Differential localization of coatomer complexisoforms within the Golgi apparatusJörg Moelleken*, Jörg Malsam*, Matthew J. Betts†, Ali Movafeghi‡, Ingeborg Reckmann*, Ingrid Meissner*,Andrea Hellwig§, Robert B. Russell†, Thomas Söllner*, Britta Brügger*, and Felix T. Wieland*¶

*Biochemistry Center and §Department of Neurobiology, University of Heidelberg, 69120 Heidelberg, Germany; †Structuraland Computational Biology Unit, European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany;and ‡Department of Plant Biology, Faculty of Natural Sciences, Tabriz University, Tabriz, Iran

Communicated by James E. Rothman, Columbia University Medical Center, New York, NY, December 20, 2006 (received for review December 8, 2006)

Coatomer, the coat protein of coat protein complex (COP)I-vesicles,is a soluble protein complex made up of seven subunits, �-, �-, ��-,�-, �-, �-, and �-COP. Higher eukaryotes have two paralogousversions of the �- and �- subunits, termed �1- and �2-COP and �1-and �2-COP. Different combinations of these subunits are known toexist within coatomer complexes, and �1/�1-, �1/�2-, and �2/�1-COP represent the major coatomer populations in mammals. Therole of COPI vesicles in the early secretory pathway is the subjectof considerable debate. To help to resolve this discussion, we usedquantitative immunoelectron microscopy and found that signifi-cant localization differences for COPI-isoforms do exist, with apreference for �1�1- and �1�2-coatomer in the early Golgi appa-ratus and �2�1-coatomer in the late Golgi apparatus. These differ-ences suggest distinct functions for coatomer isoforms in a mannersimilar to clathrin/adaptor vesicles, where different adaptor pro-teins serve particular transport routes.

coat protein complex

Intracellular vesicular transport is mediated by small protein-coated carriers. Various functions are attributed to coated vesi-cles, which are defined by their constituent coat proteins. Isoformswith different combinations of coat protein complex (COP)IIprotein paralogs have been found in the COPII coats on vesiclesbudding from the endoplasmic reticulum (ER) of mammals andyeast (1–3). Because only one direction of travel from the ER to theGolgi apparatus is known for COPII vesicles, these isoforms mightreflect differing cargo specificities (4). In the late Golgi/endocyticpathways, a variety of coat proteins are known, including fourdifferent tetrameric adaptor complexes, some of which use theclathrin system as an outer shell of their coat (5). Within thesepathways, particular transport routes are served by specific coatcomplexes (for review, see refs. 6 and 7). Thus, for particularpathways, the selection of cargo as well as the direction might bedetermined by individual coats.

Just which pathways are mediated by vesicles bearing theCOPI coat is the subject of considerable debate. The COPI coatconsists of coatomer, which is a stable cytoplasmic complex ofseven subunits, �-, �-, ��-, �-, �-, �-, and �-COP, and the smallGTP-binding protein ADP-ribosylation factor, Arf1. COPI ves-icles were found to originate from the Golgi apparatus and areimplicated in transport within the Golgi itself in anterograde andretrograde direction, and from the Golgi and the ER–Golgiintermediate compartment (ERGIC) to the ER (8–10). Theobservation that COPI vesicles participate in several transportdirections is difficult to rationalize with just a single isoform ofthe complex (reviewed in ref. 11). Questions about the functionof COPI vesicles are intimately linked to our view of the generalmechanisms of transport through the Golgi. Specifically, istransport in both directions, anterograde and retrograde, per-formed by these carriers, or are COPI vesicles exclusivelyretrograde carriers within the Golgi and from the Golgi to theER? Moreover, is anterograde transport across the Golgi exclu-sively mediated by maturation of Golgi cisternae (12, 13)?

Genome analyses identified a second isoform for each of thetwo coatomer subunits, �- and �-COP (14), the paralogs �2- and�2-COP. Previously, the corresponding cDNAs were expressedas tagged constructs (15), and the endogenous gene productswere shown to be incorporated into coatomer complexes in astoichiometric manner, which resulted in three major isoforms ofthe complex with the compositions �1�1-, �1�2-, and �2�1-COP(16). To extend this work, we have prepared polyclonal antibod-ies specific against �1-, �2-, and �2-COP and have investigatedthe localization of coatomer isoforms in NRK cells by usingimmunoelectron microscopy. We found a striking heterogeneitywithin the Golgi apparatus: whereas �1- and �2-COP seemrestricted to the early Golgi and a pre-Golgi compartment, themajority of �2-COP-containing isoforms of the complex islocalized to the trans side of the organelle. These results stronglysuggest the existence of different homogeneously coated iso-forms of COPI vesicles that might serve multiple transport routeswithin the early secretory pathway.

Results�- and �-COP Subunits from Other Species. Sequence searches of �-and �-COP subunits against completely sequenced genomes wereused to identify the protein families to which they belong. Theresulting peptide sequences were aligned [see supporting informa-tion (SI) Table 1], and phylogenetic trees were built from thesealignments. The trees strongly support concurrent duplication of �-and �-COP into two paralogs in the vertebrates, sometime aftertheir divergence from Ciona intestinalis (SI Figs. 5 and 6). Ciona,plants, insects, fungi, and the malaria parasite Plasmodium falcipa-rum all have one �- and �-COP isoform each, with very recentduplications and/or alternative transcripts (possibly annotationartifacts) that cluster together, implying that they are duplicationsspecific to a particular species. Absences of isoforms in somevertebrates might be an artifact of unfinished genomes, whereas thefish appear to have really lost �1-COP after the duplication, becausenone of the three fish species examined have �1-COP. The appear-ance of multiple isoforms of coatomer might be linked to theevolution of complex, multicellular organisms with cells processinga much wider range of cargo proteins.

Generation and Characterization of Antibodies Directed Against COPSubunit Isoforms. �-COP resembles �-adaptin, which is made ofan N-terminal ‘‘trunk’’ domain and a C-terminal ‘‘appendage’’

Author contributions: J. Moelleken, T.S., and F.T.W. designed research; J. Moelleken, J.Malsam, M.J.B., A.M., I.R., I.M., and A.H. performed research; J. Malsam, M.J.B., A.M., andA.H. contributed new reagents/analytic tools; J. Moelleken, J. Malsam, M.J.B., R.B.R., T.S.,B.B., and F.T.W. analyzed data; and J. Moelleken, R.B.R., B.B., and F.T.W. wrote the paper.

The authors declare no conflict of interest.

Abbreviations: COP, coat protein complex; ER, endoplasmic reticulum; ERGIC, endoplasmicreticulum–Golgi intermediate compartment.

¶To whom correspondence should be addressed. E-mail: felix.wieland@bzh.uni-heidelberg.de.

This article contains supporting information online at www.pnas.org/cgi/content/full/0611360104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0611360104 PNAS � March 13, 2007 � vol. 104 � no. 11 � 4425–4430

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/content/full/0611360104/DC1http://www.pnas.org/cgi/content/full/0611360104/DC1http://www.pnas.org/cgi/content/full/0611360104/DC1http://www.pnas.org/cgi/content/full/0611360104/DC1http://www.pnas.org/cgi/content/full/0611360104/DC1

domain (17–20). The appendage domains are generally the leastsimilar in sequence between �1- and �2-COP; for example, inmouse, the sequence identity is 75% in the appendage domains,compared with 81% in the trunk. We therefore expressedconstructs in Escherichia coli that contained the appendagedomains of �1- and �2-COP linked to GST. The purifiedproducts were used to raise antibodies in rabbits, and theresulting antisera (anti-�1-app and anti-�2-app) were character-ized by using immunoprecipitation experiments, Western blot-ting, and immunofluorescence microscopy. Both antisera areable to precipitate coatomer from cytosol (Fig. 1A). Antibodiesthat are directed against ��-COP (lane 1), a subunit of coatomerthat is shared by all complex isoforms, or CM1 antibodies (lane3), which are directed against native coatomer, precipitate allisoforms of coatomer (containing �1-, �2-, �1-, and �2-COP).However, anti-�1-app antibodies and anti-�2-app antibodiesspecifically precipitate coatomer containing �1-COP (lane 5) or�2-COP (lane 7), respectively. Likewise, an antibody directed

against an N-terminal sequence of �2-COP (16) showed aremarkable preference for coatomer containing �2- and �1-COP(lane 11). As a control, an unrelated antibody (lane 2 and 8) andthe corresponding preimmune sera (lanes 4, 6, and 10) do notsignificantly precipitate coatomer.

We investigated the specificity of the antibodies further by usingimmunofluorescence microscopy (Fig. 1B). For this investigation,fixed and permeabilized 3T3/NIH fibroblasts were incubated withthe various antisera, and bound antibodies were visualized by usingfluorescence-labeled secondary antibodies. The antisera against�1-COP (Fig. 1Bb), �2-COP (Fig. 1Bf), and �2-COP (Fig. 1Bk) gaverise to a pattern typical of the Golgi apparatus, which colocalizeswith coatomer markers (data not shown). The staining could beblocked by competition from the respective protein constructs (Fig.1 Bc and Bh) and by a peptide representing the N-terminal domainof �2-COP (Fig. 1Bl). Coincubation of the anti-�1-app antiserumwith GST-�2-app protein (Fig. 1Bd) and the anti-�2-app antiserumwith GST-�1-app (Fig. 1Bg) did not eliminate specific binding.Therefore, these antibodies can be used for immunocytochemistryand immunoprecipitation with high specificity for �1-COP-, �2-COP-, and �2-COP-containing coatomer. No antibody with spec-ificity for �1-COP was available for these applications. Becausecoatomer is a stable complex that is recruited to the donormembrane en bloc (21), these antibodies can be used to analyze thestoichiometry and localization of coatomer isoforms.

Stoichiometry of Coatomer Isoforms. The antibodies were used toreassess the stoichiometry of the various coatomer isoforms in amammalian cytosol. Immunoprecipitations were performed on3T3/NIH cytosol, and the individual subunits of precipitatedcoatomer were quantified by using the recombinant GST constructsthat contained the �1- or �2-appendage domain and recombinantHis-tagged versions of �1- and �2-COP as mass standards (Fig. 2).We found a 70:30 ratio of �1- to �2-COP in total coatomer afterimmunoprecipitation with the CM1 antibody (directed against allcoatomer isoforms) (Fig. 2 Aa and Ab). Similarly, immunoprecipi-tations with anti-�1-app or anti-�2-app antibodies gave a �1- and�2-COP ratio of 77:23 in �1-coatomer isoforms (Fig. 2 Ac and Ad)and a 86:14 ratio in �2-COP that contained immunoprecipitates(Fig. 2 Ae and Af). These numbers allowed the calculation of therelative proportion of coatomer isoforms, as shown in Fig. 2B.Isoforms of coatomer containing �1/�1-, �1/�2-, and �2/�1-COPwere found at a ratio of �10:3:5. The �2�2-coatomer is found in amuch smaller quantity, which is probably close to the sensitivitylimit of the analytical method used.

Intra-Golgi Localization of the Coatomer Isoforms Containing �1-, �2-,and �2-COP. The antibodies were used for immunogold labeling onultrathin cryosections of NRK cells. Primary antibodies werevisualized with protein A that was labeled with 15-nm gold. Tovalidate the polarity of the Golgi sections investigated, antibodiesthat were directed against the well established cis-Golgi markerprotein GM130 were used (22). This marker antibody was visual-ized with 10-nm gold bound to protein A. Representative labelingpatterns are shown in Fig. 3. Whereas preferential gold staining isobserved at the cis half of the Golgi with antibodies directed against�1- (Fig. 3 A and B) and �2-COP (Fig. 3E), the �2-coatomer (Fig.3 C and D) is restricted more to the trans half of the organelle. Asexpected, labels of antibodies directed against the common subunit��-COP are observed throughout the Golgi apparatus (Fig. 3F).Images that were obtained with the various antibodies were eval-uated quantitatively by separating Golgi areas along an arbitrarymiddle line, defining a cis half and a trans half of the organelle, andcounting the gold dots in each section. As shown in Fig. 4, �70%of �1-COP-containing coatomer was found in the cis half. The�2-COP coatomer showed an even stronger preference, with �80%localized to the early Golgi. The �2-coatomer, in contrast, showeda distribution of �60% trans and �40% cis. As a control, only a

Fig. 1. Specificity of the antibodies used. (A) Immunoprecipitations (IP) withcytosol were performed with anti-��-COP (lane 1), CM1 (lanes 3 and 9), anti-�1-app (lane 5), anti-�2-app (lane 7), and anti-�2-COP (lane 11) antibodies. Theeluates were separated by using SDS/PAGE and analyzed by Western blotting,using the indicated antibodies. Okt8 (lanes 2 and 8) is a CD8-specific antibody.Lanes 4, 6, and 10 reflect corresponding �1-, �2-, and �2-preimmune sera, respec-tively. (B) Indirect immunofluorescence analysis using anti-�1-app (b), anti-�2-app (f), and anti-�2-COP (k) antibodies shows a typical perinuclear staining. The�1-COPsignalcanbeeliminatedbycoincubationoftheantibodywithGST-�1-app(c) but not GST-�2-app protein (d). The �2-COP signal is unaffected by a coincu-bation of GST-�1-app (g) but is decreased drastically in the presence of GST-�2-app protein (h). That the �2-COP signal is significantly reduced in the presence ofthe corresponding peptide (l) indicates the specificity of this antibody. Micro-graphs a, e, and i show the signal background of the corresponding Alexa-fluor-conjugated secondary antibodies in the absence of primary antibodies.

4426 � www.pnas.org�cgi�doi�10.1073�pnas.0611360104 Moelleken et al.

Dow

nloa

ded

by g

uest

on

July

2, 2

021

slight preference, �55% cis and �45% trans, was observed whenoverall coatomer localization was studied with the anti-��-COPantibody (Fig. 3F). This is in agreement with a previous study inendocrinic pancreatic cells (23). The stringent localization of thecis-Golgi marker protein GM130 (22) to only one half of the Golgicorroborates the validity of the analytical method used. The markedlocalization of �2-coatomer to the early Golgi has led us toinvestigate this finding in further detail, by comparison with theKDEL receptor, a resident of the ERGIC and the cis-Golgi (24, 25).A pattern was obtained as depicted in Fig. 3G, where the antibodiesagainst the KDEL receptor were labeled with 10-nm gold, andantibodies against �2-COP with 15-nm gold. Again, a significantpreference of �2-coatomer for the ERGIC and the cis-Golgi wasobserved. Taken together, the significant differential localization ofcoatomer isoforms suggests that they serve different functionswithin the Golgi apparatus.

DiscussionWe have identified clear coatomer isoforms with different combi-nations of the �- and �-COP subunits in the vertebrates. Theseisoforms, like the other five COPs shared by all coatomers, arepresent stoichiometrically in their constituent complexes (16).Phylogeny strongly suggests that this diversification of �- and �-COPsubunits is a vertebrate innovation. Such gene family expansions arecommon in vertebrate-specific processes, such as the immunesystem, olfaction, and particular developmental processes (26).

Using new antibodies available against the two �-COP subunits,we have reassessed the relative abundance of coatomer in mam-malian cytosol and have found coatomer isoforms that contain thesubunits �1/�1-, �1/�2-, and �2/�1-COP in a ratio of �10:3:5.Coatomer with �2�2-COP, if it exits, represents �5% of the total.

Strikingly, the coatomer isoforms are differentially distributedwithin the Golgi apparatus, with �1-COP found preferentially inthe early Golgi and �2-coatomer found preferentially in the lateGolgi. A double labeling experiment with the KDEL receptor, anERGIC/cis-Golgi marker (27, 28), underscores the preference of�2-coatomer for these compartments.

The distinct localization of individual coatomer isoformsstrongly suggests the presence of various types of COPI vesiclesin the cell, each of which is homogeneously coated with a definedisotypic coatomer complex. Given the pattern emerging fromimmunoelectron microscopy, gradients are highly likely to existin the relative abundance of such vesicles, with more �1- and�2-COPI vesicles in the early Golgi and more �2 vesicles in thelate Golgi. Alternatively, or in addition, COPI vesicles that areenveloped by various mixtures of coatomer isoforms might exist(although this possibility is less likely given, e.g., the strongpreference of �2-coatomer for the early Golgi).

At present, it is unknown how the specific coatomer isoforms arerecruited to a certain Golgi cisterna. Interestingly, the difference ofthe isoforms lies in their ��-subcomplexes, and it is �-COP that hastwo binding sites for oligomeric forms of the members of the p24family of membrane proteins (29) that constitutively cycle betweenER and Golgi (30–33). Differential ��-COP combinations mightwell have different affinities to the cytoplasmic tails of these type1 transmembrane proteins or to different combinations thereof.Thus, if these proteins or their heteromeric complexes were un-equally distributed across the Golgi, their differential interactionwith coatomer isoforms could explain differential recruitment ofthe complex. Indeed, a preferential localization of p23, p24, and p27at the cis-Golgi has been shown (32–34). Additionally, individualcoatomer isoforms might differ in their affinity for the cytoplasmic

Fig. 2. Quantification of coatomer isotypes. (A) Immunoprecipitations (IPs) using CM1 (a and b), anti-�1-app (c and d), or anti-�2-app (e and f ) antibodies wereperformed with freshly prepared cytosol. Calibration curves obtained from Western blot analyses with GST-�1-app, GST-�2-app, His6-�1-COP, or His6-�2-COPproteins allowed the determination of the absolute amounts of protein in the samples. To the right of the quantification blots are the molar ratios, taking molarmasses into account. (B) Diagram showing the relative proportion of coatomer isotypes. Error bars represent the SD about the mean.

Moelleken et al. PNAS � March 13, 2007 � vol. 104 � no. 11 � 4427

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

2, 2

021

tails of various Golgi-resident membrane proteins, causing differ-ential recruitment to individual cisternae. Indeed, COPI vesiclesthat contain either p24 proteins or the Golgi-resident enzymesmannosidase I and II have been described (35). Likewise, strongevidence for functionally different COPI vesicles comes from animmunoelectron microscopy study in which two distinct popula-tions of COPI vesicles were defined by inclusion of either the KDELreceptor or proinsulin (23). Similarly, GOS28, a Golgi v-SNARE(soluble N-ethylmaleimide sensitive factor attachment protein re-ceptor), was found in COPI vesicles devoid of the KDEL receptorand vice versa (36). The existence of different populations of COPIvesicles had been deduced previously, based on the analysis ofcarriers immuno-isolated with antibodies directed against p27 (37).However, because these carriers were uncoated, it was not com-pletely clear whether they were derived from COPI-coated vesicles.

A variety of functions have been attributed to COPI vesicles.These include the transport of proteins cycling between theGolgi and the ER and the retrieval of proteins that have escapedthe ER (10, 38). Originally, COPI vesicles were defined andcharacterized as carriers for biosynthetic cargo across the Golgiapparatus, and various anterograde cargo proteins were found tobe passengers (23, 39–42). Later, in the context of explaininganterograde transport mediated by the maturation process ofcisternae of the organelle, a role was attributed to COPI vesiclesas carriers for residential proteins that need to be sorted back toallow the individual cisternae to achieve and maintain theiridentity (11, 13, 43).

Sequence similarities exist between the clathrin/adaptor proteinsand coatomer. The tetrameric adaptor complex resembles acoatomer ‘‘core’’ made up of �-, �-, �-, and �-COP (17, 42, 44, 45),and there is a distant similarity of the trimeric subcomplex �-, ��-,and �-COP with the clathrin proteins (18). Furthermore, theanalogy to the clathrin system became even more apparent whenthe appendage domain of �1-COP revealed a striking structuralsimilarity to the appendage domains of �- and �-adaptin (19, 20).

Our observations of different distributions strongly suggeststhat coatomer complex isoforms serve different functions, withinthe early secretory pathways, that are similar to the variousfunctions individual adaptor proteins exert in the late secretorypathways (5). Thus, it becomes unlikely that COPI vesiclesexclusively mediate back transport of Golgi-resident proteinsduring maturation of Golgi cisternae.

Further work will be needed to focus on the isolation andproteomics of individual COPI vesicles to elucidate their func-tions. With a preference for the cis-Golgi and ERGIC, forexample, �1�2-coatomer is a likely candidate to form vesicles forrecycling and retrieval of proteins from these organelles to theER. Determining the particular functions of the complex iso-forms will be crucial for a better understanding of the Golgiwithin the vertebrates.

Materials and MethodsAntibodies. The following polyclonal antibodies were used: anti-�-COP (CT peptide) (F.T.W.); anti-�-COP (46); anti-��-COP (47);anti-�1-COP (peptide, human, amino acids 63–74) (F.T.W.); anti-

Fig. 3. Localization of isotypic coatomer subunits within the Golgi. Immunogold labeling was performed on ultrathin slices of 2% paraformaldehyde/0.2%glutaraldehyde-fixed NRK cells. Shown are representative pictures of Golgi stacks labeled with anti-�1-app (A and B), anti-�2-app (C and D), anti-�2-COP (E),anti-��-COP (F), and anti-�2-COP antibodies (G), each bound to 15-nm protein A-gold. Double labeling was performed by using anti-GM130 (A–F) or anti-KDELreceptor antibodies (G), and 10-nm protein A-gold. Average labeling densities per �m2 of total Golgi were 25.0 � 2.2 for �1-COP, 27.7 � 2.8 for �2-COP, 13.3 �1.4 for �2-COP, 38.2 � 5.5 for ��-COP, and 40.3 � 3.8 for GM130.

4428 � www.pnas.org�cgi�doi�10.1073�pnas.0611360104 Moelleken et al.

Dow

nloa

ded

by g

uest

on

July

2, 2

021

�2-COP (16); and anti-KDEL receptor (Erd2p, kindly provided bythe late H.-D. Söling, Max Planck Institute for Biophysical Chem-istry, Göttingen, Germany). The monoclonal antibodies that wereused are as follows: anti-GM130 (BD Transduction Laboratories,San Jose, CA); Okt8 (48); CM1 (49). HRP-conjugated secondaryantibodies: goat anti-mouse and donkey anti-guinea pig (Dianova,Hamburg, Germany) and mouse anti-rabbitnative (RG16) (Sigma–Aldrich, Munich, Germany). True Blot was obtained from eBio-sciences (San Diego, CA). Alexa-fluor-conjugated secondary an-tibodies were obtained from Molecular Probes (Leiden, TheNetherlands). Preimmune serum was obtained from the rabbitsthat were used to raise antibodies.

Preparation of Cytosol from Cultured Cells. 3T3/NIH cells grown toconfluency on 15-cm dishes were washed three times with PBSand scraped off in the presence of 0.25 ml of immunoprecipi-tation buffer (25 mM Tris�HCl, pH 7.4/150 mM NaCl/1 mMEDTA) that was complemented with EDTA-free protease in-hibitor. All further steps were performed at 4°C. The cells werepassed 20 times each through 21- and 27-gauge needles. Thehomogenate was centrifuged at 800 � g for 5 min, and theresulting supernatant was subjected to a further centrifugationstep at 100,000 � g for 60 min. Typically, the high-speedsupernatant (cytosol) had a protein concentration of 2 mg/ml.

Immunoprecipitations and Western Blotting. Antibodies were incu-bated with protein A-Sepharose (CL-4B) (Amersham, Little Chal-font, U.K.) on a rotating wheel in immunoprecipitation buffer for60 min at room temperature. After washing three times withimmunoprecipitation buffer, freshly prepared cytosol (0.5 mg/10 �lof beads) was added, and incubation continued for 60 min at roomtemperature. Beads were collected by centrifugation, extensivelywashed, and eluted in reducing SDS sample buffer at 70°C. Sampleswere subjected to SDS/PAGE and analyzed by Western blotting,using the indicated antibodies. To reduce background when rabbitantibodies were used for immunoprecipitation and subsequentWestern blot analysis, monoclonal mouse anti-rabbitnative second-

ary antibodies (RG16) (Sigma–Aldrich) were applied. True Blotwas obtained from eBiosciences.

Cloning and Expression of �1- and �2-Appendage Domains. mRNAwas isolated from 3T3/NIH mouse fibroblasts with oligo(dT)25Dynabeads (Dynal Biotech, Hamburg, Germany) according to themanufacturer’s protocol. Reverse transcription was performed byusing the 5� RACE System V 2.0 kit (Life Technologies, Paisley,Scotland). The regions corresponding to the �1- and �2-COPappendage domains (nucleotides 1663–2625 and 1663–2616, re-spectively) were amplified by using PCR with the following primerpairs: 5�-TCT AGA ATC CCT GGT CTG GAG AAA GCCCTG-3� (forward) and 5�-GGA TTC CTA GCC CAC GGA TGCCAA GAT GAT-3� (reverse) for �1 and 5�-GCT AGC ATA CCAGGG ATG GAA AAG GCC TTA-3� (forward) and 5�-GGA TTCTTA TCC CAC AGA AGC CAA GAT AAC ATC-3� (reverse) for�2. �1-COP was cloned as an XbaI/BamHI- and �2-COP as anNheI/BamHI fragment in the expression vector pPro-GST-TEV(kindly provided by M. Lutzmann, Heidelberg University Biochem-istry Center). GST-�1-app and GST-�2-app were over-expressed inE. coli BL21(DE3) (Promega, Madison, WI) within 4 h at 37°C afterinduction with 1 mM isopropyl �-D-thiogalactoside. All furthersteps were performed at 4°C. Cells were washed once with PBS,lysed in lysis buffer (PBS/1 mM DTT) complemented with EDTA-free protease inhibitor by using an Emulsiflex (Avestin, Ottawa,ON, Canada) at 15,000–20,000 psi (1 psi � 6.89 kPa). The lysate wascentrifuged at 8,000 � g for 10 min, and the resulting supernatantwas subjected to a further centrifugation step at 100,000 � g for 60min. Glutathione-Sepharose beads (Glutathione-Sepharose 4 FastFlow; Amersham) were added to the high-speed supernatant, andbinding proceeded overnight. After extensive washing in lysisbuffer, the recombinant proteins were eluted in 50 mM Tris�HCl,pH 8.0/150 mM KCl/1 mM DTT/20 mM glutathione. The proteinswere concentrated by using ultrafiltration (Biomax, 30-kDa cut-off;Millipore, Billerica, MA) to 10 mg/ml. Exclusion-size chromatog-raphy (Superdex 75, XK16/30; GE Healthcare, Little Chalfont,U.K.) allowed purification to near homogeneity. Typically, the yield

Fig. 4. Quantification of coatomer isoforms in the cis and trans halves of the Golgi. The relative labeling densities within the cis half and the trans half of theGolgi with the indicated antibodies are shown. The labeling densities were normalized to allow comparison of independent experiments. The amount of Golgistacks analyzed is indicated for each antibody. The cis-Golgi marker GM130 was used as an internal standard to validate polarity of the Golgi; therefore, thenumber of Golgi stacks analyzed for GM130 reflects the sum of all Golgi stacks that were taken into account. Error bars represent the SD about the mean.

Moelleken et al. PNAS � March 13, 2007 � vol. 104 � no. 11 � 4429

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

2, 2

021

was 10 mg of GST-�1-app and 5 mg of GST-�2-app per 6 liters ofE. coli culture, each.

Generation of Anti-�1- and Anti-�2-COP Specific Antibodies. Recom-binant GST-appendage (app) domains fusion proteins (�1-COP:amino acids 555–874; �2-COP: amino acids 555–871; both mouse)were used to raise antibodies in rabbits. GST-�1-app or GST-�2-app protein (2.5 mg of each) were bound to glutathione-Sepharosebeads overnight at 4°C. The beads were washed with PBS. GST-�1-app or GST-�2-app antiserum (7.5 ml of each) was adjusted to1� PBS and was added to the GST-�2-app- or GST-�1-app-loadedglutathione-Sepharose beads, respectively (negative affinity purifi-cation). After 2 h of incubation at room temperature, the resultingflow-through (equal to anti-�1-app or anti-�2-app antibody) wascollected, adjusted to 50% glycerol, and stored at �20°C.

Quantification of Coatomer Isotypes. Total coatomer immunopre-cipitations using the CM1 antibody were performed with cytosolfreshly prepared from 3T3/NIH cells. A given sample was analyzedby Western blotting, using anti-�1-app or anti-�2-app antibody.Comparison of �1- or �2-COP Western blot signals with GST-�1-app or GST-�2-app protein calibration curves, respectively, allowedthe determination of the mass. The molar ratio of �1- and �2-COPwas calculated by taking molar masses into account. �1- or �2-coatomer immunoprecipitations were performed by using anti-�1-app or anti-�2-app antibodies, respectively. Each sample wasanalyzed by Western blotting, using anti-�1-COP- or anti-�2-COP-specific antibodies. His-tagged �1-COP or �2-COP proteins (16)were used as protein standards. The molar ratio of �1- and �2-COPwithin �1- or �2-coatomer was calculated by taking molar massesinto account. The relative proportion of coatomer isotypes matchesthe product of the corresponding molar ratios; e.g., �1�1-coatomer/total coatomer � �1-COP/total coatomer (0.7) � �1-COP/�1-coatomer (0.77) � 0.54. Data represent the mean � SEM of fiveindependent experiments.

Electron Microscopy and Immunogold Labeling. NRK cells grown to50% confluency on 10-cm dishes were fixed with 2% paraformal-dehyde/0.2% glutaraldehyde in EM buffer (250 mM Hepes/KOH,pH 7.4/100 mM NaCl) for 1 h. Incubation was extended overnightat 4°C. Cells were embedded gradually in 2%, 5%, and 10% gelatinin EM buffer, infused with 2.3 M sucrose, frozen in liquid nitrogen,

and sectioned with a cryo-ultramicrotome (50-nm slices). Ultrathinsections were picked up in a 1:1 mixture of 2.3 M sucrose in PBSand 2% methylcellulose in distilled water, collected on Formvar-and carbon-coated grids, and stored at 4°C. Sections were processedand labeled with the indicated primary antibodies and 15-nmprotein A-gold (Department of Cell Biology, Utrecht University,The Netherlands), as described in ref. 50. After fixation with 1%glutaraldehyde for 10 min at room temperature to destroy proteinA binding sites of bound antibodies (51), sections were incubatedwith anti-GM130 antibody and 10-nm protein A-gold. Contrastingwas done with 0.5% uranyl acetate/1.8% methylcellulose in distilledwater, and then the sections were air-dried and examined in anelectron microscope (EM10 or LEO 906E; Zeiss, Oberkochen,Germany). Antibodies were diluted as follows: anti-�1-app 1:100,anti-�2-app 1:300, anti-�2-COP 1:20, anti-��-COP 1:4,000, anti-KDEL receptor 1:500, and anti-GM130 1:300. Gold conjugateswere diluted according to the manufacturer’s protocol. Quantifi-cation was performed as described in ref. 23. Briefly, the cis side ofthe Golgi was identified by the presence of GM130 labeling.The stack was split into two halves by an arbitrary line drawn in themiddle of the stack. The labeling density per �m2 within each halfwas evaluated and normalized to allow comparison of differentstacks. Data represent the mean � SEM of the relative labelingdensities.

Cell Culture. NRK cells and 3T3/NIH cells were grown in DMEMcontaining 10% heat-inactivated FCS or 10% heat-inactivated calfserum, respectively, and supplemented with 100 units of penicillinper ml, 100 �g of streptomycin per ml, 2 mM L-glutamine, and 0.5mg of G418 per ml. Cells were incubated at 37°C under 5%CO2/95% air.

Bioinformatics. For bioinformatics procedures, see SI Text.

We thank Sigrid Berger-Seidel and Karin Gorgas for their help in the earlystages of this work; Hilmar Bading and Joachim Kirsch for generouslyproviding their EM facilities; George Posthuma for invaluable advice forEM; and Emily Stoops, Julian Langer, and Julien Béthune for reading themanuscript. This work was supported by German Research Council GrantSFB 638, A10 (to B.B. and F.T.W.) and The European Union Grant 3Drepertoire (M.J.B.), Contract LSHG-CT-2005-512028.

1. Pagano A, Letourneur F, Garcia-Estefania D, Carpentier JL, Orci L, Paccaud JP (1999)J Biol Chem 274:7833–7840.

2. Esaki M, Liu Y, Glick BS (2006) FEBS Lett 580:5215–5221.3. Roberg KJ, Crotwell M, Espenshade P, Gimeno R, Kaiser CA (1999) J Cell Biol 145:659–672.4. Peng R, De Antoni A, Gallwitz D (2000) J Biol Chem 275:11521–11528.5. Robinson MS (2004) Trends Cell Biol 14:167–174.6. Owen DJ, Collins BM, Evans PR (2004) Annu Rev Cell Dev Biol 20:153–191.7. Kirchhausen T (2000) Annu Rev Biochem 69:699–727.8. Rothman JE, Wieland FT (1996) Science 272:227–234.9. Martinez-Menarguez JA, Geuze HJ, Slot JW, Klumperman J (1999) Cell 98:81–90.

10. Letourneur F, Gaynor EC, Hennecke S, Demolliere C, Duden R, Emr SD, Riezman H,Cosson P (1994) Cell 79:1199–1207.

11. Pelham HR, Rothman JE (2000) Cell 102:713–719.12. Bonfanti L, Mironov AA, Jr, Martinez-Menarguez JA, Martella O, Fusella A, Baldassarre M,

Buccione R, Geuze HJ, Mironov AA, Luini A (1998) Cell 95:993–1003.13. Glick BS, Elston T, Oster G (1997) FEBS Lett 414:177–181.14. Blagitko N, Schulz U, Schinzel AA, Ropers HH, Kalscheuer VM (1999) Hum Mol Genet 8:2387–2396.15. Futatsumori M, Kasai K, Takatsu H, Shin HW, Nakayama K (2000) J Biochem (Tokyo) 128:793–801.16. Wegmann D, Hess P, Baier C, Wieland FT, Reinhard C (2004) Mol Cell Biol 24:1070–1080.17. Schledzewski K, Brinkmann H, Mendel RR (1999) J Mol Evol 48:770–778.18. Eugster A, Frigerio G, Dale M, Duden R (2000) EMBO J 19:3905–3917.19. Hoffman GR, Rahl PB, Collins RN, Cerione RA (2003) Mol Cell 12:615–625.20. Watson PJ, Frigerio G, Collins BM, Duden R, Owen DJ (2004) Traffic 5:79–88.21. Hara-Kuge S, Kuge O, Orci L, Amherdt M, Ravazzola M, Wieland FT, Rothman JE (1994)

J Cell Biol 124:883–892.22. Nakamura N, Rabouille C, Watson R, Nilsson T, Hui N, Slusarewicz P, Kreis TE, Warren

G (1995) J Cell Biol 131:1715–1726.23. Orci L, Stamnes M, Ravazzola M, Amherdt M, Perrelet A, Sollner TH, Rothman JE (1997)

Cell 90:335–349.24. Hardwick KG, Lewis MJ, Semenza J, Dean N, Pelham HR (1990) EMBO J 9:623–630.25. Lewis MJ, Pelham HR (1990) Nature 348:162–163.26. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, Devon K, Dewar K,

Doyle M, FitzHugh W, et al. (2001) Nature 409:860–921.27. Pelham HR (1988) EMBO J 7:913–918.

28. Griffiths G, Ericsson M, Krijnse-Locker J, Nilsson T, Goud B, Soling HD, Tang BL, WongSH, Hong W (1994) J Cell Biol 127:1557–1574.

29. Bethune J, Kol M, Hoffmann J, Reckmann I, Brugger B, Wieland F (2006) Mol Cell Biol26:8011–8021.

30. Gommel D, Orci L, Emig EM, Hannah MJ, Ravazzola M, Nickel W, Helms JB, WielandFT, Sohn K (1999) FEBS Lett 447:179–185.

31. Nickel W, Sohn K, Bunning C, Wieland FT (1997) Proc Natl Acad Sci USA 94:11393–11398.32. Fullekrug J, Suganuma T, Tang BL, Hong W, Storrie B, Nilsson T (1999) Mol Biol Cell

10:1939–1955.33. Blum R, Pfeiffer F, Feick P, Nastainczyk W, Kohler B, Schafer KH, Schulz I (1999) J Cell

Sci 112:537–548.34. Rojo M, Pepperkok R, Emery G, Kellner R, Stang E, Parton RG, Gruenberg J (1997) J Cell

Biol 139:1119–1135.35. Malsam J, Satoh A, Pelletier L, Warren G (2005) Science 307:1095–1098.36. Orci L, Ravazzola M, Volchuk A, Engel T, Gmachl M, Amherdt M, Perrelet A, Sollner TH,

Rothman JE (2000) Proc Natl Acad Sci USA 97:10400–10405.37. Lanoix J, Ouwendijk J, Stark A, Szafer E, Cassel D, Dejgaard K, Weiss M, Nilsson T (2001)

J Cell Biol 155:1199–1212.38. Cosson P, Letourneur F (1994) Science 263:1629–1631.39. Orci L, Glick BS, Rothman JE (1986) Cell 46:171–184.40. Nickel W, Malsam J, Gorgas K, Ravazzola M, Jenne N, Helms JB, Wieland FT (1998) J Cell

Sci 111:3081–3090.41. Pepperkok R, Whitney JA, Gomez M, Kreis TE (2000) J Cell Sci 113:135–144.42. Serafini T, Stenbeck G, Brecht A, Lottspeich F, Orci L, Rothman JE, Wieland FT (1991)

Nature 349:215–220.43. Becker B, Melkonian M (1996) Microbiol Rev 60:697–721.44. Cosson P, Demolliere C, Hennecke S, Duden R, Letourneur F (1996) EMBO J 15:1792–1798.45. Duden R, Griffiths G, Frank R, Argos P, Kreis TE (1991) Cell 64:649–665.46. Allan VJ, Kreis TE (1986) J Cell Biol 103:2229–2239.47. Harter C, Draken E, Lottspeich F, Wieland FT (1993) FEBS Lett 332:71–73.48. Zarling JM, Kung PC (1980) Nature 288:394–396.49. Palmer DJ, Helms JB, Beckers CJ, Orci L, Rothman JE (1993) J Biol Chem 268:12083–12089.50. Folsch H, Pypaert M, Schu P, Mellman I (2001) J Cell Biol 152:595–606.51. Folsch H, Pypaert M, Maday S, Pelletier L, Mellman I (2003) J Cell Biol 163:351–362.

4430 � www.pnas.org�cgi�doi�10.1073�pnas.0611360104 Moelleken et al.

Dow

nloa

ded

by g

uest

on

July

2, 2

021

http://www.pnas.org/cgi/content/full/0611360104/DC1