteins recycle between internal compartments and the apical or basolateral PM domains. Most strikingly, VAMP8 was the only SNARE protein tested in this study which displayed a clear asymmetry between apical and basolateral cycling.
In this study, we present evidence that VAMP8/endobrevin, a SNARE protein localizing to early endosomes, cycles through the apical but not through the basolateral PM. In contrast, VAMP3 cycled at a lower rate through both PM domains in polarized epithelial cells. These morphological as well as recycling studies suggest that VAMP8/endobrevin may play a role in establishing or maintaining epithelial-cell polarity. Since VAMP8 expression is not restricted to polarized cell types it can be anticipated that VAMP8's trafficking function is common to most cell types. Perhaps the apical and basolateral recycling machinery exists in all cells and is only able to be differentially utilized in polarized cells. We favor a model in which VAMP8 operates in a vesicle trafficking step between early endosomes and the plasma membrane; i.e. the recycling of apical PM proteins such as certain classes of Na+/H+ exchangers . In this scenario, VAMP8 would be present on vesicles derived from early endosomes and upon arrival at the plasma membrane interact with a PM syntaxin. In preliminary experiments, we were unable to coprecipitate VAMP8-interacting SNARE proteins in sufficient amounts to determine their identity by microsequencing (unpublished observations). However, one attractive candidate is syntaxin 3, which has been localized to the apical PM domain in polarized epithelial cells .
Another VAMP-isoform, VAMP7 (also called TI-VAMP), has been previously implicated in apical trafficking [32,33]. Antibodies against VAMP7/TI-VAMP partially inhibited the delivery of hemagglutinin to the apical PM in MDCK cells . Another post-Golgi VAMP-isoform, VAMP3/cellubrevin, localizes to recycling endosomes and is suggested to function in TfR-recycling in non-polarized cells [19,24]. We found that VAMP3 recycles through both PM domains in MDCK cells, although at rates slower than that of VAMP8, suggesting that it functions in both apical and basolateral recycling.
In conclusion, we demonstrate differential trafficking for post-Golgi SNARE proteins in apical and basolateral cycling pathways in polarized MDCK cells. Polarized epithelial cells continue to be the classic model for understanding the asymmetrical distribution of proteins in cells and we are beginning to understand the molecular mechanisms that generate and maintain this asymmetry. Nevertheless, further work will be required to determine the precise step in endocytic cycling in which VAMP8 and VAMP3 function. In addition, the cargo proteins transported by VAMP8 and VAMP3 demarcated vesicles remain to be elucidated.