The glycoprotein of VSV accumulates in a distal Golgi compartment in the
presence of CCCP
JANIS K. BURKHARDT, SUSAN HESTER and YAIR ARGON*
Department of Microbiology and Immunology, Box 3010, Duke University Medical Center, Durham, North Carolina 27710, USA
•Author for correspondence
Summary
The post-translational modifications of the G pro-tein of vesicular stomatitis virus, described in thepreceding paper, indicate that its transport is ar-rested by carbonylcyanide m-chlorophenylhydra-zone (CCCP) in or near the trans-Golgi. Immuno-fluorescence microscopy of BHK-21 cells infectedwith vesicular stomatitis virus and treated withCCCP shows an accumulation of G protein in theGolgi area. In the same cells, the morphology ofwheat germ agglutinin (WGA)-staining structuresin the perinuclear region is aberrant. Using anti-BiPantibody, there is no obvious change in the struc-ture of the endoplasmic reticulum. Electron mi-croscopy reveals that the aberrant structures in theperinuclear region result from dilation of Golgicisternae and accumulation of large vacuoles nearthe Golgi stack. The appearance of these aberrant
structures is dose-dependent and they disappearafter the protonophore is removed. The vast ma-jority of the vacuoles accumulate on the trans sideof the Golgi stack. A small fraction of them containthe marker enzyme thiamine pyrophosphatase(TPPase). By immunoelectron microscopy, most ofthe vacuoles contain G protein. We conclude thatmost of the Golgi-associated vacuoles are derivedfrom a distal Golgi transport compartment, poss-ibly the trans-Golgi reticulum, and that CCCPreversibly inhibits the transport of newly syn-thesized G protein through this distal compart-ment.
Export of membrane and secreted proteins entails theirorderly transfer between organelles of the secretoryapparatus. The first major transfer is from the endoplas-mic reticulum (ER) to the Golgi complex, and the secondis between the Golgi complex and the final destination ofthe proteins. It is during these two transitional transportsteps that exported proteins are concentrated (Salpeter &Farquhar, 1981; Quinn et al. 1984), sorted (reviewed byFarquhar, 1985; Pfeffer & Rothman, 1987), and theirtransport kinetics regulated (Fries et al. 1984; Lodish etal. 1983; Scheele & Tartakoff, 1985). As with any activeprocess, these transitional steps require ATP and aretemperature sensitive (Jamieson & Palade, 1968; Balch etal. 1986; Matlin & Simons, 1983; Tartakoff, 1986).
Relatively little is known about the membrane com-partments that mediate the transitions from the ER to theGolgi complex and from there to the plasma membrane.Transitional elements are thought to represent structuresthrough which proteins exit from the ER (Palade, 1975),but there is as yet no direct demonstration of such a role
for transitional elements. Likewise, the complicatedsystem of membrane vesicles and tubular elements,known as the /rans-Golgi reticulum (TGR; Willingham& Pastan, 1984), the trans-Go\gi network (TGN; Grif-fiths & Simons, 1986) or GERL (Novikoff, 1964), isthought to represent the structure where protein trafficexits the Golgi stack (Griffiths & Simons, 1986). There isevidence that a variety of proteins are sorted and pack-aged into vesicles in the TGR. Condensing vacuoles,which give rise to secretory granules, bud from this partof the Golgi (Hand & Oliver, 1977; Jamieson & Palade,1967; Novikoff et al. 1977). Lysosomal hydrolases andthe mannose 6-phosphate receptor have been localized toits clathrin-coated regions (Geuze et al. 1985). Proteinsdestined for plasma membrane accumulate in this organ-elle when their transport is blocked at 20°C (Griffiths etal. 1985; Saraste & Kuismanen, 1984).
To date, the TGR is defined almost exclusively bymicroscopy. Given its probable role in protein transport,the purification and characterization of the TGR is ofgreat interest. This is difficult, however, because thearchitecture of the TGR is complicated. It contains
643
numerous tubular elements and a variety of vesicles.Some of these vesicles are smooth, some are clathrin-coated, and others bear a different, unidentified coat(Griffiths et al. 1985; Orci et al. 1986). Some of thestructures in the region are endocytic (Marsh et al. 1986;Willingham et al. 1984), while others probably mediateER-to-Golgi and intra-Golgi traffic. Only a subpopu-lation is involved in mediating traffic from the Golgi toplasma membrane. Although the TGR contains sialyl-transferase and acid phosphatase (Hand & Oliver, 19846;Novikoff, 1964; Roth et al. 1985), there are at present nomarkers that are unique to this compartment. For thetime being, the TGR and other distal compartments arebest defined by the passage of transported proteins.
The protonophore CCCP inhibits the transport of avariety of membrane and secreted proteins at two stages(Tartakoff & Vassalli, 1979; Fries & Rothman, 1980;Godelaine et al. 1981; Argon & Milstein, 1984; Kabcenell& Atkinson, 1985; Argon et al. 1989; Burkhardt &Argon, 1989). The post-translational modifications of thearrested proteins indicate that these two stages corre-spond to the transitions between the ER and Golgi, andthe Golgi and the plasma membrane. Because of thespecificity of its action, CCCP provides a means ofidentifying the membrane compartments that correspondto these transport stages, by determining where thearrested proteins accumulate.
The fact that CCCP acts specifically at both transporttransitions might indicate a common underlying mechan-ism. However, because CCCP inhibits two transportstages, the analysis of either one alone is complicated. Inthe preceding paper (Burkhardt & Argon, 1989), wedescribed a system in which the G protein of vesicularstomatitis virus (VSV) is largely arrested at the latetransitional step. The G protein arrested in the presenceof CCCP bears sialylated complex oligosaccharides butdoes not appear on the cell surface. We interpret this toindicate transport of G protein as far as the trans-Golgi.
In this paper, we take advantage of the fact that Gprotein is refractory to arrest at the first CCCP-sensitivestage, in order to characterize the site of the late transportarrest by microscopy. We show that in the presence ofCCCP, G protein indeed accumulates in the trans-Go\g\region, and provide evidence that the site of accumu-lation is the TGR. In addition, we describe the disrup-tion of Golgi structure by CCCP, and show that thisstructural alteration is related to the arrest of proteintransport.
Materials and methods
Cell culture, virus infections and drug treatmentsBHK-21 cells were grown as described in the preceding paper(Burkhardt & Argon, 1989). The IgD-producing hybridomaBl-8.<5 (Neuberger & Rajewsky, 1981) was grown in RPMI-1640 supplemented with glutamine, penicillin, streptomycin,and 5% foetal calf serum. Viral infections were as described(Burkhardt & Argon, 1989). Infections of cells with VSV andtreatments with CCCP, monensin and cycloheximide weredone as described in the preceding paper (Burkhardt & Argon,1989).
Light microscopy
Fluoresceinated anti-G (DTAF-H14D5) was prepared as de-scribed (Burkhardt & Argon, 1989). Wheat germ agglutinin(WGA) was directly conjugated to rhodamine-B-isothiocyanate(TRITC) by a modification of the procedure described byGoding (1976). WGA binds to glycoproteins with complexcarbohydrates and therefore serves as a stain for the Golgicomplex, distal exocytic compartments, the plasma membraneand endocytic compartments (Virtaanen et al. 1980). To markthe endoplasmic reticulum, we used a rat monoclonal antibody(MAb) against the resident ER protein BiP (Bole et al. 1986; agenerous gift from Dr David Bole, Yale University). This MAbwas detected with MAR18.S, an anti-rat if light-chain MAb(Lanier et al. 1982), tagged with rhodamine as described forWGA. BHK cells were grown to subconfluence on sterilecoverslips. After infection and treatment with CCCP, cover-slips were rinsed with PBS to remove serum proteins, fixed for1 h in fresh 2% paraformaldehyde, 0-1% glutaraldehyde inPBS, and washed three times with 50niM-ammonium chloridein PBS. For immunofluorescence, cells were permeabilized byincubation for lOmin in 1% Triton X-100, 0-25% gelatin,PBS. For direct labelling, coverslips were incubated for 1 h withDTAF-H14D5 or TRITC-WGA, both at lO^gml"1 in0-25 % gelatin/PBS, and rinsed six times with the same buffer.For indirect labelling, coverslips were incubated with undilutedanti-BiP culture supernatant, rinsed with gelatin/PBS, andincubated for an additional hour with TRITC-MAR18.5 at10 jug ml"1. Double-labelling regimes in either order gave thesame results as with each reagent separately. Following label-ling, coverslips were rinsed briefly in water, mounted on slideswith mounting medium containing 2-5% DABCO (Poly-sciences, Warrington, PA) to minimize fading, and examinedwith a Zeiss IM35 inverted microscope equipped with epifluor-escence optics. Photographs were taken using T-MAX 400(Kodak, Rochester, NY).
Electron microscopy
Subconfluent BHK cells were washed with balanced saltsolution (BSS) and removed from culture dishes by digestionwith 50/igml"1 proteinase K for approximately 5 min at 4°C.Phenylmethylsulphonyl fluoride was added to 40j[/gml~1
) cellswere pelleted and resuspended in a small volume of BSS, andfixed in suspension at room temperature with 2% glutaralde-hyde in 150mM-sodium cacodylate, pH7-4, 0-01% CaCl2.After 10 min, the cells were centnfuged and fixed for anadditional hour as a pellet. The fixed cell pellets were embeddedin 1 % agar and postfixed for 1 h at 4°C with 2% osmiumtetroxide, 1 % potassium ferrocyanide, 150 mM-sodium cacody-late, pH7-2. Following washes with sodium cacodylate andsodium acetate, the agar blocks were stained en bloc with 1 %uranyl acetate, 0-2M-sodium acetate, pHS-2, for 1 h at roomtemperature. Following washes with sodium acetate and water,the samples were dehydrated through a graded ethanol seriesand embedded in EMBED 812 (EM Sciences, Fort Wash-ington, PA). Silver sections were contrasted with uranyl acetateand lead citrate, and observed with a Philips EM300 electronmicroscope at 80 kV.
Enzyme cytochemistry
Thiamine pyrophosphatase (TPPase) cytochemistry was doneusing the method of Novikoff & Goldfischer (1961), as modifiedby Hand & Oliver (1984«). Cells washed with BSS were fixedfor 10 min in suspension with 1% paraformaldehyde, 1%glutaraldehyde, 0-05% CaCl2, 01 M-sodium cacodylate,pH 7-2. Cells were then washed with cacodylate buffer contain-ing 7 % sucrose, embedded in agar, and stored overnight at 4CC.
644 jf. K. Burkhardt et al.
After several washes in Tris-maleate buffer, agar blocks wereincubated in complete reaction mix (2mM-TPP-HCl, 3-6 mM-lead nitrate, 5mM-MnCl2, 5% sucrose, 0-1 M-Tris-maleate,pH7-2) for 1 h at 37°C with shaking. The specificity of thisreaction was demonstrated by omission of substrate. Thereaction mixture was replaced after 30min of incubation.Following the reaction, cells were washed several times andtreated with 1 % ammonium sulphide for light microscopy.They were then postfixed with 2% osmium tetroxide, 1%potassium ferrocyanide, 7% sucrose, 0-1 M-sodium cacodylate,pH 7'2, washed with cacodylate, sucrose and water, and staineden bloc for 1 h with 0-5 % aqueous uranyl acetate. Dehydrationand embedding was as described above.
Immunoelectron microscopyImmunoelectron microscopy was done by the ultrathin frozen-sectioning procedure of Tokuyasu (1980), as modified byGriffiths et al. (19836). Cells were washed with BSS and fixedfor lOmin at room temperature with 4 % paraformaldehyde,ISOmM-Pipes, pH7-2, followed by overnight fixation in 8%paraformaldehyde in the same buffer. Cell pellets were cryopro-tected by infiltration with 2 1 M-sucrose in PBS, and frozen inliquid nitrogen. Sections were cut at — 100°C using a ReichertUltracut E equipped with an FC4E cryounit. Sections collectedon a loop containing 2-3 M-sucrose in PBS were transferred toFormvar-coated grids and floated onto 5% FCS in PBS as ablocking step. After washing with PBS, grids were incubated on5 j*l drops of H14D5 at 5 0 / i g m r ' (diluted in 5 % FCS in PBS)for 30 min at room temperature, and washed on several drops ofPBS for a total of 15 min. Grids were then incubated on drops ofappropriately diluted protein A-colloidal gold (gift from Dr G.Griffiths, EMBL, Heidelberg), and washed again on PBS for30min. After a brief water wash, grids were incubated for10min at 4°C on drops of 2% methyl cellulose, 0-3% uranylacetate. Grids were then looped out, excess methyl cellulose wasremoved, and the grids were dried in a desiccated chamber.
Results
G protein is present in the Golgi region of CCCP-treatedcellsThe post-translational modifications borne by CCCP-arrested G protein predict that most of it has reached theGolgi complex (Burkhardt & Argon, 1989). To test this,we used immunofluorescence microscopy to localize thesite of G protein accumulation. At 3 h after VSV infec-tion, cells were incubated in the presence or absence ofCCCP for 60min, fixed, permeabilized and stained withDTAF-anti-G (Fig. 1, left). In order to visualize betterthe intracellular distribution of G protein, surface G wassaturated with unlabelled antibody prior to permeabiliz-ation. To mark the Golgi complex, the plasma membraneand endocytic vesicles, the cells were counterstained withTRITC-WGA (Fig. 1, bottom panel). The localizationof G protein in CCCP-treated cells is different fromcontrol cells. In both treated and untreated cells, the Gprotein accumulates in the perinuclear region. However,whereas the perinuclear G protein labelling in controlcells is reticular, the perinuclear G protein in CCCP-treated cells shows a punctate pattern, characteristic ofcoarse vesicles. The perinuclear distribution of G proteinoverlaps the pattern of WGA staining, but the correspon-dence is not precise. The labelling pattern of G protein
also corresponds well to the distribution of TPPase (notshown, but see Fig. 6). Therefore, we interpret theperinuclear labelling of G as reflecting accumulation inthe Golgi region.
In addition to the perinuclear concentration, G proteinis present in the cell periphery of both CCCP-treated andcontrol cells. Since the nuclear envelope is often stained(see arrowhead in bottom panel), we presume that at leastpart of this peripheral G protein is present in the ER. Todemonstrate this further, we used double labelling withanti-G and a MAb against BiP, a resident ER protein(Bole et al. 1986) (Fig. 1, top panel). Unlike G protein,BiP does not accumulate in the perinuclear region. Thedistribution of BiP overlaps most, but not all of theperipheral G protein label. This pattern is not surprising:some G protein is expected to be in the ER, but inaddition, virions are known to be present in endocyticvesicles (data not shown).
Since we observed accumulation of G in the Golgiregion of CCCP-treated cells, we looked for a concomi-tant depletion of G from the ER. In an effort to 'chase' Gprotein out of the ER and into the Golgi, BHK cells weretreated with a combination of cycloheximide and CCCP.The fluorescence pattern resulting from this combinationof drugs did not differ from the pattern due to CCCPalone (data not shown). In agreement with the analogouspulse-chase experiments described in the precedingpaper (Burkhardt & Argon, 1989), these immunofluor-escence results show that some G protein remains in theER during CCCP treatment, while the majority accumu-lates in the Golgi region.
CCCP disrupts WGA-positive structuresIn the course of the co-localization studies, we observedthat CCCP treatment alters the structure of some organ-elles. The reticular anti-BiP staining pattern is not alteredsignificantly by CCCP treatment (Fig. 1, top panel),implying that ER structure is not grossly affected.Staining with TRITC-WGA, however, reveals signifi-cant effects of CCCP treatment on cell structure (Fig. 2).The typical pattern of Golgi staining with WGA showscompact perinuclear staining in addition to finer punctatestaining that is probably due to endocytic structures(Fig. 2A). Instead, CCCP-treated cells exhibit large,disperse, WGA-positive patches that are often distrib-uted far from the nucleus (Fig. 2B). This suggests thatCCCP treatment disrupts the structure of the Golgicomplex or other distal transport organelles. In addition,WGA staining of CCCP-treated cells is invariably moreintense than that of control cells, perhaps because CCCPcauses the accumulation of cellular glycoproteins in theGolgi region.
It should be noted that the disruption of WGA-positivestructures was consistently less severe in VSV-infectedcells than in uninfected cells (compare Fig. 1, lower rightpanel with Fig. 2B). The reason for this 'stabilization' isunknown.
Like the inhibition of G protein transport, the mor-phological alteration induced by CCCP treatment can bereversed by washing out the protonophore. To quantifythe reversibility, we scored the disruption of WGA-
G protein arrest in distal Golgi vacuoles 645
Q_Ooo
Q_OOO
Fig. 1. Localization of G protein by double-label immunofluorescence. VSV-infected BHK cells were either not treated ortreated for 60min with CCCP (20 /.IM). Fixed and permeabilized cells were then double-labelled with DTAF-anti-G and anti-BiP/TRITC-anti-K (top panel), or with DTAF-anti-G and TRITC-WGA (bottom panel). The reactivity due to cell surface Gwas reduced by incubation with unlabelled anti-G prior to permeabilization. Staining of the nuclear envelope is marked with anarrowhead.
646 jf. K. Burkhardt et al.
Fig. 2. TRITC-WGA labelling of uninfected cells treatedwith CCCP. Uninfected BHK cells were not treated (A), ortreated for 90min with 20/iM-CCCP (B). After fixation andpermeabilization, cells were labelled with TRITC-WGA.Note that in CCCP-treated cells (B), large, intense WGA-positive patches are dispersed throughout the cytoplasm. Incomparison, untreated cells (A) exhibit only fine punctatestaining in the cell periphery.
stained structures, by determining the percentage of cellsdisplaying large, WGA-positive patches >0-5 of a nucleardiameter away from the nucleus. Samples that weretreated with 20 JZM-CCCP for 90 min contained 84 % suchcells (« = 495). Only 8% of untreated cells had thisdistribution (n = 391). If treated cells were incubated forone hour in the absence of CCCP, this fraction decreasedto 30% (n = 217), showing that the effect is reversible.
Specific disruption of Golgi ultrastructureThe morphological effects of CCCP treatment weredefined further by analysis at the ultrastructural level. Aspredicted by the immunofluorescence results, we ob-served that CCCP alters the ultrastructure of the Golgi
complex without affecting the structure of the ER. Withthe exception of the expected changes in mitochondria(Hackenbrock, 1968) other organelles appear normal. Ascompared to untreated controls (Fig. 3A), BHK cellstreated for 90 min with CCCP exhibit both dilation ofGolgi cisternae and pronounced accumulation of largevacuoles in the Golgi area (Fig. 3B). In these cells, somevacuoles are distributed further from the Golgi region,accounting for the patchy fluorescence that we observewith WGA labelling.
The morphological effects of CCCP are not limited toBHK cells. Under conditions where CCCP arrests thesecretion of immunoglobulin by hybridoma cells (Argon& Milstein, 1984), we observe a similar disruption ofGolgi ultrastructure (Fig. 4A and B). As in BHK cells,the Golgi cisternae are dilated and vacuoles accumulate inthe Golgi region. Another feature of CCCP treatment isthat the Golgi cisternae usually remain in close appo-sition, keeping the stack structure largely intact. In thisrespect the effects of CCCP are distinct from those ofmonensin, which under similar conditions completelydisrupts the structure of the Golgi stack (compareFig. 4B and C). As compared with BHK cells, theCCCP-induced vacuoles in hybridoma cells are moreclosely confined to the Golgi region. This makes it moreapparent that the vacuoles accumulate preferentially onone side of the Golgi stack (see below).
The disruption of the Golgi complex by CCCP is notdue to the accumulation of transported proteins. Ifhybridoma cells are treated with cycloheximide to blockprotein synthesis, and then treated with CCCP, thedisruption of Golgi ultrastructure is the same as withCCCP alone (Fig. 4D). Treatment with cycloheximidealone does not alter Golgi structure (data not shown).
In order to characterize the dose-dependence ofCCCP's effects on Golgi structure, Bl-8.6 hybridomacells were treated for 90 min with CCCP at concen-trations ranging from 0-100 /ZM, and with 10/iM-CCCPfor times up to 4 h, and analysed by electron microscopy.The severity of Golgi disruption was scored according toan arbitrary scale, as described in the legend to Fig. 5. Asshown in Fig. 5A, the primary effect on Golgi structure isprobably the dilation of Golgi cisternae, because it isevident already at 1 fiM-CCCP. The vacuolization nearthe stack is already maximal at 5 /iM-CCCP, and thedisruption increases in severity over the range of5-100 fiM. Even at the highest doses and the longest timesof CCCP tested (100 JUM for 90min or 10/iMfor4h), theGolgi stack remains intact. Although the cisternae be-come quite dilated and contorted, adjacent cisternaeretain their close apposition.
As shown in Fig. SB, the disruption of Golgi structureby CCCP occurs very quickly. Abnormal stacks arevisible by 5 min, and the disruption is maximal by 30 minof treatment. Over the course of 4 h, the Golgi disruptiondoes not change significantly, but by 4 h the cells begin toshow signs of toxicity, including condensation of chroma-tin and dilation of the ER and nuclear envelope. Theseresults suggest that the Golgi complex is very sensitive tothe effects of CCCP, since the disruption occurs with aslittle as 2^M-CCCP, and within as few as 5 min.
G protein arrest in distal Golgi vacuoles 647
3A
BOi'
Fig. 3. Electron micrographs of the disrupted Golgi region in CCCP-treated BHK cells. A. Untreated cells; B, cells treatedwith 10jiM-CCCP for 90min. n, nucleus; m, mitochondrion; /, lysosome; er, endoplasmic reticulum; invb, multivesicular body;v, CCCP-induced vacuoles; arrowheads, coated vesicles. Bar, 0-5 /im.
648 jf. K. Burkhardt et al.
Fig. 4. Golgi morphology in hybridoma cells treated with transport inhibitors. Bl-8.<5 cells were untreated (A), or treated asfollows: B, 10/iM-CCCP for 90min; C, 10/iM-monensin for 90min; D, lOO^gmF1 cycloheximide for 60min followed by 10//M-CCCP for 90min in the continued presence of cycloheximide. Bar, 1 pm.
The CCCP-induced vacuoles are on the trans side of theGolgi stackAs noted above, the CCCP-induced vacuoles accumulatepreferentially on one side of the Golgi stack. Since thesite of vacuole accumulation has important implicationsfor the effects of CCCP on protein transport, we usedcytochemistry to determine whether the vacuoles ac-cumulate on the cis or the trans side of the Golgi stack.Bl-8.(5 cells were untreated, or treated for 90min with10/XM-CCCP, and the reaction product of thiaminepyrophosphatase was used to mark the trans side of theGolgi stack (Novikoff & Goldfischer, 1961; Hand &Oliver, 1984a). As an additional means of establishingGolgi polarity, we used the cluster of small, smoothvesicles that is often seen on the cis side of the Golgistack, away from the TPPase-positive cisternae (Fig. 6,arrows). In control cells (Fig. 6A), TPPase reactivity ispresent in the last one or two cisternae of the Golgi stack;the TGR is usually unreactive.
When CCCP-treated cells are assayed histochemicallyfor TPPase (Fig. 6B), it can be seen that the vacuolesaccumulate preferentially on the TPPase-positive, or
trans, side of the Golgi stack (Table 1). Of the 53 Golgistacks counted (one Golgi counted per cell), noneexhibited the vacuoles only in the cw-Golgi; 83% hadvacuoles exclusively or predominantly in the trans-Go\g\.In 11/53 Golgi complexes, one or more vacuoles were
Table 1. Distribution of CCCP-induced vacuoles
Side of Golgi*
Only transMostly traits^Cis + traits^Mostly cisOnly cis
Total
Number
IS
#3'
s•iS
Percentage
28551160
100
• Only Golgi complexes that satisfied the following criteria werescored: they showed a clearly polarized TPPase reaction product inthe stack and they were sufficiently isolated from other stacks to scorethe association of CCCP-induced vacuoles.
•f Golgi complexes, where more than 3/4 of the associated vacuoleswere on the trans side of the stack.
^ Golgi complexes, where the associated vacuoles were equallyprominent on both sides of the stack.
G protein arrest in distal Golgi vacuoles 649
30 60 90 120Time (min)
Fig. 5. Dose dependence of Golgi disruption. Bl-8.6 cellswere treated for 90 min with various concentrations of CCCP(A), or with lO^iM-CCCP for various times (B), and analysedby electron microscopy. Golgi morphology was scored asfollows: category 1, normal morphology ( • ) ; category 2,dilated cisternae without vacuoles (O); category 3, dilatedcisternae with vacuoles in the Golgi region (A) ; category 4,dilated cisternae with disperse vacuoles ( • ) . Samples werecoded and analysed randomly. Fifty stacks were scored foreach data point.
themselves TPPase positive. The majority of vacuoles,however, did not contain the TPPase reaction product.We interpret these results to mean that most of theCCCP-induced vacuoles derive from membranes distal tothe last Golgi cisterna, perhaps from the TGR.
G protein accumulates in the CCCP-induced vacuolesOur biochemical studies indicated that G protein reachesthe trans-Go\g\, even in the presence of CCCP (Burk-hardt & Argon, 1989). Since CCCP causes vacuolizationin the /raws-Golgi region, we asked whether G accumu-lates in the vacuoles. Infected BHK cells were untreatedor treated for 90 min with CCCP, and processed forcryosectioning. Sections were labelled with anti-G andprotein A-colloidal gold. Uninfected cells used to controlfor antibody specificity were essentially free of label (datanot shown). In untreated cells (Fig. 7A), G protein isdistributed throughout the secretory pathway. Label \spresent over ER (not shown), Golgi and the plasmamembrane. Virions on the plasma membrane and inendosomes are also labelled. In CCCP-treated cells, thevast majority of gold particles is present over the Golgistack and over large vacuoles in the Golgi region(Fig. 7B—D). Some of these vacuoles are clearly distinct
from the stack itself (arrowheads in Fig. 7C and D).None of the CCCP-induced vacuoles contained virions,labelled or unlabelled. This makes it unlikely that thevacuoles are endosomal structures. Lower levels of Gprotein label are present over other membranes of thesecretory pathway (not shown). This distribution isconsistent with our inability to chase G protein out of theER, as measured either by immunofluorescence or bycarbohydrate maturation. Despite the presence of Gprotein in other compartments, visual comparison ofsamples like those shown in Fig. 7 clearly shows accumu-lation of G protein in the CCCP-induced, Golgi-associ-ated vacuoles.
Discussion
We show here that the protonophore CCCP causesdistinct ultrastructural changes in the Golgi complex oftreated cells. Together with the data presented in thecompanion paper (Burkhardt & Argon, 1989), a straight-forward relation emerges between the biochemical andmorphological effects of CCCP: it arrests G proteintransport in the compartment whose structure it disrupts.
The apparent absence (or leakiness) of the earlyCCCP-sensitive stage in this system (Burkhardt &Argon, 1989) makes it possible to ask where G proteinaccumulates when arrested at the late CCCP-sensitivestage. Immunofluorescence confirms the prediction fromthe post-translational modifications of G protein, that it isenriched in the Golgi region. That other glycoproteinsalso accumulate in the Golgi regions of CCCP-treatedcells is indicated by the increased intensity of WGAlabelling, even in uninfected cells.
The Golgi complexes of CCCP-treated BHK cells donot appear normal, even at the light-microscope level.They seem vesiculated, and are no longer strictly perinu-clear. As seen by electron microscopy (EM), the Golgicisternae are dilated, and large vacuoles accumulate in theGolgi region. This morphology is characteristic not onlyof BHK cells, but also of myeloma cells under conditionswhere CCCP inhibits the secretion of Ig (Argon &Milstein, 1984). The alteration of Golgi structure isdependent on the dose of CCCP. At lower drug concen-trations, which produce only partial inhibition of Igsecretion (Argon & Milstein, 1984), only dilated cisternaeare observed. Doses of CCCP that produce completeinhibition of transport correlate with the appearance ofboth dilated disternae and Golgi-associated vacuoles. Noother morphological changes are observed at higherdoses, until toxic doses are reached.
Surprisingly, we have detected no effects of CCCP onER structure by either light or electron microscopy. At allthe doses tested (below the toxic dose of 100 /XM) there isno dilation or vesiculation of ER elements. Moreover, wecould not detect any paucity of transitional elementssimilar to that reported by Tartakoff (1986) in dinitro-phenol (DNP)-treated pancreas cells. This is true even inmyeloma cells, where the export of Ig from the ER isblocked by CCCP (Tartakoff & Vassalli, 1979; Argon &Milstein, 1984).
650 J. K. Burkhardt et al.
i
1
B »• •
Fig. 6. Thiamine pyrophosphatase activity in CCCP-treated cells. Bl-8.<5 cells were untreated (A), or treated for 90min with10//M-CCCP (B), and reacted for TPPase. Note the presence of large CCCP-induced vacuoles (v) on the TPPase-positive sideof the Golgi stack. Arrowheads, c/s-Golgi vesicles. Bar, 0 1 Jim.
Because the CCCP-induced vacuoles seem to underliethe inhibition of protein transport, we wanted to identifythe compartment(s) from which they are derived. Mostof the CCCP-induced vacuoles are WGA-positive, as seenby immunofluorescence (Fig. 1) and by immunoelectronmicroscopy (data not shown). In addition, most of theCCCP-induced vacuoles are positive for ricin communisagglutinin I binding (not shown), indicating that theirglycoproteins contain galactose (Virtaanen et al. 1980).These lectin bindings show that most of the vacuoles arederived from compartments distal to the mid-Golgi(Roth & Berger, 1982). The use of a more specific markerfor the trans-Go\g\, thiamine pyrophosphatase (TPPase)activity (Hand & Oliver, 1984*; Novikoff & Goldfischer,1961), confirmed this conclusion. The vast majority ofthe vacuoles accumulate on the TPPase-positive, ortrans, side of the Golgi complex. Most of the TPPasereactivity remains in the Golgi stack after treatment withCCCP, consistent with the observation that the stackremains intact. The CCCP-induced vacuoles themselvesare seldom positive for TPPase, making it unlikely that
they derive from the trans-Go\g\ cisternae. Takentogether, this marker analysis suggests that the Golgi-associated, CCCP-induced vacuoles represent dilatedpost-Golgi structures.
We have also examined the distribution of anotherdistal Golgi marker, acid phosphatase reactivity (Novi-koff, 1964; Hand & Oliver, 19846). However, the levelsof acid phosphatase in the Golgi complexes of BHK andmyeloma cells (even in the absence of CCCP) wereinsufficient to determine unequivocally whether thevacuoles contain this marker.
We cannot at present exclude the possibility that someof the CCCP-induced vacuoles are derived from othertransport compartments. Few, if any, of the vacuoleswould be expected to come from Golgi-proximal com-partments (such as the transition from the ER to the cis-Golgi), because most label with lectins specific forterminal sugars, and because most contain G protein thatbears complex gLycans. We also expect that few, if any, ofthe vacuoles are endocytic structures. We consider thisunlikely for three reasons. First, in our hands the
G protein arrest in distal Golgi vacuoles 651
pm
pm
• G
mvb
G
n7A B
G •
Fig. 7. EM immunolocalization of G protein in CCCP-treated BHK cells. Ultrathin frozen sections of infected BHK cells werelabelled with anti-G followed by protein A-colloidal gold. A. Untreated cells; B-D, cells treated for 90min with CCCP. G,Golgi stack; in, mitochondrion; w, nucleus; pm, plasma membrane; mvb, multivesicular body; arrowhead in A, budded virion.Arrows in C and D point to G protein-containing vacuoles that are clearly distinct from the Golgi stack. Vacuoles that are notclearly bounded by a membrane, and may therefore be a drying artifact, are marked by *. Bar, 0-5/im.
652 jf. K. Burkhardt et a!.
internalization of [lzsI]transferrin bound to its surfacereceptor is not inhibited by CCCP (Wiest & Argon,unpublished). Second, the CCCP-induced vacuoles donot contain VSV particles, only G protein, while endo-somal structures in VSV-infected cells often containvirions. Third, immunolabelling of myeloma cells treatedwith CCCP shows that the vacuoles contain high concen-trations of immunoglobulin (Burkhardt, Dul & Argon,unpublished data), and in these cells the endocytic trafficof immunoglobulin is negligible (Pernis, 1985).
The ultrastructural alteration caused by CCCP isreminiscent of the effects of another transport inhibitor,monensin. Both drugs inhibit protein transport by dis-rupting a compartment related to the Golgi complex,thereby trapping the exported proteins in the alteredcompartment. The effects of the two ionophores differ,however, in that monensin disrupts the architecture ofthe Golgi stack (Tartakoff & Vassalli, 1977; and Fig. 4),while CCCP causes vacuolization at its periphery, with-out breaking up the stack itself. These differences areconsistent with the biosynthetic stages inhibited by thetwo drugs: monensin inhibits mid-Golgi processing(Griffiths et al. 1983a) and disrupts internal Golgistructure, whereas CCCP inhibits late processing events(Argon et al. 1989; Burkhardt & Argon, 1989), anddisrupts peripheral Golgi elements. Quinn et al. (1983)showed that when the transport of Semliki Forest virus(SFV) glycoprotein is arrested in monensin-inducedvacuoles, incomplete virions bud into the disrupted Golgimembranes. No similar intracellular budding of VSV isobserved in CCCP-induced vacuoles although they con-tain high concentrations of G protein. It will be interest-ing to examine the ability of SFV to bud into the CCCP-induced vacuoles. The dilation of the CCCP-sensitivecompartment, as well as the accumulation of viral glyco-proteins within it, provide experimental means for itsisolation and future biochemical characterization.
As a protonophore, CCCP neutralizes acidic compart-ments (Poole & Okhuma, 1981). Traws-Golgi elementshave been shown to be mildly acidic (Anderson & Pathak,1985). Thus, the selective effect of CCCP on particularmembranes in the Golgi region may reflect pH differ-ences among Golgi subcompartments. Indeed, otherweak bases also cause dilation of trans-Golgi elements(Geuze et al. 1985; Thorens & Vassalli, 1986). CCCParrests transport at two stages, both of which have beenpreviously shown to require ATP, but dilates only onecompartment. It may therefore be argued that thedisruption of trans-Go\g\ elements results from pHeffects rather than ATP depletion. Whether the accumu-lation of the arrested proteins in this compartment resultsdirectly from pH or ATP perturbation remains to berigorously tested.
A likely interpretation of our data is that the compart-ment marked by sensitivity to CCCP is the distalcompartment known as the /raws-Golgi reticulum (Will-ingham & Pastan, 1984; Griffiths & Simons, 1986). Thevast majority of the dilated structures are associated withthe trans side of the stack, as expected for the TGR. Theanalyses of marker enzymes and lectin binding are alsdconsistent with this interpretation. Another supporting
observation is the occasional finding of clathrin-like coatson portions of the CCCP-induced vacuoles. One definingcharacteristic of the TGR is its high clathrin content, ascompared with other Golgi elements (Friend & Farqu-har, 1967; Orci et al. 1985; Willingham et al. 1981).Finally, the most important criterion for defining theTGR is the state of maturation of the proteins containedin it. The vast majority of CCCP-arrested G proteinbears ER and czs-Golgi modifications, such as jV-glycos-ylation and acylation with palmitate (Burkhardt & Argon,1989). Its glycans contain galactose and are partly sialyl-ated (Burkhardt & Argon, 1989). This is the expectedphenotype of a TGR-arrested protein (Fuller et al. 1985;Griffiths et al. 1985). In this paper, we show that thearrested G protein indeed accumulates in the CCCP-induced vacuoles. Thus, it seems likely that by alkaliniz-ing the TGR elements, CCCP causes their swelling sothat G protein accumulates in them and is unable tocomplete its transport to the plasma membrane.
We thank Dr D. Lyles for his generous supply of monoclonalantibodies, Dr D. Bole for MAb anti-BiP, Dr J. Keene foradvice and for gifts of VSV stocks, and Dr S. Miller forguidance in electron microscopy. We are indebted to Dr G.Griffiths for introducing us to cryoultramicrotomy and lmmu-nocytochemistry and for many helpful discussions. Our thanksalso to A. Stockdale, D. Wiest and Drs C. Lapham and J. Dulfor critical reading of this manuscript. This work is in partialfulfilment of the requirements for the Ph.D. degree of DukeUniversity (J.K.B.). It was supported by NIH grants AI-08817and AI-23282. J.K.B. was supported by NIH training grantsGM-07184 and CA-09058.
References
ANDERSON, R. G. W. & PATHAK, R. K. (1985). Vesicles andcisternae in the /ra;u-Golgi apparatus of human fibroblasts areacidic compartments. Cell 40, 635-643.
ARGON, Y., BURKHARDT, J. K., LEEDS, J. M. & MILSTEIN, C.(1989). Reversible inhibition of the intracellular transport ofimmunoglobulin D. J. lmmun. 142, 554-561.
ARGON, Y. & MILSTEIN, C. (1984). Intracellular processing ofmembrane and secreted immunoglobulin 6 chains. J. lmmun. 133,1627-1633.
BALCH, W. E., ELIOT, M. M. & KELLER, D. S. (1986). ATP-coupled
transport of vesicular stomatitis virus G protein between theendoplasmic reticulum and the Golgi. J biol Chem 261,14681-14689.
BOLE, D. G., HENDERSHOT, L. M. & KEARNEY, J. F. (1986).
Posttranslational association of immunoglobulin heavy chainbinding protein with nascent heavy chains in nonsecreting andsecreting hybridomas. J. Cell Biol. 102, 1558-1566.
BURKHARDT, J. K. & ARGON, Y. (1989). Intracellular transport of theglycoprotein of VSV is inhibited by CCCP at a late stage of post-translational processing. J. Cell Sci. 92, 633-642.
FARQUHAR, M. G. (1985). Progress in unravelling pathways of Golgitraffic. A. Rev. Cell Biol. 1, 447-488.
FRJEND, D. S. & FARQUHAR, M. G. (1967). Functions of coatedvesicles during protein absorption in the rat vas deferens. J. CellBiol. 35, 357-376.
FRIES, E., GUSTAFSSON, L. & PETERSON, P. A. (1984). Four
secretory proteins synthesized by hepatocytes are transported fromendoplasmic reticulum to Golgi complex at different rates. £JV/BOJ. 3, 147-152.
FRIES, E. & ROTHMAN, J. E. (1980). Transport of vesicular stomatitisvirus glycoprotein in a cell-free extract. Proc. natn. Acad. Sci.U.SA. 77, 3870-3874.
FULLER, S. D., BRAVO, R. & SIMONS, K. (1985). An enzymatic assay
G protein arrest in distal Golgi vacuoles 653
reveals that proteins destined for the apical and basolateral domainsof an epithelial cell line share the same late Golgi compartments.EMBOJ. 4, 297-307.
GEUZE, H. J., SLOT, J. W., HASIUK, A. & VON FIGURA, K. (1985).
Possible pathways for lysosomal enzyme delivery. J . Cell Biol. 101,2253-2262.
GODELAINE, D., SPIRO, M. J. & SPIRO, R. G. (1981). Processing of
the carbohydrate units of thyroglobuhn.^. biol. Chem. 256,10161-10168.
CODING, J. W. (1976). Conjugation of antibodies withfluorochromes: modifications to the standard methods. J. Immiin.Meth. 13, 215-226.
GRIFFITHS, G., PFEIFFER, S., SIMONS, K. & MATUN, K. (1985). Exit
of newly synthesized membrane proteins from the trans cisterna ofthe Golgi complex to the plasma membrane. .7. Cell Biol. 101,949-964.
GRIFFITHS, G., QUINN, P. & WARREN, G. (1983O). Dissection of the
Golgi complex I: Monensin inhibits the transport of viral proteinsfrom medial to trans-Golgi cisternae in baby hamster kidney cellsinfected with Semliki Forest virus. J . Cell Biol 96, 835-850.
GRIFFITHS, G. & SIMONS, K. (1986). The trans Golgi network:Sorting at the exit site of the Golgi complex. Science 234,438-443.
GRIFFITHS, G., SIMONS, K., WARREN, G. & TOKUYASU, K. T.
(19836). Immunoelectron microscopy using thin, frozen sections:application to studies of the intracellular transport of SemlikiForest virus spike glycoproteins. Meth. Enzym. 96, 435-450.
HACKENBROCK, C. R. (1968). Ultrastructural bases for metabolicallylinked mechanical activity in mitochondria: II. Electron transport-linked ultrastructural transformations in mitochondria. J. Cell Biol.37, 345-369.
HAND, A. R. & OLIVER, C. (1977). Relationship between the Golgiapparatus, GERL, and secretory granules in acinar cells of the ratexorbital lacrimal gland. J . Cell"Biol. 74, 399-413.
HAND, A. R. & OLIVER, C. (1984a). Effects of secretory stimulationon the Golgi apparatus and GERL of rat parotid acinar cells. J.Histochem. Cytochem. 32, 403-412.
HAND, A. R. & OLIVER, C. (19846). The role of GERL in thesecretory process. In Cell biology of the secretory process (ed. M.Cantin), pp. 148-170. Basel: Karger.
JAMIESON, J. D. & PALADE, G. E. (1967). Intracellular transport ofsecretory proteins in the pancreatic exocnne cell: II. Transport tocondensing vacuoles and zymogen granules. J. Cell Biol. 34,597-616.
JAMIESON, J. D. & PALADE, G. E. (1968). Intracellular transport ofsecretory proteins in the pancreatic acinar cell: IV. Metabolicrequirements. J . Cell Biol. 39, 589-603.
KABCENELL, A. K. & ATKINSON, P. H. (1985). Processing of roughendoplasmic reticulum membrane glycoproteins of rotavirus SAILJ. Cell Biol. 101, 1270-1280.
LANIER, L. L., GUTMAN, G. A., LEWIS, D. A., GRISWOLD, S. T. &
WARNER, N. L. (1982). Monoclonal antibodies against ratimmunoglobulin kappa chains. Hybridoma 1, 125-131.
LODISH, H. F., KONG, N., SNIDER, M. & STROUS, G. J. A. M.
(1983). Hepatoma secretory proteins migrate from roughendoplasmic reticulum to Golgi at characteristic rates. Nature,Land. 304, 80-83.
MARSH, M., GRIFFITHS, G., DEAN, G. E., MELLMAN, I. &
HELENIUS, A. (1986). Three dimensional structure of endosomesin BHK-21 cells. Proc. natii. Acad. Sci. U.SA. 83, 2899-2903.
MATLIN, K. S. & SIMONS, K. (1983). Reduced temperature preventstransfer of a membrane glycoprotein to the cell surface but doesnot prevent terminal glycosylation. Cell 34, 233-243.
NEUBERGER, M. S. & RAJEWSKY, K. (1981). Switch from hapten-specific immunoglobulin M to immunoglobulin D secretion in ahybrid mouse cell line. Proc natn. Acad. Sci. U.SA. 78,1138-1142.
NOVIKOFF, A. B. (1964). GERL, its form and function in neurons ofrat spinal ganglia. Biol. Bull mar. Biol. Lab., Woods Hole 127, 358.
NOVIKOFF, A. B. & GOLDFISCHER, S. (1961). Nucleosidediphosphatase activity in the Golgi apparatus and its usefulness forcytological studies. Proc. natn. Acad. Sci. U.SA. 17, 802-812.
NOVIKOFF, A. B., MORI, M., QUINTANA, N. & YAM, A. (1977).
Studies of the secretory process in the mammalian pancreas: I.
The condensing vacuoles. J Cell Biol. 75, 148-165.ORCI, L., GUCK, B. S. & ROTHMAN, J. E. (1986). A new type of
coated vesicle carrier that appears not to contain clathrin: Itspossible role in transport within the Golgi stack. Cell 46, 171-184.
ORCI, L., RAVAZZOLA, M., AMHERDT, M. & PERRELET, A. (1985).
Clathrin immunoreactive sites in the Golgi apparatus areconcentrated at the trans pole in polypeptide hormone-secretingcells. Proc. natn. Acad. Sci. U.SA. 82, 5385-5389.
PALADE, G. (1975). Intracellular aspects of the process of proteinsecretion. Science 189, 347-358.
PERNIS, B. (1985). Internalization of lymphocyte membranecomponents. Immun. Today 6, 45-49.
PFEFFER, S. R. & ROTHMAN, J. E. (1987). Biosynthetic proteintransport and sorting by the endoplasmic reticulum and Golgi. A.Rev. Biochem. 56, 829-852.
POOLE, B. & OHKUMA, S. (1981). Effect of weak bases on theintralysosomal pH in mouse peritoneal macrophages. J. Cell Biol.90, 665-669.
QUINN, P., GRIFFITHS, G. & WARREN, G. (1983). Dissection of the
Golgi complex II: Density separation of specific Golgi functions invirally infected cells treated with monensin. J. Cell Biol. 96,851-856.
QUINN, P., GRIFFITHS, G. & WARREN, G. (1984). Density of newlysynthesized plasma membrane proteins in intracellular membranes:II. Biochemical studies. J. Cell Biol. 98, 2142-2147.
ROTH, J. & BERGER, E. G. (1982). Immunocytochemical localizationof galactosyl-transferase in HeLa cells: codistribution withthiamine pyrophosphatase in (ra/is-Golgi cisternae. J. Cell Biol. 92,223-229.
ROTH, J., TAATJES, D. J., LUCOCQ, J. M., WEINSTEIN, J. &
PAULSON, J. C. (1985). Demonstration of an extensive trans-tubular network continuous with the Golgi stack that may functionin glycosylation. Cell 43, 287-295.
SALPETER, M. & FARQUHAR, M. G. (1981). High resolutionautoradiographic analysis of the secretory pathway inmammotrophs of the rat anterior pituitary. ,7. Cell Biol. 91,240-246.
SARASTE, J. & KUISMANEN, E. (1984). Pre- and post-Golgi vacuolesoperate in the transport of Semliki Forest virus membraneglycoproteins to the cell surface. Cell 38, 538-549.
SCHEELE, G. & TARTAKOFF, A. (1985). Exit of nonglycosylatedsecretory proteins from the rough endoplasmic reticulum isasynchronous in the exocrine pancreas. J. biol. Client. 260,926-931.
TARTAKOFF, A. M. (1986). Temperature and energy dependence ofsecretory protein transport in the exocrine pancreas. EMBOJ. 5,1477-1482.
TARTAKOFF, A. M. & VASSALU, P. (1977). Plasma cellimmunoglobulin secretion: Arrest is accompanied by alterations inthe Golgi complex. J. exp. Med. 146, 1332-1345.
TARTAKOFF, A. & VASSALLI, P. (1979). Plasma cell immunoglobulinM molecules: their biosynthesis, assembly, and intracellulartransport. J. Cell Biol. 83, 284-299.
THORENS, B. & VASSALU, P. (1986). Chloroquine and ammoniumchloride prevent terminal glycosylation of immunoglobuhns inplasma cells without affecting secretion. Nature, Land. 321,618-620.
TOKUYASU, K. T. (1980). Immunocytochemistry on ultrathin frozensections. Histochem. J. 12, 381-403.
VIRTAANEN, I., EKBLOM, P. & LAURILLA, P. (1980). Subcellular
compartmentalization of saccharide moities in cultured normal andmalignant cells. J. Cell Biol. 85, 429-434.
WILUNGHAM, M. C , HANOVER, J. A., DICKSON, R. B. & PASTAN, 1.(1984). Morphologic characterization of the pathway of transferrinendocytosis and recycling in human KB cells. Proc. natn. Acad.Sci. U.SA. 81, 175-179.
WILUNGHAM, M. C , KEEN, J. H. & PASTAN, I. H. (1981).
