The COG Complex, Rab6 and COPI Define a Novel Golgi Retrograde Trafficking Pathway that is Exploited by SubAB Toxin

Authors


Dr Vladimir V. Lupashin, vvlupashin@uams.edu

Abstract

Toxin trafficking studies provide valuable information about endogenous pathways of intracellular transport. Subtilase cytotoxin (SubAB) is transported in a retrograde manner through the endosome to the Golgi and then to the endoplasmic reticulum (ER), where it specifically cleaves the ER chaperone BiP/GRP78 (Binding immunoglobin protein/Glucose-Regulated Protein of 78 kDa). To identify the SubAB Golgi trafficking route, we have used siRNA-mediated silencing and immunofluorescence microscopy in HeLa and Vero cells. Knockdown (KD) of subunits of the conserved oligomeric Golgi (COG) complex significantly delays SubAB cytotoxicity and blocks SubAB trafficking to the cis Golgi. Depletion of Rab6 and β-COP proteins causes a similar delay in SubAB-mediated GRP78 cleavage and did not augment the trafficking block observed in COG KD cells, indicating that all three Golgi factors operate on the same ’fast' retrograde trafficking pathway. SubAB trafficking is completely blocked in cells deficient in the Golgi SNARE Syntaxin 5 and does not require the activity of endosomal sorting nexins SNX1 and SNX2. Surprisingly, depletion of Golgi tethers p115 and golgin-84 that regulates two previously described coat protein I (COPI) vesicle-mediated pathways did not interfere with SubAB trafficking, indicating that SubAB is exploiting a novel COG/Rab6/COPI-dependent retrograde trafficking pathway.

Many bacterial and plant toxins often exhibit a structural organization consisting of two fragments, called A and B (1). B subunits of these toxins recognize and bind to specific receptors on the surface of mammalian cells and direct the internalization and trafficking of an enzymatic A subunit. A medically important subset of bacterial toxins, called the AB5 toxins, consists of hexameric assemblies comprising a single catalytically active A subunit and a pentamer of B subunits. Subtilase cytotoxin (SubAB) belongs to a new AB5 toxin family that is produced by a strain of Shiga-toxigenic Escherichia coli(2). The catalytic A subunit of SubAB, which shares sequence homology to a subtilase-like serine protease of Bacillus anthracis, specifically cleaves and inactivates the endoplasmic reticulum (ER) lumen-localized chaperone BiP/GRP78 (Binding immunoglobin protein/Glucose-Regulated Protein of 78 kDa) (3). The SubAB-mediated GRP78 cleavage leads to a transient inhibition of protein synthesis, cell cycle arrest and 2 days later to cell death (4). SubAB is the only toxin known to target a luminal ER protein. The B subunits of SubAB bind to the surface of target cells via glycans displayed on glycoproteins, including β1 integrin (5). Surprisingly, SubAB has a high degree of binding specificity for glycans terminating with α2,3-linked residues of the non-human sialic acid N-glycolylneuraminic acid (6), and therefore in humans a SubAB receptor is generated by metabolic incorporation of an exogenous factor derived from food. Internalization of SubAB is exclusively clathrin dependent (7). The other two well-studied AB5 toxins, Shiga and Cholera toxins, are internalized in both a clathrin- and a cholesterol-dependent manner. On the way from the cell surface to the ER, all three toxins are passing through the Golgi apparatus, although the exact intra-Golgi retrograde trafficking pathways used by various toxins are not well understood.

Studies of bacterial toxins can provide valuable information about pathways of intracellular transport. Retrograde transport of both Shiga and Cholera toxins to the Golgi and the ER seems to be dependent on different Rab and ADP ribosylation factor (ARF) GTPases, SNAREs and vesicular tethering proteins (8–15). Comparison of different toxins reveals differences indicating the presence of more than one pathway between early endosomes and the ER through the Golgi apparatus (16). Specific SubAB-mediated GRP78 cleavage provides an extremely sensitive and quantitative assay (7) to measure SubAB arrival at the ER in both control cells and in cells deficient in putative components of the Golgi retrograde trafficking machinery. This should allow identification of SubAB intra-Golgi trafficking pathway(s). We have recently found that codepletion of Rab6 and conserved oligomeric Golgi (COG) vesicular tethering factor suppresses the Golgi fragmentation phenotype that is associated with the COG knockdown (KD) (17). This epistatic assay provides the evidence that two proteins act in the same pathway.

In this work, we have used both single and double KD strategies to uncover a novel Rab6, COG and COPI-dependent intra-Golgi retrograde trafficking pathway that is exploited by SubAB toxin for fast delivery to the ER.

Results

In both HeLa and Vero cells, Brefeldin A treatment completely blocks SubAB-mediated cleavage of GRP78 (7), indicating that the functional Golgi apparatus is essential for SubAB trafficking from plasma membrane to the ER. We have previously identified the COG protein complex as an important vesicle tethering component of the intra-Golgi retrograde trafficking machinery that is responsible for the localization of Golgi glycosyltransferases and for the retrograde delivery of Shiga toxin B subunit (15,18). To test the hypothesis that the COG complex is involved in the retrograde trafficking of SubAB toxin, we have utilized a short hairpin RNA (shRNA) strategy to create a stable Cog4p knockdown (Cog4 sKD) HeLa GalNAc T2-GFP cell line. The level of Cog4 protein in Cog4 sKD cells was reduced by more than 85%, as compared to the actin loading control (Figure 1A). Golgi membranes in Cog4 sKD cells appeared to be partially fragmented (Figure 1B), that is consistent with the phenotype previously described for Cog3 siRNA KD cells (15). Both control and Cog4 sKD cells were incubated with purified SubAB toxin (3 μg/mL) for 3 h. Kinetics of GRP78 cleavage was used to estimate the SubAB trafficking efficiency. In control HeLa cells, GRP78 cleavage started at the 60 min time-point, with more than 80% of GRP78 cleaved after 120 min of incubation (Figure 1C,D, control). In sharp contrast, GRP78 cleavage in Cog4 sKD cells was significantly delayed and started 120 min after toxin application (Figure 1C,D, Cog4 KD). Three hours after the start of the assay, less than 60% of the GRP78 was cleaved in Cog4 sKD cells indicating that Cog4p is required for the ’fast' delivery of the SubAB to the ER. The block in SubAB trafficking in Cog4 sKD cells was not complete and GRP78 was completely cleaved 5 h after toxin application (data not shown), indicating that in HeLa cells SubAB is also using a Cog4p-independent ’slow’ delivery pathway. Importantly, the delay in GRP78 cleavage observed in Cog4 sKD cells was not because of the Golgi fragmentation, since a treatment of HeLa cells with Nocodazole, a drug that affects microtubule function and causes severe Golgi fragmentation, demonstrated that the SubAB-mediated GRP78 cleavage kinetics were similar to that of the untreated cells (Figure 1D, Nocodazole). Incubation of HeLa cells with labeled cytotoxin at 4°C revealed that the same amount of SubAB binds to mock-treated and Cog4 KD cells (Figure 1E), indicating that COG complex KD did not compromise the plasma membrane-localized SubAB receptor(s).

Figure 1.

SubAB-mediated GRP78 cleavage is inhibited in stable Cog4 KD HeLa cells. GalNAcT2-GFP HeLa cells were mock transfected or transfected with pCog4 shRNA plasmid and maintained in selective puromycin-containing media. One month after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with the indicated antibodies (A) or fixed and visualized by fluorescent microscopy (B). Bar, 10 μm. Effect of Cog4 KD on SubAB-induced cleavage of GRP78. Control and Cog4 sKD cells were treated with SubAB (3 μg/mL) and incubated at 37°C for the time indicated. Panel (C) shows a blot representative of three such experiments. D) Graph of SubAB-induced GRP78 cleavage for control, Cog4 sKD and HeLa cells treated with 30 μm Nocodazole for 2 h before toxin application. E) Binding of SubAB-HF555 to control and Cog4 KD GalNAcT2-GFP HeLa cells. Cells were exposed to SubAB-Hylite 555 toxin for 20 min at 4°C, formalin fixed and viewed by laser confocal microscopy (63× oil objective). Bar, 10 μm.

The COG complex is composed of eight subunits that are grouped into two subcomplexes: lobe A (Cog1-4) and lobe B (Cog5-8) (19–21). To check whether other subunits of the COG complex are necessary for the fast intracellular trafficking of SubAB toxin, HeLa cells were transfected with siRNAs that specifically downregulated biosynthesis of Cog2p, Cog3p, Cog4p, Cog6p and Cog8p. In all cases, a significant (96% for COG3 KD, 92% for COG4 KD, 91% for COG6 KD and 88% for COG8 KD) reduction in the protein level of targeted proteins was achieved for all COG subunits as estimated by immunoblot analysis using Odyssey LI-COR System (Figure 2A). Because of a lack of anti-Cog2p antibodies, the efficiency of COG2 KD was estimated by reverse transcriptase-polymerase chain reaction (RT-PCR) (Figure 2B). In all COG KD cells, the rate of SubAB-mediated GRP78 cleavage was significantly decreased (Figure 2C), indicating that the entire COG complex is important for a fast SubAB intracellular delivery. Expression of siRNA-resistant mouse Cog3p (15) partially restored a SubAB-mediated GRP78 cleavage in COG3 KD HeLa cells (Figure 2D,E); similar rescue of SubAB trafficking was observed in Cog4KD cells transfected with Cog4-myc siRNA-resistant plasmid and in COG8KD cells transfected with Cog8-myc siRNA-resistant plasmid (Figure S1). To test for possible synergy between lobe A and lobe B subunits in SubAB delivery, GRP78 cleavage in Cog3/Cog6 and Cog3/Cog4 double KD cells was analyzed (Figure 2F). The delay in GRP78 cleavage was found to be more significant in double COG subunit KD HeLa cells as compared with single COG subunit KD cells. It was more pronounced in Cog3/Cog4 KD cells lacking two subunits of the same lobe, indicating that the increased delay in SubAB trafficking was due to more efficient silencing achieved in a double KD procedure, rather than to the synergy between COG complex subcomplexes. Very limited (less than 20%) GRP78 cleavage was observed in COG double KD cells 2 h after incubation with toxin, signifying the essential role of the COG complex in a fast PM-Golgi-ER delivery of SubAB toxin. Importantly, we found that at the 180 min incubation time-point, up to 20% of GRP78 was cleaved even in the most severe Cog3/Cog4 double KD cells (Figure 2F). SubAB-mediated GRP78 cleavage continued slowly, reaching up to 80% after 6 h of incubation (data not shown). This slowly sustained GRP78 cleavage is likely to indicate existence of a secondary ’slow’ COG complex-independent SubAB delivery pathway in HeLa cells.

Figure 2.

Both single and double KD of COG subunits cause a similar delay in the SubAB-mediated GRP78 cleavage. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Cog2, Cog3, Cog4, Cog6 or Cog8. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with indicated antibodies (A). Total RNA was purified from control, Cog2 KD and Cog3 KD cells and the level of Cog2 and Cog3 RNA was estimated by RT-PCR (B). C) Graph of SubAB-induced GRP78 cleavage for control, Cog2 KD, Cog3 KD, Cog4 KD, Cog6 KD and Cog8 KD in HeLa cells. D–E) COG3 siRNA-induced block in SubAB delivery is reversible. GalNAcT2-GFP HeLa cells were transfected with siRNA to human Cog3 or transfected with siRNA to human Cog3 then (24 h later) transfected with a plasmid encoding mouse Cog3. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS and separated by SDS–PAGE and analyzed by immunoblotting with either the polyclonal Cog3 antibody that recognizes human and mouse Cog3 or with the monoclonal Cog3 that only recognizes human Cog3 (D). E) Graph of SubAB-induced cleavage of GRP78 for Cog3 KD cells and for Cog3 KD transfected with siRNA-resistant mouse Cog3. F) GalNAcT2-GFP HeLa cells were mock transfected or double transfected with siRNA to Cog3/Cog4 or Cog3/Cog6. Graph of SubAB-induced GRP78 cleavage for control, Cog3/6 KD and Cog3/4 KD in HeLa cells.

COG complex deficiency could affect SubAB cytotoxicity either directly, via regulation of retrograde membrane carriers that carry SubAB to ER, or indirectly, by affecting localization and/or expression of SubAB receptors on the cell surface. We have shown previously that the depletion of Cog3p did not block the delivery of both plasma membrane and lysosomal proteins (15,18). In agreement to this result, the depletion of other COG subunits did not significantly effect the anterograde protein trafficking in HeLA cells (Figure S2) To visualize SubAB trafficking, we have labeled SubAB with HiLyte Fluor 555 (HF555) dye and with Alexa Fluor 647(AF647) and analyzed SubAB-HF555 and SubAB-AF647 intracellular trafficking by the immunofluorescence (IF) approach. Labeling of SubAB with either HiLyte Fluor 555 or Alexa Fluor 647 dyes did not change the kinetics of SubAB-mediated GRP78 cleavage (data not shown); hence SubAB-HF555/AF647 trafficking is likely to be identical to the trafficking of the unlabeled SubAB.

We have initially tried to use SubAB-HF555 in HeLa cells and found that after 60 and 120 min of incubation, the majority of SubAB-HF555 did not localize with GalNAc-T2-GFP-positive Golgi structures, indicating that in HeLa cells SubAB is being distributed throughout a variety of endocytic compartments (data not shown). As a result, the pattern of SubAB intracellular trafficking in HeLa cells was not interpretable with a straightforward IF approach. This is in agreement with an earlier observation that in several cell types, including HeLa cells, not more than 5–10% of endocytosed toxin is transported in the direction of the Golgi apparatus (22).

A less complicated intracellular trafficking pattern of SubAB was observed previously in Vero cells (3). In Vero cells, significantly lower SubAB initial concentration (0.05 μg/mL) was sufficient to completely cleave GRP78 in 60 min (Figure 3B, control). Vero cells transfected with COG3 siRNA demonstrated substantial downregulation of Cog3p (Figure 3A). Again, as in the HeLa case, depletion of Cog3p in Vero cells significantly delayed SubAB-mediated GRP78 cleavage (Figure 3B). Incubation of both control and Cog3 KD Vero cells with SubAB-HF555 at 4°C resulted in a comparable binding of labeled toxin to the plasma membrane (Figure 3C,D, 0 min row). Upon incubation at 37°C, the SubAB-HF555 was rapidly internalized and in more than 80% of control cells, the toxin was mostly located in a perinuclear region where it colocalized with cis Golgi marker giantin (Figure 3C, 30 min, arrows). This major colocalization of internalized SubAB-HF555 with cis Golgi was persistent in control Vero cells 60 min after toxin trafficking started (Figure 4C, 60 min). Labeled SubAB toxin was rapidly endocytosed in Cog3 KD Vero cells; however, in a striking contrast with control cells, the majority of internalized cytotoxin was found at the cell periphery at both 30 and 60 min time-points (Figure 3D). Less than 15% of Cog3 KD cells demonstrated perinuclear localization of SubAB-HF555. However, even in these cells the perinuclear toxin was not colocalized with the cis Golgi marker giantin (Figure 3D, 60 min, arrowhead), indicating that COG complex malfunction causes accumulation of internalized cytotoxin in pre-cis-Golgi trafficking intermediates.

Figure 3.

Cog3 depletion in Vero cells partially inhibits SubAB cytotoxicity and blocks SubAB-HF555 delivery to cis Golgi. Vero cells were mock transfected or transfected with siRNA to Cog3. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with the indicated antibodies (A). Control and Cog3 KD cells were exposed to SubAB for the time indicated and SubAB-induced GRP78 cleavage was calculated for each time-point (B). Subcellular localization of SubAB-HF555 in control (C) and Cog3 KD (D) Vero cells. Cells were exposed to SubAB-HF555 toxin for 20 min at 4°C and then either immediately fixed (0 min) or incubated for 30 or 60 min at 37°C (B). For immunofluorescence, cells were formalin fixed, permeabilized with Triton-X-100, treated with anti-giantin and developed using HiLyte-488-conjugated anti-rabbit antibodies. Samples were viewed by dual-laser confocal microscopy (63× oil objective). Bar, 10 μm. Arrows indicate colocalization of giantin and SubAB-HF555 in control cells. Arrow heads show perinuclear area localization of SubAB-HF555 in Cog3 KD cells. Bar, 10 μm.

Figure 4.

Fluorescently labeled SubAB partially localizes with TGN46-positive compartments in Cog3-depleted cells. Vero cells were mock transfected or transfected with siRNA to Cog3. Seventy-two hours after transfection, the cells were treated with 1 μg/mL SubAB HF555 for 1 h. Control (A) and Cog3 cells (B) were fixed in 4% paraformaldehyde, permeabilized with Triton-X-100, stained with anti-TGN46, anti-EEA1, anti-Rab7 and anti-LAMP2, and developed using HiLyte-488-conjugated-anti-mouse and HiLyte-488-conjugated-anti-rabbit antibodies. Samples were viewed by dual-laser confocal microscopy (63× oil objective). Bar, 10 μm. Arrows demonstrate colocalization between TGN46 and SubAB HF555.

To gain more understanding about SubAB-positive pre-cis-Golgi compartments, both COG3 KD and mock-treated Vero cells were incubated with fluorescently labeled SubAB (SubAB-HF555 and SubAB-AF647) and stained for the markers for both endocytic and secretory pathways (Figure 4). In control cells, SubAB upon incubation for 60 min at 37°C was partially colocalized with trans Golgi marker TGN46 and did not show any significant colocalization with endosomal-lysosomal compartments. In COG3 KD Vero cells, some (∼15–20%) of internalized SubAB was colocalized with TGN46 (Figure 4, TGN46 row, arrow), indicating that COG depletion did not block SubAB delivery to trans Golgi network (TGN)/trans Golgi. No significant colocalization of SubAB was observed with early endosomal protein EEA1 (Early Endosomal Antigen I) (Figure 4, EEA1 row), late endosomal protein Rab7 (Figure 4, rab7 row) and lysosomal marker Lamp2 (Figure 4, Lamp2 row). We also were unable to find a significant SubAB colocalization with Snx2, Vps26, Rab11 and PDI (Protein Disulfide Isomerase) (Figure S3 and data not shown). Similar results were obtained for COG8 KD cells (data not shown).

To identify other Golgi proteins that may regulate intracellular trafficking of SubAB, we have analyzed known COG complex-interacting trafficking factors.

We have demonstrated previously that the COG complex is acting downstream of the trans Golgi-localized Rab6 protein and that the epistatic depletion of Rab6 inhibited the Golgi-disruptive effect of COG inactivation by siRNA or antibodies (17). Here we found that the depletion of Rab6 in HeLa cells affected the kinetics of SubAB-mediated GRP78 cleavage in a manner analogous to COG depletion (Figure 5). Efficient double Rab6/Cog3 KD (Figure 5A) resulted in a delay in SubAB delivery similar to that observed in the single Rab6 KD cells (Figure 5B), indicating that both Rab6 and the COG complex operate on the same retrograde pathway of SubAB delivery. Rab6 siRNA was not efficient in Vero cells (data not shown), therefore, we tested if the overexpression of GDP-restricted Rab6 mutant would interfere with SubAB trafficking. Vero cells were cotransfected with pCFP and pRab6S21N plasmids and treated with fluorescently labeled SubAB. We found that in cyan fluorescent protein (CFP)-positive cells internalized SubAB was preferentially colocalized with EEA1 and not with giantin (Figure 5C), indicating that Rab6 malfunction was interfered with the endosome-to-Golgi leg of SubAB retrograde trafficking itinerary.

Figure 5.

Rab6 and Rab6/Cog3 depletions as well as overexpression of Rab6-GDP cause similar delays in the SubAB-mediated GRP78 cleavage. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Rab6 or Rab6/Cog3. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with the indicated antibodies (A). B) Graph of SubAB-induced GRP78 cleavage for control, Rab6 KD or Rab6/Cog3 double KD cells. C) Vero cells were transfected with pRab6-S21N and 24 h later treated with 1 μg/mL SubAB-HF555 for 20 min. Toxin was additionally internalized for 60 min, after that the cells were fixed in 4% paraformaldehyde, permeabilized with Triton-X-100, treated with anti-EEA1 and anti-giantin, and developed using HiLyte-488-conjugated anti-mouse antibodies and HiLyte-647-conjugated anti-rabbit antibodies. Samples were viewed by dual-laser confocal microscopy (63× oil objective). Bar, 10 μm. Arrows indicate colocalization of EEA1 and SubAB.

The COG complex genetically and physically interacts with subunits of COPI vesicular coat that operates in both intra-Golgi and Golgi to ER retrograde membrane trafficking steps (15,23,24), and COG and COPI may work in concert to ensure the proper retention and retrieval of a subset of proteins in the Golgi (25). To downregulate β-COP, we used a previously characterized siRNA in HeLa cells (26). Efficient (more than 90%) siRNA-mediated depletion of β-COP did not affect the intracellular level of Cog3p (Figure 6A), but caused a significant delay in SubAB-mediated GRP78 cleavage (Figure 6B). Again, double β-COP/Cog3 KD resulted in a delay in SubAB delivery similar to that observed in the single β-COP KD cells (Figure 6B), demonstrating that both COPI and COG complexes control the same retrograde SubAB delivery pathway.

Figure 6.

β-COP andβ-COP/Cog3 knockdowns along with overexpression of Arf1-Q71L cause delays in the SubAB-mediated GRP78 cleavage and block the traffic of SubAB. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to β-COP or β-COP/Cog3. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with the indicated antibodies (A). B) Graph of SubAB-induced GRP78 cleavage for control, β-COP KD or β-COP/Cog3 double KD cells. C) Vero cells were transfected with the CFP-Arf1-Q71L plasmid and 24 h later treated with 1 μg/mL SubAB-HF555 for 20 min. Toxin was additionally internalized for 60 min, after that the cells were fixed in 4% paraformaldehyde, permeabilized with Triton-X-100, treated with anti-giantin and developed using HiLyte-647-conjugated anti-rabbit antibodies. Samples were viewed by dual-laser confocal microscopy (63× oil objective). Bar, 10 μm. The asterisk indicates a control cell not transfected with CFP-Arf1-Q71L.

KD of β-COP was in some way detrimental for HeLa cells (only 50% of cells were recovered after 72 h of β-COP KD) and resulted in an extensive Golgi fragmentation (data not shown). Therefore, we have tested if the less invasive COPI malfunction caused by overexpressioin of GTP-locked mutant for an Arf1 (27) would interfere with SubAB trafficking. Vero cells were transfected with pArf1Q71L-CFP plasmid and treated with fluorescently labeled SubAB. We found that CFP-Arf1Q71L expression blocked fluorescently labeled SubAB trafficking to giantin-positive cis Golgi membranes (Figure 6C), confirming the essential role of COPI in retrograde trafficking of SubAB. As in the case of COG3 KD cells, in Arf1Q71L expressing Vero cells SubAB was trapped in trafficking intermediates that did not colocalize with any tested endocytic markers (data not shown).

The COG complex genetically and physically interacts with a subset of Golgi-operated SNARE molecules (15,23,24). In mammalian cells, hCog4p and hCog6p interact with Qa-SNARE Syntaxin 5a (28). We have used previously described siRNA (29) to KD both long and short isoforms of Syntaxin 5 (Figure 7A) and found that SubAB-mediated GRP78 cleavage was almost completely blocked (Figure 7B). Less than 8% of GRP78 was cleaved even after 180 min of incubation of HeLa cells with SubAB toxin, indicating that Syntaxin 5 is required for both fast (COG complex-dependent) and slow (COG complex-independent) intracellular delivery of SubAB. In Syntaxin 5 KD Vero cells, fluorescently labeled SubAB failed to colocalize with both cis (giantin) and trans (P230) Golgi compartments (Figure 7C) and was accumulated in peripherally located small membrane structures, indicating an early block in retrograde trafficking.

Figure 7.

SubAB-mediated GRP78 cleavage is completely blocked in Syntaxin 5 KD cells. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Syntaxin 5. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with indicated antibodies (A). Note that both long and short isoforms of Syntaxin 5 were depleted, while the non-specific protein band (indicated with an asterisk) was not. B) Graph of SubAB-induced GRP78 cleavage for control and Syntaxin 5 KD cells. C) Vero cells were transfected with siRNA to Syntaxin 5. Seventy-two hours later, the cells were treated with 1 μg/mL SubAB-HF555, fixed in 4% paraformaldehyde, permeabilized with Triton-X-100, treated with anti-P230 and anti-giantin, and developed using HiLyte-488-conjugated anti-mouse antibodies and HiLyte-647-conjugated anti-rabbit antibodies. Samples were viewed by dual-laser confocal microscopy (63× oil objective). Bar, 10 μm.

KD of Cog2, Cog3, Cog6, Cog8 and Rab6 resulted in a fourfold increase of potency or inhibitory concentration of 50% (IC50) for SubAB in HeLa cells (Table 1 and Figure S4). More dramatic IC50 increase was detected in β-COP KD cells, while even a significant increase (up to 6 μg/mL) in SubAB concentration applied on Syntaxin 5 KD cells has not resulted in any GRP78 cleavage.

Table 1.  Potency and efficacy of SubAB on control and single KDs in HeLa cells treated for 3 h.
 IC50 (μg/mL)r2Imax (%)
  1. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Cog2, Cog3, Cog4, Cog6, Cog8, Rab6, Syntaxin 5, β-COP and Vps26. Seventy-two hours after transfection, the cells were intoxicated with SubAB for 3 h at 0, 0.074, 0.222, 0.667, 2 and 6 μg/mL. Cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE with the GRP78 antibody. Results of GRP78 cleavage by SubAB were calculated using the GraphPad Prism software to demonstrate IC50, r2 and Imax values.

Control0.1530.95480
Cog2 KD0.7880.99879
Cog3 KD0.5790.99966
Cog4 KD0.3880.93633
Cog6 KD0.6450.97970
Cog8 KD0.6560.98863
Rab6 KD0.7390.96662
β-COP KD3.650.89737
Syntaxin 5 KD>6No fit0
Vps26 KD0.4170.99656

It was reported recently that Cog2p physically interacts with another cis Golgi vesicle-tethering factor p115 (30), and that p115 stimulates formation of Syntaxin 5-containing SNARE complex in vitro(31). In addition, p115/SNARE interaction was shown to be essential for Golgi biogenesis (32). On the other hand, we have shown previously that the epistatic depletion of Rab6 did not inhibit the Golgi-disruptive effect of p115 inactivation by siRNA (17), indicating that Rab6 and p115 operate on independent Golgi trafficking pathways. Here we used the same siRNA-mediated p115 silencing to test for a possible p115 requirement in SubAB trafficking in HeLa cells. P115 KD cells lacked the signal for p115 protein and demonstrated the characteristic Golgi fragmentation phenotype (Figure 8A,C), but rather surprisingly, did not show any inhibition or delay in SubAB-mediated GRP78 cleavage (Figure 8B), indicating that SubAB is delivered to the ER via a p115-independent pathway.

Figure 8.

The retrograde traffic of subtilase cytotoxin is independent of p115 and golgin-84. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to COG3, p115 or golgin-84. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS–PAGE and analyzed by immunoblotting with the indicated antibodies (A). A non-specific protein band on a golgin-84 blot is indicated with an asterisk. Effect of protein depletion on SubAB-induced cleavage of GRP78. Control, COG3 KD, p115 KD and golgin-84 KD cells (B) were treated with SubAB (3 μg/mL) and incubated at 37°C for the time indicated. Each lane on the graph is calculated from three independent experiments. C) GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Golgin-84 or p115 cells. Seventy-two hours later, the cells were fixed and visualized by fluorescent microscopy. Bar, 10 μm.

The Warren lab recently proposed that p115-giantin and golgin-84-CASP (CASP = CDP/cut alternatively spliced product) Golgi tethers define two distinct subpopulations of COPI vesicles (33). As p115 was dispensable for the fast COPI-dependent delivery of SubAB through the Golgi stack, we tested golgin-84 involvement in cytotoxin trafficking. Interestingly, golgin-84 KD cells lacked the signal for golgin-84 protein and demonstrated the characteristic Golgi fragmentation phenotype in HeLa cells (Figure 8A,C), but again, as in the case of p115 KD cells, did not show any inhibition or delay in SubAB-mediated GRP78 cleavage (Figure 8B), indicating that SubAB is delivered to the ER via a golgin-84-independent pathway. This data indicate that SubAB is exploiting a novel p115/golgin-84-independent intra-Golgi trafficking pathway. Another possibility is that the small residual amounts of coiled-coil Golgi tethers in p115 KD and golgin 84 KD cells were sufficient to fully support SubAB retrograde delivery.

We have reported previously that Cog3 KD HeLa cells are deficient in retrograde trafficking of Shiga-like toxin B (SLTB) subunit (15). Previous reports suggested that SubAB and SLTB retrograde trafficking pathways are similar, but not identical (7). Indeed, we have observed that in Vero cells incubated with both SubAB-HF555 and SLTB-AF647, SLTB accumulation in perinuclear Golgi presided the arrival of SubAB to the same area (data not shown). We also found different sensitivity of toxins retrograde trafficking to the level of COG3 KD (Figure 9). At the intermediate level of COG3 KD, fast delivery of SubAB to Golgi was blocked, while SLTB was delivered normally (Figure 9, partial COG3 KD row). In a contrast, cells that maintain a very low level of Cog3p, both SubAB and SLTB were not delivered to the Golgi (Figure 9, COG3 KD).

Figure 9.

Correct SubAB and SLTB delivery require different Cog3 protein level. Vero cells were mock transfected or transfected with siRNA to Cog3. Seventy-two hours after transfection, the cells were pulse chased with 1 μg/mL of SLTB-Alexa fluor 647 and SubAB-HiLyte fluor 555 for 1 h. Cells were formalin fixed, permeabilized with Triton-X-100, treated with anti-Cog3 and developed using HyLite-488-conjugated anti-rabbit antibodies. Samples were viewed by triple-laser confocal microscopy (63oil objective). Level of Cog3 expression in individual cells (left column) is shown using a rainbow palette. Bar, 10 μm.

Endosomal-to-Golgi trafficking of Shiga toxin is also regulated by Rab6 (9,34) and by sorting nexins SNX1 and SNX2 (35,36). Combined depletion of SNX1 and SNX2 in Vero cells gave a total inhibition of Shiga toxin transport to the TGN by 80% (36). Having established the requirement for both COG complex and Rab6 in SubAB trafficking, we tested whether SNX1 and/or SNX2 are essential for this process in HeLa cells. siRNA-mediated inhibition was efficient in knocking down SNX1, SNX2 and both SNX1/SNX2 (Figure 10A), but in all cases SubAB-mediated GRP78 cleavage was indistinguishable from control cells (Figure 10B), indicating that SubAB is delivered to the ER via the SNX1/SNX2-independent pathway. Interestingly, we found that the KD of another retromer component, Vps26 in HeLa cells (Figure 10A), moderately inhibited SubAB-mediated GRP78 cleavage (Figure 10C), indicating that the retromer complex is not completely disposable for SubAB retrograde delivery.

Figure 10.

The retrograde traffic of SubAB is independent of SNX1/2 and partially inhibited in cells depleted for Vps26. GalNAcT2-GFP HeLa cells were mock transfected or transfected with siRNA to Snx1, Snx2, both Snx1 and Snx2, and Vps26. Seventy-two hours after transfection, the cells were solubilized in PBS with 2% SDS, and the lysates were separated by SDS—PAGE and analyzed by immunoblotting with the indicated antibodies (A). Effect of protein depletion on SubAB-induced cleavage of GRP78. Control, Snx1 KD, Snx2 KD, Snx1/2 double KD and Vps26 KD cells (B and C) were treated with SubAB (3 μg/mL) and incubated at 37°C for the time indicated. Each lane on the graph is calculated from three independent experiments.

Discussion

Analysis of SubAB trafficking in HeLa and Vero cells revealed several key Golgi-resident proteins that are indispensable for the efficient delivery of the SubAB to the ER. The COG complex has been previously implicated in the retrograde trafficking of several Golgi enzymes, SNAREs and Shiga toxin B subunit (15,18,28). Our experiments revealed that downregulation of any subunit of the COG complex significantly delayed SubAB delivery to the ER. The delay in SubAB delivery was very similar in cells deficient for either lobe A (Cog 2, 3 and 4) or lobe B (Cog 6 and 8) of the COG complex, suggesting that the entire COG complex is required for the toxin delivery. The defect in SubAB trafficking was most apparent in cells deficient for two COG subunits, indicating that the residual COG complex activity that remained in cells after the KD of any single subunit was still sufficient to support at least some SubAB trafficking. Binding to the cell surface and initial internalization of labeled SubAB were normal in COG-deficient cells, whereas trafficking to cis Golgi was dramatically inhibited, indicating that the COG complex is required for either intra-Golgi or endosome-to-Golgi trafficking steps. In COG3 KD Vero cells, the internalized SubAB did not show any significant colocalization with markers of endosomal/lysosomal pathway indicating a post-endosomal trafficking block. Some SubAB was able to arrive at the perinuclear area and colocalize with TGN46 in Cog3 KD cells, but failed to colocalize with the cis Golgi marker giantin, indicating an intra-Golgi trafficking defect.

Importantly, COG deficiencies only notably delayed but did not completely block the SubAB-mediated GRP78 cleavage, indicating the existence of multiple retrograde pathways exploited by SubAB toxin. The incomplete SubAB trafficking block in COG KD cells was actually beneficial to this study, and allowed us to analyze other components of the intra-Golgi trafficking machinery that may work on the same or parallel pathway. KD of either trans Golgi Rab6 or β-COP delayed SubAB delivery in a manner similar to that observed in COG KD cells, suggesting that these trafficking regulators may work on the same pathway. Block in SubAB delivery to the ER in cells that overexpress either Rab6-GTP or Arf1-GTP further underlined the importance of Rab6 and COPI for SubAB trafficking. Most importantly, simultaneous KD of either Rab6/Cog3 or β-COP/Cog3 did not change the severity of the toxin trafficking block, supporting the notion that the COG complex, Rab6 and COPI regulate the same fast intra-Golgi trafficking pathway exploited by SubAB cytotoxin. This is in good agreement with previously described interactions between the COG and COPI complexes (15,23–25), as well as with the recently uncovered epistatic interaction between Rab6 and the COG complex (17). SubAB was preferentially accumulated in EEA1-positive endosomal compartment in cells that overexpress Rab6GDP mutant, but not in COG3 KD cells, indicating that Rab6 functions upstream of the COG complex in SubAB retrograde trafficking pathway.

It is possible that all SubAB trafficking in normal cells is COG/Rab6/COPI dependent, but in cells deficient for this major retrograde trafficking pathway, the alternative endosome to ER transport is activated. The induction of routes that are normally non-existent was observed before in the case of ricin trafficking in cells that express a temperature-sensitive mutant of ε-COP (37). We favor the alternative hypothesis that SubAB toxin normally travels through the Golgi by two separate routes: fast, COG/Rab6/COPI-dependent and slow, COG/Rab6/COPI-independent. This interpretation is supported by the toxin trafficking results obtained in cells deficient in Qa-SNARE Syntaxin 5.

SubAB trafficking block in Syntaxin 5 KD cells was similar to a block observed in cells pretreated with Brefeldin A [data not shown and (7)]. In both cases, no GRP78 cleavage was observed after 3 h of incubation with SubAB, indicating that Brefeldin A treatment and Syntaxin 5 KD inhibit both fast and slow routes through the Golgi. Interestingly, that among all COG complex KD cells the strongest inhibition of SubAB trafficking was associated with depletion of Cog4p, a lobe A subunit that directly interacts with Syntaxin 5 (28). Syntaxin 5 was implicated in both intra-Golgi (38) and early/recycling endosome to the TGN (39) retrograde trafficking steps. One possibility is that SubAB is delivered from endosomes to the ER via two distinct mechanisms by exploiting either the fast COG complex-dependent inter-cisternae vesicular trafficking pathway or by hopping on to the slow direct TGN to ER membrane flow (Figure 11). In specific experimental conditions, the retrograde transport of Cholera toxin from the plasma membrane to the ER was shown to require the TGN, but not the Golgi complex (40). Existence of the endogenous direct TGN-to-ER trafficking route was also proposed (41). Syntaxin 5 KD most likely will eliminate both routes for SubAB delivery.

Figure 11.

Model of retrograde transport of subtilase cytotoxin. The solid arrows indicate the Rab6/COG complex/COPI-dependent pathway that SubAB traffics through the HeLa cell. The dotted arrow indicates that Syntaxin 5 will completely inhibit both the Rab6/COG complex/COPI-dependent and -independent pathways.

Alternatively, the complete SubAB trafficking block in Syntaxin 5 KD cells is associated with the dual role of Syntaxin 5 in both anterograde and retrograde trafficking. Indeed, some of the endogenous CD44 and the majority of the model (VSVG-GFP) anterograde cargo proteins were accumulated in the Golgi area in Syntaxin 5 KD cells, while both control and COG-deficient HeLa cells correctly delivered both proteins to the plasma membrane (Figure S2).

Depletion of several Golgi factors (Cog3, Cog4, β-COP and Syntaxin 5) in HeLa cells leads to an intense Golgi fragmentation, but the Golgi breakup observed in KD cells seemed to be not linked to the inhibition of SubAB trafficking. Indeed, the trafficking of SubAB to the fragmented Golgi continued in the presence of Nocodazole (7), and the Golgi fragmentation observed in p115 KD and golgin-84 KD cells did not interfere with cytotoxin arrival in the ER.

Rather surprisingly, we found that SubAB intra-Golgi trafficking does not require two previously described Golgi-localized coiled-coil vesicle tethers p115 and golgin-84. Recent work from Warren lab proposed that p115-giantin and golgin-84-CASP Golgi tethers define two distinct subpopulations of COPI vesicles (33). Because only a subfraction of Golgi-derived COPI vesicles was efficiently captured by both tethering pairs, Warren lab suggested that ’other tethers should identify other populations, allowing us to map the flow patterns of resident, cargo and recycling molecules'. Our finding clearly reinforces this prediction, suggesting that SubAB is using a novel p115/golgin-84-independent and COG/Rab6/COPI-dependent intra-Golgi trafficking pathway.

The nature of the molecule (or molecules) that carry SubAB through the retrograde Golgi pathways is currently unknown. It could be a SubAB plasma membrane glycoprotein receptor itself, analogous to a lipid-raft ganglioside GM1 (monosialotetrahexosylganglioside) that was proposed to carry Cholera toxin all the way from plasma membrane to ER (42). Alternatively, SubAB could use distinct carriers for different trafficking steps ’jumping’ from one to another in sorting compartments. While the nature of the intracellular SubAB carrier is not known at present, it is clear that SubAB retrograde trafficking is quite distinct from the trafficking of other known AB5 toxins. Both Cholera and Shiga toxins enter target cells via both clathrin-dependent and clathrin-independent mechanisms, while SubAB internalization is strictly clathrin dependent (7). Unlike Shiga toxin, SubAB trafficking does not require sorting nexins SNX1/SNX2, is more sensitive to COG complex depletion and is dependent on COPI vesicular coat.

The initial characterization of the COG/Rab6/COPI-dependent retrograde pathway may provide a useful framework for better understanding of intra-Golgi trafficking in mammalian cells.

Materials and Methods

Reagents and antibodies

Reagents were as follows: Nocodazole (stock solution = 10 mm) (Sigma), SubAB (1.49 mg/mL) (2) and SLTB subunit prepared as described previously (43). Antibodies used for IF microscopy or western blotting (WB) were purchased through commercial sources, gifts from generous individual investigators or generated by us. Antibodies (and their dilutions) were as follows: rabbit affinity purified anti-hCog3 [WB 1:10 000; (44)], hCog4 (WB 1:1000, this lab), hCog6 (WB 1:1000, this lab), hCog8 (WB 1:1000, this lab), Syntaxin 5 (WB 1:1000, this lab), p115 [IF 1:100; (45)], Rab6 (WB 1:200; Santa Cruz), Giantin (IF 1:1000; Covance), TGN46 (IF 1:300; AdB Serotec), Vps26 (WB 1:500; Abcam), mouse monoclonal anti-β-actin (WB 1:10 000; Sigma), β-COP (WB 1:500; Sigma), glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (WB 1:1000; Santa Cruz), golgin-84 (WB 1:500; BD Biosciences), Snx1 (WB 1:1000; BD Biosciences), Snx2 (WB 1:1000; BD Biosciences), p115 [IF 1:100; (46)], EEA1 (IF 1:200; BD Biosciences), P230 (IF 1:200; BD Biosciences), GPP130 (IF 1:1000; Alexis Biochemicals), LAMP2 (IF 1:100; Developmental Studies Hybridoma Bank, University of Iowa), Rab7 (IF 1:200; Cell Signaling Technology), CD44 (IF 1:50; Developmental Studies Hybridoma Bank, University of Iowa) and goat anti-GRP78 (WB goat 1:1000; Santa Cruz).

Secondary anti-goat-horseradish peroxidase (HRP), anti-mouse-HRP and anti-rabbit-HRP for WB were obtained from Jackson ImmunoResearch Laboratories. IRDye 680 goat anti-rabbit, IRDye 700 goat anti-mouse and IRDye 800 donkey anti-goat for WB were obtained from LI-COR Biosciences. Anti-rabbit HiLyte 488 and anti-rabbit HiLyte 555 for IF were obtained from AnaSpec, Inc.

Cell culture

HeLa cells stably expressing GalNAcT2-GFP were cultured in DMEM/F-12 medium supplemented with 15 mm HEPES, 2.5 mml-glutamine, 10% FBS and 0.4 mg/mL G418 sulfate. In addition, the stable Cog4 KD cell line was cultured in DMEM/F-12 medium supplemented with 15 mm HEPES, 2.5 mml-glutamine, 10% FBS and 1.0 μg/mL puromycin. GalNAcT2-GFP HeLa cells were a gift from Dr Storrie (UAMS).

Vero cells were grown in DMEM/F-12 (50:50) with 15 mm HEPES, 2.5 mml-glutamine, 10% FBS and 1% antibiotic/antimycotic (100 U/mL penicillin G, 100 μg/mL streptomycin and 0.25 μg/mL amphotericin B). All cell culture media and sera are from Invitrogen.

shRNA- and siRNA-induced knockdowns

To generate a stable KD of Cog4, HeLa T2-GFP cells (15) were transfected with the shRNA plasmid (OriGene) (target sequence: TGACATCTTGGACCTGAAGTTCTGCATGG) using Lipofectamine 2000, according to a protocol from the manufacturer (Invitrogen). Medium containing 1 μg/mL puromycin was used to select for the plasmid. Colonies were chosen according to Cog4 downregulation as judged by immunoblotting.

Human Cog2 (target sequence: GGGCAGTTGATGAACGAAT), Cog4 (target sequence: GTGCTGAAATCCACCTTTA), Cog6 (target sequence: AGATATGACAAGTCGCCTA), Cog8 (target sequence: GATGATCCCTATTTCCATA), Cog3 (15), Snx1 and Snx2 (47), p115 (32), β-COP (26), Rab6 (17), golgin-84 (33), Vps26 (48) and Syntaxin 5 (49) siRNA duplexes were obtained from Dharmacon and Sigma-Proligo. Transfection was performed using Oligofectamine (Invitrogen), following a protocol recommended by Invitrogen. Two cycles of transfections (100 nm siRNA each) were performed and cells were analyzed 72 h after the second cycle.

For Vero cells, we used RNAiMAX to transfect Cog3 and Syntaxin 5 siRNA duplexes using a protocol from the supplier (Invitrogen).

Plasmid preparation and transfection

Plasmids for Rab6-GDP (pRab6S21N), CFP-Arf1-Q71L and mouse Cog3 were described previously (9,15,50). Mutant (siRNA resistant) versions of Cog4 and Cog8 were generated by GeneScript using plasmids pCOD1-myc and pDOR1-myc (51).

Plasmid transfections were performed with Lipofectamine 2000 (Invitrogen) or with FuGene (Roche), according to the manufacturer's protocol.

Immunofluorescence microscopy

For IF, cells are grown on 12-mm glass coverslips 1 day before transfection. Seventy-two hours after siRNA transfection, the cells are fixed and stained with antibodies as previously described (18,52). In short, cells were fixed in 4% paraformaldehyde (16% stock solution; Electron Microscopy Sciences). Cells were then treated with 1% Triton-X-100 for 1 min. After that the cells were treated with 50 mm ammonium chloride for 5 min. Cells were washed with PBS and blocked twice for 10 min with 1% BSA, 0.1% saponin in PBS. Cells were stained in primary antibody for 1 h (1% cold fish gelatin, 0.1% saponin in PBS) at room temperature. Cells were washed 4× with PBS and incubated for 30 min with fluorescently tagged secondary antibody (1:400 HiLyte; AnaSpec). After that the cells were washed 4× with PBS and rinsed with ddH2O. Cells were mounted on microscope slides using Prolong® Gold antifade reagent (Invitrogen). Cells were imaged with the 63× oil 1.4 numerical aperture (NA) objective of an LSM510 Zeiss Laser inverted microscope outfitted with confocal optics for image acquisition. Image acquisition is controlled with LSM510 software (Release Version 4.0 SP1).

Subtilase cytotoxin assay

Both control and KD GalNAcT2-GFP HeLa cells were grown on 12-well culture plates to 70–80% of confluency. Cells were placed in 10% FBS in DMEM/F-12 (50:50) without antibiotics and antimycotics for at least 1 h before the assay was performed. The SubAB was diluted in the same medium (without antibiotics) and warmed up to 37°C. The time–course is 0, 30, 60, 120 and 180 min. Cells were incubated with the SubAB for the appropriate times in the 37°C incubator with 5% CO2. Cells were then washed off with PBS and lysed with 2% SDS warmed up to 95°C. Ten microliters of cell lysates was loaded on a 12% gel and immunoblotted with an anti-GRP78 (Santa Cruz; C-20) antibody. All experiments were performed in triplicates. The blots were scanned and analyzed with an Odyssey Infrared Imaging System (LI-COR).

For Vero cells, the concentration of SubAB was changed to 0.05 μg/mL and the incubation was performed in a 32°C incubator.

Fluorescent labeling of the subtilase cytotoxin and Shiga-like toxin B

SubAB was labeled with HiLyte Fluor 555 (Dojindo Laboratories) and the SLTB was labeled with Alexa Fluor 647 (Invitrogen), according to the protocols supplied by the manufacturers.

Reverse transcriptase-polymerase chain reaction (RT-PCR)

Control, Cog2 KD and Cog3 KD HeLa cells were lysed and homogenized passing 5× using a 22-gauge needle. The mRNA from control and COG2 KD cells was isolated using the RNeasy Plus kit (Qiagen) using their protocol. RNA samples were translated to cDNA by using multiscribe reverse transcriptase (Applied Biosystems) in a BioRad MJ Mini Thermal cycler machine. The cDNAs were amplified using 25 cycles of polymerase chain reaction (PCR), using Taq DNA polymerase and primers for COG2 and COG3. Ten microliters of PCR product was loaded to a 1% agarose gel and imaged with BioRad Quantity One software.

SDS–PAGE and western blotting

SDS–PAGE and WB were performed as described previously (24). The blots were incubated first with primary antibodies, and then with a secondary immunoglobulin G (IgG) antibody conjugated with IRDye 680 or IRDye 800 dyes. The blots were scanned and analyzed with an Odyssey Infrared Imaging System (LI-COR). Alternatively, a signal was detected using an enhanced chemiluminescent reagent for HRP from Pierce and quantified using imagej software (http://rsb.info.nih.gov/ij/).

Acknowledgments

We are thankful to S. Munro, B. Storrie, E. Sztul and others who provided reagents and critical reading of the manuscript. We also would like to thank Paul Prather for entering data in the GraphPad Prism for calculating the IC50 and Imax values. Supported by grants from the NSF (MCB-0645163) and the NIH (1R01GM083144-01A1).

Ancillary