SEARCH

SEARCH BY CITATION

Keywords:

  • Golgi apparatus;
  • membrane trafficking;
  • organelle homeostasis;
  • Rab6;
  • Rab33b;
  • Shiga toxin;
  • siRNA

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

We used multiple approaches to investigate the coordination of trans and medial Rab proteins in the regulation of intra-Golgi retrograde trafficking. We reasoned that medially located Rab33b might act downstream of the trans Golgi Rab, Rab6, in regulating intra-Golgi retrograde trafficking. We found that knockdown of Rab33b, like Rab6, suppressed conserved oligomeric Golgi (COG) complex- or Zeste White 10 (ZW10)-depletion induced disruption of the Golgi ribbon in HeLa cells. Moreover, efficient GTP-restricted Rab6 induced relocation of Golgi enzymes to the endoplasmic reticulum (ER) was Rab33b-dependent, but not vice versa, suggesting that the two Rabs act sequentially in an intra-Golgi Rab cascade. In support of this hypothesis, we found that overexpression of GTP-Rab33b induced the dissociation of Rab6 from Golgi membranes in vivo. In addition, the transport of Shiga-like toxin B fragment (SLTB) from the trans to cis Golgi and ER required Rab33b. Surprisingly, depletion of Rab33b had little, if any, immediate effect on cell growth and multiplication. Furthermore, anterograde trafficking of tsO45G protein through the Golgi apparatus was normal. We suggest that the Rab33b/Rab6 regulated intra-Golgi retrograde trafficking pathway must coexist with other Golgi trafficking pathways. In conclusion, we provide the first evidence that Rab33b and Rab6 act to coordinate a major intra-Golgi retrograde trafficking pathway. This coordination may have parallels with Rab conversion/cascade events that regulate endosome, phagosome and exocytic processes.

The Golgi apparatus is a central hub for membrane trafficking in the mammalian cell. It receives newly synthesized proteins and lipids from the endoplasmic reticulum (ER), modifies many of these by the addition of sugars, for example as they transit cis to trans through the Golgi apparatus, and finally sorts them to various destinations as they exit the Golgi. Together, these processes constitute forward or anterograde membrane trafficking (for review, see 1). At the same time, the Golgi is a hub for retrograde trafficking. At one extreme, the trans Golgi network (TGN), accessory transport-targeting proteins such as Wntless (2,3) or mannose 6-phosphate receptors (MPRs) are received from endosomes (4,5). Protein trafficking from endosomes to the TGN is necessary for normal metazoan development. TGN delivered Wntless is required for Wnt secretion that in turn triggers concentration-dependent responses that function in development from sea anemones to humans (for review, see 6). MPRs function in lysosomal enzyme delivery (for review, see 7). In addition, various proteins of unknown function such as TGN38/46 or the cis Golgi protein GPP130 cycle continuously between the Golgi apparatus, plasma membrane and endosomes (8,9). Furthermore, bacterial toxins such as Shiga and cholera that bind to glycolipids also traffic from endosomes to TGN, presumably, by piggybacking on normal cellular pathways (for review, see 10,11).

The Golgi apparatus itself is the site of considerable intra-Golgi, retrograde membrane trafficking that may well balance the anterograde transport of newly synthesized cargo proteins and lipids through the organelle. Resident proteins such as GPP130 are transported from cis to trans cisternae and recycled back via intra-Golgi retrograde transport (9). Bacterial protein toxins such as cholera toxin, pseudomonas toxin and Shiga toxin are transported in a retrograde manner through the Golgi apparatus before being exported to the ER (for review, see 10,11). If, indeed, secreted proteins are conveyed through the Golgi apparatus by cisternal maturation, then all Golgi resident proteins must be rapidly recycled to maintain their residency within the Golgi apparatus (for review, see 12). Considerable evidence suggests that intra-Golgi resident protein recycling is via coat protein I (COPI) vesicles (for review, see 1). However, contrary to expectation, the αCOP mutant ret1-1 that blocks COPI vesicle formation only partially inhibited cisternal maturation in yeast indicating, as suggested by the authors, that multiple intra-Golgi mechanism(s) may operate in the retrieval of resident Golgi proteins (13). Tubular connections have been shown to form between Golgi cisternae in mammalian cells (14,15) and in yeast Golgi cisternae may form transient direct contacts between each other (13). These may provide the alternate recycling routes. Finally, Golgi enzymes appear to cycle continuously on a slower timescale within the Golgi apparatus and between the Golgi and ER (for reviews, see 1,16).

Key to regulating the balance between retrograde and anterograde Golgi trafficking are Rab proteins, small GTPases of the Ras superfamily, that often act together in a coordinated manner (for review, see 17). Like other members of the Ras superfamily, Rab GTPases are membrane associated and active in effector recruitment in the GTP-bound state, and inactive cytosolic proteins in the GDP-bound state. Rab1 and Rab2 have been known since the early 1990s to regulate ER to Golgi trafficking (for review, see 18). Disruption of Rab1 in various ways blocks ER to Golgi trafficking and results in Golgi proteins redistributing to the ER, i.e. a brefeldin A-like phenotype (19,20). Rab9 is required for MPR recycling from late endosomes to Golgi (21). Rab6 or more precisely the Rab6a/Rab6a′ isoforms that differ in only three amino acids have been implicated in multiple Golgi trafficking pathways including MPR and Shiga toxin trafficking from endosomes to Golgi (4), COPI-independent Golgi to ER trafficking as exemplified by Shiga toxin and Golgi enzymes (22–24) and most recently, Golgi to plasma membrane trafficking (25). Knockdown of Rab6a/Rab6a′ levels inhibits Golgi enzyme recycling to the ER (24). Genetically, Rab6 interacts with the retrograde Golgi tether complexes, conserved oligomeric Golgi (COG) and Zeste White 10 (ZW10)/RINT-1 (for reviews, see 26,27); Rab6 depletion selectively suppresses COG or ZW10/RINT-1 knockdown induced Golgi disruption and vesiculation (28). The multiplicity of roles that Rab6 plays relative to Golgi trafficking may well reflect both the existence of distinct isoforms and the preferential location of Rab6 to the trans Golgi/TGN (28,29), the interface between the Golgi and the multiple trafficking pathways. In addition to Rab6, the medial Golgi Rab, Rab33b (30), has also been implicated in Golgi to ER cycling of Golgi enzymes (23,31). Rab33b is one of two members of the Rab33 subfamily and, in contrast to its neural and immune cell homolog, Rab33b, is universally expressed in mammalian tissues (32).

In the present study, we reasoned that medially located Rab33b might overlap functionally with the trans Golgi Rab, Rab6, to regulate Golgi homeostasis and coordinate intra-Golgi retrograde trafficking. We found that knockdown of Rab33b, like Rab6, suppressed COG- or ZW10-depletion induced disruption of the Golgi ribbon in HeLa cells. Moreover, efficient Rab6 induced relocation of Golgi enzymes to the ER was Rab33b-dependent while Rab6 depletion had little, if any, effect on GTP-Rab33b induced Golgi enzyme redistribution, suggesting that the two Rabs act sequentially in an intra-Golgi trafficking pathway. In support of this hypothesis, we found that the transport of Shiga toxin from the trans to cis Golgi and ER required Rab33b. To probe for a possible mechanistic basis for a linkage between the two Rabs, we tested if overexpression of GTP-Rab33b might induce dissociation of Rab6 from Golgi membranes prior to any significant effect on Golgi enzyme distribution. This is the predicted outcome (33) if the downstream Rab, Rab33b, recruits a GAP (GTPase-activating protein) for the upstream Rab, Rab6. We found a twofold decrease in Golgi associated Rab6 in such an experiment. Surprisingly, depletion of Rab33b had little, if any, immediate effect on cell growth and multiplication. Furthermore, the Golgi apparatus supported anterograde trafficking of tsO45G protein through the organelle. We conclude that Rab33b and Rab6 overlap functionally in regulating Golgi homeostasis relative to COG and ZW10 and may well act sequentially to coordinate a retrograde Golgi trafficking pathway that can be uncoupled from anterograde transport.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

The efficacy of four siRNAs targeted against human Rab33b was tested. Of these, siRNAs corresponding to coding nucleotides 384–403 [siRab33b(4)] and 418–439 [siRab33b(5)] were the most effective in lowering Rab33b protein levels. As shown in Figure 1A, a minor drop in Rab33b protein levels was apparent as early as 1 day post siRab33b(5) transfection with more substantial decreases apparent at 2 and 3 days posttransfection. Densitometric measurements indicated that Rab33b protein levels were depressed at least 80% 3–4 days post initial transfection. In striking contrast, knockdown of Rab33b proteins levels had no immediate effects on the growth and multiplication of HeLa cells (Figure 1B). Cell morphology was unperturbed by phase-contrast microscopy and no statistically significant differences in cell generation time were observed up to 3 days posttransfection whether the cells were exposed to one or two transfection cycles with the siRNA (Figure 1B). Only with longer time intervals posttransfection, e.g. 4 days, were significant effects on cell multiplication observed (Figure 1B). Four days posttransfection cell number was 50–80% of control (mock transfection). However, even at this time point, the cells remained well spread with no change in morphology by phase-contrast microscopy (data not shown). In sum, the data suggest that cell multiplication and accompanying anterograde trafficking through the Golgi apparatus necessary to support normal cell multiplication are not tightly coupled to levels of Rab33b, a known regulator of Golgi retrograde trafficking (23,31).

image

Figure 1. Treatment of HeLa cells with siRNAs directed against Rab33b(5) effectively lowered Rab33b protein levels with no immediate effects on the growth and multiplication of HeLa cells. HeLa cells stably expressing GalNAcT2-GFP were transfected with either nontargeting siRNA (Control) or siRab33b(5) at a concentration of 200 nm in the absence of fetal bovine serum for 4 h and then cultured for 96 h. To achieve maximal knockdown, two cycles of siRNA transfections were performed at 24 h (Day 0) and 48 h (Day 1) post cell plating. (A) To obtain equivalent cell yields, cell numbers plated per dish were increased appropriately for 0, 24 and 48 h. Seventy thousand cells were plated in each 35 mm tissue cell culture dish for 72 h and 96 h. Total cell lysates were collected at 0, 24, 48, 72 or 96 h posttransfection. Western blotting using human Rab33b D5 monoclonal antibody showed substantial knockdown of Rab33b relative to β-tubulin (as control) at days 2 and 3 posttransfection with a small reduction in protein levels seen at 1 day posttransfection. (B) For quantifying effect of siRab33b treatment on cell multiplication, 70,000 cells were plated in each 35 mm tissue culture dish. Manual cell counts of randomly selected cell fields on each day indicated that growth and multiplication of HeLa cells were not immediately affected by one or two siRab33b transfection cycles. Only at 4 days were significant effects on cell multiplication observed.

Download figure to PowerPoint

We next tested for an involvement of Rab33b in Golgi homeostasis. As a starting point, we hypothesized that depletion of the medial Rab, Rab33b, might share Golgi phenotypic effects to those observed for knockdown of the trans Golgi Rab, Rab6, as both are known regulators of retrograde Golgi trafficking (23,24,31,33). As shown previously, Rab6 interacts genetically with two putative tether complexes in Golgi retrograde trafficking, COG and ZW10/RINT-1 (28 and references therein). In epistatic knockdown experiments, both COG3- and ZW10/RINT-1-depletion induced disruption of Golgi organization was suppressed by Rab6 knockdown (28). In similar experiments, we found that knockdown of Rab33b suppressed both COG3- and ZW10-depletion induced disruption of Golgi organization (Figure 2, 4 days post initial siRNA treatment; Figure 3, 3 days post initial siRNA treatment). Knockdown of Rab33b produced a Golgi ribbon that was, if anything, more continuous as indicated by Golgi segments per cell (Figure 2C,F). As indicated by immunofluorescence staining, the penetrance of the Rab33b knockdown was high (Figure 2G). Furthermore, taking the alternate approach of overexpression of a dominant negative mutant, we found that GDP-restricted Rab33b significantly suppressed ZW10-depletion induced Golgi disruption; 64% incidence of compact, juxtanuclear Golgi apparatus versus 11% in ZW10-depleted cells not overexpressing GDP-restricted Rab33b. In an important control experiment, we observed qualitatively that knockdown of Rab33b had no detectable effect on the association of Rab6 with the Golgi (data not shown). Moreover, we have shown previously that depletion of an endosomal Rab such as Rab5 has no effect on COG- or ZW10-knockdown induced Golgi disruption (28). We conclude that Rab33b does, indeed, have an important and specific role in Golgi homeostasis and may well act either in parallel or in series with Rab6 in maintaining Golgi organization and homeostasis.

image

Figure 2. Treatment of NAGT-1-myc cells with siRNAs directed against COG3 and Rab33b suppressed COG3-depletion induced Golgi disruption. (A–D) NAGT-1 myc cells double-treated with Rab33b + COG3 siRNAs show compact Golgi phenotype. NAGT-1 myc cells were mock (A) or COG3 (B), Rab33b (C) and double (Rab33b + COG3) (D) treated with two cycles of siRNA for 96 h and fixed. Cells were stained with anti-myc antibody. Micrographs were taken with a 63×/1.4 numerical aperture objective under identical exposure conditions for both mock cells and cells treated with siRab33b, siCOG3 or siRab33b + COG3. All micrographs shown in (A–D) are single-plane projections of full cell depth, confocal image stacks. (E) Total cell lysates from (A–D) were immunoblotted with anti-COG3 (upper row), anti-Rab33b (middle row) and anti-tubulin (lower row) antibodies. (F) Knockdown of Rab33b produced a Golgi ribbon that was more continuous than mock as indicated by the average number of Golgi segments per cell. (G) Penetrance of Rab33b knockdown was high (a). Mock cells stained with Rab33, D5 monoclonal antibody illustrate compact Golgi phenotype. (b) siRab33b treated cells stained with Rab33b D5 monoclonal antibody shows negligible staining indicating that Rab33b expression in the cells is reduced. C, bar = 10 µm; Gb, bar = 20 µm.

Download figure to PowerPoint

image

Figure 3. Rab33b depletion suppressed ZW10 knockdown induced Golgi dispersal. HeLa cells stably expressing GalNAcT2-GFP were either (A) Mock transfected or transfected with (B) ZW10(102) siRNA, (C) Rab33b siRNA(4 or 5, 5 shown), or (D) Rab33b siRNA(4 or 5, 5 shown) plus ZW10(102) siRNA. The concentration of each siRNA was 200 nm. After 72 h, cells were fixed and analyzed by confocal microscopy, 63× objective, maximum intensity projection stacked view, for the distribution of GalNAcT2-GFP. Immunoblotting with anti-ZW10 (E) and anti-Rab33b antibodies (F) were done to evaluate the corresponding protein knockdown efficiency. α-Tubulin was used as a loading control. Similar results were observed with Rab33b siRNA(4 or 5). Cells double-treated with Rab33b + ZW10(102) siRNAs showed a compact Golgi phenotype indicative of phenotype suppression. Bar = 15 µm.

Download figure to PowerPoint

We hypothesized that the medial Rab, Rab33b, might act downstream of more trans Rab, Rab6, in a sequential (serial) intra-Golgi retrograde trafficking pathway. If so, depletion of Rab33b should inhibit GTP-Rab6 induced redistribution of Golgi enzymes to the ER. To test this, GTP-Rab6 was overexpressed in control and siRab33b(5) GalNAcT2-GFP HeLa cells (3 day knockdown period) and the redistribution of GalNAcT2-GFP to the ER assessed. As shown in Figure 4, treatment of HeLa cells with siRab33b(5) inhibited by ∼50% the GTP-Rab6 induced redistribution of GalNAcT2-GFP into the ER. When the experiment was reversed and the effect of Rab6 depletion on the slow GTP-Rab33b induced redistribution of Golgi enzymes to the ER (23,31) was assayed, there was little, if any, inhibition observed (data not shown). At moderate overexpression levels, GDP-Rab33b has little to no effect on Rab6 induced redistribution of Golgi enzymes (23), presumably indicating that GDP-Rab33b is a poor competitor for the needed guanine nucleotide exchange factor. Consistent with this, only at very high overexpression levels does GDP-Rab33b inhibit GalNAcT2-GFP redistribution to the ER (31). Overall, these results suggest that a significant portion of Rab6-induced Golgi enzyme relocation to the ER is the result of sequential Rab-dependent intra-Golgi trafficking. However, as inhibition is far from complete, there may well be a second pathway. Young et al. have presented evidence for a trans-first Rab6-regulated pathway from the Golgi to proximal trans Golgi-inserted ER (24). To probe for a possible mechanistic linkage between Rab33b and Rab6, we tested whether overexpression of GTP-restricted Rab33b might affect the level of Golgi associated Rab6. Novick et al. (33) have presented evidence of a Rab GAP cascade in which the downstream Rab, in this case Rab33b, recruits a GAP for the upstream Rab, in this case Rab6. Consistent with this possibility, when GTP-restricted Rab33b was expressed briefly in HeLa cells, there was a 49% decrease in Rab6 associated with Golgi membranes prior to any significant relocation of a Golgi glycosyltransferase to the ER (Figure 5).

image

Figure 4. Rab33b knockdown suppressed GTP-restricted Rab6 induced redistribution of Golgi proteins to the ER. HeLa cells stably expressing GalNAcT2-GFP were either transfected with control siRNA or with Rab33b siRNA(5). After 72 h, cells were microinjected with diluted plasmid DNA encoding a GTP-restricted Rab6a mutant protein. Following a 6 h expression period, cells were fixed, stained with anti-Rab6 antibody (Cy3 second antibody) to identify injected cells, and the distribution of GalNAcT2-GFP between the Golgi apparatus and ER scored as previously described (23). At lower stock plasmid concentrations (10–20 ng/µL), GTP-Rab6 induced redistribution of GalNAcT2-GFP into the ER was inhibited by ∼ 50% in Rab33b depleted cells. The smooth curve plot function of Kadiagraph software was used to plot siControl results, a functional equivalent to using a french curve. Examples of Rab induced GalNAcT2-GFP redistribution to the ER have been published previously (23).

Download figure to PowerPoint

image

Figure 5. GTP-Rab33b overexpression displaces Rab6 from Golgi membranes prior to any major redistribution of Golgi glycosyltransferase to the ER. HeLa cells stably expressing GalNAcT2-GFP (note some cell to cell variation in expression level (A,B) were microinjected with plasmids encoding GTP-restricted Rab33b (B,D,F, 20 ng/µL stock plasmid concentration) and after a 6 h expression period were fixed and stained for antibody Rab6 (C,D, Cy3 second antibody) or Rab33b (Cy5 second antibody). Protein distributions in control (A,C,E) and GTP-restricted Rab33b expressing cells (B,D,F) are shown either qualitatively (A–D) or quantitatively (XY intensity plots, E,F). All cells in B,D,F are overexpressing GTP-restricted Rab33b. Quantitatively, there was a twofold to threefold decrease in Rab6 staining with GTP-Rab33b expression. Images are maximum intensity proteins of confocal image stacks. A representative field of cells is shown. Bar = 10 µm.

Download figure to PowerPoint

As a third test for retrograde Golgi trafficking levels in Rab33b depleted cells, we used internalized Shiga-like toxin B fragment (SLTB) as a tracer for intra-Golgi trafficking. SLTB is the targeting subunit of Shiga-like toxin. It binds to the glycolipid GB3 at the cell surface, is internalized into endosomes, is later transferred from endosomes to the Golgi apparatus, and subsequently accumulates in the ER. To synchronize SLTB accumulation in endosomes, wild-type (WT) HeLa cells were incubated with Cy3-SLTB at 4°C and then transferred to 19.5°C, a permissive temperature for internalization, but nonpermissive temperature for subsequent transport of Cy3-SLTB from endosomes to the Golgi apparatus (9). Cells were three color labeled for p115 (cis Golgi protein, green), Rab6 (trans Golgi protein, Cy5) and Cy3-SLTB (red). As shown in Figure 6A–D, Cy3-SLTB accumulated in endosomes at 19.5°C irrespective of siRab33b treatment (3 day knockdown period, Figure 7E blotting control). With shift to permissive conditions, 37°C, 20 min chase, Cy3-SLTB accumulated juxtanuclearly in the Golgi apparatus (Figure 6E–H) as indicated by the distributions of Rab6 and p115, trans and cis Golgi markers, respectively, that separate slightly from one another (see Figure 6L′ for higher magnification view of * proximal area). Importantly, the intensity of network-like, ER-type cytoplasmic staining was decidedly higher in the cells treated with siCONTROL1 RNA (Figure 6E versus G) than in Rab33b depleted cells where no ER accumulation was apparent. With further incubation at 37°C, 60 min chase, the level of ER associated Cy3-SLTB continued to increase in the control siRNA treated cells while little to no ER associated SLTB signal was apparent in siRab33b treated cells (Figure 6I–L). In both cases, consistent with previous results (22), significant Golgi associated SLTB remained.

image

Figure 6. Rab33b depletion suppressed the transport of SLTB from the Golgi to ER. Wild-type HeLa cells were either transfected with control siRNA or Rab33b siRNA(5). After 72 h cells were incubated with Cy3-SLTB at 4°C for 30 min followed by a 1 h incubation at 19.5°C to accumulate Cy3-SLTB (red) in endosomes. The temperature block was then released. Cells were fixed at the end of the endosome accumulation period (0), 20 and 60 min post temperature block release. Cells were antibody stained for the Golgi proteins p115 (green) and Rab6 (blue). (A–D) 0 min; (E–H) 20 min post temperature release; (I–L) 60 min post temperature release. (A,E,I) SLTB, control, gray-scale. (B,F,J) SLTB, p115 and Rab6, control, 3-color. (C,G,K) SLTB, siRab33b, gray-scale. (D,H,L) SLTB, p115, Rab6, siRab33b, 3-color. L′, 2× blow-up of area next to asterisk in L showing a linescan example. SLTB accumulated in the juxtanuclear Golgi apparatus 20 min post temperature release in either control or siRab33b cells. In control cells, there was subsequent SLTB accumulation in the ER as indicated by increased cytoplasm brightness. However, in the siRab33b cells, the cytoplasm was comparatively dark indicating that ER accumulation of SLTB was inhibited. Bar = 10 µm.

Download figure to PowerPoint

image

Figure 7. Rab33b depletion suppresses the intra-Golgi transport of SLTB from trans to cis Golgi. Wild-type HeLa cells were treated with control and Rab33b siRNAs and incubated with Cy3-SLTB as described in caption of Figure 6. As shown in Figure 6L′, images were linescanned perpendicular to the juxtanuclear Golgi staining to determine the relative distribution of the cis marker, p115, the trans marker, Rab6 and Cy3-SLTB. At each individual time point (A–D), the pixel intensity versus distance of p115 (•), Rab6 (□) and SLTB-Cy3 (▴) was plotted and scored for peak intensity position. (E) Immunoblot showing Rab33b knockdown. The mean distance between the peak intensity of Rab6 and p115 and that of Rab6 and Cy3-SLTB was determined for each time point and condition post temperature release. (F) Distance between peak intensities, control; (G) Distance between peak intensities, Rab33b siRNA. N > 40 for each determination. Golgi associated SLTB peak intensity shifted to near correspondence with p115 in the control case but showed little, to no, shift in the Rab33b siRNA case indicating a strong inhibition of intra-Golgi trafficking of SLTB.

Download figure to PowerPoint

Based on the above results, Rab33b is a strong candidate for modulating intra-Golgi retrograde trafficking. If so, the intra-Golgi distribution of SLTB should be profoundly affected by Rab33b knockdown. To test this prediction, we quantified the distribution of Golgi associated SLTB relative to Rab6 and p115 during a chase sequence in WT and Rab33b depleted HeLa cells (see Figure 6). We used a laser confocal linescan approach (9,34) in which a vector is placed perpendicular to the Golgi ribbon (e.g. Figure 6L′) and the intensity profile of individual proteins determined in multilabeled samples at each pixel along the vector (Figure 7). As shown in the example in Figure 7A,B, the intensity profile of Golgi associated SLTB at a 20 min chase time corresponded closely to that of the trans marker, Rab6, in both siCONTROL1 and siRab33b treated cells. Later, at a 60 min chase time (Figure 7C,D), the intensity profile of Golgi associated SLTB in the siCONTROL1 treated cells now corresponded closely to that of the cis marker, p115, while that in siRab33b treated cells was found to approximate closely that of Rab6 indicating that in the control case SLTB had trafficked in a retrograde manner from trans to cis, while in the Rab33b knockdown intra-Golgi retrograde trafficking was strongly inhibited. To quantify this effect, the midpoint profile distributions were compared for 40+ Golgi examples from control and siRab33b treated cells. The trans to cis distribution of SLTB was found to change significantly in control cells but not in siRab33b cells (Figure 7F,G). In siRab33b treated cells, Golgi associated SLTB remained trans associated, consistent with the strong inhibition of SLTB transfer into the ER in siRab33b HeLa cells (see Figure 6).

As intra-Golgi retrograde trafficking is significantly inhibited under conditions that have at most limited effect on cell growth and multiplication, there may well be significant uncoupling between retrograde and anterograde trafficking through the Golgi apparatus. As an explicit test for anterograde trafficking, we compared transport from the ER to cell surface for tsO45G protein in control and Rab33b knockdown cells. tsO45G protein is a temperature sensitive form of VSV-G protein. tsO45G folds abnormally at 39.5°C and is retained in the ER at this nonpermissive temperature. When cells are shifted to permissive temperature, 32°C, the protein folds normally and is transported from the ER to Golgi apparatus and from there to the cell surface. As shown in Figure 8, the redistribution pattern of tsO45-GFP tagged following temperature release was normal in cells 3 days post initial siRab33b treatment. Qualitatively, substantial transport of tsO45 to the cell surface was observed in cells 4 days post siRab33b treatment (Figure 8B,C). The Rab33b-depleted Golgi apparatus is competent to support anterograde trafficking through the organelle.

image

Figure 8. Rab33b knockdown has little, if any, effect on the redistribution pattern of a membrane protein from the ER to the cell surface. Wild-type HeLa cells were transfected with either scrambled siRNA (siControl, A,C) or siRab33b(5, B,D) at a concentration of 200 nm for either 3 or 4 days (A–D, two transfection cycles in each case) before plasmids encoding tsO45-GFP (20 ng/µL) were microinjected into the nuclei. tsO45-GFP accumulated in the ER at 39.5°C before being trafficked to the Golgi apparatus and plasma membrane at 32°C. Redistribution of tsO45-GFP from the ER (A,B) to the Golgi and cell surface is qualitatively visible following shift to permissive temperature in siControl and siRab33b treated cells (C,D). (E) Following shift to permissive temperature, 32°C, the frequency of cells with either ER (red), Golgi (blue) or surface (green) tsO45-GFP distribution pattern was scored at 0, 40, 85 and 120 min for both siControl (open symbols) and siRab33b (closed symbols) transfected cells. Quantification shown for 3 days posttransfection and qualitative images were examples from 4 days posttransfection. All images were deconvolved before distribution pattern scoring. Bar = 20 µm.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

We investigated the hypothesis that the small GTPases, Rab6 and Rab33b, located in the trans Golgi/TGN and medial Golgi, respectively, are functionally overlapping regulators of Golgi organization, i.e. homeostasis. We found that depletion of Rab33b, like that of Rab6 (28), epistatically suppressed the Golgi fragmentation effects of siRNA knockdown of the retrograde Golgi tether proteins, COG3 and ZW10. We propose that this functional overlap is a consequence of complementary roles in Golgi membrane trafficking. In separate assays, Rab33b depletion inhibited GTP-Rab6 induced redistribution of Golgi enzymes to the ER, but not vice versa, and the intra-Golgi trafficking of Shiga toxin. To provide in vivo evidence for a possible mechanistic link between the two Rabs, we tested if overexpression of GTP-Rab33b might induce the displacement of Rab6 from Golgi membranes prior to any major effort on the distribution of Golgi glycosyltransferases. This would be the expected effect if Rab33b were the downstream Rab that recruited a Rab6 GAP (e.g. 35). An ∼50% decrease in Golgi associated Rab6 was indeed observed. In sum, we conclude that Rab33b and Rab6 overlap functionally in regulating Golgi homeostasis and suggest that this may be a consequence of the two regulating sequential steps in Golgi retrograde trafficking.

We chose Rab33b expression as a lever to characterize Golgi homeostasis and intra-Golgi retrograde trafficking relationships because of its medial location in the organelle (30,32). By virtue of its location, it would be expected to be much Golgi-centric in its action. To date, only three Rab proteins, Rab33b (23,31), Rab6 (22–24,33) and Rab18 (30), have been implicated in intra-Golgi retrograde trafficking. The precise Golgi localization of Rab18 is presently unknown while Rab6 is preferentially localized at the trans Golgi/TGN interface between the organelle and plasma membrane/endosomes (29,30, present work). Rab6 has also been implicated in endosome to Golgi (4) and Golgi to plasma membrane trafficking (25). We reasoned that depletion of the medial Rab33b should provide an intra-Golgi specific link between membrane trafficking and any epistatic suppression effects. In stating this reasoning, we note that Rab33b has been recently implicated in other pathways, i.e. autophagosome formation through interaction with Atg16L (36,37).

Consistent with the expectation that Rab33b is a significant regulator of Golgi homeostasis and intra-Golgi retrograde trafficking, we found Rab33b depletion (i) suppresses Golgi disruption induced by knockdown of either of the two retrograde Golgi tethers, COG and ZW10, (ii) inhibited, albeit partially, GTP-restricted Rab6 induced retrograde trafficking of Golgi enzymes to the ER and (iii) blocked the intra-Golgi trafficking of SLTB, a marker for constitutive retrograde trafficking. Rab6 induced redistribution of GalNAcT2-GFP to the ER was inhibited with no apparent loss of Golgi associated Rab6 indicating that the two Rabs may well be acting sequentially, trans-to-medial, in a Rab conversion/cascade, and moreover that a substantial portion of Rab6-dependent Golgi to ER trafficking must be intra-Golgi (compare also, 24,38). Consistent with a possible in vivo Rab GAP cascade, overexpression of GTP-Rab33b induced the displacement of Rab6 from Golgi membranes prior to any significant redistribution of Golgi glycosyltransferases to the ER. The possibility of a Rab cascade in which Rab33b acts to recruit a Rab6 GAP is important both functionally and structurally as a way to limit the distribution of Rab6 into medial Golgi cisternae. In the GDP state GAP induced activation, Rab6 associates with a Rab GDI and is no longer Golgi associated. Such principles are clearly important in yeast exocytosis (for review, see 17) and likely important for the mammalian Golgi apparatus. Yeast have fewer Rabs, and applicability to the yeast Golgi remains to be shown. We conclude that Rab33b is a bona fide regulator of intra-Golgi retrograde trafficking that may well act sequentially (i.e. in series) with Rab6 and could conceivably regulate the intra-Golgi distribution of Rab6.

Contrary to the simplest predictions of a cisternal maturation/progression model of Golgi function in secretion, inhibition of Golgi retrograde trafficking through Rab33b depletion had little, if any, immediate affect on cell multiplication or Golgi competency in anterograde trafficking. With a standard 3 day Rab33b knockdown, tsO45G protein was transported normally from the ER to the Golgi apparatus and from the Golgi apparatus to the cell surface upon shift to permissive temperature. Cell number was normal. Only with the more prolonged 4 day Rab33b knockdown was any inhibition of cell multiplication noted. tsO45G transport to the cell surface was still rapid. Presumably, the delayed onset of growth inhibition is an indirect consequence of the retrograde Golgi trafficking defect. We propose that the Rab6/Rab33b regulated pathway of retrograde Golgi enzyme and toxin trafficking must exist in parallel with other Golgi retrograde trafficking pathways.

In conclusion, we provide the first in vivo evidence that Rab33b and Rab6 functionally overlap in the regulation of Golgi homeostasis. Genetically, both are upstream of COG and ZW10 tether proteins. Likely, the two coordinate a major intra-Golgi retrograde trafficking pathway. This coordination may have parallels with Rab conversion/cascade events that regulate endosome (39), phagosome (40) and exocytic processes (41) and the distribution of ‘cognate’ Rabs between organelles. Finally, our results point to future experiments to establish the direct biochemical sequence of steps involved. Rab33b may well recruit a Rab6 GAP as part of a Rab GAP cascade to limit the redistribution of Rab6 to the medial Golgi. These experiments may well prove to be a difficult task as the Rab6 GAP has been elusive with no candidate protein being identified on the basis of TBC motif homology (42,43).

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Cell culture

WT HeLa cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (Atlanta Biologicals) in a humidified incubator at 37°C and 5% CO2. HeLa cells stably expressing GalNAcT2-GFP (44) were maintained in complete culture media in the presence of 0.45 mg/mL of Geneticin sulfate (Sigma-Aldrich). For expression of tsO45G protein, WT HeLa cells were cultured at 32°C to produce accumulation of the protein in the ER and then transferred to media pre-warmed to 39.5°C, conditions permissive for anterograde transport of tsO45G to the cell surface.

Antibodies

Affinity purified rabbit polyclonal antibodies directed against ZW10 were prepared as described previously (27). Affinity purified anti-COG3p was prepared as described previously (45). Mouse monoclonal anti-p115 was produced as described (46). Several antibodies were purchased commercially: rabbit anti-Rab6 (C-19 peptide) was from Santa Cruz Biotechnology; mouse antibodies directed against Rab33b were either from Abnova or Frontier Science Co., Ltd.; rabbit anti-α-tubulin was from Sigma-Aldrich; Cy2-, Cy3- or Cy5-conjugated donkey anti-mouse or anti-rabbit immunoglobulin (IgG) antibodies were from Jackson ImmunoResearch Laboratories; Hilyte 647-conjugated, highly cross-absorbed, goat anti-rabbit secondary antibodies were from AnaSpec.

siRNA treatment

Dharmacon RNA Technologies manufactured all small interfering RNAs (siRNAs). The siRab6 (28, depletes both Rab6a and Rab6a′), siZW10(102, 47) and siCOG3 (52) sequences have been published previously. The sense sequences of the four Rab33b siRNAs were (Dharmacon 1): CACAAACCAUUAAUGCUUAUU; (2): GACCAACAUGGCUAGUUUUU; (4): GAUAGAAGAAUGCAAACAAUU; (5): GAUAUACCACGGUUCUUGUU starting at nucleotide positions 604, 348, 384 and 418 of the Rab33b coding sequence, respectively. Control siRNAs were siCONTROL1, Dharmacon siGenomics nontargeting siRNA 1 (UAGCGACUAAACACAUCAA). siRNA duplexes were transfected at a final concentration of 200 nm, using Oligofectamine (Invitrogen) according to the manufacturer's protocol with minor modifications (24,27). For double siRNA transfections, cells were transfected with a mixture of the siRNA duplexes with the concentration of each duplex at 200 nm. In some cases, two cycles of siRNA transfections were done to achieve maximal knockdown (24,28). For the second cycle of siRNA transfections, cells were treated with the corresponding siRNAs typically 24 h after the initial transfection.

For quantification of the effect of siRNA treatment on cell multiplication, six randomly selected cell fields were photographed with a 10× phase-contrast objective every 24 h. Cells were plated at a density of 70,000 per 35 mm tissue culture dish. The number of cells per field was counted manually. Cells were collected at 0, 24, 48, 72 or 96 h posttransfection for western blotting or immunofluorescence staining. For immunoblotting experiments, cell numbers plated per dish were increased appropriately for growth periods shorter than 72 h to give equal cell yield.

Microinjection

An Eppendorf FemtoJet/InjectMan NI 2 microinjection system was used as described previously (23,48). For Rab6 and Rab33b overexpression experiments, GalNAcT2-GFP HeLa cells were used. Plasmids encoding GTP-Rab6a were microinjected into the cell nucleus 72 h post siRNA treatment. The plasmid concentration range used was based on preliminary experiments to optimize conditions. The Rab6 expression period was 6 h at 37°C while plasmids encoding GDP-restricted Rab33b (stock plasmid concentration, 100 ng/µl) were expressed for 16 h as previously described (23). GTP-Rab33b (stock plasmid concentration 20 ng/µl) was expressed for either 6 or 16 h. After 6 h expression, GTP-Rab33b has little, if any, effect on the distribution of GalNAcT2-GFP between the Golgi apparatus and ER. After 16 h expression, GalNAcT2-GFP was redistributed to the ER. Cells were fixed and stained with respective Rab antibodies to identify Rab overexpressing cells. The distribution of GalNAcT2-GFP was then scored in the Rab overexpressing and control cell population. For tsO45-GFP expression experiments, WT HeLa cells were used and the length of siRNA treatment was 72 h. Injection into the nucleus was at room temperature and cells were subsequently incubated at 39.5°C to accumulate tsO45G-GFP in the ER and then at 32°C in the presence of cycloheximide to chase newly synthesized tsO45G into the Golgi apparatus and plasma membrane (23,49). Following deconvolution with Huygens Professional software (Scientific Volume Imaging), tsO45G-GFP distribution patterns were scored visually as previously described (31). Rab33b knockdown at the individual cell level was confirmed by immunofluorescence staining for the Rab protein.

SLTB internalization

WT HeLa cells plated on glass coverslips were treated with siRNA for 72 h, as described above, before being transferred to wells of a 24-well plate containing ice-cold CO2-independent DMEM (Invitrogen) supplemented with 10% fetal bovine serum. Cy3-conjugated SLTB was prepared as described previously (50) and incubated with cells at 4°C at a final concentration of 70 µg/mL for 30 min to allow binding of SLTB to the cell surface. Cells were washed with ice-cold CO2-independent media before adding 19.5°C CO2-independent media to each well. The plate was placed in a 19.5°C water bath for 1 h to allow Cy3-SLTB to accumulate in endosomes. Coverslip cultures were dipped 4× in complete DMEM growth media and transferred to warm 37°C media to allow Cy3-SLTB transport from endosomes to the Golgi apparatus or ER. The cells were then fixed with formaldehyde at 20 or 60 min following transfer for 37°C. SLTB incubation and chase steps were in the presence of cycloheximide.

Immunoblot analysis

HeLa cells were lysed in 2% SDS, followed by standard SDS–PAGE (∼12% acrylamide) and western blotting as described previously (51). Antibodies and dilutions used for immunoblotting were anti-ZW10 (affinity purified, 1:100), anti-Rab6 (1:1000), anti-COG3 (1:1000), anti-Rab33b (D5,1:500) and anti-α-tubulin (1:2000).

Wide field and spinning disk fluorescence microscopy

Widefield microscopy and spinning disk confocal imaging were performed with a Zeiss Axiovert 200M microscope (Carl Zeiss) fitted with high numerical aperture (NA) 20×/0.8 NA, 63×/1.40 NA and 100×/1.40 NA objectives and a CARV II spinning disk confocal accessory (BD Bioimaging) mounted to the sideport of the microscope. Widefield images were taken with a CoolSnap HQ camera (Roper Scientific). Spinning disk confocal images were captured with a Retiga EXi camera (Qlmaging). The microscope was controlled with Ivision Mac (BioVision). Image stacks for the Golgi apparatus collected through the entire cell depth were compressed into a single plane using maximum intensity projection (MIP) and Ivision software. Alternatively, MIPs were viewed as XY intensity plots using the View as Surface option in Ivision (e.g. Figure 5E,F). Deconvolution was done with Huygens software (SVI). Images were prepared for publication with Adobe Photoshop software (Adobe Systems). Matched image sets were mapped to a common linear gray-scale range.

Laser scanning confocal microscopy and Golgi linescanning

Single plane images were taken using a 63×/1.40 NA objective and a Zeiss LSM 510 microscope operated in multitrack configuration with excitation alternately from the Argon laser using the 488 nm laser line or the 561 nm HeNe laser or the 633 nm HeNe laser. Laser output was attenuated to 5, 10 or 18% power level, respectively. Pixel dwell time was 2.69 µs and XY pixel size was 0.14 µm. Images were collected in 12-bit mode with 4× averaging, conditions that minimized photobleaching while maximizing signal quality. There was no pixel shift between channels. The distributions of GPP130, Rab6 and p115 were analyzed from peak intensities along a line track drawn perpendicular to the Golgi apparatus ribbon as described previously (9,34). In brief, ribbon-like Golgi apparatus structures in a plane giving brightest Rab6 staining were identified for analysis versus more globular Golgi structures where the orientation of the individual cisternae was harder to discern. The pixel intensities along the line track for Cy3-SLTB, Rab6 and p115 were plotted versus distance using Zeiss software, and the distances between peak intensities were determined. Two separate distance samplings per cell (same Golgi region) were made for 30 cells at 20 and 60 min following a temperature increase to 37°C. The initial 20 min time point was used to allow SLTB to be transferred from endosomes to Golgi apparatus.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

We gratefully acknowledge helpful discussions with members of the Lupashin laboratory in the Department of Physiology and Biophysics, University of Arkansas for Medical Sciences. We greatly appreciate the gifts of antibodies and reagents by colleagues. This work was supported in part by the United States National Science Foundation grant MCB-0549001.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References