COPI (coat protein I)-coated vesicles are implicated in various transport steps within the early secretory pathway. The major structural component of the COPI coat is the heptameric complex coatomer (CM). Recently, four isoforms of CM were discovered that may help explain various transport steps in which the complex has been reported to be involved. Biochemical studies of COPI vesicles currently use CM purified from animal tissue or cultured cells, a mixture of the isoforms, impeding functional and structural studies of individual complexes. Here we report the cloning into single baculoviruses of all CM subunits including their isoforms and their combination for expression of heptameric CM isoforms in insect cells. We show that all four isoforms of recombinant CM are fully functional in an in vitro COPI vesicle biogenesis assay. These novel tools enable functional and structural studies on CM isoforms and their subcomplexes and allow studying mutants of CM.
Protein-coated vesicles function in the transport of soluble material and of membranes at various intracellular routes. COP (coat protein) II vesicles, coated with the heterodimeric complexes Sec23/24 and Sec13/31, serve the exit from the endoplasmic reticulum (ER) of newly synthesized proteins and lipids (1). Isoforms of these Sec proteins exist and appear to serve the transport of disparate subsets of cargo molecules (2–4). Likewise, isoforms of clathrin-coated vesicles (CCVs) are defined by adaptor protein (AP) complexes that are involved in transport in the late secretory pathway (reviewed in 5). Here, the AP1 complex serves transport from the trans-Golgi network (TGN) to endosomes, AP2 is involved in endocytosis, AP3 functions in biogenesis of lysosome-related organelles and AP4 has a presumed role in basolateral and TGN–endosome sorting (reviewed in 6–8). Other coats involved in intracellular transport that are presently being defined in more detail functionally are e.g. retromer (9), Golgi-localized γ-adaptin ear-containing ARF-binding proteins (GGAs) (10) and the BBSome (11).
COPI vesicles are covered with coatomer (CM), a soluble and stable heptameric protein complex. They serve transport in the early secretory pathway. Although their function in retrograde transport from the Golgi apparatus to the ER is well established, less agreement exists about their possible roles in transport steps between Golgi cisternae in an anterograde and retrograde direction, as well as anterograde transport from the intermediate compartment (12–18). With the discovery of four isoforms of CM that represent four stoichiometric combinations of the subunits α-, β-, β′-, γ1-, γ2-, δ-, ε-, ζ1- and ζ2-COPs, it appears to be possible that individual isoforms may serve different transport routes, underlining a potential functional diversity. CM isoforms were found by immuno-electron microscopy (EM) to be unequally distributed within the Golgi stack (19). It is not clear to date, however, whether COPI vesicles exist uniformly coated with individual isoforms of CM, and which individual function the isoforms of the coat complex would serve. The subtle differences between γ1- and γ2-COP, and between ζ1- and ζ2-COP have not allowed separation of CM isoforms from the endogenous mixture obtained from animal cytosol. This lack of individual and pure heptameric isoforms has impeded experimental dissection of their potential functions.
Most vesicle coats are assembled by the stepwise recruitment of first a membrane–proximal complex, which builds the inner layer of the coat, followed by an additional, membrane–distal coat complex, which forms the outer layer. For example, the inner layer of the adaptor–clathrin coat is formed by the adaptor complexes, whereas the outer layer is formed by clathrin (20). In contrast, CM is incorporated en bloc into the COPI coat, and thus builds up the equivalent of both layers in one step (21). Interestingly, the heptameric CM (CM7) complex can be chemically dissociated into subcomplexes, among others a tetrameric βγδζ-COP (CM4) and a trimeric αβ′ε-COP (CM3) subcomplex. Under the conditions used, CM4 is unstable and partially dissociates further into smaller subcomplexes (22,23). Structural data of the γ1-appendage domain and of ζ1-COP, and sequence analyses of the CM4 subunits, suggest a structural relationship between the tetrameric subcomplex of CM and the adaptor complexes (24–28). Recent structural data on the trimeric subcomplex show that it contains structural motifs that also occur in Sec13/31, the outer layer of the COPII coat, and clathrin (29,30). As such, CM4 is thought to represent the inner layer of the COPI coat, whereas CM3 represents the outer layer. The current methods to obtain CM subcomplexes by dissociation of heptameric CM are restricted to an analytical scale, as they neither allow purification of larger amounts of subcomplexes, nor their purification to homogeneity. Thus, the lack of pure CM subcomplexes has severely hampered investigations aimed at deciphering the functions of the CM subcomplexes.
Here, we describe the expression in insect cells of the four isoforms of mouse heptameric CM and their purification to homogeneity. We used a single virus containing all seven genes to produce the individual complexes. We show that the resulting CM complexes are functional in the formation of COPI vesicles. Furthermore, we show the expression and purification of tetrameric and trimeric CM subcomplexes. These isoformic CM holo- and subcomplexes now provide, for the first time, a completely recombinant toolbox for dissecting the molecular mechanisms of CM action in the secretory pathway.
Cloning of CM subunit cDNAs and assembly of bacmids
To obtain the recombinant heptameric complexes of each of the four known CM isoforms (19,31), we used the MultiBac baculoviral expression system (32). This system allows the combination of various subunits into a single plasmid using a so-called multiplication module, and is flexible toward constructing different combinations of subunits to assemble subcomplexes and mutants of CM. Furthermore, using the MultiBac system, the expression of the heptameric CM holocomplex from a single virus (instead of using different viruses for the different subunits) is possible, which ensures that all cells express the seven subunits in stoichiometric levels, and assemble the heptameric complex under native conditions.
Mouse CM subunit cDNAs were obtained from I.M.A.G.E. clones (33) (Table S1, Supporting Information), amplified by polymerase chain reaction (PCR) using primers with unique restriction sites (Table S2) and cloned into pCR-Blunt cloning vectors (Figure 1A). 6×His-tags, followed by a tobacco etch virus (TEV) protease cleavage site (HT-tags), were inserted at the N-termini of α- and β-COP for nickel-affinity purification of CM holo- and subcomplexes (trimeric αβ′ε-COP and tetrameric βγδζ-COP). The various γζ-COP combinations were subcloned into pUCDM transfer vectors (Figure 1B) and the four isoform-specific pUCDM-γxζx transfer vectors were each recombined into the MultiBac bacmid by Cre/Lox recombination (Figure 1C). Thus, we generated four different MultiBacs, each encoding one particular γζ-COP combination.
The remaining five subunits, which are shared by all isoforms (α-, β-, β′-, δ- and ε-COP), were subcloned into pFBDM transfer vectors (Figure 1B) and combined into a single pFBDM vector using the multiplication module (Figure 1D). This strategy ensured that every subunit was under control of its own strong viral promoter, either the polyhedrin (polh) or the p10 promoter.
Finally, the expression cassette from the pFBDM-αββ′δε vector was inserted into the isoform-specific bacmids through Tn7 transposition (Figure 1E). As a result, MultiBac bacmids for production of the four isoform-specific CM holocomplexes were obtained.
Expression and Purification of isoforms of heptameric CM
Sf9 insect cells were transfected with the resulting four bacmids, each with the defined combination of a heptameric CM isoform. The virus stocks were amplified in two further rounds to obtain large volumes of high-titer virus stocks. The virus stocks were used to express each heptameric isoform of CM in Sf9 insect cells. Efficient expression of CM subunits can already be observed in Sf9 lysates (Figure 2A). After lysis, cell debris and insoluble material were removed by centrifugation. CM complexes were isolated from the soluble protein fraction by nickel-affinity purification, eluted with imidazole, desalted and analyzed by SDS–PAGE (Figure 2B). This quick purification yielded CM of sufficient purity for many biochemical assays (e.g. in vitro COPI vesicle biogenesis). As depicted in Figure 2C,D, the four isoformic CM complexes contain all seven subunits in a stoichiometric ratio (compare with rabbit liver-derived CM in Figure 2C). CM from natural sources (rabbit liver, rat liver, rat hepatocytes and mouse 3T3 cells) mainly represent a mixture of the three isoforms γ1ζ1, γ1ζ2 and γ2ζ1, with only a minor (<5%) content of γ2ζ2(19,31). Therefore, we mixed freshly purified recombinant CM isoforms to yield a 3T3 CM-like mixture consisting of 50% γ1ζ1, 20% γ1ζ2 and 30% γ2ζ1, thereby mimicking the endogenous material.
Recombinant CM isoforms were further purified by ion exchange chromatography on a ResourceQ column (GE Healthcare). A typical elution profile is shown in Figure 3A. A major peak eluting at ∼0.4 m KCl was pooled that contains all seven COPs (Figure 3C) and was further subjected to size exclusion chromatography on a Superose 6 column (GE Healthcare). A representative elution profile shows a main peak eluting at an apparent molecular mass of ∼700 kDa that represents CM (Figure 3B). The symmetric peak profile indicates a highly homogeneous population of CM complexes. On the basis of SDS–PAGE and Coomassie staining, we estimate the purity of isoforms after nickel-immobilized metal affinity chromatography (IMAC), anion exchange chromatography and gel filtration (peak fractions) to be >95% (Figure 3C). Table S4 displays the yield and purity of CM at the different steps of a typical CM purification.
With the development of sensitive and quantitative methods, interest is growing in lipid analysis of various in vitro reconstitution experiments with liposomes and membranes. This raises the question as to the purity of proteins used in such experiments with regard to lipid contaminations. In general, lipid contaminations are not assessed in protein samples purified from cell extracts by standard affinity chromatography methods. Therefore, we performed quantitative lipid analysis by nano-electrospray ionization tandem mass spectrometry (nano-ESI–MS/MS). To this end, nickel affinity-purified recombinant CM samples were subjected to lipid extraction in the presence of lipid standards, and the amount of phosphatidylcholine (PC) as the major membrane lipid was determined by MS/MS as described (34,35). The amount of PC co-purified with CM varied greatly between individual purifications from ∼1.5 to ∼23 nmol PC/mg protein, corresponding to ∼1–13 nmol PC/nmol protein. The co-purification of lipids was not a phenomenon specific for recombinant CM, but also occurred when purifying CM from rabbit liver or when affinity-purifying unrelated proteins from insect cells. Preliminary analyses suggest that the amount of lipids co-purified correlates with the amount of nickel-IMAC beads used, and that part of the lipids can be removed by further purification steps (e.g. size exclusion chromatography).
Functional characterization of recombinant CM in an in vitro COPI vesicle reconstitution assay
Next, we investigated the functional state of the CM isoforms. To this end, their capability to form COPI vesicles in vitro from Golgi-enriched membranes was compared with CM preparations from rabbit liver. Rat liver Golgi was incubated with recombinant myristoylated ADP-ribosylation factor 1 (Arf1), GTPγS and the various CMs, and vesicles formed were purified by sucrose density gradient centrifugation and analyzed by western blotting (Figure 4A) and by EM (Figure 4B) as described in Beck et al. (36). Each of the four isoforms was able to support the formation of Golgi-derived COPI vesicles at a similar extent (compare lanes 6, 8, 10 and 12 in Figure 4A). EM analysis of the vesicle preparations with the recombinant CMs showed populations homogeneous in size and highly similar in shape, and not distinguishable from vesicles derived from rabbit liver CM. We conclude that all four recombinant isoforms of the complex are in a native state after purification and support COPI vesicle formation.
Generation of recombinant CM subcomplexes
To obtain recombinant CM subcomplexes, we used a strategy similar to the one described above for the heptameric CM isoforms. For the isoformic tetrameric CM subcomplexes (referred to as CM4), δ-COP and His-tagged β-COP were subcloned into a single pFBDM transfer vector using the multiplication module, and recombined into MultiBac bacmids containing the four different γζ-COP combinations (Figure 5A). For the trimeric subcomplex, we combined β′-COP, ε-COP and His-tagged α-COP in a single pFBDM transfer vector using the multiplication module, and recombined this into a MultiBac bacmid using Tn7 transposition (Figure 5B). Viruses for the expression of CM subcomplexes were generated using the bacmids and the subcomplexes were expressed in insect cells (Figure 5C). Like heptameric CM, the CM4 and CM3 subcomplexes could be readily purified by nickel-affinity chromatography (Figure 5D,E).
Here we show the production of recombinant heptameric isoform-specific CM complexes in insect cells. This is the first time that the recombinant production and purification to homogeneity of all four heptameric CM isoforms have been achieved. Each complex was expressed from a single virus encoding all subunits, each under control of a strong viral promoter. A clear advantage of this approach is that all infected cells express all subunits and assemble the complex in its natural environment, i.e. the cytosol. Alternative approaches would be to express and purify all subunits individually and then mix them to assemble complexes in vitro, or to express complexes by infecting the cells with a mixture of viruses encoding single subunits. These approaches, however, are obviously hampered by problems with solubility and stability of individual subunits, and a low efficiency of multiple infections of a cell with all the viruses. Our approach, in contrast, allows the reproducible production of stoichiometric CM complexes for all four isoforms, in a quality and quantity allowing functional and structural studies on a molecular level. The recombinant CMs, homogenous in their composition, can be routinely produced and purified following the protocol described here, and are functional in an in vitro vesicle budding assay. Our approach now provides a well-defined replacement for heterogeneous CM from animal tissues that has been used before for functional studies.
In summary, we present here a new toolbox that opens up, for the first time, the possibility to dissect functions of CM isoforms as well as their subcomplexes. Our approach is robust, and at the same time highly flexible because of the iterative assembly of the expression cassettes by multiplication into a single multigene baculovirus. Thus, for a host of applications including detailed structure–function analyses, this method enables rapid generation of many variants of CM complexes. This includes the generation of CM mutants with altered subunits that allow studying effects of a mutation in the context of the holocomplex rather than in a single subunit or domain. Finally, our work shows that the MultiBac expression system is highly suitable for the recombinant expression of large protein complexes.
Materials and Methods
Cloning of mouse CM subunits
Mouse CM subunits were amplified from I.M.A.G.E. cDNA clones (33) [imaGenes; former German Resource Center for Genome Research (RZPD)] (Table S1) by PCR using Stratagene PfuTurbo DNA polymerase (Agilent Technologies) and specific oligonucleotides for each cDNA that included restriction sites for subsequent cloning steps (Table S2) (Figure 1A). The blunt-end PCR products were purified using the QIAquick PCR Purification Kit (QIAGEN) and ligated into the pCR-Blunt vector using the Zero Blunt PCR Cloning Kit (Invitrogen). The cloned cDNAs were sequenced completely to confirm that no mutations were introduced by the PCR process and that no missense mutations were introduced as compared to the genomic sequence. Indeed, the cDNAs contained only silent changes originating from the template clones. Only in α-COP, we detected a threonine in position 761 that was already present in the I.M.A.G.E. clone. This T761 is different from the serine in this position in the reference genome of the C57BL/6J strain, but matches the genomic sequence of the FVB strain.
Cloning of the isoforms of ζ-COP and γ-COP to the pUCDM transfer vector
ζ1- and ζ2-COP were excised from the respective pCR-Blunt vectors and inserted into the multiple cloning site (MCS) 1 of pUCDM (Figure 1B) to put them under control of the p10-promoter using the XmaI/NheI (ζ1-COP) and XhoI/SphI (ζ2- COP) sites. Next, γ1-COP and γ2-COP were excised with BamHI and BssHII from the respective pCR-Blunt plasmids and ligated into the MCS2 of the pUCDM-ζ1-COP and pUCDM-ζ2-COP vectors to put them under control of the polh-promoter. Like this, four individual pUCDM-γxζx-COP plasmids were generated with each a different γζ-COP combination. All steps involving pUCDM plasmids were performed in the pir+Escherichia coli strain BW23473.
Cloning of β′-COP, δ-COP and ε-COP to the pFBDM transfer vector
To allow use of the multiplication module in the transfer vector pFBDM (32,37) (Figure 1B), we had to eliminate the SpeI site in the MCS2 by site-directed mutagenesis using the Stratagene QuikChange Site-directed Mutagenesis Kit (Agilent Technologies) and the oligonucleotides 5′-GCCTACGTCGACGAGCTCTCTAGTCGCGGCCGCTTTCG-3′ and 5′-CGAAA GCGGCCGCGACTAGAGAGCTCGTCGACGTAGGC-3′. All further use of the name pFBDM refers to the version in which the SpeI site in MCS2 has been eliminated.
The cDNA of β′-COP was excised from the pCR-Blunt construct with XmaI and NheI, and ligated into pFBDM digested with the same enzyme combination, thus placing the cDNA in the MCS1 under control of the strong viral p10 promoter. δ-COP and ε-COP were both excised from the respective pCR-Blunt constructs and cloned into the MCS2 of a pFBDM vector using the EcoRI site to place them under control of the strong viral polh-promoter and the orientation of insertion was checked by PCR and sequencing.
Addition of histidine tags to β-COP and α-COP and cloning to pFBDM
A HT-tag for nickel-affinity purification was inserted at the N-terminus of β-COP. To this end, pCR-Blunt–β-COP was digested with XhoI and NdeI. An insert encoding the tag was generated by annealing two complementary oligonucleotides (Table S3), inserted using overhangs compatible with the overhangs generated by XhoI and NdeI and the in-frame insertion of the HT-tag was checked by sequencing. Cleavage of the tag by the TEV–protease will leave a glycine–histidine rest at the N-terminus of β-COP. HTβ-COP was subsequently cloned into the MCS1 of pFBDM using XmaI (present directly upstream of the HT-tag in the annealed oligonucleotides) and NheI to put it under control of the p10 promoter. To increase the efficiency of nickel-affinity purification and to prepare the constructs for future use in the separate expression and purification of CM subcomplexes, an HT-tag was also cloned at the N-terminus of α-COP. To this end, the HT-tag insert was ligated into the MCS2 of a BssHII/EcoRI digested pFBDM vector and subsequently the α-COP cDNA was cloned from pCR-Blunt–α-COP to the pFBDM with the HT-tag using EcoRI and NotI. In-frame cloning of HT-tag and α-COP was confirmed by sequencing. Thus, HTα-COP was cloned into the MCS2 of pFBDM and put under control of the polh promoter. Cleavage of the HT-tag by the TEV–protease results in an additional glycine–glutamate–phenylalanine tripeptide at the N-terminus of α-COP.
Combination of HTα-COP, HTβ-COP, β′-COP, δ-COP and ε-COP into a single pFBDM transfer vector
The multiplication module of the pFBDM vector was exploited stepwise to combine the five CM subunits HTα-COP, HTβ-COP, β′-COP, δ-COP and ε-COP into a single pFBDM transfer vector (Figure 1D). To this end, the fragment of interest was excised from the donor pFBDM vector with PmeI and AvrII and inserted into the multiplication module of the acceptor pFBDM vector digested with Bsp68I/NruI and SpeI. First, δ-COP was inserted into pFBDM-ε-COP. Next, the δε-COP cassette was added to pFBDM–HTα-COP. Then, β′-COP was cloned into pFBDM-αδε-COP, and finally HTβ-COP was inserted into pFBDM-αβ′δε-COP to yield the five subunit transfer vector pFBDM-αββ′δε-COP in which each subunit is under the control of its own promoter.
Generation of bacmids encoding heptameric CM isoforms
To recombine the γζ-COP isoforms into the MultiBac bacmid using the Cre/Lox recombination site (Figure 1C), the isoformic pUCDM-γζ-COP plasmids were transformed into electrocompetent DH10MultiBacCrecells (DH10βE. coli cells containing the bacmid and the Tn7- and Cre-helper plasmids) as described (32,37). Colonies growing under chloramphenicol selection in the pir− host contained the pUCDM transfer vector recombined into the bacmid. To insert the expression cassette for the remaining five CM subunits using the Tn7 transposition sites into the same bacmids as the γζ-COP isoformic bacmids (Figure 1E), single isoformic DH10MultiBac–γζ-COP clones were made electrocompetent as described (37), and the pFBDM-αββ′δε-COP plasmid was transformed into the cells by electroporation. Colonies with the pFBDM plasmid inserted into the bacmid were selected for growth on gentamycin and by blue/white screening (transposition disrupts the lacZ gene). Bacmid DNA was isolated using the PureLink HiPure Plasmid DNA Miniprep Kit (Invitrogen).
Transfection of insect cells and generation of virus stocks
Spodoptera frugiperda Sf9 cells cultured in suspension at 27°C in Sf-900 II serum-free medium (Invitrogen) supplemented with 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco, Invitrogen). For transfection, 2.25 × 106cells in 3 mL Sf-900 II were seeded into 25 mL cell culture flasks and allowed to attach for at least 1 h. After attachment, the Sf-900 II medium was aspirated, 5 mL Grace's medium (Invitrogen) was added to wash the cells briefly and the medium was aspirated again. Bacmid DNA (2.5 µg) was mixed with 250 µL Grace's medium, then 15 µL Cellfectin (Invitrogen) in 250 µL Grace's medium was added and the mixture was incubated for 30 min at room temperature. After the incubation, an additional 2 mL Grace's medium was added, the DNA/Cellfectin mixture was added to the cells and the transfection was performed for 5 h at 27°C. Then, the transfection medium was replaced by 5 mL Sf-900 II medium with penicillin/streptomycin and the cells were cultured for 48 h. The supernatant containing the primary virus particles (V0) was harvested and stored at 4°C protected from light. Fresh medium was added to the cells and they were cultured for another 2–3 days to allow protein expression. The cells were detached by vigorous shaking, pelleted, resuspended in Laemmli sample buffer and the expression of CM was analyzed by SDS–PAGE and Coomassie-blue staining or western blotting.
Amplification of the virus stocks in Sf9 cells cultured in suspension was done as published (37). In brief, 50 mL of 0.5 × 106 Sf9 cells/mL were infected with 3 mL V0 and incubated for 2–3 days. Cells were pelleted by centrifugation (5 min at 1500 ×g 4°C), the supernatant (V1 virus stock) was stored at 4°C protected from light. For the generation of V2 virus stocks, 2.5–5 mL V1 virus stock were added to 200 mL Sf9 cells, the cells were cultured for 2–3 days, the cells were pelleted by centrifugation, and the supernatant (V2 virus stock) was stored at 4°C protected from light. For long-term storage of virus stocks, 10% fetal calf serum (FCS) (Gibco) was added. Samples from the infected cultures were taken directly before harvesting of the virus to check for the expression of CM subunits by SDS–PAGE and Coomassie staining or western blotting.
Expression of isoform-specific heptameric CM complexes in Sf9 cells
The optimal amount of each V2 virus to be used for the expression was determined in 40 mL cultures with 1.5 × 106 cells/mL infected with 100 µL–2 mL V2. The smallest amount of virus that led to an inhibition of cell division after 24 h was used for further expression. The optimal time for expression was determined by taking samples at various time points and analyzing them by SDS–PAGE and Coomassie staining. For larger scale expression, 200 mL of 1.5 × 106 cells/mL were infected with an appropriate amount of V2 and incubated for the appropriated time (typically 3 days). Cells were harvested by centrifugation, resuspended in cold PBS and pelleted again. The supernatant was discarded, and the cell pellets were shock-frozen in liquid nitrogen and stored at −80°C.
Nickel-affinity purification of heptameric CM complexes
Pellets were thawed and resuspended in lysis buffer (see below). Cells were lysed using a high-pressure emulsifier (Avestin) and cell debris was pelleted at 1 00 000 ×g. Nickel-IMAC sepharose beads (GE Healthcare) were washed twice by pelleting the beads (5 min at 1000 ×g) and resuspending them in lysis buffer. Beads were then incubated with the cleared cell lysate on an overhead-rotator at 4°C for 1 h. The beads were washed 3× with 50 mL wash buffer (i.e. lysis buffer with 40 mm imidazole), then packed in an Econo-Pac chromatography column (Bio-Rad Laboratories) and eluted in fractions with elution buffer (lysis buffer with 250 mm imidazole). The fractions were analyzed by SDS–PAGE and Coomassie staining, and the CM-containing fractions were pooled. This nickel-affinity purified CM was desalted on PD10 desalting columns (GE Healthcare) to storage buffer (see below), divided in aliquots, snap-frozen in liquid nitrogen and stored at −80°C. Protein concentrations were determined using Bradford reagent (Bio-Rad Laboratories). For our initial analyses, we used a lysis buffer containing 50 mm HEPES–NaOH pH 7.4, 300 mm NaCl, 20 mm imidazole, 1 mm DTT and a storage buffer containing 50 mm HEPES–NaOH pH 7.4, 150 mm NaCl and 1 mm DTT. However, as we noted that a significant amount of CM stored in this buffer precipitated in further assays, we decided to change the buffer system to resemble the one in which rabbit liver CM is stored (lysis buffer 25 mm HEPES–KOH pH7.4, 200 mm KCl, 10% glycerol, 20 mm imidazole and 0.02% v/v monothioglycerol; storage buffer 25 mm HEPES–KOH pH 7.4, 200 mm KCl, 10% glycerol and 1 mm DTT). This resulted in stably stored CM.
Purification of heptameric CM complexes by anion exchange and size exclusion chromatography
For the purification of CM by anion exchange and size exclusion chromatography, nickel-affinity purified CM was directly loaded onto a column packed with 1.5 mL ResourceQ anion exchange resin (GE Healthcare). The column was washed with 100% buffer A (25 mm HEPES–KOH pH7.4, 100 mm KCl, 10% glycerol and 1 mm DTT) until baseline was reached and CM was eluted with an increasing gradient of buffer B (Buffer A, but with 1 m KCl). CM typically eluted at ∼35% buffer B (∼0.4 m KCl). The peak fractions were pooled, loaded on a 10/300GL Superose 6 size exclusion chromatography column (GE Healthcare) and eluted using storage buffer (see above). Size exclusion purified CM was concentrated using Amicon Ultra-4 (1 00 000 MWCO) centrifugal filter units (Millipore). Aliquots were snap-frozen in liquid nitrogen and stored at −80°C.
Analysis of lipids co-purified with CM
Lipids were extracted from 0.1 mg recombinant CM in the presence of 100 pmol PC standard according to the method of Bligh and Dyer (38). Quantitative nano-ESI–MS/MS was performed as previously published (34,35).
In vitro COPI vesicle biogenesis
Expression and purification of myristoylated Arf1, preparation of rat liver Golgi membranes, in vitro COPI vesicle biogenesis and negative-staining EM analysis of COPI vesicles were performed as described previously (39).
We thank Evelyn Bauer and Ingrid Meißner for technical assistance, and the members of the Wieland lab for fruitful discussions. We thank Dr. Hilmar Bading, in whose laboratory the EM experiments were carried out. This research was supported by grant SFB 638/TP A10 from the German research foundation (DFG). I. B. acknowledges support from the European Commission (EC) through the FP7 I3 project P-CUBE (www.p-cube.eu).