Mechanistic and structural studies of large multi-subunit assemblies are greatly facilitated by their reconstitution in heterologous recombinant systems. In the present paper, we describe the generation of recombinant human APC/C (anaphase-promoting complex/cyclosome), an E3 ubiquitin ligase that regulates cell-cycle progression. Human APC/C is composed of 14 distinct proteins that assemble into a complex of at least 19 subunits with a combined molecular mass of ~1.2 MDa. We show that recombinant human APC/C is correctly assembled, as judged by its capacity to ubiquitinate the budding yeast APC/C substrate Hsl1 (histone synthetic lethal 1) dependent on the APC/C co-activator Cdh1 [Cdc (cell division cycle) 20 homologue 1], and its three-dimensional reconstruction by electron microscopy and single-particle analysis. Successful reconstitution validates the subunit composition of human APC/C. The structure of human APC/C is compatible with the Saccharomyces cerevisiae APC/C homology model, and in contrast with endogenous human APC/C, no evidence for conformational flexibility of the TPR (tetratricopeptide repeat) lobe is observed. Additional density present in the human APC/C structure, proximal to Apc3/Cdc27 of the TPR lobe, is assigned to the TPR subunit Apc7, a subunit specific to vertebrate APC/C.
- anaphase-promoting complex/cyclosome (APC/C)
- cell cycle
- recombinant expression
- single-particle electron microscopy
- three-dimensional structure
Multi-protein complexes co-ordinate virtually every life process by functioning as molecular machines that perform sophisticated biological reactions [1–3]. Approximately 80% of budding yeast proteins are components of multimeric complexes . Understanding the mechanisms of such assemblies requires structural, biophysical and biochemical analyses of entire complexes and defined subcomplexes. Frequently, their low natural abundance and heterogeneity when isolated from endogenous sources limits the opportunities for structural and biophysical studies, and thus methods to reconstitute recombinant complexes significantly enhances the scope of mechanistic analysis. Methodologies such as protein crystallography require large quantities of highly concentrated homogeneous sample to obtain high-resolution structural information. Electron microscopy and single-particle analysis methods do not require such high concentrations or total amounts of sample, but do still require high sample homogeneity. Likewise, native MS which allows determination of subunit stoichiometry and insights into multimeric complex assembly processes is also dependent on a highly concentrated sample in defined buffer conditions . The ability to specify the subunit composition of multi-protein complexes and to engineer and mutate specific subunits offers the potential to test biological hypotheses, including examining the roles and locations of specific subunits within the context of the whole complex. Furthermore, it allows the introduction of chemical probes enabling single-molecule analysis to explore the relative dynamics of individual subunits.
The APC/C (anaphase-promoting complex/cyclosome) is an unusually large multi-subunit E3 ubiquitin ligase which regulates cell-cycle progression through the proteasome-dependent proteolysis of cell cycle regulatory proteins [5–9]. The core APC/C, formed from at least 13 different proteins, is activated on association of a regulatory co-activator subunit, either Cdc20 (Cdc is cell division cycle) or Cdh1 (Cdc20 homologue 1) (Table 1). Together, the core subunits and co-activators perform catalytic, regulatory, substrate recognition and scaffolding functions. Many of the scaffolding subunits are comprised of multiple repeat motifs including the TPR (tetratricopeptide repeat) and PC (proteasome cyclosome) repeat . APC/C substrate recognition is primarily determined by two degrons, the D (destruction)-box and KEN-box [10–12], present in most substrates, which are recognized by the co-activator subunits [13–17], with the core subunit Apc10 contributing to D-box recognition [18–21]. APC/C regulation is primarily exerted at the level of co-activators by controlling co-activator abundance, their capacity to bind the APC/C, and through co-activator inhibitors such as the MCC (mitotic checkpoint complex). The APC/C is highly conserved across eukaryotes, and 12 out of the 13 Saccharomyces cerevisiae APC/C subunits are conserved in Schizosaccharomyces pombe and humans (Table 1). The latter two species lack Apc9 of S. cerevisiae APC/C, but instead incorporate the species-specific subunits Apc14 and Apc16 respectively [22,23]. Vertebrate APC/C also differs from yeast APC/C through an additional TPR subunit, Apc7, a paralogue of Apc3/Cdc27. Because of the presence of two copies of the TPR and some of the smaller accessory proteins, the holo APC/C is comprised of between 18 and 19 subunits with an overall molecular mass ranging from 1 to 1.2 MDa (Table 1) .
Recently we reported the expression and reconstitution of recombinant S. cerevisiae APC/C  using the first-generation MultiBac cloning system for insect cell–baculovirus expression [25–31]. The reconstituted S. cerevisiae APC/C was correctly assembled, as judged by its structural correspondence to native APC/C, and its capacity to ubiquitinate mitotic cyclin in the presence of co-activator in a D-box- and KEN-box-dependent manner . The ability to recapitulate the endogenous APC/C catalytic and regulatory activity using recombinant reconstituted APC/C provided strong evidence that the molecular composition of S. cerevisiae APC/C had been completely defined, a crucial prerequisite for understanding the complete system. Interestingly, despite the characterization of yeast and human APC/C subunits some 14 years ago , the human Apc15  and Apc16 [22,23] subunits have only been identified relatively recently.
In the present paper we describe an approach for generating recombinant human APC/C. The 14 co-expressed proteins human APC/C reconstituted efficiently, allowing us to isolate milligram quantities of the complex from a few litres of insect cells. Its three-dimensional structure, as determined by electron microscopy and single-particle analysis of negatively stained samples, is similar to that reported for endogenous human APC/C [34,35], and ubiquitination assays demonstrate its capacity to recognize and ubiquitinate the budding yeast APC/C substrate Hsl1 (histone synthetic lethal 1), dependent on Cdh1. The successful reconstitution of catalytically active recombinant human APC/C provides validation of the subunit composition that determines its co-activator-dependent catalytic activity.
The approach used in the present study is based on modified MultiBac pFBDM and pUCDM vectors to allow USER ligation-independent cloning [31,36–38] thereby generating multi-gene-containing baculovirus transfer vectors for insect cell co-expression. The vectors pF1 and pU1 were modified from pFBDM and pUCDM  respectively. pF1 incorporates an Nb.BbvCI/AsiSI endonuclease site within a MUM1 cassette, whereas pU2 incorporates an Nb.BsmI/SwaI endonuclease site within its MUM2 cassette. The 14 human APC/C genes were cloned into two pF1 and two pU1 vectors as indicated in Figure 1. Details of the design of pF1 and pU1 together with cloning procedures will be published elsewhere (Ziguo Zhang and David Barford). The resultant recombinant transfer vectors were transformed into MultiBacDH10α cells  to generate bacmids through in vivo recombination using the Tn7 site of the pF1 vector and the Cre-Lox site of the pU1 vector. The two bacmids incorporate 23.9 and 20.7 kb of human APC/C cDNAs respectively. Sf9 cells were transfected with the resultant bacmids to generate recombinant baculoviruses. For affinity purification, Apc4 was fused to a C-terminal TEV (tobacco etch virus)-cleavable StrepIIx2 tag.
High Five insect cells (Invitrogen) were co-infected with the two recombinant baculoviruses at an MOI (multiplicity of infection) of 2 at a cell density of 2.0×106 cells per ml. High Five cells were incubated at 27°C at 150 rev./min for 72 h. The cells were harvested, flash-frozen in liquid nitrogen and stored at −80°C.
All purification steps were performed at 4°C. Cell pellets were thawed on ice and resuspended in APC/C lysis buffer [50 mM Tris/HCl (pH 8.3), 250 mM NaCl, 5% glycerol, 2 mM DTT (dithiothreitol), 1 mM EDTA, 0.1 mM PMSF, 2 mM benzamidine, 5 units/ml benzonase (Novagen) and Complete™ EDTA-free protease inhibitors (Roche)]. After sonication the lysate was centrifuged for 60 min at 48000 g and the soluble supernatant was bound to a 5 ml StrepTactin Superflow Cartridge (Qiagen) with a flow rate of 1 ml/min. The column was washed with APC/C wash buffer [50 mM Tris/HCl (pH 8.0), 250 mM NaCl, 5% glycerol, 2 mM DTT, 1 mM EDTA and 2 mM benzamidine]. Recombinant APC/C was eluted with APC/C wash buffer supplemented with 2.5 mM desthiobiotin (Sigma). StrepTactin elution fractions were incubated with TEV protease at 4°C overnight, then diluted 2-fold into buffer A without NaCl [buffer A: 20 mM Hepes-NaOH (pH 8.0), 125 mM NaCl, 5% glycerol, 2 mM DTT and 1 mM EDTA] and loaded on to a ResourceQ anion-exchange column (GE Healthcare). The column was washed with buffer A and eluted with a gradient of buffer B [20 mM Hepes-NaOH (pH 8.0), 1 M NaCl, 5% glycerol, 2 mM DTT and 1 mM EDTA]. ResourceQ peak fractions were concentrated and loaded on to a Superose 6 10/300 GL column (GE Healthcare) equilibrated in APC/C size-exclusion buffer [20 mM Hepes-NaOH (pH 8.0), 200 mM NaCl and 2 mM DTT].
In vitro APC/C ubiquitination assays
APC/C ubiquitination assays were adopted and modified from . Each ubiquitination reaction contained approximately 1 μg of recombinant human APC/C or recombinant S. cerevisiae APC/C, 0.625 μg of human ubiquitin-activating enzyme E1 (Boston Biochem), 0.5 μg of UbcH10 (Boston Biochem) for human reactions or 0.5 μg of Ubc4 for yeast reactions, 0.5 μg of FLAG-tagged Hsl1 (residues 667–872), and a molar excess (judged by Coomassie-Blue-stained gel) of purified recombinant human or S. cerevisiae Cdh1 in a 10 μl reaction volume with 40 mM Tris/HCl (pH 7.5), 10 mM MgCl2, 0.6 mM DTT, 2.7 mM ATP, 6.6 μg of methyl-ubiquitin, 200 ng of ubiquitin aldehyde (Enzo Life Science) and 2 mM LLnL (N-acetyl-Leu-Leu-Norleu-aldehyde) (Sigma). Reaction mixtures were incubated at room temperature (22 °C) and terminated by adding SDS/PAGE loading buffer at 0, 15, 30 and 60 min. Reactions were analysed by SDS/PAGE (8% gels) followed by Western blotting with an antibody against the FLAG tag on Hsl1.
Human APC/C at ~0.1 mg/ml was loaded on to glow-discharged Quantifoil 2/2 electron microscopy grids coated with a second layer of thin carbon. After 30 s the grids were washed twice with water and negatively stained with 2% (w/v) uranyl acetate. Data were collected at room temperature in an FEI Tecnai TF20 electron microscope at an accelerating voltage of 200 kV under low-dose conditions with an exposure of ~100 e−/Å2 (1 Å=0.1 nm), a nominal magnification of 50000 and an underfocus of ~1.2 μm generating a first minimum in the contrast transfer function at ~17 Å. Images were recorded using a Tietz F415 CCD (charge-coupled device) camera and adjacent boxes of 2 pixels×2 pixels were averaged, resulting in a calibrated sampling of 3.47 Å/pixel.
Particles were manually selected using the EMAN  boxer software. A total of 5639 particles were selected. The human APC/C structure was determined using the S. cerevisiae APC/CCdh1 reconstruction obtained from negatively stained images  as the initial reference for refinement. Multiple rounds of multi-reference alignment were performed using SPIDER  software, and angular assignment was performed by projection matching in IMAGIC . All 5639 individual particles were used to calculate three-dimensional reconstructions using a locally developed Fourier space algorithm, as described previously [43,44]. Atomic co-ordinates of the S. cerevisiae APC/C homology model  were docked into the human APC/C maps using URO .
We previously reported the expression and reconstitution of recombinant S. cerevisiae using the first-generation MultiBac cloning system [24,25,28]. For human APC/C comprising 14 proteins we have constructed two bacmids for baculovirus generation as indicated in Figure 1. For expression of human APC/C we combined both viruses for co-infection of High Five insect cells. The APC/C was purified by means of StrepTactin, ion-exchange and size-exclusion chromatography. SDS/PAGE of recombinant human APC/C showed all 14 APC/C subunits, migrating at their correct apparent molecular masses (Figure 2A).
Using an in-vitro-based ubiquitination assay, we found that recombinant human APC/C was active as an E3 ubiquitin ligase, dependent on co-activator, towards Hsl1, previously shown to be a substrate of endogenous human APC/C . Figure 2(B) shows that recombinant human Cdh1 activated human recombinant APC/C to catalyse multi-ubiquitination of Hsl1, as judged by the formation of higher-molecular-mass species of Hsl1. The activity of recombinant human APC/C was comparable with that of recombinant S. cerevisiae APC/C, although interestingly the human APC/C–UbcH10 system catalysed higher-molecular-mass multi-ubiquitinated species of Hsl1 compared with the budding yeast APC/C–Ubc4 system. Thus the recombinant human APC/C recapitulates the activity of endogenous APC/C, indicating functional assembly of the complex in the baculovirus–insect cell system.
To further assess the functional assembly of human APC/C, we recorded negatively stained electron microscopy micrographs of the complex. These micrographs, and representative class-averages from the image data set, showed the characteristic triangular shape of the APC/C in projection [21,24,34,35,47,48] (Figures 3A and 3B). A three-dimensional reconstruction of human APC/C to a resolution of 20 Å (Figure 3C) was determined using the S. cerevisiae APC/CCdh1 binary complex  as a starting reference model (Figure 4A). This revealed an asymmetric structure, some 250 Å in its longest dimension, sharing a strong resemblance to the endogenous human APC/C determined by cryo-electron microscopy and single-particle analysis [34,35] (Figure 4B). A lattice-like outer shell delineates a central cavity. Importantly, differences between human apo APC/C and S. cerevisiae APC/CCdh1, notably in human APC/C the absence of Cdh1 density but additional density due to Apc7, indicate the absence of bias in the human APC/C reconstruction. The structural similarity between recombinant human APC/C and endogenous human APC/C (Figure 4B), and its capacity to ubiquitinate Hsl1 (Figure 2B), demonstrates the correct assembly of the recombinant human APC/C.
On the basis of the 10 Å resolution cryo-EM reconstruction of the endogenous S. cerevisiae APC/C–Cdh1 complex, we had previously constructed a structural model of 70% of APC/C residues, corresponding to the majority of the large subunits (Apc3/Cdc27, Apc6/Cdc16, Cdc26, Apc8/Cdc23, Apc2, Apc10 and Cdh1) . This model, with the exclusion of Cdh1, could be readily docked into our three-dimensional reconstruction of recombinant human APC/C, requiring no adjustment of the relative juxtapositions of Apc3, Apc6 and Apc8 of the TPR lobe, or of Apc2 and Apc10 of the combined catalytic and substrate-recognition module  (Figure 4A). Additional density above Apc3 at the top of the TPR lobe, which is absent from S. cerevisiae APC/C, can be assigned to the TPR subunit Apc7 (69 kDa), a paralogue of Apc3 that is unique to vertebrate APC/C. The volume of the Apc7-assigned density is consistent with Apc7 being a homodimer .
The successful assembly and reconstitution of human APC/C using the insect cell–baculovirus expression system indicates that all human APC/C subunits necessary for a functional and catalytically active APC/C have been identified. Interestingly Apc15  and Apc16 [22,23] were only discovered as human APC/C subunits relatively recently. Our efforts to generate fully assembled recombinant human APC/C prior to the identification of Apc16 were unsuccessful. In the absence of Apc16, we could only generate subcomplexes as indicated by incomplete subunit composition on SDS/PAGE and negatively stained electron micrographs (results not shown). This indicated that Apc16, located within the TPR subcomplex , is necessary for the optimal assembly of recombinant human APC/C in insect cells. Its requirement for APC/C assembly, and its location at the top of the TPR lobe , presumably close to Apc3/Cdc27, suggests an analogous role to the Cdc27-associated Apc9 subunit of S. cerevisiae APC/C . Apc16 is required for the metaphase-to-anaphase transition in human cells, and its loss stabilizes APC/C substrates in mitosis . In the context of human cells, Apc16 was not required for the assembly of conserved APC/C subunits , raising the question of what role Apc16 plays to mediate APC/C-dependent substrate ubiquitination. Further emphasizing its importance, Apc16 is a conserved subunit of Caenorhabditis elegans and Danio rerio APC/C [23,50], although it does not share recognizable sequence similarities with yeast APC/C subunits. Another recently identified human APC/C subunit, Apc15, that functions to regulate APC/C–MCC interactions , was not required for the correct assembly of the recombinant complex (results not shown). This is reminiscent of the S. cerevisiae Apc15 orthologue Mnd2, whose deletion also did not affect the subunit composition of budding yeast APC/C .
The recombinant human APC/C three-dimensional reconstruction strongly resembles one of the conformers reported for endogenous human apo APC/C (Figure 4B). Conformational flexibility of the TPR lobe (also referred to as the arc lamp domain) relative to the platform was described for the apo states of both human and Xenopus APC/C . Association of either the co-activator Cdh1 to Xenopus APC/C  or MCC to human APC/C  rigidified the TPR lobe to adopt a more closed conformation. The results of the present study do not support the notion that the TPR lobe undergoes large conformational changes relative to the platform. First, we do not observe conformational variations of the TPR lobe of recombinant human apo APC/C, either through assessment of two-dimensional class averages (which show clear detail over the entire particle and no smearing as seen for endogenous human apo APC/C ), or based on examination of the reconstructed three-dimensional volumes (Figures 3 and 4). Secondly, we could readily fit the pseudo-atomic structure of the TPR lobe of S. cerevisiae APC/C comprising Apc3/Cdc27, Apc6/Cdc16 and Apc8/Cdc23 into the recombinant human apo APC/C reconstruction without any adjustments of the relative positions of the three TPR subunits. Moreover, this model could also be docked into the APC/C density of the APC/C–MCC reconstruction of Herzog et al.  (see ). Thirdly, the pseudo-atomic model was generated using the 10 Å cryo-electron microscopy reconstruction of the S. cerevisiae APC/C–Cdh1–D-box ternary complex [21,24], and its close correspondence to an apo-state of human APC/C is not consistent with large conformational differences within the TPR lobe between apo and co-activator-bound states of human APC/C.
To our knowledge, the human APC/C with its 14 distinct proteins is the largest multi-subunit complex expressed using heterologous expression systems, and its size exceeds that of S. cerevisiae APC/C . On the basis of the subunit stoichiometry of S. cerevisiae APC/C , and assuming unit stoichiometry for Apc16, human APC/C would be composed of 19 distinct polypeptide chains with a combined molecular mass of 1.2 MDa. Generation of recombinant human APC/C provides a platform for further structural and mechanistic studies of the complex.
The present study, together with the generation of recombinant S. cerevisiae APC/C, the seven-subunit mediator head module , and eight-subunit CSN (COP9 signalosome) complex , exemplify the efficacy of the MultiBac system  for complete reconstitution of functional multi-subunit complexes in insect cells. Reconstitution of other multi-subunit assemblies, for example the 11-subunit eukaryotic exosome, has been described . However, differing from the APC/C, mediator complex and CSN expressed using the MultiBac system, the exosome was reconstituted in vitro from subcomplexes and individual subunits expressed in Escherichia coli. Generation of the multi-gene vectors using USER [36–38] cloning strategies was achieved within a few weeks, and the separate steps of the procedure affords the flexibility to generate combinations of subunits for formation of distinct subcomplexes, important for understanding multi-subunit complex assembly processes.
Ziguo Zhang generated clones and viruses, Jing Yang expressed and purified APC/C. William Chao performed ubiquitination assays. Eric Kong carried out electron microscopy data collection and analysis. Paula da Fonseca performed electron microsopy analysis and produced Figures 3 and 4. Edward Morris supervised the electron microscopy work. David Barford directed the project and wrote the paper.
This work was was supported by Cancer Research UK [grant number C576/A14109 (to D.B.)].
The electron microscopy map has been deposited in the Electron Microscopy Data Bank (EMDB) under accession number EMD-2226.
Abbreviations: APC/C, anaphase-promoting complex/cyclosome; Cdc, cell division cycle; Cdh1, Cdc20 homologue 1; CSN, COP9 signalosome; D-box, destruction box; DTT, dithiothreitol; Hsl1, histone synthetic lethal 1; MCC, mitotic checkpoint complex; PC, proteasome cyclosome; TEV, tobacco etch virus; TPR, tetratricopeptide repeat
- © The Authors Journal compilation © 2013 Biochemical Society