γ-Secretase is a multi-subunit membrane protease complex that catalyses the final intramembrane cleavage of the β-amyloid precursor protein (APP) during the neuronal production of amyloid-β peptides (Aβ), which are implicated as the causative agents of Alzheimer's disease (AD). In the present study, we report the reconstitution of a highly purified, active γ-secretase complex into proteoliposomes without exogenous lipids and provide the first direct evidence for the existence of a microenvironment of 53 molecular species from 11 major lipid classes specifically associated with the γ-secretase complex, including phosphatidylcholine and cholesterol. Importantly, we demonstrate that the pharmacological modulation of certain phospholipids abolishes both the integrity and the enzymatic activity of the intramembrane protease. Together, our findings highlight the importance of a specific lipid microenvironment for the structure and function of γ-secretase.

INTRODUCTION

The neuropathological feature of Alzheimer's disease (AD) is the extracellular deposition of 40–43 amino acid (a.a.) long amyloid-β (Aβ) peptides into senile plaques and the intracellular accumulation of hyperphosphorylated tau proteins in neurofibrillary tangles (NFTs) [1]. Although not definitive, evidence collected over the past 20 years has implicated the Aβ peptides as the causative agents in the pathogenesis of AD. Based on this observation, the Aβ cascade hypothesis of AD now proposes that polymerization of Aβ into soluble oligomeric and/or insoluble amyloid deposits is a critical early event that triggers a sequence of pathological reactions including hyperphosphorylation of the protein tau and formation of neurofibrillary lesions, neuroinflammation and neuronal death, ultimately leading to dementia [25]. Aβ peptides are derived from the amyloid precursor protein (APP), a type-I membrane protein that undergoes two successive cleavage steps by β-secretase and by γ-secretase respectively. Together, the two steps generate the APP intracellular domain (AICD) peptide, which is released into the cytoplasm, and Aβ peptides, which are secreted into the extracellular compartment [6,7]. Thus, γ-secretase is responsible for the production of the toxic Aβ peptides and, being so, has become an important therapeutic target for the development of drugs to slow down or prevent the pathogenesis of AD [8].

At the biochemical level, γ-secretase is the founding member of a new class of intramembrane-cleaving proteases that includes the site 2 protease (S2P), rhomboids and the signal peptide peptidase (SPP) [9,10]. Like γ-secretase, S2P, rhomboid and SPP all hydrolyse their substrates within their transmembrane domains (TMDs), whereas the catalytic residues are either submerged in the membrane bilayer or located at the membrane/cytosol interfaces. γ-Secretase is a multi-subunit protease complex composed of four integral membrane proteins: presenilin (PSEN, nine TMDs); nicastrin (NCT, one TMD); Pen-2 [PSEN enhancer 2; three TMDs]; and Aph-1 (anterior pharynx-defective 1; seven TMDs). PSEN is the catalytic component of the complex whereas the other three proteins act as essential cofactors [11,12]. γ-Secretase is an aspartyl protease, characterized by two conserved intramembrane aspartate residues found within PSEN [13]. Importantly, the molecular mechanisms by which γ-secretase processes its substrates remain unknown. Previous findings from biochemical assays with APP-based substrates support a sequential model of substrate processing [14,15]: endoproteolytic activity of γ-secretase at the epsilon (ε) cleavage site near the predicted membrane/cytosol interface first releases 50- and 51-a.a. long AICDs and 49- and 48-a.a. long Aβ peptides. Next, these long Aβ peptides are successively shortened every three or four residues by γ-cleavages along two production lines: Aβ49 > Aβ46 > Aβ43 > Aβ40 and Aβ48 > Aβ45 > Aβ42 > Aβ38, leading to the production of Aβ peptides of different length (38–43 a.a. residues).

The mechanism by which γ-secretase cleaves its transmembrane-spanning substrate would be easier to elucidate if an atomic resolution 3D structure of this membrane-embedded protein complex was available. We succeeded in reconstituting intact, proteolytically active γ-secretase complex and determined its 3D structure by EM and cryo-EM and single particle image analysis at the resolution of 15 Å (1 Å=0.1 nm) [16] and 12 Å [17] respectively. 3D structures revealed the presence of a low-density interior chamber and apical and basal pores that could allow the entry of water molecules required to accomplish peptide bond hydrolysis and provided plausible explanations for how γ-secretase releases its cleavage products into distinct subcellular compartments [16]. The cryo-EM structure also revealed a potential substrate-binding surface groove in the transmembrane region of the complex [17]. Recently, Lu et al. [18] determined the 4.5 Å 3D structure of solubilized γ-secretase by cryo-EM and image processing, revealing a horseshoe-shaped structural organization of the TMDs and a structural homology between the NCT extracellular domain and the glutamate carboxypeptidase PSMA. This new structure also showed the arrangement of the α-helical structural components of the γ-secretase complex. More recently, Bai et al. [19] determined the 3.4 Å 3D structure of solubilized γ-secretase by high-resolution cryo-EM. The distance of 10.6 Å between the two catalytic aspartate residues found by Bai et al. [19] indicates that the catalytic site does not adopt an active conformation in this structure. However, this structure revealed two PC molecules embedded into the complex and mediating hydrophobic interactions between its subunits. More specifically, the first PC molecule interacts with TMDs 1/4/7/8 of Aph-1 and the TMD of NCT, whereas the second PC molecule interacts with TMDs 1 and 8 of PS and TMD 4 of Aph-1 [19]. Despite this progress, the high-resolution structure of the γ-secretase complex in its active conformation is required to gain functional insights into its biochemistry.

Interestingly, previous studies have shown that folding, structure and function of membrane proteins or multiprotein complexes are influenced by their lipid environments [20,21]. MS was employed to assess the specificity and stoichiometry of complex protein–protein and other ligand–protein interactions [20,21]. Robinson and co-workers used MS to study the binding selectivity and interaction stoichiometry of membrane proteins and lipids [20]. To this end, they measured the intact mass of proteolipid complexes generated in solution (by using a high-concentration of non-ionic detergent; [22]) or in the gas phase (ion mobility; [23]). For instance, they observed that the water channel Aquaporin Z from Escherichia coli was mainly stabilized by exogenous lipids including phosphatidylcholines, phosphatidylserines (PSs) and phosphatidylglycerols (PGs) [24].

Recent studies also demonstrated the altered lipid composition of brain, cerebrospinal fluid and plasma of AD patients. Changes in sphingolipids and cholesterol content during neurodegeneration have received particular attention [2530]. Also, phospholipid levels were decreased in brain regions highly affected by the pathology of AD [31] and phospholipid profiles were altered in the brain, cerebrospinal fluid and plasma at different stages of AD [32,33]. It is therefore conceivable that lipids could modulate the activity of proteins involved in AD. Indeed, γ-secretase is an integral membrane protein and membrane lipid composition could affect proteolytic processing of APP and other substrates. The localization of γ-secretase in membrane rafts was reported [34] and, conceivably the enrichment of raft lipids such as sphingolipids, cholesterol or phospholipids with predominantly saturated fatty acid moieties could enhance its activity.

In the present study, we report the reconstitution of the purified active γ-secretase complex into proteoliposomes, without adding exogenous lipids. Consistent with this observation, we further demonstrated that the γ-secretase complex co-purifies with a specific constellation of endogenous lipids. Previous in vitro studies investigated the influence of defined exogenous lipids on reconstituted γ-secretase complexes [3537] and showed that γ-secretase activity is abolished when the complex is embedded in a lipid membrane with less fluidity and made exclusively of saturated lipids [38]. However, these studies did not reveal the composition of the endogenous lipid microenvironment of γ-secretase. In our study, we determined the molecular composition of the lipidome associated with the γ-secretase complex. We further demonstrate that a pharmacological modulation of the lipid environment of γ-secretase abolishes both the integrity and the activity of the protease complex. We provide the first model for the structural re-arrangement of the γ-secretase complex and the dissociation of the subunits caused by changes in the composition of associated lipids.

EXPERIMENTAL

Materials

3-([3-Cholamidopropyl]dimethylammonio)-2-hydroxy-1-propanesulfonate (CHAPSO), egg yolk 1,2-diacyl-sn-glycero-3-phosphocholine (PC) and 1,2-diacyl-sn-glycero-3-phosphoethanolamine (PE) were purchased from Sigma–Aldrich. 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC) was purchased from Avanti Polar Lipids, Inc. Dulbecco's modified Eagle's medium (DMEM), FBS, penicillin-streptomycin, G418 sulfate, zeocin, hygromycin and blasticidin were purchased from Invitrogen. Puromycin was purchased from VWR. Gel filtration calibration standards blue dextran 2000, thyroglobulin and ferritin were purchased from GE Healthcare. The E. coli lipid extract used for the reconstitution of γ-secretase into proteoliposomes was purchased from Avanti Polar Lipids, Inc.

Cell lines and cultures in suspension

Suspension-adapted Chinese hamster ovary (CHO) DG44 cells stably overexpressing the human γ-secretase complex [39] 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. Next, cells were cultured in 400 ml of ProCHO5 medium supplemented with 1% FBS medium in a 1-litre square-shaped glass bottle with orbital shaking at 110 rpm. At the final 10-litre stage, the culture was divided into six 5-litre cylindrical glass bottles (1.7-litre per bottle) and agitated at 110 rpm at 37°C. After 2–3 days, cells reached a density of 1.8–2.0×106 cells/ml and were harvested by centrifugation at 350 g for 5 min. Cell pellets were washed once with PBS and stored at–80°C.

γ-Secretase purification

Briefly, whole membrane lysates were prepared by solubilizing CHO DG44 cells stably overexpressing the human γ-secretase complex [39] in 1% CHAPSO–HEPES (pH 7.4) containing a protease inhibitor cocktail (Roche) and centrifuged at 100000 g for 1 h to collect the lysate. Anti-Flag M2 affinity beads (Sigma–Aldrich) were added to the membrane protein lysate, incubated overnight at 4°C, washed three times with 0.1% digitonin-TBS buffer and eluted in 20 ml of 0.1% digitonin-TBS buffer containing 0.2 mg/ml FLAG peptide. Next, gel filtration chromatography (GFC) of the γ-secretase complex was performed on a Superdex 200 10/300 GL column (GE Healthcare). Three-hundred microlitres of 50-fold concentrated M2-flag affinity eluted γ-secretase were loaded and eluted with 0.05% digitonin-TBS at 0.3 ml/min, in 0.5 ml of fractions. The column was calibrated with the soluble standards blue dextran 2000 (void volume), thyroglobulin (669 kDa) and ferritin (440 kDa).

γ-Secretase activity assays

γ-Secretase assays using recombinant APP–C100–Flag were performed as reported previously [4042]. Briefly, purified γ-secretase was 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 μM substrate, 0.1% (w/v) phosphatidylcholine and 0.025% (w/v) phosphatidylethanolamine (PE). The resulting products, AICD–Flag and Aβ, were detected with anti-Flag (M2, Sigma–Aldrich) and anti-Aβ (6E10, Covance) specific antibodies respectively or analysed by MS as described below.

Western blotting and antibodies

Native purified γ-secretase samples were run on NativePAGE™ Novex® Bis-Tris 4%–16% gels for blue-native (BN)/PAGE analysis (Invitrogen), transferred on to PVDF membranes and probed with antibodies NCT164 (NCT-specific, 1:1000, BD Biosciences) or stained with Coomassie to reveal the high-molecular-mass γ-secretase complex (Figures 1 and 8). Samples from γ-secretase activity assays were run on 4%–12% bis-tris gels and transferred on to PVDF membranes to detect Aβ and AICD–Flag with Aβ-specific 6E10 (1:1000; Covance) and Flag-specific M2 antibodies respectively (Figures 1 and 8). Anti-mouse/rabbit/rat IgGs conjugated to Alexa 680 were purchased from Invitrogen and the Odyssey infrared imaging system (LI-COR) was used to detect the fluorescent signal. GFC γ-secretase samples were separated by SDS/PAGE on NuPAGE® Novex® 4%–12% bis-tris gels (Invitrogen) and probed with antibodies targeting PS1–NTF (MAB1563, Chemicon), PS1–CTF (C-terminal fragment; MAB5232, Chemicon), NCT (BD Biosciences), Pen-2 (UD-1, a gift from H. Karlstrom), Flag–Pen-2 (anti-Flag M2, Sigma–Aldrich), Aph-1 (O2C2, Covance) and Aph-1-HA (3F10, Roche) (Figures 68; and Supplementary Figure S4). For the 2D BN/ and SDS/PAGE analysis of γ-secretase in the presence or in the absence of PLA2 treatments (Figure 8B), samples were first run on NativePAGE™ Novex® Bis-Tris 4%–16% gels as described above, followed by SDS/PAGE on 4%–12% bis-tris gels and transferred on to PVDF membranes to detect all subunits.

Large-scale preparation of purified and active γ-secretase

Figure 1
Large-scale preparation of purified and active γ-secretase

(A) The integrity, homogeneity and purity of γ-secretase purified from large-scale suspension cultures were assessed by GFC (left) or Coomassie-stained native PAGE (right). GFC fraction B14 was used for both γ-secretase reconstitution into proteoliposomes and for the mass spectrometric analysis of lipids binding to the complex. F, molecular mass standards ferritin (440 kDa); T, molecular mass standards thyroglobulin (669 kDa); V, void volume. (B and C) Activity assays performed with purified γ-secretase before (E1: M2 anti-Flag affinity purification first eluate) and after GFC (GFC fraction B14). Control reactions were performed in the absence of substrate (E-S) or in the absence of enzyme (S-E). APP cleavage products AICD and Aβ were detected by western blot analysis (B) or by MALDI–TOF MS (C). The cleavage product peaks are labelled according to human APP-C99 numbering. *Flag-tagged Pen-2 detected by the anti-Flag M2 antibody used to detect APP–C100–Flag and AICD–Flag.

Figure 1
Large-scale preparation of purified and active γ-secretase

(A) The integrity, homogeneity and purity of γ-secretase purified from large-scale suspension cultures were assessed by GFC (left) or Coomassie-stained native PAGE (right). GFC fraction B14 was used for both γ-secretase reconstitution into proteoliposomes and for the mass spectrometric analysis of lipids binding to the complex. F, molecular mass standards ferritin (440 kDa); T, molecular mass standards thyroglobulin (669 kDa); V, void volume. (B and C) Activity assays performed with purified γ-secretase before (E1: M2 anti-Flag affinity purification first eluate) and after GFC (GFC fraction B14). Control reactions were performed in the absence of substrate (E-S) or in the absence of enzyme (S-E). APP cleavage products AICD and Aβ were detected by western blot analysis (B) or by MALDI–TOF MS (C). The cleavage product peaks are labelled according to human APP-C99 numbering. *Flag-tagged Pen-2 detected by the anti-Flag M2 antibody used to detect APP–C100–Flag and AICD–Flag.

Immunoprecipitation–MS analysis of Aβ and AICD–Flag

Aβ and AICD–Flag generated in γ-secretase in vitro assays were analysed as described previously [43]. 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 (IP) at 4°C with anti-Flag M2 resin (Invitrogen). Aβ and AICD–Flag were eluted with 1:20:20 (v/v/v) mixture of 1% (v/v) trifluroacetic acid/acetonitrile/H20 mixed 1:1 (v/v) with saturated CHCA (α-cyano 4-hydroxy cinnaminic acid) and analysed by MALDI–TOF MS in reflecting mode on an ABI 4800 MALDI–TOF/TOF mass spectrometer (Applied Biosystems). Molecular masses were accurately measured and searched against a.a. sequences of human APP-C99 with addition of a methionine residue at the N-terminus and a Flag-tag sequence at the C-terminus (C100–Flag).

Reconstitution of γ-secretase into proteoliposomes

Purified γ-secretase at the concentration of 0.3 mg/ml in 50 mM Tris/HCl, pH 7.4, 150 mM NaCl, digitonin 0.1% (w/v) was incubated for 1 h at 4°C in the presence or absence of a phospholipidic extract from E. coli containing 67.0% (w/w) of PE, 23.2% (w/w) of PG and 9.8% (w/w) of cardiolipin (CL), at lipid to protein ratios (LPR) of 0.5 or 0.3. Detergent was then removed over a period of 48 h at room temperature by dialysis, using a cellulose acetate dialysis membrane with a cut-off of 100 kDa (Harvard Apparatus) or over a period of 3–4 h using polystyrene beads (Biobeads from BioRad). Reconstituted proteoliposomes were harvested and examined by EM, as described below.

TEM

For imaging of negatively stained samples, 4 μl aliquots of the reconstituted γ-secretase samples were adsorbed on to carbon film coating copper EM grids, washed with three droplets of pure water and subsequently negatively stained with 2% (w/v) uranyl-acetate. The prepared grids were imaged using a Philips CM10 TEM instrument (FEI Company) operating at 80 kV. The images were recorded by the 2k × 2k side-mounted Veleta CCD camera (Olympus, Germany).

Lipid extraction

In Figures 4 and 5, 100 μl of the selected γ-secretase GFC fraction B14 (equivalent to 10 μg of total protein) were mixed with 100 μl of 150 mM ammonium bicarbonate buffer. In Figure 5, CHO cells were quickly washed with 150 mM ammonium bicarbonate buffer and ∼100000 cells were homogenized with zirconia beads (0.5 mm) for 30 s on a TissueLyserII (QUIAGEN). Ten microlitres of cell lysate (70 μg of equivalent protein) were mixed with 100 μl of 150 mM ammonium bicarbonate buffer. In Figures 6 and 7, 100 and 150 μl of the GFC fractions A12 to C3 were mixed with 100 and 50 μl in 150 mM ammonium bicarbonate buffer, respectively. Prior to lipid extraction, 10 μl of an internal standard mixture composed of 50 pmol of PE 12:0/13:0, PI (phosphatidylinositol) 12:0/13:0 and Chol-d7; 40 pmol of PC 12:0/13:0, PS 12:0/13:0, LPE 13:0, LPC 13:0 and SM d18:1/12:0; 20 pmol of Cer d18:1/12:0, TAGd516:0/18:0/16:0, DAG-d5 17:0/17:0; 10 pmol of PG 12:0/13:0 (all from Avanti Polar Lipids) were spiked into the sample. Lipids were extracted using a modified Folch protocol [44,45]. Briefly, 265 μl of methanol were added to the sample and agitated for 10 min at 4°C. Then, 730 μl of chloroform were added and the sample was agitated in a shaker for 1 h at 4°C. The lower organic phase was collected and evaporated in a vacuum evaporator at 4°C.

Shotgun lipidomics

Lipid extracts were reconstituted in 100 μl of chloroform/methanol mixture (1:2, v/v). For the analysis, 10 μl were transferred into a 96-well plate (Eppendorf) and diluted with 20 μl of 0.05% triethylamine in methanol. The plate was sealed with aluminium foil and kept at 4°C. Lipid species of PS, PE, PC, PG, PI, SM, PE-O, PC-O, LPC and LPE classes were quantified in negative ion mode and of PE, PC, PE-O, PC-O and SM in positive ion mode. To quantify cholesterol, a separate 70 μl aliquot of the lipid extract was acetylated and acetylcholesterol quantified in positive ion mode [46]. Samples were infused into a hybrid quadrupole Orbitrap tandem mass spectrometer Q Exactive (Thermo Fisher Scientific) via a robotic nanoflow electrospray ion source Triversa NanoMate (Advion BioSciences). The NanoMate was controlled by Chipsoft 8.3.1 software; the backpressure was 0.8 psi (1 psi ≈ 6.9 kPa) and the ionization voltage 1.2 kV. The Q Exactive mass spectrometer was operated under the following settings: ion transfer capillary temperature was 200°C; S-lens level was set to 50 arbitrary units. Full MS spectra were acquired under the target mass resolution Rm/z200=140000 (full width at half maximum at m/z 200); the target value for the automated gain control (AGC) was 1×106 and the maximum ion injection time was 50 ms. MS/MS spectra were acquired with Rm/z200=70000 and AGC of 1×105 and the maximum ion injection time of 1000 ms. In t-MS2 experiments the width of the precursor isolation window was set to 1 Th and centred at each half integer m/z (e.g. 400.5; 401.5) by using an inclusion list covering the m/z range of 400.5–1000.5. The normalized collision energy was 30%.

Data processing

Lipids analysed in FT-MS mode were identified by LipidXplorer software [47] by matching the m/z of their monoisotopic peaks to the corresponding elemental composition constraints. Lipid molecular species were identified by t-MS2 considering the m/z of precursor ions and of acyl anions fatty acid moieties. Identifications in both FT-MS and t-MS2 modes were supported by lipid-class specific queries compiled in Molecular Fragmentation Query Language (MFQL), which are available at the LipidXplorer wiki-site: https://wiki.mpicbg.de/wiki/lipdx/index.php/Main_Page. The mass tolerance was 5 ppm and the intensity threshold was adjusted individually according to the noise level reported by the Xcalibur software (Thermo Fisher Scientific).

Lipid modifications

Briefly, purified γ-secretase (100 nM) was subjected to three different lipid modification treatments: cholesterol oxidase (CO; Sigma–Aldrich), phospholipase A2 (PLA2; Sigma–Aldrich) and cyclodextrin (CD; Sigma–Aldrich). The purified γ-secretase complex was solubilized in 0.05% (w/v) digitonin-TBS and was incubated with the lipid modifying compounds at 37°C for 4 h. γ-Secretase was subjected to treatment with CO (130 nM) in a 50 mM KH2PO4 buffer, pH 7.0. Phospholipase A2 treatment (1.3–130 nM) was performed in 20 mM Tris/HCl with 40 mM CaCl2 buffer at pH 7.5 in the presence or in the absence of the inhibitor chloroquine (800 μM). Cholesterol depletion by CD (1 mM) was performed in 25 mM HEPES buffer, pH 7.0. The entity of the γ-secretase complex was assessed after each treatment by BN/PAGE analysis, as described above and by affinity pull-down (Monoclonal Anti-HA affinity resin, Sigma–Aldrich) of 20 μl γ-secretase and SDS/PAGE/Western-blot detection of all subunits.

RESULTS

Purification of active γ-secretase

To improve large-scale and low-cost production of highly purified γ-secretase, we have adapted a CHO cell line stably overexpressing γ-secretase to grow in serum-free suspension culture based on orbital shaker technology [39,48]. More recently, we used the PiggyBac transposon multigene transfer system to generate CHO cell lines co-expressing the four γ-secretase subunits that assemble to form the functional protease complex [39]. To produce active γ-secretase in amounts compatible with 2D crystallization trials and lipidomic analyses, we purified the protease complex from 10 litres of suspension cultures using the above-described multi-step affinity purification procedure [39,40]. Next, we assessed the purity and integrity of the γ-secretase preparation by (i) GFC and (ii) BN/PAGE gel which was probed with an antibody against NCT or stained with Coomassie Blue (Figure 1A). Under native conditions (GFC and BN/PAGE), the purified enzyme behaved as a single high molecular weight complex (HMWC) with an apparent molecular weight of ∼350–400 kDa, consistent with the previous reports for the active γ-secretase complex [35,49] and confirming the purity of the recombinant complex. Next, the enzymatic activity of the purified complex and the specificity of APP processing were assessed by Western blot and mass spectrometric analysis of both the Aβ and AICD cleavage profiles, as described previously [42,43]. As shown in Figures 1(B) and 1(C), the purified γ-secretase used in the present study was active and generated AICD and Aβ profiles similar to those observed under physiological conditions, i.e. close to equimolar AICD 50-99 and AICD 49-99 production and Aβ production with a large excess of Aβ40 when compared with the more amyloidogenic Aβ42 involved in the formation of AD senile plaques [50].

The final yield obtained from 10 litres of suspension culture after the GFC step, was ∼3 mg of highly purified and active protease complex. Since this amount is compatible with modern 2D crystallization trials (the protein concentration needed for crystallization samples varies between 0.3–1.0 mg/ml, depending on protein solubility), the purified enzyme was used for the reconstitution of γ-secretase into proteoliposomes (see below).

The reconstitution of γ-secretase into proteoliposomes does not require exogenous lipids

We employed a 2D crystallization strategy, which is a detergent-mediated reconstitution of membrane proteins into liposomes at low LPR [51]. Briefly, the detergent was carefully removed from purified and digitonin-solubilized γ-secretase in the presence or in the absence of a digitonin-solubilized phospholipidic extract from E. coli (see Experimental). Digitonin was efficiently removed within 2 days when this was achieved by dialysis using membranes with a cut-off value of 100 kDa and took 3–4 h when BioBeads were employed. Dialysis membranes with a 9–14 kDa cut-off have been successfully used to crystallize many membrane proteins, but were not employed here as removal of the low critical micellar concentration (cmc) detergent, digitonin, would have been unacceptably slow and detrimental to protein stability.

Negative stain TEM was used to visualize the resulting structures, revealing reconstituted proteoliposomes. Also, the characteristic extra-membrane part of the γ-secretase complex was evident in some views. Figure 2 shows examples of γ-secretase reconstitution into an exogenous polar lipid extract from E. coli containing 67.0% (w/w) PE, 23.2% (w/w) PG and 9.8% (w/w) CL at an LPR of 0.5 using polystyrene Bio-Beads (A, B) and at an LPR of 0.3 by dialysis (C, D). In both cases, the γ-secretase particles are homogeneously distributed in the membranes and their stain-excluding top-views are the size expected if all four subunits, NCT, PS1, Aph-1 and Pen-2, are present. Very interestingly, the γ-secretase complex also reconstituted into membranes as densely packed proteins, without the addition of an exogenous phospholipidic extract from E. coli (Figure 3). The results obtained using Bio-Beads (Figures 3A and 3B) and by dialysis (Figures 3C and 3D) to remove the detergent were similar and again top-views of the complexes have the expected dimensions. Lipid bilayers without reconstituted protein were not detected for these samples.

γ-Secretase reconstituted into proteoliposomes in the presence of exogenous lipids

Figure 2
γ-Secretase reconstituted into proteoliposomes in the presence of exogenous lipids

(A) Overview negative stain TEM image of γ-secretase particles reconstituted in the presence of E. coli polar lipid at a LPR of 0.5 using biobeads. (B) Enlarged view of a specific region of (A). (C) Overview negative stain TEM image of γ-secretase particles reconstituted in the presence of E. coli polar lipid at a LPR of 0.3 by dialysis. (D) Enlarged view of a specific region of (C). Black arrows indicate reconstituted protein particles. White arrows indicate protein-free lipid membranes. Arrowheads indicate the extra-membrane domains of the complex.

Figure 2
γ-Secretase reconstituted into proteoliposomes in the presence of exogenous lipids

(A) Overview negative stain TEM image of γ-secretase particles reconstituted in the presence of E. coli polar lipid at a LPR of 0.5 using biobeads. (B) Enlarged view of a specific region of (A). (C) Overview negative stain TEM image of γ-secretase particles reconstituted in the presence of E. coli polar lipid at a LPR of 0.3 by dialysis. (D) Enlarged view of a specific region of (C). Black arrows indicate reconstituted protein particles. White arrows indicate protein-free lipid membranes. Arrowheads indicate the extra-membrane domains of the complex.

γ-Secretase reconstituted into proteoliposomes in the absence of exogenous lipids

Figure 3
γ-Secretase reconstituted into proteoliposomes in the absence of exogenous lipids

(A) Overview negative stain TEM image of γ-secretase particles reconstituted in the absence of exogenous lipids using biobeads. (B) Enlarged view of a specific region of (A). (C) Overview negative stain TEM image of γ-secretase particles reconstituted in the absence of exogenous lipids by dialysis. (D) Enlarged view of a specific region of (C). Black arrows indicate the protein particles. White arrows indicate protein-free lipid membranes. Arrowheads indicate the extra-membrane domains of the complex.

Figure 3
γ-Secretase reconstituted into proteoliposomes in the absence of exogenous lipids

(A) Overview negative stain TEM image of γ-secretase particles reconstituted in the absence of exogenous lipids using biobeads. (B) Enlarged view of a specific region of (A). (C) Overview negative stain TEM image of γ-secretase particles reconstituted in the absence of exogenous lipids by dialysis. (D) Enlarged view of a specific region of (C). Black arrows indicate the protein particles. White arrows indicate protein-free lipid membranes. Arrowheads indicate the extra-membrane domains of the complex.

Together, the above-described results suggest that the amount of endogenous lipids co-purifying with the γ-secretase complex is sufficient to generate liposomes in which the purified protease particles are inserted.

MS analysis of lipids co-purifying with the γ-secretase complex

To characterize lipids co-purifying with the γ-secretase complex, we analysed its highly enriched GFC fraction B14 by quantitative shotgun lipidomics [52]. We quantified a total of 53 species from 11 major lipid classes (Figure 4; Supplementary Figure S1). In total, the fraction contained 106 pmol of lipids bound to 1 μg of purified γ-secretase, the equivalent of 2.5-2.8 pmol of the protein complex assuming its MW of 350 to 400 kDa. The stoichiometric ratio between the γ-secretase complex and specific lipid species was approximately 1:3 for cholesterol and 1:20 for phosphatidylcholines (moles proteins/moles lipids).

Molecular lipid species bound to the γ-secretase complex

Figure 4
Molecular lipid species bound to the γ-secretase complex

Lipids attached to purified γ-secretase (10 μg of proteins from the GFC fraction B14; Figure 1A) were extracted and analysed by shotgun lipidomics. The abundance of molecular species is expressed in picomoles per microgram of the protein content. Error bars indicate the S.D. calculated on experimental duplicates.

Figure 4
Molecular lipid species bound to the γ-secretase complex

Lipids attached to purified γ-secretase (10 μg of proteins from the GFC fraction B14; Figure 1A) were extracted and analysed by shotgun lipidomics. The abundance of molecular species is expressed in picomoles per microgram of the protein content. Error bars indicate the S.D. calculated on experimental duplicates.

To identify the molecular lipid species, their precursor ions were subjected to MS/MS in negative ion mode. Upon collision-induced fragmentation, molecular anions of glycerophospholipids produced abundant acyl anions of their fatty acid moieties. If both independent determinations were correct, the relative abundances of individual species quantified by top-down (FT MS) and bottom-up (FT MS/MS) analyses were expected to corroborate, which we also observed as shown in Supplementary Figure S2.

At the lipid species level, the majority of PC comprised saturated and mono-unsaturated FA moieties, i.e. PC 18:1/18:1, PC 16:1/18:1 and PC 16:0/18:1 (Figure 4; Table 1). Lipids comprising the moieties of palmitoleic acid (16:1), palmitic acid (16:0) and oleic acid (18:1) were most abundant. The same trend was observed for PE, PI and PS classes (Figure 4; Table 1). We only detected two SM species, i.e. SM d18:1/16:0 and SM d18:1/24:1 that represented approximately 5 mol% of the total lipid content. We did not detect phosphatidic acid (PA), DAG, ceramides or hexosylceramides.

Table 1
Description and percentage of the predominant lipids bound to the γ-secretase complex.
Lipid class Major lipid
species 
FA name Concentration
(pmol/ μg proteins) 
Abundance1 (%) 
Phosphatidylcholine 18:1/18:1 Oleic acid 15.31 13 
Phosphatidylcholine 16:1/18:1 Palmitoleic acid/ Oleic acid 10.18 
Phosphatidylcholine 16:0/18:1 Palmitic acid/ Oleic acid 10.07 
Sterol cholesterol 7.95 
Phosphatidylethanolamine 18:1/18:1 Oleic acid 7.76 
Phosphatidylinositol 18:1/18:1 Oleic acid 6.77 
Ether lipid
Phosphatidylcholine 
O-18:1/16:0 O- alkyl C18:1/ Palmitic acid 6.43 
Phosphatidylserine 18:0/18:1 Stearic acid/ Oleic acid 4.38 
Phosphatidylcholine 14:0/18:1 Myristic acid / Oleic acid 4.30 
Phosphatidylcholine 16:0/16:1 Palmitic acid/ Palmitoleic acid 4.27 
Lipid class Major lipid
species 
FA name Concentration
(pmol/ μg proteins) 
Abundance1 (%) 
Phosphatidylcholine 18:1/18:1 Oleic acid 15.31 13 
Phosphatidylcholine 16:1/18:1 Palmitoleic acid/ Oleic acid 10.18 
Phosphatidylcholine 16:0/18:1 Palmitic acid/ Oleic acid 10.07 
Sterol cholesterol 7.95 
Phosphatidylethanolamine 18:1/18:1 Oleic acid 7.76 
Phosphatidylinositol 18:1/18:1 Oleic acid 6.77 
Ether lipid
Phosphatidylcholine 
O-18:1/16:0 O- alkyl C18:1/ Palmitic acid 6.43 
Phosphatidylserine 18:0/18:1 Stearic acid/ Oleic acid 4.38 
Phosphatidylcholine 14:0/18:1 Myristic acid / Oleic acid 4.30 
Phosphatidylcholine 16:0/16:1 Palmitic acid/ Palmitoleic acid 4.27 

#, normalization to the total concentration of the 10 major lipid species identified

Next we compared the composition of γ-secretase-associated lipids with the total lipid composition of CHO cells (Figure 5). The γ-secretase lipidome was strongly (by approximately 6-fold) depleted with cholesterol and enriched with PC and PE species comprising monounsaturated fatty acid moieties (Figure 5).

Comparative profiles of lipids associated with either purified γ-secretase or CHO whole cell extract

Figure 5
Comparative profiles of lipids associated with either purified γ-secretase or CHO whole cell extract

Lipids from CHO cell lysates (70 μg of proteins equivalent) or attached to purified γ-secretase (GFC fraction B14) were extracted and analysed by shotgun lipidomics as described in the ‘Experimental’ section. The abundance of molecular species is expressed in mol%. Error bars indicate the S.D. calculated on experimental duplicates. For clarity, only 19 species whose relative abundances exceed a threshold of 0.5 mol% are shown.

Figure 5
Comparative profiles of lipids associated with either purified γ-secretase or CHO whole cell extract

Lipids from CHO cell lysates (70 μg of proteins equivalent) or attached to purified γ-secretase (GFC fraction B14) were extracted and analysed by shotgun lipidomics as described in the ‘Experimental’ section. The abundance of molecular species is expressed in mol%. Error bars indicate the S.D. calculated on experimental duplicates. For clarity, only 19 species whose relative abundances exceed a threshold of 0.5 mol% are shown.

To test if the identified lipids specifically bound the γ-secretase complex, we quantified lipids in each GFC fraction, including control fractions devoid of γ-secretase. As shown in Figures 6(A) and 6(C), concentration profiles of cholesterol and γ-secretase correlated in GFC fractions A14 to B9. The total lipid concentration follows exactly the same trend, showing a bell-shaped curve between fractions A14 and B9 that peaked in B14 and B13 fractions (Figure 6B). In contrast with cholesterol (compare Figures 6B and 6C; Supplementary Figure S3), the same lipids were also detected in fractions depleted of γ-secretase (B8 to B1). This indicates that during the purification procedure, only cholesterol remained stably bound to the protein complex and was only carried with the intact complex, whereas other lipids might dissociate from the cholesterol–protein assembly. Remarkably, lipids (except cholesterol) of the same molecular composition maintained their association and were eluted together in γ-secretase-free fractions. It may suggest that, once assembled, a lipid shell present in a large molar excess over proteins, may maintain its integrity in the aqueous environment without the protein core. In contrast, the core constellation of proteins was only found in association with lipids.

Specificity of total lipids and cholesterol bound to the γ-secretase complex: first characterization

Figure 6
Specificity of total lipids and cholesterol bound to the γ-secretase complex: first characterization

(A) Western-blot analysis of all γ-secretase subunits in fractions collected during the first GFC purification experiment. Twenty microliters of each GFC fraction was loaded on to a NuPAGE® Novex® 4-12% bis-tris gel. Proteins separated by SDS/PAGE were transferred on to a nitrocellulose membrane and probed with antibodies targeting all γ-secretase subunits (see ‘Experimental’). (B and C) Total lipid and cholesterol contents of the GFC eluted fractions. The lipids contained in the GFC fractions (A12–C3) were extracted and analysed by shotgun lipidomics. Total lipid content (B) and cholesterol (C) content are expressed in picomoles. Error bars indicate the S.D. calculated on experimental duplicates.

Figure 6
Specificity of total lipids and cholesterol bound to the γ-secretase complex: first characterization

(A) Western-blot analysis of all γ-secretase subunits in fractions collected during the first GFC purification experiment. Twenty microliters of each GFC fraction was loaded on to a NuPAGE® Novex® 4-12% bis-tris gel. Proteins separated by SDS/PAGE were transferred on to a nitrocellulose membrane and probed with antibodies targeting all γ-secretase subunits (see ‘Experimental’). (B and C) Total lipid and cholesterol contents of the GFC eluted fractions. The lipids contained in the GFC fractions (A12–C3) were extracted and analysed by shotgun lipidomics. Total lipid content (B) and cholesterol (C) content are expressed in picomoles. Error bars indicate the S.D. calculated on experimental duplicates.

We further tested the lipid association specificity in two ways. First, we collected active GFC fractions enriched in γ-secretase (B15 to B11) from the first round of chromatography (Figure 6A) and subjected them to a second round of chromatography (Figure 7). Prior to chromatography, we reduced the total volume by 5-fold by spinning it through 100 K cut-off membrane for 20 min. Note that to prepare the sample for the first round of chromatography, we reduced its volume by 70-fold in 6 h. As shown in Figure 7, cholesterol, other lipids and secretase proteins fully co-purified, did not dissociate and therefore no lipids were detectable in the late-eluting fractions.

Specificity of total lipids and cholesterol bound to the γ-secretase complex: second characterization

Figure 7
Specificity of total lipids and cholesterol bound to the γ-secretase complex: second characterization

(A) Western-blot analysis of the γ-secretase subunits in fractions collected during the second GFC purification. Fractions from the first GFC run that were enriched in γ-secretase (B15 to B11; see Figure 6A) were concentrated and subjected to a second GFC purification step. γ-Secretase subunits were detected by Western blot analysis as described in Figure 6. (B and C) Total lipid and cholesterol contents of the fractions eluted during the second GFC. The lipids contained in GFC fractions A12-C3 were extracted and analysed by shotgun lipidomics. Total lipid content (B) and cholesterol (C) content are expressed in pmol. Error bars indicate the standard deviation calculated on experimental duplicates.

Figure 7
Specificity of total lipids and cholesterol bound to the γ-secretase complex: second characterization

(A) Western-blot analysis of the γ-secretase subunits in fractions collected during the second GFC purification. Fractions from the first GFC run that were enriched in γ-secretase (B15 to B11; see Figure 6A) were concentrated and subjected to a second GFC purification step. γ-Secretase subunits were detected by Western blot analysis as described in Figure 6. (B and C) Total lipid and cholesterol contents of the fractions eluted during the second GFC. The lipids contained in GFC fractions A12-C3 were extracted and analysed by shotgun lipidomics. Total lipid content (B) and cholesterol (C) content are expressed in pmol. Error bars indicate the standard deviation calculated on experimental duplicates.

Second, we tested if lipids were bound in a non-specific manner to the M2 anti-Flag affinity resin. To do so, we repeated the purification procedure as described in Figure 6, except that untransfected CHO cells were used, instead of the CHO cells overexpressing the Flag-tagged protease complex. As expected, no endogenous γ-secretase was detected in the GFC fractions (Supplementary Figure S4) and no lipids were found in the GFC fraction B14 (Supplementary Figure S5).

Altogether, our analyses confirmed the specific co-purification of the identified lipid species, including cholesterol, with the γ-secretase complex.

Altering the composition of lipids bound to γ-secretase affects its integrity and activity

We demonstrated above that a constellation of lipids specifically bound γ-secretase. In order to assess whether these lipids are essential for the structural and catalytic properties of γ-secretase, we treated them either with CO, CD or PLA2. CO specifically oxidizes the hydroxyl group of cholesterol into a keto group, whereas CD is a cholesterol binder, which may also bind other lipids with lower affinity [53]. PLA2 specifically cleaves fatty acid moieties at the sn-2 position off the glycerol backbone.

Upon treatment, we first analysed the structural properties of the HMWC γ-secretase by native PAGE. As shown in Figure 8(A), CO did not affect the entity of the complex, whereas CD led to its aggregation. In contrast, a low molecular mass γ-secretase complex (LMWC) migrating at the apparent Mr of 320 kDa was found after PLA2 treatment. Next, the γ-secretase activity was assessed in a cell-free activity assay with the recombinant substrate APP–C100–Flag. As shown in Figure 8(A), the analysis of the AICD–Flag cleavage product revealed full functionality for the CO and CD treated γ-secretase. In contrast, PLA2 treatment abolished the γ-secretase activity completely.

The pharmacological modulation of γ-secretase lipidic micro-environment affects both the structure and the function of the complex

Figure 8
The pharmacological modulation of γ-secretase lipidic micro-environment affects both the structure and the function of the complex

(A) Effects on the γ-secretase quaternary structure and activity of CO, CD and PLA2-based modifications of the lipid microenvironment of the protease complex. Upper panel: the integrity of the complex was analysed by Native-PAGE, followed by Western blot analysis to detect PS1 (antibody Mab5232 for CO and CD treatments) or NCT (antibody NCT164 for PLA2-treated samples). The enzymatic activity of the complexes with modified lipids was further assessed by using the cell-free γ-secretase activity assay with recombinant APP–C100–Flag. Lower panel: the cleavage product AICD–Flag was revealed by Western-blot analysis using the anti-Flag M2 antibody. (B) The integrity of PLA2-treated γ-secretase was determined by native PAGE (left) and by 2D gel electrophoresis combining native PAGE with SDS/PAGE (right). All subunits were detected with antibodies described under ‘Experimental’. (C) Dose-dependent dissociation of the HMWC by PLA2, as analysed by native PAGE and Western blot to detect NCT (antibody NCT164).*Partial dissociation of γ-secretase, at a PLA2 concentration of 13 nM. (D) The PLA2 inhibitory compound chloroquine prevents the PLA2-dependent dissociation of γ-secretase. Top panel: BN/PAGE analysis of γ-secretase pre-treated with 800 μM chloroquine and further treated with PLA2. Bottom panel: anti haemagglutinin (HA)-tag affinity purification of Aph-1–HA associated subunits, after chloroquine and PLA2 treatments. S, starting material; U, unbound fraction.

Figure 8
The pharmacological modulation of γ-secretase lipidic micro-environment affects both the structure and the function of the complex

(A) Effects on the γ-secretase quaternary structure and activity of CO, CD and PLA2-based modifications of the lipid microenvironment of the protease complex. Upper panel: the integrity of the complex was analysed by Native-PAGE, followed by Western blot analysis to detect PS1 (antibody Mab5232 for CO and CD treatments) or NCT (antibody NCT164 for PLA2-treated samples). The enzymatic activity of the complexes with modified lipids was further assessed by using the cell-free γ-secretase activity assay with recombinant APP–C100–Flag. Lower panel: the cleavage product AICD–Flag was revealed by Western-blot analysis using the anti-Flag M2 antibody. (B) The integrity of PLA2-treated γ-secretase was determined by native PAGE (left) and by 2D gel electrophoresis combining native PAGE with SDS/PAGE (right). All subunits were detected with antibodies described under ‘Experimental’. (C) Dose-dependent dissociation of the HMWC by PLA2, as analysed by native PAGE and Western blot to detect NCT (antibody NCT164).*Partial dissociation of γ-secretase, at a PLA2 concentration of 13 nM. (D) The PLA2 inhibitory compound chloroquine prevents the PLA2-dependent dissociation of γ-secretase. Top panel: BN/PAGE analysis of γ-secretase pre-treated with 800 μM chloroquine and further treated with PLA2. Bottom panel: anti haemagglutinin (HA)-tag affinity purification of Aph-1–HA associated subunits, after chloroquine and PLA2 treatments. S, starting material; U, unbound fraction.

Further analysis by Native-PAGE and by 2D (Native+SDS)-PAGE of PLA2-treated γ-secretase revealed a clear loss of the Aph1, PSEN–CTF (PS–CTF and PSEN–NTF (PS–NTF) subunits for the LMWC (Figure 8B). Consistently, a PLA2 dose-dependent generation of LMWC was associated with concomitant losses of γ-secretase activity (Figure 8C). Further supporting the PLA2-specific effect, pre-treatment of the γ-secretase complex with the PLA2 inhibitor chloroquine prevented its dissociation and consequently the formation of the LMWC (Figure 8D).

Hence stably associated cholesterol was not essential for γ-secretase activity, whereas dissociating phospholipid core was essential for both activity and integrity.

DISCUSSION

As no high-resolution crystal structure of γ-secretase is available, we have recently initiated 2D crystallization trials of this membrane protease complex. Although well-ordered 2D crystals of γ-secretase were not obtained, reconstitution of the purified protein was achieved in both the presence and the absence of exogenous lipids from E.coli (Figures 2 and 3). The latter unexpected observation strongly suggests that a considerable amount of endogenous lipids are firmly bound to the purified γ-secretase and indicates that lipids might be essential for its stability and activity.

Using quantitative shotgun lipidodmics [4556], we obtained the first direct evidence of the existence of a lipid microenvironment stably associated with γ-secretase. We observed that PC, PE, PS, PI and cholesterol were the major lipid classes associated with the complex (Figure 4; Supplementary Figure S1; Table 1). We also identified minor lipids such as SM, ether lipids, lysolipids and PG. These findings are consistent with previous studies demonstrating a relationship between the lipid environment surrounding γ-secretase and its enzymatic activity [3537]. The composition of the γ-secretase associated lipids differed from the composition of the whole CHO cell lipidome in several ways, most notably by having a much lower content of cholesterol (Figure 5), which was stoichiometrically associated with the protease complex in an approximate 1:3 molar ratio. This also suggested that the lipid shell was not composed from randomly recruited cellular lipids, but instead, a specific segregation of lipids occurred during the assembly and maturation of the active complex. Fraering et al. [35] showed that PC and SM help to regulate the active site conformation when the γ-secretase complex is reconstituted in proteoliposomes, markedly improving the efficiency with which it processes APP substrates. They also found that SM, in the presence of 0.1% PC+0.025% PE and cholesterol increased the overall γ-secretase activity. Another study showed that the addition of PC, PE and SM, PA and PS together with sphingolipids also regulated γ-secretase activity [36]. Further, the authors observed that the class of anionic phospholipids PI was the only phospholipid that decreased γ-secretase activity over the PC-only control at tested concentrations. Interestingly, in our study, PI lipids represent one of the major phospholipid classes interacting with the complex.

Taken together these studies provide clear evidence that the composition of the lipid shell has an impact on the activity of γ-secretase. In our study PC, PE species and cholesterol are present in the associated microenvironment in 39, 13 and 7 mol% quantity (Supplementary Figure S1). Species comprising monounsaturated oleoyl moieties (C18:1) were both in sn-1 and in sn-2 positions of PC, PE and PI (PC 18:1/18:1, PE 18:1/18:1, PI 18:1/18:1). At the same time PS comprised both saturated and monounsaturated C18 fatty acid moieties (PS 18:0/18:1) fatty. We also observed that on average lipids associated with the complex are more of medium (C16–C18) rather than short (C14) or long (C20–C24) chain length. This is consistent with the observation that monounsaturated PC with a chain length of C16–C20 enhance γ-secretase activity in a cell-free assay [38].

Treating γ-secretase with CO and CD did not dissociate the complex, but the latter stimulated its aggregation (Figure 8). Strikingly, PLA2 treatment dissociated γ-secretase and abrogated its activity (Figure 8). More specifically, PLA2 dissociated the complex into a LMWC made of NCT and Pen2, followed by the loss of Aph1, PS–CTF and PS–NTF (Figure 8 and model depicted in Figure 9). Recently, high-resolution Cryo-EM showed that two PCs are involved in the arrangement of the TMDs of γ-secretase [19]. Specifically, one PC molecule mediates the interaction between Aph-1 and NCT, whereas the second PC molecule mediates the interaction between Aph-1, PS–CTF and PS–NTF (Figure 9). We show that the loss of these interactions upon PLA2 activity causes the dissociation from the complex of the subunits Aph-1, PS–CTF and PS–NTF (Figures 8 and 9). As the NCT–Pen-2 interaction discovered recently implicates only hydrophilic domains [19], it is not affected by changes in the lipid environment, consistent with its stability towards PLA2 treatment (Figures 8 and 9). Interestingly, the γ-secretase catalytic site does not adopt an active conformation in the same cryo-EM 3.4 Å 3D structure [19]. Our findings suggest that the lipid microenvironment is important for the structural determination of γ-secretase in an active conformation. They further suggest that the structure of γ-secretase in its active conformation should be determined in the presence of its lipid environment, rather than using an artificial system with a surfactant that maintains the solubilized membrane complex in a detergent- and lipid-free solution.

Model for the rearrangement and dissociation of the γ-secretase complex, mediated by the pharmacological modulation of the lipid micro-environment

Figure 9
Model for the rearrangement and dissociation of the γ-secretase complex, mediated by the pharmacological modulation of the lipid micro-environment

In this model, the two previously reported phospholipid molecules interacting with the TMDs of γ-secretase [19] mediate hydrophobic interactions between the subunits Aph-1 and NCT and between Aph-1, PS–CTF and PS–NTF (top left). These hydrophobic interdomain interactions are displayed with blue squares (top left). Phospholipase A2 cleaves fatty acid moieties from the glycerophospholipids, which leads to a local redistribution of free fatty acids (red circle; top right) and lysolipids (green circle; top right). In addition, the hydrophobic interactions between the TMDs are distorted by the local degradation of lipids, inducing the destabilization of the native γ-secretase structure (bottom right). This ultimately leads to the dissociation of the complex into single Aph-1, PS–CTF and PS–NTF subunits and a NCT–Pen-2 sub-complex stabilized through a previously reported hydrophilic extracellular interaction (bottom left) [19].

Figure 9
Model for the rearrangement and dissociation of the γ-secretase complex, mediated by the pharmacological modulation of the lipid micro-environment

In this model, the two previously reported phospholipid molecules interacting with the TMDs of γ-secretase [19] mediate hydrophobic interactions between the subunits Aph-1 and NCT and between Aph-1, PS–CTF and PS–NTF (top left). These hydrophobic interdomain interactions are displayed with blue squares (top left). Phospholipase A2 cleaves fatty acid moieties from the glycerophospholipids, which leads to a local redistribution of free fatty acids (red circle; top right) and lysolipids (green circle; top right). In addition, the hydrophobic interactions between the TMDs are distorted by the local degradation of lipids, inducing the destabilization of the native γ-secretase structure (bottom right). This ultimately leads to the dissociation of the complex into single Aph-1, PS–CTF and PS–NTF subunits and a NCT–Pen-2 sub-complex stabilized through a previously reported hydrophilic extracellular interaction (bottom left) [19].

Overall, certain features of the surrounding lipid environment appear to be important for the activity of γ-secretase, such as total lipid class composition ([3537]; present study), fatty acyl chain length ([37]; present study), unsaturation of individual molecular species ([37]; present study) and the preferred type of isomerization [37]. Changes in the lipid composition of membranes in the aged brain contribute to cognitive decline by affecting synaptic functions and neuronal survival [57]. It is entirely possible that the same changes in the lipid composition of membranes and, in particular, within the microenvironment directly surrounding γ-secretase, affect the overall activity of this protease during aging or alter the specificity of γ-cleavage leading to pathological production of the longer and more toxic Aβ species including Aβ42 and Aβ43 and finally cause AD. In the latter case, and as suggested by many different studies in the past, dietary intervention could help to prevent or slow down the pathogenesis of this devastating disease.

AUTHOR CONTRIBUTION

Sophie Ayciriex performed the shotgun lipidomics mass spectrometric analyses of the lipids associated with γ-secretase. Hermeto Gerber purified γ-secretase and the APP–C100–Flag substrate, performed BN/PAGE and Western blot analyses and carried out γ-secretase activity assays and IP–MS analysis of the cleavage products. Hermeto Gerber and Guillermo Osuna performed all lipid modifications and analysed their effects on both integrity and activity of γ-secretase by BN/PAGE, 2D/PAGE and IP with Western blot analyses. Mohamed Chami reconstituted the γ-secretase into proteoliposomes and carried out the EM. Patrick Fraering initiated and supervised the project. Henning Stahlberg, Andrej Shevchenko and Patrick Fraering designed the research. Sophie Ayciriex, Hermeto Gerber, Andrej Shevchenko and Patrick Fraering wrote the manuscript. All authors edited the manuscript.

The authors thank Shirley Müller (C-CINA, Biozentrum, University of Basel) for critical reading and editing of the manuscript. The authors further thank J.R. Alattia, J. Pascual and Virginie Braman for assistance with γ-secretase purification and M. Dimitrov for assistance with IP–MS analyses of AICD and Aβ peptides. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

FUNDING

This work was supported by the Swiss National Science Foundation (grant numbers 31003A_152677/1 (to H.G. and P.C.F.) and 15230 146929 (to H.S. and M.C.) and NCCR TransCure]; and the Deutsche Forschungsgemeinschaft [grant number TRR83 project A17 (to A.S.)].

Abbreviations

     
  • a.a.

    amino acid

  •  
  • AD

    Alzheimer's disease

  •  
  • AGC

    automated gain control

  •  
  • AICD

    APP intracellular domain

  •  
  • Aph-1

    anterior pharynx-defective 1

  •  
  • APP

    amyloid precursor protein

  •  
  • amyloid-β peptides

  •  
  • BN

    blue-native

  •  
  • CD

    cyclodextrin

  •  
  • Cer

    ceramide

  •  
  • CHAPSO

    3-([3-cholamidopropyl]dimethylammonio)-2-hydroxy-1-propanesulfonate

  •  
  • CHO

    Chinese hamster ovary

  •  
  • CL

    cardiolipin

  •  
  • CO

    cholesterol oxidase

  •  
  • CTF

    C-terminal fragment

  •  
  • DAG

    diacylglycerol

  •  
  • EM

    electron microscopy

  •  
  • FT

    Fourier transform

  •  
  • GFC

    gel filtration chromatography

  •  
  • HMWC

    high molecular weight complex

  •  
  • IP

    immunoprecipitation

  •  
  • LMWC

    low molecular weight complex

  •  
  • LPC

    lysophosphatidylcholine

  •  
  • LPE

    lysophosphatidylethanolamine

  •  
  • LPR

    lipid to protein ratio

  •  
  • NCT

    nicastrin

  •  
  • PA

    phosphatidic acid

  •  
  • PC

    phosphatidylcholine

  •  
  • PC-O

    ether PC

  •  
  • PE

    phosphatidylethanolamine

  •  
  • Pen-2

    presenilin enhancer 2

  •  
  • PE-O

    ether PE

  •  
  • PG

    phosphatidylglycerol

  •  
  • PI

    phosphatidylinositol

  •  
  • PLA2

    phospholipase A2

  •  
  • PS

    phosphatidylserine

  •  
  • PS1–CTF

    presenilin1–C-terminal fragment

  •  
  • PS1–NTF

    presenilin1–N-terminal fragment

  •  
  • PSEN

    presenilin

  •  
  • S2P

    site 2 protease

  •  
  • SM

    sphingomyelin

  •  
  • SPP

    signal peptide peptidase

  •  
  • TAG

    triacylglycerol

  •  
  • TMD

    transmembrane domain

References

References
1
Goedert
M.
Spillantini
M.G.
A century of Alzheimer's disease
Science
2006
, vol. 
314
 (pg. 
777
-
781
)
[PubMed]
2
Selkoe
D.J.
Alzheimer's disease: genes, proteins and therapies
Physiol. Rev.
2001
, vol. 
81
 (pg. 
742
-
761
)
[PubMed]
3
Hardy
J.
Selkoe
D.J.
The amyloid hypothesis of Alzheimer's disease: progress and problems on the road to therapeutics
Science
2002
, vol. 
297
 (pg. 
353
-
356
)
[PubMed]
4
Karran
E.
Mercken
M.
De Strooper
B.
The amyloid cascade hypothesis for Alzheimer's disease: an appraisal for the development of therapeutics
Nat. Rev. Drug Discov.
2011
, vol. 
10
 (pg. 
698
-
712
)
[PubMed]
5
Benilova
I.
Karran
E.
De Strooper
B.
The toxic Abeta oligomer and Alzheimer's disease: an emperor in need of clothes
Nat. Neurosci.
2012
, vol. 
15
 (pg. 
349
-
357
)
[PubMed]
6
Hardy
J.
Framing b-amyloid
Nat. Genet.
1992
, vol. 
1
 (pg. 
233
-
234
)
[PubMed]
7
Haass
C.
Hung
A.Y.
Schlossmacher
M.G.
Oltersdorf
T.
Teplow
D.B.
Selkoe
D.J.
Normal cellular processing of the b-amyloid precursor protein results in the secretion of the amyloid b-peptide and related molecules
Ann. N.Y. Acad. Sci.
1993
(pg. 
109
-
116
)
8
Golde
T.E.
Koo
E.H.
Felsenstein
K.M.
Osborne
B.A.
Miele
L.
γ-Secretase inhibitors and modulators
Biochim. Biophys. Acta
2013
, vol. 
1828
 (pg. 
2898
-
2907
)
[PubMed]
9
Wolfe
M.S.
Selkoe
D.J.
Biochemistry. Intramembrane proteases–mixing oil and water
Science
2002
, vol. 
296
 (pg. 
2156
-
2157
)
[PubMed]
10
Tolia
A.
De Strooper
B.
Structure and function of gamma-secretase
Semin. Cell Dev. Biol.
2009
, vol. 
20
 (pg. 
211
-
218
)
[PubMed]
11
De Strooper
B.
Aph-1, Pen-2, and Nicastrin with Presenilin generate an active γ-secretase complex
Neuron
2003
, vol. 
38
 (pg. 
9
-
12
)
[PubMed]
12
Fraering
P.C.
Structural and functional determinants of γ-secretase, an intramembrane protease implicated in Alzheimer's disease
Curr. Genomics
2007
, vol. 
8
 (pg. 
531
-
549
)
[PubMed]
13
Wolfe
M.S.
Xia
W.
Ostaszewski
B.L.
Diehl
T.S.
Kimberly
W.T.
Selkoe
D.J.
Two transmembrane aspartates in presenilin-1 required for presenilin endoproteolysis and g-secretase activity
Nature
1999
, vol. 
398
 (pg. 
513
-
517
)
[PubMed]
14
Sato
T.
Dohmae
N.
Qi
Y.
Kakuda
N.
Misonou
H.
Mitsumori
R.
Maruyama
H.
Koo
E.H.
Haass
C.
Takio
K.
, et al. 
Potential link between amyloid beta-protein 42 and C-terminal fragment gamma 49-99 of beta-amyloid precursor protein
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
24294
-
24301
)
[PubMed]
15
Takami
M.
Nagashima
Y.
Sano
Y.
Ishihara
S.
Morishima-Kawashima
M.
Funamoto
S.
Ihara
Y.
γ-Secretase: successive tripeptide and tetrapeptide release from the transmembrane domain of beta-carboxyl terminal fragment
J. Neurosci.
2009
, vol. 
29
 (pg. 
13042
-
13052
)
[PubMed]
16
Lazarov
V.K.
Fraering
P.C.
Ye
W.
Wolfe
M.S.
Selkoe
D.J.
Li
H.
Electron microscopic structure of purified, active gamma-secretase reveals an aqueous intramembrane chamber and two pores
Proc. Natl. Acad. Sci. U.S.A.
2006
, vol. 
103
 (pg. 
6889
-
6894
)
[PubMed]
17
Osenkowski
P.
Li
H.
Ye
W.
Li
D.
Aeschbach
L.
Fraering
P.C.
Wolfe
M.S.
Selkoe
D.J.
Li
H.
Cryoelectron microscopy structure of purified gamma-secretase at 12 A resolution
J. Mol. Biol.
2009
, vol. 
385
 (pg. 
642
-
652
)
[PubMed]
18
Lu
P.
Bai
X.C.
Ma
D.
Xie
T.
Yan
C.
Sun
L.
Yang
G.
Zhao
Y.
Zhou
R.
Scheres
S.H.W.
Shi
Y.
Three-dimensional structure of human gamma-secretase
Nature
2014
, vol. 
512
 (pg. 
166
-
170
)
[PubMed]
19
Bai
X.C.
Yan
C.
Yang
G.
Lu
P.
Ma
D.
Sun
L.
Zhou
R.
Scheres
S.H.
Shi
Y.
An atomic structure of human gamma-secretase
Nature
2015
, vol. 
525
 (pg. 
212
-
217
)
[PubMed]
20
Sharon
M.
Robinson
C.V.
The role of mass spectrometry in structure elucidation of dynamic protein complexes
Annu. Rev. Biochem.
2007
, vol. 
76
 (pg. 
167
-
193
)
[PubMed]
21
Brown
S.H.
Mitchell
T.W.
Oakley
A.J.
Pham
H.T.
Blanksby
S.J.
Time to face the fats: what can mass spectrometry reveal about the structure of lipids and their interactions with proteins?
J. Am. Soc. Mass. Spectrom.
2012
, vol. 
23
 (pg. 
1441
-
1449
)
[PubMed]
22
Barrera
N.P.
Di Bartolo
N.
Booth
P.J.
Robinson
C.V.
Micelles protect membrane complexes from solution to vacuum
Science
2008
, vol. 
321
 (pg. 
243
-
246
)
[PubMed]
23
Wang
S.C.
Politis
A.
Di Bartolo
N.
Bavro
V.N.
Tucker
S.J.
Booth
P.J.
Barrera
N.P.
Robinson
C.V.
Ion mobility mass spectrometry of two tetrameric membrane protein complexes reveals compact structures and differences in stability and packing
J. Am. Chem. Soc.
2010
, vol. 
132
 (pg. 
15468
-
15470
)
[PubMed]
24
Laganowsky
A.
Reading
E.
Hopper
J.T.
Robinson
C.V.
Mass spectrometry of intact membrane protein complexes
Nat. Protoc.
2013
, vol. 
8
 (pg. 
639
-
651
)
[PubMed]
25
van Echten-Deckert
G.
Walter
J.
Sphingolipids: critical players in Alzheimer's disease
Prog. Lipid. Res.
2012
, vol. 
51
 (pg. 
378
-
393
)
[PubMed]
26
Dietschy
J.M.
Central nervous system: cholesterol turnover, brain development and neurodegeneration
Biol. Chem.
2009
, vol. 
390
 (pg. 
287
-
293
)
[PubMed]
27
He
X.
Huang
Y.
Li
B.
Gong
C.X.
Schuchman
E.H.
Deregulation of sphingolipid metabolism in Alzheimer's disease
Neurobiol. Aging
2010
, vol. 
31
 (pg. 
398
-
408
)
[PubMed]
28
Haughey
N.J.
Bandaru
V.V.
Bae
M.
Mattson
M.P.
Roles for dysfunctional sphingolipid metabolism in Alzheimer's disease neuropathogenesis
Biochim. Biophys. Acta
2010
, vol. 
1801
 (pg. 
878
-
886
)
[PubMed]
29
Cheng
H.
Wang
M.
Li
J.L.
Cairns
N.J.
Han
X.
Specific changes of sulfatide levels in individuals with pre-clinical Alzheimer's disease: an early event in disease pathogenesis
J. Neurochem.
2013
, vol. 
127
 (pg. 
733
-
738
)
[PubMed]
30
Lütjohann
D.
Meichsner
S.
Pettersson
H.
Lipids in Alzheimer's disease and their potential for therapy
Clin. Lipidol.
2012
, vol. 
7
 (pg. 
65
-
78
)
31
Prasad
M.R.
Lovell
M.A.
Yatin
M.
Dhillon
H.
Markesbery
W.R.
Regional membrane phospholipid alterations in Alzheimer's disease
Neurochem. Res.
1998
, vol. 
23
 (pg. 
81
-
88
)
[PubMed]
32
Fonteh
A.N.
Chiang
J.
Cipolla
M.
Hale
J.
Diallo
F.
Chirino
A.
Arakaki
X.
Harrington
M.G.
Alterations in cerebrospinal fluid glycerophospholipids and phospholipase A2 activity in Alzheimer's disease
J. Lipid. Res.
2013
, vol. 
54
 (pg. 
2884
-
2897
)
[PubMed]
33
Kosicek
M.
Hecimovic
S.
Phospholipids and Alzheimer's disease: alterations, mechanisms and potential biomarkers
Int. J. Mol. Sci.
2013
, vol. 
14
 (pg. 
1310
-
1322
)
[PubMed]
34
Hur
J.Y.
Welander
H.
Behbahani
H.
Aoki
M.
Franberg
J.
Winblad
B.
Frykman
S.
Tjernberg
L.O.
Active gamma-secretase is localized to detergent-resistant membranes in human brain
FEBS J.
2008
, vol. 
275
 (pg. 
1174
-
1187
)
[PubMed]
35
Fraering
P.C.
Ye
W.
Strub
J.M.
Dolios
G.
LaVoie
M.J.
Ostaszewski
B.L.
van Dorsselaer
A.
Wang
R.
Selkoe
D.J.
Wolfe
M.S.
Purification and characterization of the human gamma-secretase complex
Biochemistry
2004
, vol. 
43
 (pg. 
9774
-
9789
)
[PubMed]
36
Osenkowski
P.
Ye
W.
Wang
R.
Wolfe
M.S.
Selkoe
D.J.
Direct and potent regulation of gamma-secretase by its lipid microenvironment
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
22529
-
22540
)
[PubMed]
37
Holmes
O.
Paturi
S.
Ye
W.
Wolfe
M.S.
Selkoe
D.J.
Effects of membrane lipids on the activity and processivity of purified gamma-secretase
Biochemistry
2012
, vol. 
51
 (pg. 
3565
-
3575
)
[PubMed]
38
Winkler
E.
Kamp
F.
Scheuring
J.
Ebke
A.
Fukumori
A.
Steiner
H.
Generation of Alzheimer disease-associated amyloid beta42/43 peptide by gamma-secretase can be inhibited directly by modulation of membrane thickness
J. Biol. Chem.
2012
, vol. 
287
 (pg. 
21326
-
21334
)
[PubMed]
39
Alattia
J.R.
Matasci
M.
Dimitrov
M.
Aeschbach
L.
Balasubramanian
S.
Hacker
D.L.
Wurm
F.M.
Fraering
P.C.
Highly efficient production of the Alzheimer's gamma-secretase integral membrane protease complex by a multi-gene stable integration approach
Biotechnol. Bioeng.
2013
, vol. 
110
 (pg. 
1995
-
2005
)
[PubMed]
40
Cacquevel
M.
Aeschbach
L.
Osenkowski
P.
Li
D.
Ye
W.
Wolfe
M.S.
Li
H.
Selkoe
D.J.
Fraering
P.C.
Rapid purification of active gamma-secretase, an intramembrane protease implicated in Alzheimer's disease
J. Neurochem.
2008
, vol. 
104
 (pg. 
210
-
220
)
[PubMed]
41
Wu
F.
Schweizer
C.
Rudinskiy
N.
Taylor
D.M.
Kazantsev
A.
Luthi-Carter
R.
Fraering
P.C.
Novel gamma-secretase inhibitors uncover a common nucleotide-binding site in JAK3, SIRT2, and PS1
FASEB J.
2010
, vol. 
24
 (pg. 
2464
-
2474
)
[PubMed]
42
Alattia
J.R.
Kuraishi
T.
Dimitrov
M.
Chang
I.
Lemaitre
B.
Fraering
P.C.
Mercury is a direct and potent gamma-secretase inhibitor affecting Notch processing and development in Drosophila
FASEB J.
2011
, vol. 
25
 (pg. 
2287
-
2295
)
[PubMed]
43
Dimitrov
M.
Alattia
J.R.
Lemmin
T.
Lehal
R.
Fligier
A.
Houacine
J.
Hussain
I.
Radtke
F.
Dal Peraro
M.
Beher
D.
Fraering
P.C.
Alzheimer's disease mutations in APP but not gamma-secretase modulators affect epsilon-cleavage-dependent AICD production
Nat. Commun.
2013
, vol. 
4
 pg. 
2246
 
[PubMed]
44
Folch
J.
Lees
M.
Sloane Stanley
G.H.
A simple method for the isolation and purification of total lipides from animal tissues
J. Biol. Chem.
1957
, vol. 
226
 (pg. 
497
-
509
)
[PubMed]
45
Carvalho
M.
Sampaio
J.L.
Palm
W.
Brankatschk
M.
Eaton
S.
Shevchenko
A.
Effects of diet and development on the Drosophila lipidome
Mol. Syst. Biol.
2012
, vol. 
8
 pg. 
600
 
[PubMed]
46
Liebisch
G.
Binder
M.
Schifferer
R.
Langmann
T.
Schulz
B.
Schmitz
G.
High throughput quantification of cholesterol and cholesteryl ester by electrospray ionization tandem mass spectrometry (ESI-MS/MS)
Biochim. Biophys. Acta
2006
, vol. 
1761
 (pg. 
121
-
128
)
[PubMed]
47
Herzog
R.
Schwudke
D.
Schuhmann
K.
Sampaio
J.L.
Bornstein
S.R.
Schroeder
M.
Shevchenko
A.
A novel informatics concept for high-throughput shotgun lipidomics based on the molecular fragmentation query language
Genome Biol.
2011
, vol. 
12
 pg. 
R8
 
[PubMed]
48
Alattia
J.R.
Schweizer
C.
Cacquevel
M.
Dimitrov
M.
Aeschbach
L.
Oulad-Abdelghani
M.
Fraering
P.C.
Generation of monoclonal antibody fragments binding the native gamma-secretase complex for use in structural studies
Biochemistry
2012
, vol. 
51
 (pg. 
8779
-
8790
)
[PubMed]
49
Fraering
P.C.
Ye
W.
LaVoie
M.J.
Ostaszewski
B.L.
Selkoe
D.J.
Wolfe
M.S.
gamma-Secretase substrate selectivity can be modulated directly via interaction with a nucleotide-binding site
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
41987
-
41996
)
[PubMed]
50
Bibl
M.
Mollenhauer
B.
Esselmann
H.
Lewczuk
P.
Klafki
H.W.
Sparbier
K.
Smirnov
A.
Cepek
L.
Trenkwalder
C.
Rüther
E.
, et al. 
CSF amyloid-beta-peptides in Alzheimer's disease, dementia with Lewy bodies and Parkinson's disease dementia
Brain
2006
, vol. 
129
 (pg. 
1177
-
1187
)
[PubMed]
51
Rigaud
J.
Chami
M.
Lambert
O.
Levy
D.
Ranck
J.
Use of detergents in two-dimensional crystallization of membrane proteins
Biochim. Biophys. Acta
2000
, vol. 
1508
 (pg. 
112
-
128
)
[PubMed]
52
Schuhmann
K.
Herzog
R.
Schwudke
D.
Metelmann-Strupat
W.
Bornstein
S.R.
Shevchenko
A.
Bottom-up shotgun lipidomics by higher energy collisional dissociation on LTQ Orbitrap mass spectrometers
Anal. Chem.
2011
, vol. 
83
 (pg. 
5480
-
5487
)
[PubMed]
53
Sanchez
S.A.
Gunther
G.
Tricerri
M.A.
Gratton
E.
Methyl-beta-cyclodextrins preferentially remove cholesterol from the liquid disordered phase in giant unilamellar vesicles
J. Membr. Biol.
2011
, vol. 
241
 (pg. 
1
-
10
)
[PubMed]
54
Papan
C.
Penkov
S.
Herzog
R.
Thiele
C.
Kurzchalia
T.
Shevchenko
A.
Systematic screening for novel lipids by shotgun lipidomics
Anal. Chem.
2014
, vol. 
86
 (pg. 
2703
-
2710
)
[PubMed]
55
Schuhmann
K.
Almeida
R.
Baumert
M.
Herzog
R.
Bornstein
S.R.
Shevchenko
A.
Shotgun lipidomics on a LTQ Orbitrap mass spectrometer by successive switching between acquisition polarity modes
J. Mass Spectrom.
2012
, vol. 
47
 (pg. 
96
-
104
)
[PubMed]
56
Shevchenko
A.
Simons
K.
Lipidomics: coming to grips with lipid diversity
Nat. Rev. Mol. Cell Biol.
2010
, vol. 
11
 (pg. 
593
-
598
)
[PubMed]
57
Ledesma
M.D.
Martin
M.G.
Dotti
C.G.
Lipid changes in the aged brain: effect on synaptic function and neuronal survival
Prog. Lipid. Res.
2012
, vol. 
51
 (pg. 
23
-
35
)
[PubMed]

Author notes

1

These authors contributed equally to this work.

Supplementary data