The TGN is most sensitive to redistribution by rab6A-GTP. To delineate the kinetics of Golgi redistribution by rab6, double-label immunofluorescence was carried out on NRK cells 4-6 h postinjection of DNA encoding rab6A-GTP. Retrograde movement seemed to occur in a stepwise redistribution occurring from TGN to medial- to cis-Golgi. Arrows indicate cells in which a sequential redistribution can be seen. Thus, rab6A-GTP stimulated the initial disappearance of immunostaining for TGN38 before p24γ3, which resides in the cis-Golgi (A) and before medial/trans-localized glycosyltransferase Mann-II (B). Consistent with this, Mann-II also was redistributed before p24 γ3 (C). Similarly, TGN38 and GalT staining disappeared before giantin (D and E). TGN38 (F), MannII (G), and golgin-84 (H) all redistributed before the cis-Golgi marker GM130. Bars, 10 μm.

Both rab6A' and rab6A induce transport of Golgi enzymes to the ER, despite that rab6A' does not bind rabkinesin6. HeLa cells stably expressing GalNAc-T2^GFP (A) or Mann-II^VSV (B) were injected with 80 ng/μl DNA encoding isoform rab6A-GTP or rab6A'-GTP and incubated for 7 h at 37°C. Cells were then fixed and labeled by indirect immunofluorescence for rab6 and Mann-II or in the case of GalNAc-T2^GFP, which can be visualized directly, only with rab6. Both rab6A-GTP- and rab6A'-GTP-expressing cells (asterisks) showed the ability to completely redistribute GalNAc-T2^GFP or Mann-II normally located in the Golgi (G) to the ER (arrows). An example of an intermediate phenotype used in the scoring presented below is indicated (arrowhead). (C) HeLa GalNAc-T2^GFP cells were injected in two separate regions on a scratched coverslip with 20 ng/μl DNA encoding rab6A-GTP or rab6A'-GTP. After 7 h of expression time, cells were fixed and immunostained with anti-rab6 antibodies that recognize both isoforms. Total rab6 fluorescence from individual cells (in arbitrary units) was measured using OpenLab software as indicated in Materials and Methods and plotted on the y-axis. The plot shows the correlation of expression level with the resulting phenotype which was categorized based on the redistribution pattern of the Golgi marker GalNAc-T2^GFP in that cell. Each dot represents the measurement from a single cell; the bar represents the median expression level from all cells within a classification. The number of cells in each category is expressed as a percentage, and the number of counted microinjected cells is indicated (n). (D) Cells microinjected with a rab6A-GTP-encoding expression plasmid were fixed and immunostained with the ER marker sec61β and viewed for GalNAc-T2^GFP. In the “Merge” image, orange indicates colocalization of sec61β and GalNAc-T2^GFP. Bars, 10 μm.

KIF3 is not required for rab6-induced Golgi-to-ER transport. (A) HeLa cells were treated with KAP3 siRNA or control duplexes for 48 h and then lysed in sample buffer. Western Blot analysis was performed with antibodies to KAP3 and α-tubulin for normalization of protein levels. KAP3 RNAi interference specifically decreases KAP3. (B) KAP3-depleted cells show a typical redistribution of GalNAcT2 ^GFP to the ER, 7 h after injection of DNA encoding rab6-GTP (asterisks).

Disruption of the dynein/dynactin complex inhibits rab6A-GTP-mediated ER accumulation of Golgi enzymes. The experimental procedure is as described in Figure 6, except a p50-encoding expression construct was microinjected into the nucleus of HeLa GalNAc-T2^GFP cells. After 24 h of expression, a rab6A-GTP plasmid construct was injected into the same cells. At 7 h postinjection, cells were fixed and processed for indirect immunofluorescence with antibodies against p50 and rab6. (A) Graphical depiction of such an experiment with levels of p50 protein shown by squares on the y-axis, rab6 levels by diamonds on the x-axis, and p50/rab6A-GTP-coexpressing cells by circles. Cells were quantified for the level of rab6 and/or p50 protein and scored for the presence of GaNAc-T2^GFP in the juxtanuclear Golgi or in scattered mini-Golgi in the case of p50-expressing cells (open symbols), partially in the ER (gray symbols), or totally localized to the ER (black symbols). Note that p50 overexpression results in the same scattered Golgi phenotype as produced by nocodazole treatment due to inhibition of the dynein motor. (B-D) Morphological validation of the quantitative data in A. HeLa GalNAc-T2^GFP cells (C) stained for rab6 (A) and p50 (D). Microinjected cells are labeled with asterisks in each panel. Cells expressing only p50 show the typical Golgi fragmentation phenotype that indicates disruption of the dynein/dynactin complex (drawing pin). At high levels of coexpressed p50 and rab6A-GTP, GalNAc-T2^GFP remains in the scattered Golgi (arrowhead), whereas cells expressing equivalent high levels of rab6-GTP only show GaNAc-T2^GFP relocalized to the ER (arrow). This figure is representative of three independent experiments. Bar, 10 μm.

Overexpression of rab6A disrupts the steady state of the TGN and perinuclear recycling compartments. HeLa cells expressing rab6A-GTP (asterisks) were analyzed by immunofluorescence microscopy by using antibodies to the indicated endogenous proteins. Noninjected cells serve as an internal control. (A-C) Cells were triple immunostained for TGN46, MPR46, and TfR. In rab6A-GTP-expressing cells (asterisks), TGN46 redistributed to an ER-like reticulum (A), whereas MPR46 and TfR colocalized in accumulations at extreme peripheral points of the cell (arrows in B and C). (D) Similarly to MPR46 and TfR, γ-adaptin also redistributed to these peripheral points. (E) In response to rab6A-GTP overexpression, early endosomes defined by the nonrecycling protein EEA-1 were found to be slightly larger and abnormally distributed away from the juxtanuclear region compared with the situation in control cells. (F) In contrast, rab6A-GTP had no effect on the late endosome/lysosome compartment defined by LAMP1 antibody staining. Bars, 10 μm.

Dynamics of Golgi fragments in rab6A-GTP-expressing cells after nocodazole removal. Time-lapse images (time indicated in hours:minutes) show the initial pattern of dispersed Golgi fragments in nocodazoletreated GalNac-T2^GFP stably transfected HeLa cells and their movement upon the repolymerization of microtubules in the presence and absence of rab6A-GTP. (A) Rab6A-GTP-expressing cells (box C) show a similar pattern of scattered Golgi elements as in control cells (box B) at t = 0. A few cells with particularly high levels of rab6A-GTP show partial localization of GalNAc-T2^GFP in the ER as well as in mini-Golgi (arrowhead). This shows that rab6-directed Golgi-to-ER transport can occur in the absence of microtubules but only inefficiently. (B) During nocodazole washout in control cells, the scattered GalNAc-T2^GFP labeled Golgi fragments cluster (t = 20 min) and net movement is inward (t = 1h) toward the juxtanuclear region denoted by N. After 2 h 40 min of incubation without nocodazole, most Golgi fragments are concentrated at the cell center. The outline indicates the cell periphery (C) In rab6-GTP-expressing cells, there is no recovery of the Golgi apparatus. Rather, the Golgi fragments (arrows) partially regroup (t = 20 min) but then melt into the underlying ER network, which becomes increasingly distinct (t = 1 h to 2 h 40 min). Bars, 10 μm.

Reduction of both rab6A and A' by RNA interference (RNAi) effects Golgi structure and transport. (A) Western blot demonstration of reduced total rab6 protein levels by two different siRNA sequences (labeled A/A'-siRNA1 and 2) or reduced rab6A or A' protein upon addition of the respective isoform-specific duplex (labeled A-siRNA and A'-siRNA). Whole cell extracts were prepared 92 h after repeated transfection with the appropriate siRNA and analyzed by Western blotting with anti-rab6 mAb. Levels of rab6 in cells treated with control-siRNA (Ctrl-siRNA), and levels of α-tubulin were not affected. (B) Suppression of individual rab6 isoforms by isoform-specific siRNA sequences. Cells treated with siRNA for 22 h were transiently transfected with plasmids encoding GFP-rab6A or GFP-rab6A' and prepared for immunoblot 28 h postplasmid transfection. The anti-GFP immunoblot shows the efficiency and specificity of each siRNA under these conditions. Thus, each GFP-rab6 isoform is not expressed when targeted by its siRNA and is expressed when incubated with siRNA against the other isoform. siRNA targeting both rab6A and A' inhibited expression from GFP-rab6A and GFP- rab6A' plasmids as expected. GFP-rab6 expression in cells treated with an unrelated control siRNA and α-tubulin were not affected. (C-J) Immunofluorescence images of cells expressing the Golgi marker GalNAc-T2^GFP after double treatment with the corresponding siRNAs for 92 h and immunostaining with rabbit anti-rab6 antibody. These illustrations were imaged with the same camera exposure settings, allowing a direct comparison of rab6 levels after RNAi treatment (C-F). The minor effect of reduced levels of rab6A (D and H) or A' (E and I) on Golgi structure. Reduction of rab6A and A' combined (F and J) affects overall Golgi morphology compared with control-treated cells (C and G). Arrowheads indicate the cells in which there was a major reduction in Golgi size. (K-N) Relocalization of GalNAc-T2^GFP to the ER after the introduction of Sar1p (GTP) is delayed in rab6 knockdown cells. Stably transfected GalNAc-T2^GFP cells treated with Ctrl-siRNA (K and L) or A/A'-siRNA1 (M and N) for 92 h were injected with DNA encoding dominant negative Sar1p (asterisks) and incubated for 4 h. Cells were then processed for double labeling with antibodies directed against rab6 (our unpublished data) and Sar1p (L and N) and observed by microscopy. Cells were analyzed for the amount of GalNAc-T2^GFP redistribution (K and M). Bar, 10 μm.

A rab6 and p50 binding fragment of BicD inhibits rab6A-GTP-directed Golgi-to-ER transport. (A) HeLa GalNAc-T2^GFP stably transfected cells were injected with an expression vector encoding mycBicD-C. After 24 h, the same cells were injected again with DNA encoding rab6A-GTP. Seven hours later, cells were fixed and stained with antibodies directed to myc-tag and rab6. Each injected cell was then scored for the level of BicD-C and rab6 protein and analyzed for the extent of Golgi-to-ER transport of GalNAc-T2^GFP. Rab6-GTP-expressing cells are represented by diamonds on the x-axis, BicD-C-expressing cells by squares on the y-axis, and double-injected coexpressing cells by circles. Using the same phenotypic classification of GalNAc-T2^GFP distribution as shown in Figure 4, an intact Golgi localization is represented by open symbols, partial GalNAc-T2^GFP in the ER by gray symbols, and full ER localization by black symbols. Note that exogenous expression of BicD-C alone does not affect Golgi integrity. (B-D) Representative image of HeLa GalNAc-T2^GFP cells (C) stained for rab6 (A) and BicD-C (D). Microinjected cells are labeled with asterisks in each panel. In the coexpressing BicD-C/rab6A-GTP cell (arrowhead), GalNAc-T2^GFP return to the ER is inhibited compared with the cell expressing rab6A-GTP alone, which shows the characteristic accumulation of GalNAc-T2^GFP in the ER (arrow). This figure is representative of three independent experiments. Bar, 10 μm.