Highly efficient production of the Alzheimer's γ-Secretase integral membrane protease complex by a multi-gene stable integration approach§

Authors

  • Jean-René Alattia,

    1. Laboratory of Molecular & Cellular Biology of Alzheimer's Disease, Brain Mind Institute and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH1015 Lausanne, Switzerland; telephone: 41-21-693-96-51, fax: 41-21-693-95-72
    Search for more papers by this author
  • Mattia Matasci,

    1. Laboratory for Cellular Biotechnology, Institute of Bioengineering and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for more papers by this author
  • Mitko Dimitrov,

    1. Laboratory of Molecular & Cellular Biology of Alzheimer's Disease, Brain Mind Institute and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH1015 Lausanne, Switzerland; telephone: 41-21-693-96-51, fax: 41-21-693-95-72
    Search for more papers by this author
  • Lorène Aeschbach,

    1. Laboratory of Molecular & Cellular Biology of Alzheimer's Disease, Brain Mind Institute and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH1015 Lausanne, Switzerland; telephone: 41-21-693-96-51, fax: 41-21-693-95-72
    Search for more papers by this author
  • Sowmya Balasubramanian,

    1. Laboratory for Cellular Biotechnology, Institute of Bioengineering and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for more papers by this author
  • David L. Hacker,

    1. Laboratory for Cellular Biotechnology, Institute of Bioengineering and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for more papers by this author
  • Florian M. Wurm,

    1. Laboratory for Cellular Biotechnology, Institute of Bioengineering and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for more papers by this author
  • Patrick C. Fraering

    Corresponding author
    1. Laboratory of Molecular & Cellular Biology of Alzheimer's Disease, Brain Mind Institute and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH1015 Lausanne, Switzerland; telephone: 41-21-693-96-51, fax: 41-21-693-95-72
    • Laboratory of Molecular & Cellular Biology of Alzheimer's Disease, Brain Mind Institute and School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH1015 Lausanne, Switzerland; telephone: 41-21-693-96-51, fax: 41-21-693-95-72
    Search for more papers by this author

  • Author contributions: J.R.A, M.M., M.D., L.A., and S.B. performed experiments; D.H., F.W., and P.C.F. designed and supervised the study; D.H. and P.C.F. wrote the manuscript. All authors edited the article.

  • The authors declare having no competing financial interests.

  • §

    Jean-René Alattia and Mattia Matasci contributed equally to this work.

Abstract

Inefficient production of membrane-embedded multi-protein complexes by conventional methods has largely prevented the generation of high-resolution structural information and the performance of high-throughput drug discovery screens for this class of proteins. Not exempt from this rule is γ-secretase, an intramembrane-cleaving protease complex regulating a multitude of signaling pathways and biological processes by influencing gene transcription. γ-Secretase is also implicated in the pathogenesis of Alzheimer's disease and several types of cancer. As an additional challenge, the reconstitution of the protease complex in its active form requires an intricate assembly and maturation process, including a highly regulated endoproteolytic processing of its catalytic component. In this article we report the application of a transposon-mediated multigene stable integration technology to produce active γ-secretase in mammalian cells in amounts adequate for crystallization studies and drug screening. Our strategy is expected to help elucidate the molecular mechanisms of intramembrane proteolysis. It is further expected to be widely used for the production of other multi-protein complexes for applications in structural biology and drug development. Biotechnol. Bioeng. 2013; 110: 1995–2005. © 2013 Wiley Periodicals, Inc.

Introduction

γ-Secretase is the founding member of a new class of intramembrane-cleaving proteases including site 2 protease (S2P), rhomboids, and signal peptide peptidase (SPP), all of which hydrolyze their respective substrates within the transmembrane regions (Wolfe and Selkoe, 2002). γ-Secretase is a high molecular weight complex composed of four integral membrane proteins, presenilin (PS), nicastrin (NCT), anterior pharynx defective 1 (Aph1), and presenilin enhancer 2 (Pen2), with a total of 19 transmembrane-domains (TMDs; Fig. 1; for a review, see Fraering, 2007). PS constitutes the catalytic component and the other three subunits act as necessary cofactors. The maturation of this aspartyl protease is accompanied by the highly regulated endoproteolytic processing of PS to generate N- and C-terminal fragments termed, respectively, PS-NTF and PS-CTF, which are associated with the biological activity of γ-secretase (Fig. 1 and Esler et al., 2002; Thinakaran et al., 1996). γ-Secretase is involved in the regulated intramembrane proteolysis of a growing list of single-pass type-I transmembrane proteins including the amyloid precursor protein (APP), the Notch receptor, and the Neurexin synaptic protein (Bot et al., 2011; Haapasalo and Kovacs, 2011). Importantly, the γ-secretase-dependent processing of substrates triggers signaling cascades by releasing intracellular domains (ICDs) which, once cleaved from its protein precursor at the cell surface, translocate to the nucleus and selectively regulate gene expression (for a review, see Beckett et al., 2012). γ-Secretase is also responsible for the final step in production of amyloid-beta peptides (Aβ), the key causative agents of Alzheimer's disease (AD). Since mutations in PS cause most cases of early-onset familial AD (Bertram et al., 2010), a detailed understanding of γ-secretase structure–function relationships is crucial for the elucidation of the intramembrane protein cleavage activity and the design of therapeutic compounds for the safe treatment of AD.

Figure 1.

Assembly and activation of the γ-secretase complex. These processes involve the early formation in the endoplasmic reticulum (ER) of an intermediate subcomplex of Aph1 and the “immature” N-linked isoform of NCT (iNCT/Aph1 subcomplex). PS-holoprotein (PS-holo), by entering into the iNCT/Aph1 subcomplex, forms a trimeric intermediate (iNCT/Aph1/PS-holo subcomplex). In the latter, PS-holo adopts a conformation in which the hydrophobic domain containing the endoproteolysis site is membrane-embedded in between TMD6 and TMD7, occluding the active site. The association of Pen2 in the ER/cis-Golgi compartments to the trimeric and inactive iNCT/Aph1/PS-holo subcomplex drives a conformational change in PS-holo that promotes (1) the endoproteolysis of PS-holo and subsequently (2) the activation and trafficking of γ-secretase, with (3) the conversion from an immature to a mature form of NCT (mNCT) having a more complex glycosylation pattern. The fully mature and active γ-secretase (mNCT/Aph1/PS-NTF/PS-CTF/Pen2) is mainly localized in the trans-Golgi network (TGN), the endosomes, the lysosomes, and the plasma membrane.

The high-resolution crystal structure of γ-secretase has not been determined because the limited amounts of purified γ-secretase available up to now have been insufficient to support macromolecular X-ray crystallography. As a consequence, structural studies have been restricted to the imaging of single particles by electron microscopy (EM) and cryo-EM, providing only low-resolution structural information of the protease complex (Fraering, 2007; Lazarov et al., 2006; Osenkowski et al., 2009). Clearly, challenges in active γ-secretase overexpression and purification must be overcome in order to crystallize this membrane protein complex and solve its high-resolution structure.

In this study, we addressed these challenges using the PiggyBac (PB) transposon-mediated multigene transfer system to generate recombinant Chinese hamster ovary (CHO) cell lines stably co-expressing the four γ-secretase subunits. The PB transposon occurs naturally in the cabbage looper moth (Trichoplusia ni) and was recently modified to allow transposition in mammalian cells (Cary et al., 1989; Ding et al., 2005). PB-mediated gene transfer has been applied to the generation of recombinant CHO cell lines producing the tumor necrosis factor receptor as an Fc-fusion protein (TNFR-Fc) (Matasci et al., 2011). In this example, the TNFR-Fc gene, placed between the PB-specific terminal repeat sequences on the donor plasmid, was efficiently transposed from the plasmid DNA into the CHO genome via a cut-and-paste mechanism catalyzed by the PB transposase that was transiently expressed from the helper plasmid. This approach to recombinant cell line generation in CHO cells has now been extended to demonstrate the co-transposition of multiple PB transposons each carrying the genes of two γ-secretase subunits plus a gene for selection. Simultaneous multigene transfer resulted in a drastic improvement in active γ-secretase production as compared with the two other available systems, that is, γ-secretase stably overexpressed in a CHO cell line generated by successive integration of four different plasmids each carrying a gene for one γ-secretase subunit and a selection marker (Cacquevel et al., 2008) and γ-secretase transiently reconstituted in Sf9 cells using baculovirus infection (Ogura et al., 2006). Indeed, a yield of ∼10.2 mg of highly purified and active γ-secretase was achieved from individual 10-L suspension cultures of two PB-mediated CHO cell lines expressing the active protease complex. We conclude that transposon-based multigene transfer is a powerful approach for the production of milligram amounts of highly purified multi-protein complexes, including γ-secretase, for structure determination and drug discovery applications.

Materials and Methods

Construction of the p2MPT and p2MDT Transposon-Based Helper Vectors

In order to remove the BamHI and PstI restriction sites present in the puromycin resistance (pac) gene expression cassette from pMP-PB (Matasci et al., 2011), the expression cassette consisting of the herpes simplex virus thymidine kinase promoter (HSV-TK), the pac gene and SV40 polyA was amplified by PCR using the forward and reverse primers: 5′-TAATACTAGTGCTAGCTTCATCCCCGTGGCC-3′ and 5′-TAATGAATTCTCATGAACTTGTTTATTGCAGCTTATAATGG-3′, respectively. The PCR product was digested with SpeI and EcoRI and cloned into the same sites of pBluescript II KS(+) (Stratagene, La Jolla CA) to generate pBSK-RC. The pac gene was amplified by PCR from pMP-PB using the forward and reverse primers 5′-ATTAATGCATCGCCACCATGACCGAGTACAAGCCCACGGTGCGCCTCG-3′ and 5′-GCCGTGATCATTATCAGGCACCGGGCTTGCGGGTCATGCACC-3′, respectively. The PCR product was digested with NsiI and BclI and inserted into the BamHI and PstI sites of pBSK-RC to generate pBSK-RC-PURO. The dihydrofolate reductase (DHFR) gene was amplified by PCR from pMZ-59 (Miescher et al., 2000) using the forward and reverse primers 5′-TAATATGCATCGCCACCATGGTTCGACCATTGAACTGC-3′ and 5′-GCCGTGATCATTATTAGTCTTTCTTCTCGTAGACTTCAAACTTAT-3′, respectively. The PCR product was digested with NsiI and BclI and inserted into the BamHI and PstI sites of pBSK-RC giving rise to pBSK-RC-DHFR. The newly constructed pac expression cassette was then recovered from pBSK-RC-PURO by digestion with NheI and BspHI and inserted into the same sites of pMP-PB, giving pMP-PB-puro. The multi-cloning site of pMP-PB-puro was modified by inserting into the BamHI and BclI sites of this plasmid, the annealed primers 5′-GATCAGCGGCCGCTGAAGGATCCCTTGACTAGTGCCGT-3′ and 5′-GATCACGGCACTAGTCAAGGGATCCTTCAGCGGCCGCT-3′. The resulting plasmid was called pMP-PB-puro-NBS. A second mammalian expression cassette was amplified by PCR from pMP-PB with forward and reverse primers 5′-CGGCGCTAGCAGTTTAAACAACAGGAAAGTTCCATTGG-3′ and 5′-ATTACCTAGGTTCGCGACCATAGAGCCCACCGCATC-3′, respectively, and T/A subcloned into pGem-T Easy (Promega, Madison, WI) to generate pGemT-mCMV. The multiple cloning site of this plasmid was then modified by inserting the annealed primers 5′-GATCAGAATTCTGAATCTAGACTTGACTAGCTAGTGCCGT-3′ and 5′-GATCACGGCACTAGCTAGTCAAGTCTAGATTCAGAATTCT-3′ into its BamHI and BclI restriction sites. The resulting plasmid was called pGemTeasy-mCMV-EXB. The gene expression cassette from this plasmid was digested with NheI and AvrII and inserted into the NheI site of pMP-PB-puro-NBS, generating p2MPT. Finally, the expression cassette for the DHFR gene from pBSK-RC-DHFR was digested with NheI and BspHI and inserted into the corresponding sites of p2MPT, giving rise to p2MDT.

Construction of the γ-Secretase Multi-gene Transfer Vectors

The genes for ApH1aL-HA, NCT-V5His6, PS1, and Flag-Pen2 were amplified by PCR using the oligonucleotide primers and DNA templates described in Table I. The resulting PCR products were cloned into the p2MPT or p2MDT bigenic PB donor vectors as follows. p2MPT-Aph1aL-NCT was generated by first inserting the Aph1aL gene into the EcoRI and XbaI sites of p2MPT, followed by the insertion of the NCT gene into the NotI and SpeI sites. Similarly, p2MDT-PS1-Pen2 was constructed by the insertion of the Pen2 gene into the NotI and BamHI sites followed by the insertion of the PS1 gene into the EcoRI and XbaI sites of p2MDT. All the constructs were sequence verified. Plasmid DNA for transfection was prepared using the NucleoBond® PC 500 kit (Macherey-Nagel, Düren, Germany).

Table I. Primers and templates used to amplify the genes Flag-Pen2, PS1, Aph1aL-HA, and NCT-V5His6 for sub-cloning into the p2MPT or p2MDT bigenic PB donor vectors.
PrimerDirectionSequenceTemplatePCR product
Flag-Pen2-FForwardTAATGCGGCCGCCATGGACTACAAGGACGACpcDNA3.1/Zeo(+)-en2NotI-FLAG-Pen2-BamHI
Flag-Pen2-RReverseTAATGGATCCTATCAGGGGGTGCCCAGGGGTATG
PS1-FForwardATTAGAATTCACCATGACAGAGTTACCTGCpcDNA3.1/Zeo(+)-PS1EcoRT-PSl-Xbal
PS1-RReverseCGGCTCTAGACTACTAGATATAAAATTGATGGAA
AphlaLHA-FForwardATTAGAATTCACCATGGGGGCTGCGGTGTTTTTCGGCTGCACTpcDNA3.1(+)-AphlaL-HAEcoRI-AphlaL-HA-Xbal
AphlaLHA-RReverseCGGCTCTAGACTATCAAGCGTAGTCTGGGACGTCGTATGGGTAGT
NCT-VH-FForwardATTAGCGGCCGCCATGGCTACGGCAGGGGGTGGCTCTGGGGCTGpcDNA6.0-NCT-VSNotI-NCT-V5-His6-Spel
NCT-VH-RReverseGCCGACTAGTTATCAATGGTGATGGTGATGATGACCGGTACGCGTAG

Cell Culture

CHO-derived S20 cells stably overexpressing human APP751, Aph1aL-HA, Flag-Pen2, PS1, and NCT-V5His6 (Cacquevel et al., 2008) were either grown on plates in DMEM containing 10% fetal bovine serum (FBS), penicillin/streptomycin, 150 µg/mL G418, 2.5 µg/mL puromycin, 250 µg/mL zeocin, 250 µg/mL hygromycin, and 10 µg/mL blasticidin, or in ProCHO5 medium (Lonza Verviers, Verviers, Belgium) containing 1% FBS and 2 mM L-glutamine. Small-scale suspension cultures of S20 cells were performed in TubeSpin® Bioreactor 50 tubes (TPP, Trasadingen, Switzerland) and agitated by orbital shaking at 180 rpm in an ISF-4-W incubator (Kühner AG, Birsfelden, Switzerland) at 37°C in the presence of 5% CO2 (Muller et al., 2005). Large-scale 10-L suspension cultures in ProCHO5 medium were performed in six 5-L bottles (1.7 L per bottle) and agitated at 110 rpm at 37°C as above.

Suspension-adapted CHO DG44 cells were maintained at 37°C with 5% CO2 in serum-free ProCHO5 medium supplemented with 13.6 mg/L hypoxanthine, 3.84 mg/L thymidine, and 4 mM glutamine. The cells were maintained in 100–200 mL of medium in TubeSpin® Bioreactor 600 tubes (Maxi-TubeSpins, TPP AG) with orbital shaking as previously described (Muller et al., 2005). The cells were transfected using linear 25 kDa polyethylenimine (PEI; Polysciences, Eppenheim, Germany) according to a protocol previously described (Matasci et al., 2011). Selection was performed in ProCHO5 supplemented with 4 mM glutamine, 10 µg/mL puromycin (Alexis Biochemicals, Lausen, Switzerland) and 100 nM methotrexate (SAFC Biosciences, Saint Louis, MS). After 10 days of selection, single cell cultures were initiated in 96-well plates by limiting dilution. All subsequent cell cultivation was performed in the absence of selection.

Western Blotting and Antibodies

Whole cell extracts prepared in 50 mM HEPES buffer containing 1% NP40 or purified γ-secretase samples were separated by electrophoresis on NuPAGE® Novex® 4–12% Bis–Tris gels (Invitrogen, Carlsbad, CA) for SDS–PAGE analysis (12% Tris-glycine), transferred onto nitrocellulose membranes, and probed with antibodies NCT164 (NCT-specific, 1:1,000, BD Biosciences, Franklin Lakes, NJ), MAB1563 (PS1-NTF-specific, 1:1,000, Merck Millipore, Billerica, MA), 3F10 (HA-specific, 1:2,000, Roche Applied Science, Penzberg, Germany), UD-1 (Pen2-specific, 1:500, gift from Helena Karlström), M2 (Flag-specific, 1:1,000, Sigma–Aldrich, Saint Louis, MS), or anti-actin (1:5,000, A2066, Sigma–Aldrich). Native samples were run on NativePAGE™ Novex® Bis–Tris 4–16% gels for BN–PAGE analysis (Invitrogen) and γ-secretase was revealed with NCT- or Flag-Pen2-specific antibodies NCT164 and M2, respectively. Samples from γ-secretase activity assays were run on 4–12% Bis–Tris gels and transferred onto PVDF membranes to detect Aβ and AICD-Flag with Aβ-specific 6E10 (1:1,000; Covance, Berkeley, CA) and Flag-specific M2 antibodies, respectively. Anti-mouse/rabbit/rat IgGs conjugated to Alexa 680 were purchased from Invitrogen and the Odyssey infrared imaging system (LI-COR, Lincoln, NE) was used to detect the fluorescent signal.

γ-Secretase Purification

Figure 3d and e: Whole cell lysates were prepared by solubilizing equal amounts of cells in 1% CHAPSO-HEPES (pH 7.4) containing a protease inhibitor cocktail (Roche), and centrifuged at 100,000 g for 1 h to collect the lysate. Anti-Flag M2 affinity beads (Sigma–Aldrich) were added to 400 µg of protein lysates, incubated overnight at 4°C, washed three times in 1% CHAPSO-HEPES buffer, resuspended in 0.25% CHAPSO-HEPES buffer, and subjected to in vitro γ-secretase activity assays (Fig. 3e). Alternatively, immunoprecipitated proteins were eluted in 1% CHAPSO-HEPES buffer containing 200 mg/mL Flag peptide, and analyzed by immunoblot (Fig. 3d). Figure 4a and b: γ-secretase was purified from S20 or PB suspension cultures following the protocol described previously (Cacquevel et al., 2008). Eluates 1 and 2 (E1 and E2) were used for Western blot analysis or activity assays (Fig. 4a) or concentrated 30-fold on filter concentrators with molecular weight cutoff of 30,000 Da before Coomassie staining (Fig. 4b).

γ-Secretase Activity Assays

γ-Secretase assays using recombinant APP-C100Flag were performed as previously reported (Alattia et al., 2011; Cacquevel et al., 2008; Wu et al., 2010). Briefly, semi- or highly-purified γ-secretase from S20 or PB cell lines were solubilized in 0.2% (w/v) CHAPSO, 50 mM HEPES (pH 7.0), 150 mM NaCl, 5 mM MgCl2, and 5 mM CaCl2 and incubated at 37°C for 4 h with 1 mM substrate, 0.1% (w/v) phosphotidylcholine, and 0.025% (w/v) phosphotidylethanolamine. The resulting products, AICD-Flag and Aβ, were detected with specific antibodies or analyzed by mass spectrometry as described below.

Immunoprecipitation–Mass Spectrometry (IP–MS) Analysis of Aβ and AICD-Flag

Aβ and AICD-Flag generated in γ-secretase in vitro assays were analyzed as previously described (Alattia et al., 2011). Briefly, Aβ was immunoprecipitated overnight using monoclonal anti-Aβ antibody 4G8 and protein G-coupled agarose (Roche). For AICD-Flag detection, Triton-X100 was added after the enzymatic reaction at a final concentration of 1% (w/v), and incubated for 20 min at 55°C prior to overnight immunoprecipitation at 4°C with anti-Flag M2 resin (Invitrogen). Aβ and AICD-Flag were eluted with 1:20:20 (v:v:v) mix of 1% (v/v) trifluoroacetic acid:acetonitrile:H2O mixed 1:1 (v:v) with saturated CHCA (a-cyano 4-hydroxy cinnaminic acid), and analyzed by MALDI-TOF mass spectrometry in reflectron mode on an ABI 4800 MALDI-TOF/TOF mass spectrometer (Applied Biosystems, Carlsbad, CA). Molecular masses were accurately measured and searched against amino acid sequences of human APP-C99 with addition of a methionine residue at the N-terminus and a Flag tag sequence at the C-terminus (C100Flag).

Estimation of Ampicillin Resistance (bla), Puromycin Resistance (pac), and Mouse DHFR Gene Copy Numbers by qPCR

The integrated gene copy number of the beta-lactamase gene (bla), the mouse DHFR gene, and the puromycin resistance (pac) gene was determined by quantitative real-time PCR (qPCR) using the LightCycler1 480 real-time PCR system (Roche Applied Science, Basel, Switzerland). From each cell line, 1 × 106 cells were collected by centrifugation, washed twice with cold PBS, flash frozen in liquid nitrogen, and stored at −80°C. Genomic DNA isolation was performed using a DNeasy Blood & Tissue kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The PCR was performed using the ABsoluteTM qPCR SYBR Green ROX mix (ABgene, Epsom, Surrey, UK) and 200 nM of each oligonucleotide primer. The forward and reverse primers for the bla gene were 5′-ACGATCAAGGCGAGTTACATGA-3′ and 5′-ACACTGCGGCCAACTTACTTCT-3′, respectively. For the DHFR gene the forward and reverse primers were 5′-ATTCCTGCATGATCCTTGTCAC-3′ and 5′-GGATAGTCGGAGGCAGTTCTGT-3′, respectively. For the pac gene the forward and reverse primers were 5′-TGGAACCGCTCAACTCGGCCAT-3′ and 5′-AAGGTGTGGGTCGCGGACGA-3′, respectively. qPCR was performed with a pre-incubation cycle at 95°C for 15 min followed by 35 cycles of amplification (8 s at 95°C, 8 s at 57°C, and 12 s at 72°C). To estimate the transgene copy number a standard curve was prepared by qPCR using genomic DNA from CHO-DG44 cells containing various known amounts of either p2MDT-PS1-Pen2 for the DHFR and bla genes and p2MPT-Aph1a1(L)-NCSTN for the pac gene.

Results

Construction of the Transposon-Based Multi-gene transfer Vectors

First, we created two PB multi-gene transfer vectors p2MPT and p2MDT carrying the puromycin resistance (pac) and mouse dihydrofolate reductase (DHFR) genes, respectively, for selection in the presence of puromycin and methotrexate (MTX; Fig. 2a; see also the full description under Materials and Methods Section). Next, the two expression vectors carrying the four γ-secretase genes were generated by pairing the subunit genes based on the order of assembly and maturation of the γ-secretase complex (Fig. 2b; see also the full description under Materials and Methods Section). Thus, the NCT and Aph1aL genes (V5His6- and HA-tagged, respectively) were expressed from separate promoters in the donor vector p2MPT-Aph1aL-NCT (Fig. 2b). The PS1 and Flag-tagged Pen2 genes were expressed from separate promoters in p2MDT-PS1-Pen2 (Fig. 2b). In each plasmid, the three individual gene expression cassettes were bordered by the PB right (RTD) and left terminal domains (LTD).

Figure 2.

Transposon-mediated multi-gene transfer for the generation of stable cell lines overexpressing the γ-secretase multi-subunit protease complex. a: Maps of the p2MPT and p2MDT transposon-based donor vectors. LTD, left terminal domain of the PB transposon; RTD, right terminal domain of the PB transposon. P-m.CMV, murine cytomegalovirus promoter; EF1-a intron A, elongation factor-1 alpha intron A; BGH pA, bovine grow hormone polyA. ColE1, ColE1 origin of replication; SV40-pA, simian virus 40 polyA; Puro-r, puromycin N-acetyl-transferase gene; P-HSV-tk, herpes simplex virus-thymidine kinase promoter. b: DHFR-deficient CHO DG44 cells were co-transfected with a plasmid expressing the PiggyBac (PB) transposase (PBase; yellow) and two plasmids carrying an artificial PB transposon delivering simultaneously the Aph1aL-HA (blue) and NCT-V5 (green) genes (p2MPT-Aph1aL-NCT; Donor 1) and the PS1 (red) and Flag-Pen2 (violet) genes (p2MDT-PS1-Pen2; Donor 2). The PB LTD and RTD elements are shown in brown. The four γ-secretase genes were expressed from the mCMV promoter while the selection genes (pac in Donor 1 and DHFR in Donor 2) were expressed from the HSV TK promoter. Promoter elements and selection genes are shown as white boxes.

Generation and Selection of Stable Cell Lines Overexpressing γ-Secretase

Suspension-adapted CHO-DG44 cells were co-transfected with the donor vectors p2MDT-PS1-Pen2, p2MPT-Aph1aL-NCT, and the helper vector pmPBase for the transient expression of the PB transposase. Recombinant cells were selected in the presence of puromycin and MTX, cloned by limiting dilution, and expanded in small-scale suspension cultures in the absence of selection. To recover cell lines co-expressing high levels of functional γ-secretase complex, 108 clones were first analyzed for NCT and PS1-NTF accumulation. Sixteen cell lines displaying high NCT and PS1-NTF levels were further characterized by Western blot for the accumulation of all γ-secretase subunits. The results revealed a substantial increase in the levels of mature PS1 heterodimers (PS1-NTF and PS1-CTF), mature NCT, Aph1, and Pen2 in clones #2, 6, 7, 8, 12, 14, and 15 when compared to the previously characterized reference cell line S20 (Fig. 3a). After 1 month of culture (10 passages) without selection, the stable expression of γ-secretase in these seven clones was re-analyzed both by denaturing SDS–PAGE and by Blue Native gel electrophoresis (BN–PAGE), which allows for the detection of the γ-secretase complex in its native conformation. As shown in Figure 3b and c, PB clones #2 and #15 stably expressed strikingly high levels of γ-secretase when compared to the S20 cell line. In an effort to examine the levels of complex activity in these cell lines, semi-pure γ-secretase was prepared by affinity chromatography (see Materials and Methods Section) and used for activity assays with the recombinant APP-based C100-Flag substrate as previously reported (Alattia et al., 2011; Cacquevel et al., 2008; Wu et al., 2010). A dramatic increase in γ-secretase levels and activity was found in PB clones #2 and #15 when compared to S20 cells (Fig. 3d and e). Next, we determined the copy number of each transposon in PB clones #2 and #15, by qPCR amplification of the pac and mouse DHFR genes using genomic DNA from the two clones as templates. The copy number of the gene for ampicillin resistance (bla) was further estimated to determine the number of transgenes integrated by recombination rather than transposition. As shown in Figure 3f, a total of about six copies of the two selection genes (pac and DHFR) were present in the PB clone #2, while about three were present in the PB clone #15. In each cell line, up to one copy of the bla gene was observed, indicating that there may have been one PB donor or PB helper plasmid integrated by recombination (Fig. 3f). Together, our data indicate that the majority of integrated transgenes in each clonal cell line resulted from transposition rather than recombination. As expected, none of the three transgenes were detected in untransfected CH DG44 cells (Fig. 3f).

Figure 3.

Analysis of CHO-based cell lines stably overexpressing active γ-secretase. a: Expression levels of mature PS1 heterodimers (PS1-NTF and PS1-CTF), mature and immature NCT (mNCT, iNCT), Aph1, and Pen2 were compared by immunoblot analysis to these in the previously characterized S20 cell line (Cacquevel et al., 2008) prepared either from adherent cultures (Ctl1) or from suspension cultures (Ctl2). Cell lines expressing high levels of all γ-secretase subunits were further passaged 10 times (one passage every 3-days when confluent) and harvested for SDS–PAGE analysis of γ-secretase subunits (b) or for BN–PAGE analysis of the γ-secretase high molecular weight complex (HMWC) detected with the Flag-Pen2-targeting antibody M2 (c). γ-Secretase immunoprecipitated from selected cell lines was either eluted and subjected to Western blot analysis (d) or used for APP-C100Flag activity assays (e). The resulting cleavage products, AICD-Flag and Met-Aβ, were detected with anti-Flag M2 and anti-Aβ (6E10) antibodies, respectively. Levels of β-actin served as loading controls. f: The copy number of the integrated ampicillin resistance (bla), puromycin resistance (pac), and mouse DHFR genes in clones #2 and #15 and in the parental CHO-DG44 (control) cells was determined by qPCR using genomic DNA as a template. The standard curves for transgene copy number were generated by qPCR with known amounts of p2MPT-Aph1aL-NCT and p2MDT-PS1-Pen2.

In summary, our PB-mediated multigene delivery system resulted in the generation of stable CHO cell lines (PB clones #2 and #15) overexpressing strikingly high levels of the four different γ-secretase subunits (with PS1 endoproteolysis leading to PS1-NTF and PS1-CTF generation) assembled to form a mature and functional intramembrane-cleaving protease complex.

Purification of Active γ-Secretase in Amounts Adequate for Crystallization Studies or Drug Screening

To evaluate the large-scale production of γ-secretase from PB clones #2 and #15, we performed purifications of the protease complex from 10-L suspension cultures using a previously described multi-step affinity purification procedure (Cacquevel et al., 2008). The biochemical characterization of γ-secretase purified from equal amounts of cells (6.0 × 109) revealed a approximately seven- to eightfold increase in both the amounts of γ-secretase components (Fig. 4a, top panels) and relative γ-secretase activity (Fig. 4a, bottom panels) in the preparations from the PB clones compared to the control S20 cell line. In addition, a 2.1-fold reduced cellular aggregation was observed in PB suspension cultures, resulting in a ∼15-fold improvement in active γ-secretase production. The purity and integrity of the γ-secretase preparations were assessed by electrophoresis on 4–16% BN–PAGE gels which were probed with an antibody against NCT or stained with Coomassie blue. As shown in Figure 4b (left panel), both analyses of the purified fractions revealed a single high molecular weight complex (HMWC) with an apparent molecular mass of ∼350 to 400 kDa, consistent with the previously reported mass for the active γ-secretase complex (Fraering et al., 2004, 2005). The densitometric analysis of the first elution (E1 bands) confirmed the approximately seven- to eightfold increase in γ-secretase levels in the two PB cell clones compared to the control S20 cell line. To further assess the purity of our preparations, the protein profile of the recovered γ-secretase complex was analyzed by SDS–PAGE (4–12%). Each visible Coomassie blue-stained band on the gel was identified by mass spectrometry (Fig. 4b, right panel, and Supplementary Fig. S1). All γ-secretase components were identified in both preparations, confirming the purity and composition of the recombinant complex (Fig. 4b, right panel). Next, the specificity of the enzymatic activity of the purified complex was assessed by mass spectrometric analysis of both Aβ and AICD cleavage profiles, as previously described (Alattia et al., 2011). Mass spectra showed no significant modification of APP cleavage profiles relative to S20 controls (Fig. 4c). Finally, the yield of γ-secretase purified from a 10-L suspension culture (PB clones #2) was estimated on a Coomassie stained BN-gel (by densitometry against ferritin calibration samples—Supplementary Fig. S2) to be ∼10.2 mg. This value is similar to the expected yield (∼12 mg), calculated by multiplying the yield of γ-secretase purified from S20 control cells (∼0.8 mg/10 L; Alattia et al., 2012) by the fold improvement in γ-secretase production (∼15). Together, our detailed biochemical characterization of γ-secretase purified from stable CHO cell lines generated with recombinant PB transposons clearly demonstrates the power of this technology to produce active multi-protein membrane complexes in amounts compatible with crystallization studies or drug screening.

Figure 4.

Large-scale purification and biochemical characterization of γ-secretase. a: γ-Secretase purified from 10-L suspension cultures of S20 (Ctl) and PB cell lines #2 and #15 was immunoprobed for all subunits (upper panels) or used for APP-C100Flag activity assays (lower panels). Both γ-secretase components and APP-C100Flag cleavage products AICD-Flag and Met-Aβ were quantified by densitometry (right panels). Data are presented as means ± SD (n = 2). b: Immunoblot or Coomassie stained BN–PAGE and SDS–PAGE analyses of purified γ-secretase. c: MALDI-TOF spectra showing APP-C100-Flag cleavage products Aβ (upper panel) and AICD-Flag (lower panel). E1 and E2: Eluates 1 and 2.

Discussion

The assembly and function of multi-protein membrane complexes in eukaryotic cells are still poorly understood. This clearly emanates from the very significant challenges on various fronts including their production, functional assembly and maturation, purification and crystallization. Indeed, only a few examples of the production of multi-protein complexes in significant quantities have been described using either transient or stable gene expression strategies. In this study we describe for the first time the application of the PB transposon-based multigene transfer system for the generation of stable CHO cell lines producing functional γ-secretase protease complex. Our results clearly demonstrate a considerable improvement in active γ-secretase production over available systems, with a final yield of ∼10.2 mg of highly purified and active γ-secretase protease complex from a 10-L suspension culture.

We previously reported that the presence of the PB helper plasmid increased the frequency of stable cell line generation and both the productivity and long-term stability of the cell lines (Matasci et al., 2011). In the latter study, a higher average transgene copy number was indeed detected in stable cell lines generated in the presence as compared to the absence of the PB helper plasmid. However, it was not shown if the transgenes integrated by transposition or recombination in the presence of the PB helper plasmid. Here, the transgene copy number was determined for clones #2 and #15. In both cell lines, there were more copies of the pac gene from p2MPT-Aph1aL-NCT than the DHFR gene from p2MDT-PS1-Pen2. For clones #2 and #15 there were six and three copies of these two selection genes, respectively. Yet, the presence of one copy of the bla gene for ampicillin resistance in each cell lines suggests that at least one copy of a plasmid was integrated by recombination rather than transposition. Nevertheless, the majority of the transgenes in each cell line were integrated by transposition.

The γ-secretase-producing cell lines described in this study are expected to greatly contribute to efforts aimed at solving the high-resolution structure of the complex. Detailed structural information is indeed essential for gaining functional insights into the biochemistry of γ-secretase and for exploring new therapeutic strategies (rational structure-based drug design) not only for safely treating AD but also for treating many types of cancer in which γ-secretase and its Notch substrates play an important role. γ-Secretase-mediated Notch signaling molecules are crucial for regulating the fate of a wide variety of cell types. Indeed, gain-of-function mutations in the Notch1 receptor are the main cause of T-cell acute lymphoblastic leukemia, while elevated expression of different Notch receptors are associated with cancers of the breast, pancreas, lung, brain and colon (reviewed in Groth and Fortini, 2012; Wang et al., 2010). Consequently, therapeutic inhibition of Notch signaling, for example by using γ-secretase inhibitors, has tremendous potential to treat Notch-associated cancers (reviewed in Groth and Fortini, 2012; Purow, 2012).

To prevent AD or slow its progression, several Aβ-based therapeutic strategies have emerged from the growing knowledge of Aβ generation and its effects on the pathogenesis of the disease (reviewed in Huang and Mucke, 2012). So far, none of these approaches often associated with unwanted side effects reported in the corresponding clinical trials, has been successful. Preliminary results from two ongoing Phase 3 trials with the γ-secretase inhibitor Semagacestat have showed that this compound not only failed to slow cognitive decline in patients with mild to moderate AD, but it actually enhanced the disease (Extance, 2011). According to the trial investigator, one possible explanation for the Semagacestat-based toxicity is that it may have affected the cleavage of Notch. Indeed, Semagacestat is not a Notch-sparing compound. The main conclusion from this and other clinical studies is that Notch-sparing γ-secretase modulators (GSMs) should be the focus of further investigations.

Together, our production of γ-secretase in amounts adequate for structural studies and drug screening will thus greatly contribute to efforts aimed at solving the high-resolution structure of the protease complex, but also at the identification of GSMs and to the elucidation of the molecular mechanisms by which such compounds specifically modulate the processing of APP.

Acknowledgements

We thank Justine Pascual, Virginie Bachmann, and the EPFL Protein Expression Core Facility for excellent technical assistance. The authors also thank Romain Hamelin and Marc Moniatte (EPFL Proteomics Core Facility) for excellent technical assistance with mass spectrometry.

This work was supported by the Swiss National Science Foundation (to L.A. and P.C.F., grant 31003A_134938/1), the Swiss National Centres of Competence in Research, Neural Plasticity and Repair (to P.C.F.), the Strauss foundation (to P.C.F.), and a KTI grant 10203.1 PFLS-LS (to F.W.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Ancillary

Advertisement