Center for Neurologic Diseases, Brigham and Women's Hospital, Boston, MA, USA
Harvard Medical School, Boston, MA, USA
Address correspondence and reprint requests to Dennis J. Selkoe, Brigham and Women's Hospital, 77 Avenue Louis Pasteur, Harvard Institutes of Medicine, Room 730, Boston, MA 02115, USA. E-mail: email@example.com
The 19-transmembrane, multisubunit γ-secretase complex generates the amyloid β-peptide (Aβ) of Alzheimer's disease (AD) by an unusual intramembrane proteolysis of the β-amyloid precursor protein. The complex, which similarly processes many other type 1 transmembrane substrates, is composed of presenilin, Aph1, nicastrin, and presenilin enhancer (Pen-2), all of which are necessary for proper complex maturation and enzymatic activity. Obtaining a high-resolution atomic structure of the intact complex would greatly aid the rational design of compounds to modulate activity but is a very difficult task. A complementary method is to generate structures for each individual subunit to allow one to build a model of the entire complex. Here, we describe a method by which recombinant human Pen-2 can be purified from bacteria to > 95% purity at milligram quantities per liter, utilizing a maltose binding protein tag to both increase solubility and facilitate purification. Expressing the same construct in mammalian cells, we show that the large N-terminal maltose binding protein tag on Pen-2 still permits incorporation into the complex and subsequent presenilin-1 endoproteolysis, nicastrin glycosylation and proteolytic activity. These new methods provide valuable tools to study the structure and function of Pen-2 and the γ-secretase complex.
We present a method by which an MBP-tagged form of the Pen-2 subunit of γ-secretase may be purified from bacteria to mg quantities at greater than 95% purity. We show that this construct is incorporated into and allows full activity of γ-secretase in a mammalian system. These methods provide valuable tools to study the structure and function of Pen-2 and γ-secretase.
Alzheimer's disease (AD) is a common neurodegenerative disorder defined by abundant extracellular deposits of the amyloid β-protein (Aβ) and intracellular neurofibrillary tangles of the tau protein and leading to progressive loss of memory and cognition with age (Selkoe et al. 2012). Regulated intramembrane proteolysis of the C-terminal portion of the β-amyloid precursor protein (APP) by γ-secretase leads to the release of the APP intracellular domain (AICD) and either Aβ peptides or p3 peptides (Haass et al. 1992). Aβ peptides produced by this proteolysis can vary in length from ~ 36 to 49 residues with the ratio of the Aβ42 to Aβ40 peptides being a commonly used marker of Aβ aggregation potential and therefore pathogenicity. The γ-secretase complex comprised presenilin (PS1 or PS2 isoforms), nicastrin (Nct), anterior pharynx defective-1 (Aph1αL, Aph1αS or Aph1β isoforms), and presenilin enhancer-2 (Pen-2), all of which are necessary and sufficient for γ-activity (Edbauer et al. 2003; Kimberly et al. 2003; Takasugi et al. 2003).
Pen-2 is very highly conserved, being an invariable 101 amino acid residues in length with 70% identity (87% similarity) in all vertebrates. The functionally essential role of the Pen-2 subunit was illustrated in vivo by Bammens et al. (Bammens et al. 2011), who described the phenotype of Pen-2 knockout mice as being very similar to those of PS1/PS2 double knockout mice or a Notch1-deficient mouse and causing embryonic lethality by embryonic day 11. Biochemical studies have previously suggested that the Pen-2 subunit is the final component added to γ-secretase (Takasugi et al. 2003) and that only after this addition is PS able to undergo autoproteolysis within a hydrophobic domain of the PS cytosolic loop between transmembrane domain (TMD) 6 and 7 (Thinakaran et al. 1996; Wolfe et al. 1999). However, it has recently been shown that purified PS1 holoprotein may undergo endoproteolysis to a minor extent in the presence of purified Pen-2 alone without a need for Aph1 or Nct (Ahn et al. 2010). Another report used Pen-2 knock-down in cells to suggest that PS1 can undergo some degree of endoproteolysis in the absence of Pen-2 (Mao et al. 2012).
A critical piece of missing information that would be invaluable for both basic and applied research is a high-resolution structure of the 19-TMD γ-secretase complex. As a potentially more achievable alternative to attempting to crystallize the intact complex, individual subunits could be studied in isolation by X-ray crystallography, NMR and/or 2D-crystallography. This report demonstrates how the Pen-2 subunit can be purified in quantities sufficient for structural analysis and also reveals a potential new tool for other functional analyses of γ-secretase.
The human Pen-2 cDNA sequence was subcloned with an N-terminal FLAG tag into the pMAL-p2x vector (New England Biolabs, Ipswich, MA, USA) leading to the addition of an N-terminal maltose binding protein (MBP) tag connected via a Factor Xa cleavage site. For expression in mammalian cells, MBP-FLAG-Pen-2 was cloned into pcDNA3.1(+) with a hygromycin selection marker.
Cell lines and transfection
Pen-2 knockout mouse embryonic fibroblasts were a kind gift of B. De Strooper (KU Leuven, Belgium). Cells were transfected using the 4D-Nucleofector X unit and P3 Primary Cell 4D-Nucleofector X kit (Lonza, Hopkinton, MA, USA), recovered in RPMI media + 10% fetal bovine serum (FBS) for 15 min, then plated in 60-mm dishes with 4 mL Dulbecco's modified Eagle's medium + 10% FBS. Six hours after plating, media were replaced with fresh Dulbecco's modified Eagle's medium + 10% FBS. Conditioned media were harvested after 18 h and stored at −80°C. Cells were lysed in 1% 3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonic acid (CHAPSO), 50 mM HEPES, pH 7.2, 150 mM NaCl with complete protease inhibitors (Roche Molecular Biochemicals, Indianapolis, IN, USA). Monoclonal stable cell lines were generated by clonal dilution and selection with hygromycin.
SDS–PAGE and ELISA
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) was performed by electrophoresing cell lysates on 4–12% Bis/Tris polyacrylamide gels and staining with GelCode Blue (Thermo Scientific, Tewksbury, MA, USA). Quantification of Coomassie-stained gels was performed with ImageJ (Rasband, 2008). Aβ40 and Aβ42 peptides released into conditioned media were measured by a triplex ELISA (Aβ captured by antibody 4G8) and read on a Sector Imager 2400 (Mesoscale Discovery, Rockville, MD, USA).
Immunoprecipitation activity assay
In vitro Immunoprecipitation (IP) activity assays were performed by capturing γ-secretase complexes using either an anti-nicastrin antibody (Sigma N1660; Sigma, St Louis, MO, USA) bound to protein A Dynabeads or an anti-MBP antibody (New England Biolabs E8032) bound to protein G Dynabeads at 4°C for 1.5 h, followed by three washes with 0.25% CHAPSO, 50 mM HEPES, pH7.2, 150 mM NaCl. The resin was resuspended in 50 μL of phosphatidylcholine (1 mg/mL), 0.25 mg/mL phosphatidylethanolamine, 0.25% CHAPSO (in 50 mM HEPES, pH7.2, 150 mM NaCl), and 1 μM C100-FLAG substrate was added and incubated at 37°C for 3 h. Bound samples were eluted with Laemmli sample buffer. Input lysate, eluted material and activity assay supernatant were all analyzed by SDS–PAGE western blots probed with mouse anti-Nicastrin (1 : 1000; BD Transduction Laboratories, San Jose, CA, USA), rabbit anti-Nicastrin (1 : 2000, Sigma N166), mouse anti-MBP (1 : 10 000 NEB E8032), rabbit anti-PS1-NTF (1 : 5000 B19; kind gift of B. De Strooper), rabbit anti-PS1-CTF (1 : 2000; Abcam 76083), rabbit anti-Pen-2 (1 : 5000; Abcam 154830, Abcam, Cambridge, MA, USA), or mouse anti-FLAG (Sigma F3165). All western blots were scanned on an Odyssey Infrared Imaging System (Li-Cor, Lincoln, NE, USA), and densitometry analysis using Odyssey software.
Fast performance liquid chromatography
The fast protein liquid chromatography steps during purification were performed using the ÄKTAprime and ÄKTApurifier systems and analyzed using Unicorn software (GE Healthcare, Pittsburgh, PA, USA).
Statistical analyses of Aβ secretion from cotransfected cells were performed using the Student's t-test with Prism software (GraphPad, La Jolla, CA, USA).
Pen-2 KO cells can be rescued by expressing F-Pen-2 and MBP-F-Pen-2 constructs
Previous reports have shown that adding peptide tags to the C-terminus of Pen-2 leads to a loss of function (Hasegawa et al. 2004; Prokop et al. 2005). Short N-terminal tags allow activity, although highly charged tags lead to an increase in Aβ42 generation by cells (Isoo et al. 2007). To test whether an MBP tag on the N-terminus of Pen-2 still allows incorporation into the γ-secretase complex, a mouse embryonic fibroblast (MEF) cell line generated from a Pen-2 KO mouse (Bammens et al. 2011) was used. We found that these cells could be rescued using Pen-2 N-terminally tagged constructs containing either FLAG (F-Pen-2) or MBP + FLAG (MF-Pen-2); rescue of γ-complex maturation was observed, as shown by the occurrence of PS1 endoproteolysis (Fig. 1a). A trace signal could be seen at the F-Pen-2 position after rescue with MF-Pen-2, suggesting minor cleavage of the fusion protein. To rule out the possibility of this released F-Pen-2 being the cause of complex maturation, IP was performed via the MBP tag of MF-Pen-2 (Fig. 1b). In the non-rescued cell line, no γ-secretase components were pulled down, but in the MF-Pen-2-transfected cells, strong signals were observed for mature Nct, PS1-NTF, and PS1-CTF with no signal at F-Pen-2 as expected, demonstrating the assembly of a MF-Pen-2-containing γ-complex.
γ-secretase containing MBP-tagged Pen-2 is active in vitro
We next asked whether MF-Pen-2 supports substrate cleavage when incorporated into γ-secretase. This was tested using an in vitro IP activity assay to enrich for γ-secretase and then adding lipids, detergent and a canonical substrate comprising the C-terminal 99 residues of APP with an N-terminal methionine (to initiate translation) and a C-terminal FLAG tag (C100-F). This approach was first carried out using anti-nicastrin to IP γ-secretase, showing clear pull-down of PS1-CTF from the lysates of cells rescued with either F-Pen-2 or MF-Pen-2 (Fig. 2a). A trace signal for F-Pen-2 could also be detected in the lane of the IP of the MF-Pen-2 rescued cells, but much weaker than the signal for MF-Pen-2 found in the same lane, and less than 10% of that found in the F-Pen-2 rescued lane. This band may be a background signal from the IP as it is also partially apparent in the blank IP lane. Upon incubation of the pulled-down γ-complexes at 37°C for 3 h, the C-terminal product generated by the cleavage of the C100-F substrate (i.e., AICD-F) was observed from both the F-Pen-2 and MF-Pen-2 immunoprecipitates. We then measured the levels of an N-terminal γ-cleavage product of C100-F (i.e., Aβ40) by ELISA. By normalizing the cleavage product amounts to the amount of generated PS1-CTF, it was possible to assess the relative proteolytic activity per unit of γ-secretase complexes containing either F-Pen-2 or MF-Pen-2; these were approximately equal between the two rescues (Fig. 2b and c). To rule out the possibility of cleavage being mediated by F-Pen-2-containing complexes or partial complexes arising from the trace amounts of F-Pen-2 observed, a separate in vitro IP activity assay was performed using anti-MBP to pull down only complexes containing MF-Pen-2. In this MBP pull-down, no IP of γ-secretase complexes could be observed from lysates of the F-Pen-2-rescued cells as expected, but mNct, MF-Pen-2, and PS1-CTF could all be seen from lysates of the MF-Pen-2 rescued cells (Fig. 2d and f). These MF-Pen-2 complexes, which after anti-MBP IP do not contain any non-MBP-tagged F-Pen-2, still show substrate cleavage, as measured by both AICD-F western blot densitometry (Fig. 2e) and Aβ40 ELISA (Fig. 2f).
γ-secretase containing MBP-tagged Pen-2 is active in intact cells
To extend this evidence for γ-activity observed by in vitro IP activity assays, the Pen-2 KO MEF cells were transiently cotransfected with APP-C99 (the C-terminal 99 residues of APP with the APP signal sequence to traffic correctly) and either F-Pen-2, MF-Pen-2, or vehicle. Aβ40 and Aβ42 concentrations in the media conditioned by these cells were measured by ELISA and normalized to the amount of PS1-CTF (Fig. 3a) and expressed as a percent of readings from cells rescued with F-Pen-2. Interestingly, a clear trend toward increase in Aβ40 and a highly significant (p < 0.001) increase in Aβ42 generation were observed when rescuing with MF-Pen-2 (vs. F-Pen-2), with no alteration in the Aβ42/40 ratio (Fig. 3b).
Multistep purification of recombinant MBP-tagged Pen-2
No high-resolution structural information about the γ-secretase complex is available, in part because of the difficulties of expressing and purifying sufficient γ-secretase. An alternative structural approach is to tackle each subunit individually to learn more about specific parts of the complex. We will now describe a method (summarized in Fig. 4) to express and purify an MBP-FLAG-tagged form of Pen-2 (MF-Pen-2) to high concentrations and purity, with the aim of facilitating structural studies of this subunit.
C43(DE3) E. coli cells were transformed with F-Pen-2 in the pMAL-p2x vector, which inserts an N-terminal MBP tag followed by a Factor Xa cleavage site. Transformants were selected on Luria Bertanyi (LB) agar containing 100 μg/mL ampicillin.
LB media inoculated with multiple colonies were grown at 37°C to saturation.
The bacterial cultures were scaled sequentially to keep them in the exponential growth phase using a 10-fold and then 20-fold dilution to a maximum volume per flask of 1 L. During scale-up, media contained a minimum of 0.2% of 10 mM D-glucose to repress transcription of the malT gene, leading to inhibition of expression of the malS gene that encodes periplasmic α-amylase (Boos and Shuman 1998), thus preventing endogenous maltose production.
Bacterial expression of MF-Pen-2 was induced with 0.1 mM isopropyl β-d-1-thiogalactopyranoside at 37°C for 4 h.
Cell pellets were resuspended in Lysis Buffer (30 mL/L 40 mM Tris, pH 8, 350 mM NaCl, 0.007 % (w/v) β-mercaptoethanol, protease inhibitor cocktail) and lysed by sonication.
n-Dodecyl β-D-maltoside (DDM) in Lysis Buffer was added to lysate to a final concentration of 0.5 % (w/v) and mixed at 4°C for 1 h to solubilize membranes and insoluble proteins.
The solution was clarified by centrifugation at 12 000 g for 1 h and filtering supernatant through a 0.22 μm Durapore membrane.
Sample was immobilized by circulating a minimum of two lysate volumes through an MBPTrap column equilibrated with 0.5% DDM/Lysis buffer (Fig. 5).
Unbound material was washed off the column with 10 column volumes 0.5 % DDM/Lysis Buffer followed by 10 column volumes of Binding Buffer (20 mM Tris, pH7.4, 200 mM NaCl, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride).
Stepwise elution was performed with 5 column volumes 0.2 mM maltose/Binding buffer, then 5 column volumes of 10 mM maltose/Binding buffer in the absence of DDM (Fig. 5a). At this stage, Coomassie-stained SDS–PAGE (Fig. 5b) showed that the low maltose elution contained MF-Pen-2 and a large amount of contaminants, while the high maltose peak predominantly contained MF-Pen-2.
Fractions from the high maltose elution were pooled and concentrated using a 30 kDa MWCO spin concentrator.
A further purification step of the concentrated sample utilized size exclusion chromatography (Superdex 200, GE Healthcare, Pittsburgh, PA, USA) in LDAO buffer (0.05 % LDAO, 20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride) to separate MF-Pen-2 from free MBP and other trace contaminating proteins (Fig. 6a).
Fractions from the SEC elution (Fig. 6b) were pooled and concentrated as before, then analyzed for purity and concentration by Coomassie staining (Fig. 7a and b) and Nanodrop spectrophotometry (data not shown), respectively.
To confirm the identity of the purified protein, the final sample was resolved by SDS–PAGE and the major band excised for trypsin digestion followed by LC-MS/MS. This band was confirmed to contain both MBP and human Pen-2 (Fig. 7c), whereas the two minor bands migrating faster on SDS–PAGE were found to be OmpF and MBP alone, respectively (data not shown). Using the above method, we obtained 1-1.5 mg of MF-Pen-2 per liter of culture, which can then be concentrated to > 15 mg/mL at ~ 95 % purity (Fig. 7a and b).
In this study, a method is described to enable expression of Pen-2 with a Factor Xa-cleavable solubility tag (MBP) and purification to sufficiently high concentration and purity to enable structural studies. The functional ability of this construct is validated in a Pen-2 KO cell system, showing incorporation into γ-secretase complexes with subsequent proper maturation and cleavage activity comparable to non-MBP-tagged Pen-2 construct.
Given prior reports about the importance of the length of Pen-2 (Hasegawa et al. 2004; Isoo et al. 2007), the lack of significant negative impact by addition of the 43 kDa maltose binding protein to the N-terminus on the function of γ-secretase was unexpected. Interestingly, rescue of Pen-2 KO cells with MBP-tagged Pen-2 actually elevated Aβ secretion from per mature γ-complex relative to rescue with the non-MBP-tagged Pen-2. There was no accompanying change in the Aβ42/40 ratio, suggesting that the overall increase in proteolytic activity in cells may be because of altered trafficking or stability rather than a conformational change in γ-secretase.
Given our validation of the MF-Pen-2 construct for mammalian and bacterial expression, the system we describe may offer a powerful tool to investigate other aspects of γ-secretase. Using the MBP tag allows a simple and inexpensive way to isolate γ-secretase complexes with amylose resin and maltose elution, even from cell lines expressing endogenous Pen-2. This could potentially facilitate experiments such as cross-linking followed by tryptic digests and LC-MS/MS, to identify closely adjacent regions within γ-secretase. Another potential use is identification of the location of Pen-2 within the γ-secretase complex and the orientation of the complex using cryo-electron microscopy.
The need for new approaches to obtain greater structural detail about the γ-secretase complex is apparent. To date, the highest resolution structures of the full complex have been obtained using electron microscopy, with the greatest resolution of 12 Å obtained by us for γ-secretase alone (Osenkowski et al. 2009), and a more recent 17.4 Å resolution obtained with the inhibitor Compound E bound altering structural conformation (Li et al. 2013b). Much of our current knowledge of how regulated intramembrane proteolysis may occur comes from 3D crystallographic structures of other intramembrane proteases such as bacterial rhomboid (Wang et al. 2006; Wu et al. 2006; Lemieux et al. 2007), an archaeal S2P (Feng et al. 2007), and an archaeal presenilin homologue (Li et al. 2013a). While the collective data on the γ-secretase complex provide some structural insights into possible enzyme mechanisms, such as potential water-accessible pores and a conformational change upon inhibitor binding, they will not enable structure-based design of compounds to modulate γ-secretase proteolytic function. An alternative approach to gaining atomic resolution details about the γ-secretase complex is to analyze individual subunits or subcomplexes, which could then be used to generate a model of the entire complex. A partial NMR structure of the C-terminal fragment of PS1 has revealed an unexpected feature of TMD 7 being only a partial helix, suggesting a mechanism by which a water- and substrate-accessible pore could be formed (Sobhanifar et al. 2010). Using the method of expression and purification reported here, it may be possible to gain high-resolution structural details by X-ray crystallography or NMR of Pen-2 with or without the cleavable MBP solubility tag. Such structural detail could reveal accessible amino acid side chains capable of hydrogen bonding that could be potential sites of interaction between the Pen-2 TMDs and presenilin. This could add to the prior knowledge that Pen-2 TMD 1 interacts with PS1 TMD 4 (Kim and Sisodia 2005a,b; Watanabe et al. 2005) and in turn how binding of Pen-2 could facilitate presenilin endoproteolysis and stabilization of the γ-secretase complex.
Acknowledgments and conflict of interest disclosure
This work was supported by National Institutes of Health Grant P01 AG015379 to DJS and MW. We thank Bart De Strooper for providing the Pen-2 knockout MEF cell line.
All experiments were conducted in compliance with the ARRIVE guidelines. No conflicts of interest to be declared.