Golgin Tethers Define Subpopulations of COPI Vesicles

See allHide authors and affiliations

Science  18 Feb 2005:
Vol. 307, Issue 5712, pp. 1095-1098
DOI: 10.1126/science.1108061


Coiled-coil proteins of the golgin family have been implicated in intra-Golgi transport through tethering coat protein complex I (COPI) vesicles. The p115-golgin tether is the best studied, and here we characterize the golgin-84–CASP tether. The vesicles bound by this tether were strikingly different from those bound by the p115-golgin tether in that they lacked members of the p24 family of putative cargo receptors and contained enzymes instead of anterograde cargo. Microinjected golgin-84 or CASP also inhibited Golgi-enzyme transport to the endoplasmic reticulum, further implicating this tether in retrograde transport. These and other golgins may modulate the flow patterns within the Golgi stack.

The flow pattern of materials within the Golgi stack is governed by COPI vesicles. In one view, anterograde vesicles deliver newly synthesized protein and lipid cargo to successive cisternae, where they can undergo posttranslational modifications. Retrograde vesicles would then salvage components of the fusion machinery as well as those Golgi enzymes that have strayed beyond the cisterna(e) in which they function. In another view, the protein and lipid cargo do not move forward in vesicles; rather, the retrograde COPI vesicles carry Golgi enzymes and fusion machinery back to the preceding cisterna, which matures into the next one (1). A recent modification to this view suggests that tubular continuities mediate the retrograde transport of Golgi enzymes, leaving COPI vesicles to carry the fusion machinery (2). These views are not mutually exclusive, but they do require considerable coordination of the COPI vesicle flow pattern (3). The key question, then, is the mechanism that underlies the targeting of COPI vesicles to particular cisterna(e).

Targeting is a multilayered process, involving tethers and SNAREs (soluble N-ethylmaleimide–sensitive factor attachment protein receptors) (4). Tethers include multi-protein complexes and coiled-coil proteins of the golgin family, the latter regulated by small guanosine triphosphatases (GTPases) of the Rab and Arl families (57). The p115-golgin tether uses giantin, GM130, and Rab1 to tether COPI vesicles to the cis-Golgi network (CGN), preparing p115 for its role in SNARE pairing that leads to vesicle fusion (710).

Golgin-84 and CASP are recently characterized golgins that interact with each other (1114). Immunoprecipitation of golgin-84 brought down CASP (20 ± 3%, n = 4) from detergent extracts of Golgi membranes, and immunoprecipitation of CASP brought down golgin-84 (12 ± 2%, n = 4). Neither antibody brought down components of the p115-golgin tether (Fig. 1A) (12).

Fig. 1.

The golgin-84–CASP tether. (A) Triton extracts of Golgi membranes (10% input at left) were immunoprecipitated (IP) with the indicated antibodies or preimmune serum, and bound proteins were analyzed by SDS–polyacrylamide gel electrophoresis, followed by immunoblotting (IB) for the indicated proteins. (B) Salt-washed Golgi membranes were incubated with purified coatomer, ARF1, and guanosine triphosphate (GTP) for 10 min at 37°C, followed by salt treatment to release the COPI vesicles and centrifugation to remove large membrane fragments. The remaining membranes were centrifuged to equilibrium, and fractions were analyzed for the indicated proteins. (C and D) The gradient-purified vesicles (40% sucrose) were processed for (C) negative staining, (inset) conventional EM, or (D) pre-embedding labeling by using antibodies to golgin-84 and 5 nm protein A-gold (arrowheads). (E and F) Incubation as in (B), but stopped after 3 min to enrich for COPI-coated buds. Cryosections were double-labeled for β′-COP (5 nm gold) and either (E) golgin-84 or (F) CASP (both 10 nm gold).

The topology of the golgin-84–CASP tether was investigated with a cell-free budding assay and density-gradient centrifugation to isolate highly purified COPI vesicles (15). These peaked at ∼40% (w/v) sucrose (Fig. 1B), and electron microscopy (EM) revealed a homogeneous population of coated vesicles, ∼75 nm in diameter (Fig. 1C). Golgin-84, but not CASP, cofractionated with the COPI-coated vesicles (Fig. 1B), and immunolabeling localized golgin-84 to some of them (Fig. 1D).

Samples from the budding assay were also processed for immunolabeling, and golgin-84 was found on 77 ± 15% of those buds that labeled for β′-COP (Fig. 1E). The opposite result was obtained for CASP, which was found on only 7 ± 3% of those buds that labeled for β′-COP, most being present on cisternal membranes (Fig. 1F). Thus, the tether is asymmetric, with CASP in Golgi membranes and golgin-84 on the vesicles.

This asymmetry was confirmed by using soluble, recombinant CASP to coat glass slides that were incubated with partially purified COPI vesicles and then processed for EM (Fig. 2A). Uncoating was necessary for binding (fig. S1), and although these uncoated vesicles, ∼60 nm in diameter, were only a small fraction of the starting material (arrows in Fig. 2F), they alone were selectively bound (Fig. 2A). Binding was not observed to slides coated with bovine serum albumin (BSA)/soybean trypsin inhibitor (STI) (Fig. 2B) or to those coated with another coiled-coil protein, the endosomal antigen EEA1 (Fig. 2C) (16). Binding was also inhibited by soluble CASP (Fig. 2D) and soluble golgin-84 (Fig. 2E). Quantitation confirmed these results (Fig. 2G).

Fig. 2.

Tethering of COPI vesicles to CASP-coated slides. (A to E) Golgi membranes and cytosol were used to prepare partially purified and uncoated COPI vesicles that were incubated on glass slides precoated with [(A), (D), and (E)] recombinant CASP, (B) BSA and STI, or (C) recombinant EEA1 in the absence [(A), (B), and (C)] or presence of (D) soluble CASP or (E) soluble golgin-84. Slides were then washed and processed for EM. (F) Input membranes. (G) Quantitation of the results in [(A) to (E)], presented as the mean of the number of the bound vesicles per μm2 ± SD (n = 5).

The composition of the bound vesicles was then analyzed using CASP-coated glass beads instead of slides and gradient-purified and uncoated COPI vesicles as the starting material. Bound proteins could be competed with soluble CASP, showing that they were a specific component of these vesicles. About 10% of the input golgin-84 was retrieved by the CASP beads, and this was used as the standard to compare the retrieval of other proteins (Fig. 3A).

Fig. 3.

Tethering of COPI vesicles to CASP- or p115(TA)-coated glass beads. (A) Gradient-purified COPI vesicles were uncoated (10% input at left) and incubated for 30 min at 4°C with glass beads precoated with recombinant CASP (lanes 2 and 3) or p115(TA) (lanes 4 and 5) in the absence (lanes 2 and 4) or presence of soluble CASP (lane 3) or TA (lane 5). Bound proteins were analyzed for the indicated proteins. (B) As in (A), except that partially purified and uncoated COPI vesicles were used. (C) As in (A), using p115(TA) beads incubated with gradient-purified and uncoated COPI vesicles in the absence (lane 2) or presence (lanes 3 to 6) of the indicated recombinant proteins. Blots were probed for giantin.

The most notable difference between the starting and bound population of vesicles was the absence of any member of the p24 family of proteins. A representative of each of the four subfamilies was absent from the bound vesicles, a striking result because they have been implicated in the biogenesis of COPI vesicles, the cytoplasmic tails acting to bind coatomer and, thus, nucleate budding (17). Other proteins must clearly nucleate the budding of COPI vesicles that use the golgin-84–CASP tether; examples include the tails of Golgi enzymes (18). Mannosidase I and II were enriched ∼5-fold and ∼2-fold respectively, consistent with their movement from medial and/or trans cisternae (7, 19) back to the CGN, where the p24 proteins are localized (20). A representative cargo protein, the polymeric immunoglobulin A receptor (pIgR), was not enriched in the CASP-binding vesicles, further supporting a role in retrograde transport (Fig. 3A).

The lowered levels of adenosine diphosphate ribosylation factor–GTPase activating protein (ARF-GAP) in the salt-washed Golgi membranes used to prepare gradient-purified COPI vesicles helped to prevent uncoating and thus increased the yield. ARF-GAP has, however, been implicated in sorting (21, 22), so similar experiments were carried out with partially purified and uncoated COPI vesicles, and the compositional analysis was repeated for key proteins. CASP beads still retrieved vesicles containing Golgi enzymes but not the pIgR cargo or the p24 family member, p26 (Fig. 3B).

Another subpopulation of COPI vesicles was isolated by using the p115-golgin tether. A functional, truncated form of p115, TA (7), was used to coat glass beads that bound vesicles containing giantin (Fig. 3C). Binding was inhibited both by TA and by a soluble fragment of giantin but not by CASP, distinguishing this tether from that mediated by golgin-84–CASP.

The composition of the bound vesicles was then compared with the starting population and with those bound to CASP beads (Fig. 3A). In each case, the bound proteins were competed by TA and, although the efficacy was lower than when soluble CASP was used to compete binding to CASP beads, at worst binding was diminished by a factor of 3 and at best by at least a factor of 10. About 20% of the giantin in the input COPI vesicles was retrieved by p115(TA) beads, and this was used as the standard to compare the retrieval of other proteins.

All of the p24 family members were enriched, although to differing extents; the least was p23, and the most was p26, which was three times as high as that in the input vesicles. The cargo protein, pIgR, was also enriched by at least a factor of 3, but both Mannosidase I and II were depleted. This is consistent with the vesicles budding from the CGN, although a role in anterograde transport would have to be limited to the early part of the Golgi stack (from the CGN to cis cisternae, for example), because p115 is restricted to these membranes (23). COPI vesicles have been isolated by using antibodies to the tail of the p24 protein, and these are depleted for both enzymes and cargo (22). Thus, more than one population of p24-containing COPI vesicles likely exists, with only some carrying cargo molecules.

Each subpopulation of COPI vesicle incorporated the vesicle tether that operated in the other vesicle. The p115(TA)-binding vesicles (using giantin) contained golgin-84, and the CASP-binding vesicles (using golgin-84) contained giantin (Fig. 3A). This cannot be the consequence of cross-contamination, given the complete absence of the p24 family of proteins in the CASP-binding vesicles. A possible explanation is that both golgin-84 and giantin are membrane proteins, so if they are incorporated into vesicles that move to another part of the stack, then there must be a means of recycling them back to their original membrane. Golgin-84, for example, after fusion of the vesicles with cis membranes, would need to be recycled back to trans membranes. The obvious mechanism would be to use vesicles moving in the opposite direction and to inactivate the passenger golgin by using a reversible modification such as phosphorylation (7).

The cycling of Golgi enzymes through the endoplasmic reticulum (ER) (24, 25) provided a means of testing the role of the golgin-84–CASP tether in retrograde transport. BSC-1 cells, stably expressing GalNAc-T2-YFP (26), were microinjected with Sar1dn to inhibit ER export, which relocated this Golgi enzyme to the ER over a 5-hour period, the Golgi fluorescence dropping from 80% to 5% (Fig. 4B) (25, 27). Neither CASP nor golgin-84 changed their locations appreciably during this time period (fig. S2). Coinjection of either soluble golgin-84 or soluble CASP substantially inhibited this retrograde movement (Fig. 4A), such that 40 to 55% of the fluorescence remained in the Golgi region (Fig. 4B). Kinetic analysis showed an increase in the halftime for relocation from ∼75 min (Sar1dn alone) to ∼235 min (+ golgin-84 or CASP) (Fig. 4C). Coinjection of EEA1 had no effect (Fig. 4, A and B).

Fig. 4.

Soluble golgin-84 and CASP inhibit retrograde transport of Golgi enzymes. (A) BSC-1 cells, stably expressing GalNAc-T2-YFP, were injected with Sar1dn (top row) or coinjected with soluble golgin-84 (second row), soluble CASP (third row), or EEA-1 (bottom row). After the injection, cells were imaged at the indicated time points. Asterisks mark the uninjected cells. Bar, 10 μm. (B) Quantitation of results in (A), as well as additional experiments in which components of the p115-golgin tether were either coinjected with Sar1dn or, in the case of Δ63GM130, expressed before injection. Results presented as the mean of fluorescence in the Golgi region ± SD (n ranges from 4 to 12 cells sampled). (C) Quantitation of the results in (A), expressed as the time taken to relocate half the GalNAc-T2 to the ER. Results presented as mean ± SD (n ranges from 4 to 7 cells sampled). (D) BSC-1 cells were microinjected with cDNA encoding tsO45 VSV-G protein and incubated at 2 hours at 40°C. Soluble golgin-84 or CASP were microinjected before shift to the permissive temperature of 32°C. The time taken for half the G protein to reach the cell surface is presented as the mean ± SD (n = 3 cells per condition).

Agents that disrupted the p115-golgin tether had only a modest effect on the retrograde transport of GalNAc-T2-YFP. Coinjection of an N-terminal fragment of giantin or GM130 or expression of Δ63-GM130, which lacks the N-terminal domain, left 10 to 15% of fluorescence in the Golgi region, compared with 5% for Sar1dn alone (Fig. 4B).

Expression of Δ63-GM130 inhibits anterograde transport of VSV-G protein by up to 60% (28). In contrast, microinjection of soluble golgin-84 had no significant effect on the halftime for VSV-G protein to reach the cell surface (41 min versus 38 min), and soluble CASP had only a modest effect (41 min versus 54 min) (Fig. 4D). RNA interference (RNAi) depletion of golgin-84 also has a partial effect on VSV-G protein transport (12).

Finally, soluble golgin-84 and CASP did not affect the retrograde movement of ERGIC53 from the CGN to the ER (fig. S3) (29), which suggests that the CASP-binding vesicles are moving within the Golgi, not from the Golgi to the ER. This was also supported by the relative absence of the p24 proteins and the KDEL-R in the vesicles isolated on CASP beads (Fig. 3, A and B). Earlier work did not uncover a role for COPI vesicles in the retrograde transport of Golgi enzymes to the ER (24), despite the presence of enzymes in isolated COPI vesicles (9, 22). The present work provides a possible resolution, suggesting a two-step retrograde pathway, the first step mediated by intra-Golgi COPI transport to the CGN, the second by a COPI-independent pathway to the ER.

The ability to isolate and analyze subpopulations of COPI vesicles brings a much-needed biochemical approach to the study of intra-Golgi transport. Other tethers should identify other subpopulations, allowing us to map the flow patterns of resident, cargo, and recycling molecules. The intersections will dictate the nature and extent of the modifications to the transiting cargo, which should help us understand the functioning of the Golgi in different cells, tissues, and organisms.

Supporting Online Material

Materials and Methods

Figs. S1 to S3


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article