Update your browser to view this website correctly. Update my browser now
Recently, it was shown that the γ-secretase activating protein (GSAP) regulates specifically the γ-cleavage of the β-amyloid precursor protein (APP), whose amyloidogenic processing is causatively linked to Alzheimer's disease (AD). This paper was contradicted by many groups refuting the reproducibility of the core findings; and recently, it was shown that GSAP did not specifically affect the γ-cleavage of APP but also that the levels of the β-cleaved product of APP (sAPPβ) was reduced in cells depleted of GSAP using siRNA. Here we confirm these findings that GSAP silencing also reduced Aβ and sAPPβ levels but also showed, for the first time, that rather than specifically regulating γ-secretase cleavage of APP, GSAP regulates the levels of full-length APP, thus affecting the processing of APP by all three proteases, i.e., α-, β- and γ-secretases. This could explain why GSAP silencing reduces both Aβ and sAPPβ levels. Furthermore, imatinib, an anti-cancer drug that selectively reduces Aβ levels, which was shown to occur by inhibiting the GSAP-γ-secretase interaction, reduced Aβ levels also in the absence of GSAP. Thus, our results not only uncover a new aspect of GSAP function but also continue to caution the validity of GSAP as a specific therapeutic target for Aβ production in AD.
GSAP was identified to be a specific regulator of γ-secretase cleavage of the amyloid precursor protein that is linked to Alzheimer's disease. This paper published in Nature in 2010was refuted by several groups questioning the validity and reproducibility of the findings. Recently, Udayar and Rajendran showed that GSAP depletion did not only reduce Aβ but also reduced sAPPβ. Here we replicate these findings now but also show that silencing of GSAP reduced APP levels, thus explaining how GSAP could have regulated Aβ levels. These results question once again the validity of GSAP's specific role in γ-secretase cleavage of APP but also uncovers new insights into the role of GSAP in regulating APP levels.
To study the role of γ-secretase activating protein (GSAP) in regulating APP processing and to study the reproducibility of GSAP's role in specific γ-secretase cleavage of APP.
He et al. showed that GSAP silencing specifically inhibited γ-cleavage of APP. This study was recently refuted by Udayar and Rajendran showing that GSAP silencing did not specifically inhibit Aβ, but it also lowered the sAPPβ levels, which is a product of β-cleavage of APP and not γ-cleavage. To reproduce this finding, we silenced GSAP using a pool of four siRNAs and assayed for secreted Aβ and sAPPβ levels. As controls, we used siRNAs against APP, BACE1, Pen-2, and a scrambled medium GC oligo (MedGC). The measurements to monitor APP processing were performed using a multiplexed system to quantitatively measure both Aβ and sAPPβ levels from the sample. We observed that GSAP silencing led to a decrease in both Aβ and sAPPβ levels, in agreement with the previously published finding (Fig. 1A). To study how GSAP affected both Aβ and sAPPβ levels, we first examined APP processing by Western blotting with anti-APP C-terminus-specific antibody. Surprisingly, GSAP silencing dramatically decreased full-length APP levels (Fig. 1B) (see independent replicate data in supplementary information; Suppl. Fig. 1A). As a control, we used siRNAs against APP and found that it significantly reduced APP protein levels (Fig. 1B), as expected. To rule out that the reduction in APP was due to general cell toxicity or defective cell viability, we performed a number of controls. First, we analyzed the levels of a cellular protein, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and found these to be unchanged in GSAP-silenced cells (Fig. 1B; Suppl. Fig. 1A) in the same blot along with anti-APP antibody. Second, the levels of Nicastrin, another type 1 transmembrane protein, were not affected in GSAP-silenced cells (Suppl. Fig. 1B). Third, cell viability measurements (of 20 replicates) by Alamar Blue™ from GSAP-silenced and scrambled oligo-treated control cells did not show any difference demonstrating that GSAP silencing did not induce any cell toxicity (Suppl. Fig. 1C). However, silencing Kif11, a kinesin involved in cell division, severely induced cell death, serving as a positive control for the cell toxicity assay measured by Alamar Blue (Suppl. Fig. 1C). Fourth, general morphological characterization showed no distinguishable effect of GSAP silencing on cell shape, structure, and number (Suppl. Fig. 1D). Fifth, release of the cytosolic protein, lactate dehydrogenase (LDH), was monitored to detect any membrane leakages or subtle cellular membrane defects. GSAP silencing did not affect the release of LDH (Suppl. Fig. 1E). We confirmed the efficiency of GSAP silencing by RT-PCR (Suppl. Fig. 1F). Together, our results conclusively show that GSAP silencing reduced full-length APP levels and thus altered both β- and γ-cleavage and not specifically γ-cleavage of APP or induced any cytotoxicity. To independently validate the effect of GSAP on APP levels and also to rule out any epitope-specific effects due to the use of the C-terminal antibody of APP that we used for Western blotting, we checked the cellular levels of APP using immunofluorescence with another anti-APP antibody, 6E10, that recognises the N-terminus of the Aβ domain. Again, silencing of GSAP substantially reduced APP levels (Fig. 1C) demonstrating that indeed GSAP regulates APP levels. Consistent with the observation that GSAP silencing reduces full-length APP levels, GSAP silencing also significantly reduced the levels of sAPPα, which is produced by the non-amyloidogenic (α-secretase) pathway (Fig. 1D). This reduction was not due to cell death, as the viability of GSAP-silenced cells was not affected (Fig. 1D; Suppl. Fig. 1). To rule out any off-target effects, we silenced GSAP using four different siRNAs singly and analyzed its effect on APP levels. We observed that each of the four siRNAs reduced APP levels individually (Suppl. Fig. 2A and Suppl. Fig. 2B), thus demonstrating the specificity of the observed effect of GSAP silencing on APP levels. Thus, GSAP regulates full-length APP levels and not specifically γ-secretase activity. To further validate our findings, we used yet another approach by using the anti-cancer drug, imatinib. Imatinib inhibits Aβ production, which was concluded to occur via binding to GSAP by blocking its activity. Since our results indicate that GSAP does not specifically regulate the γ-secretase activity, we then asked whether imatinib could regulate γ-secretase activity in a GSAP-independent manner. Treatment of cells with imatinib indeed selectively decreased Aβ levels without affecting sAPPβ levels or cell viability, as has been previously shown by several laboratories(Fig. 1E; Suppl. Fig. 1G). However, while cells silenced with GSAP siRNA again showed a reduction in both Aβ and sAPPβ levels, treatment with imatinib (10 mM, same concentration used in He et al.) caused a further significant and specific decrease in Aβ levels, suggesting that imatinib regulates Aβ levels independently of GSAP. Recent studies on imatinib and Aβ also fail to see an effect, further casting doubt that GSAP is an imatinib target. We have tried to reproduce the results using the cells, plasmids, experimental conditions, concentrations of inhibitors as identical as those employed by He et al., but we failed to reproduce many of their findings. Due to the unavailability of a working anti-GSAP antibodies, we were unable to look at the cleavage of endogenous GSAP. However, in large-set experiments, using RNAi against GSAP, we studied the functionality of the endogenous GSAP. The only one result that we can reproduce is the effect of GSAP on Aβ levels, which we discovered to be not due to the effect of GSAP on γ-secretase activity as claimed by He et al., but we have shown that this is rather due to GSAP’s effect on APP.
These results conclusively shows that processing of GSAP is not necessary for Aβ production and that GSAP is inappropriately named as it affects Aβ levels via regulating the levels of full-length APP and not through activating the γ-secretase, as reported earlier. Since APP and the fragments of APP have physiological functions, our work cautions the validity of GSAP as a therapeutic target for AD. This work also highlights the need for more efforts into reproducibility of the published data.
It is largely a cell culture study.
With all the published data on this subject, it is clear that GSAP is not a specific activator of γ-secretase cleavage of APP and further characterization of GSAP function is needed. However, it will be interesting to pursue how GSAP silencing reduces APP levels. One possibility is that APP transcript levels are reduced or APP protein degradation is enhanced.
Hela-sweAPP are cultured and used as described.
siRNAs were purchased from Invitrogen (stealth siRNA). The sequences and the gene IDs are supplied in supplementary table 1.
siRNA transfection for HeLa-swAPP cells
Cells were transfected with a final concentration of 5 nM using Oligofectamine (Invitrogen) as a transfection reagent at a concentration of 0.3 mL in a total volume of 100 mL following the manufacturer’s instructions. Each siRNA transfection was performed in quadruplicate. After 24 h the transfection mix was replaced with fresh culture medium. 69 h after transfection, medium was again replaced with 100 mL fresh medium containing 10% Alamar Blue™ (AbD Serotec). Supernatants were collected and assayed for Aβ and sAPPβ, as described below. The cells were lysed with 50 mL lysis buffer (1% NP-40, 0.1% SDS and protease inhibitor cocktail tablet) for 20 min on ice.
Alamar Blue assay
For cell viability measurements using Alamar Blue, the medium of transfected cells (siRNA or plasmid) was replaced with normal medium containing 10% Alamar Blue. The final volume in each well was 100 mL. 3 h after the medium change, cell viability was monitored using Fluoroscan Ascent Cf (Labsystems), with excitation wavelength 544 nm and emission at 590 nm.
Electrochemiluminescence (ECL) detection of Aβ, sAPPβ, and sAPPα
An electrochemiluminescence assay (Meso Scale Discovery, MD) was performed to determine the amount of secreted Aβ40, sAPPα, and sAPPβ in the cell culture medium 10. For the measurement of Aβ38, 40 and 42, triplex plates were used from conditioned supernatants collected for 12 h. Pre-coated plates were blocked with Tris Buffered Saline containing Tween, containing 3% Blocker A, for 1 h at room temperature on a shaker at 750 rpm. After washing, 10 mL of the cell culture supernatant was added to each well along with 10 mL of detection antibody followed by incubation for 2 h at room temperature on a shaker at 750 rpm. After washing detection was performed in 35 mL 2X MSDT read buffer and read with the Sector Imager 6000.
After 72 h of siRNA transfection, cells were lysed in buffer containing 1% NP-40 and 0.1% SDS and protease inhibitors (Roche). Equal amounts of the lysate (according to the protein content quantified by BCA assay (Pierce)) were run on 4–12% BIS-TRIS gels (Invitrogen). The gel was blotted onto a nitrocellulose membrane (BioRad) and probed with the respective antibodies: anti-APP, C-terminal antibody: SIGMA (F3165-1MG); 6E10 anti-Aβ recognizing APP antibody: Covance; anti-Nicastrin antibody (2332-1): Epitomics; GAPDH antibody: Meridian Science.
Cells were reverse-transfected either with MEDGC or siRNAs against APP or GSAP on chambered coverslides and 72 h later fixed, permeabilised, blocked and immunostained with 6E10 anti-APP antibody. The signal was visualised by the use of Alexa488 or Alexa546-coupled anti-mouse secondary antibody. Nuclei were visualised by DAPI staining.
Cells were reverse-transfected either with MEDGC or siRNAs against GSAP, and 69 h later, cells were treated with DMSO or Imatinib (10 μM, Enzo Life Sciences) for 3 h. Treatment was done with culture medium containing 10% Alamar Blue. The final volume of culture medium in each well was 100 μL. After 72 h, cell viability was assessed by Alamar Blue assay, supernatants were collected and assayed for Aβ and sAPPβ.
Cell cytotoxicity assay
The cell cytotoxicity assay was carried out using the Cytotoxicity Detection Kit (Roche) by measuring the activity of released lactate dehydrogenase into the culture medium. The culture supernatant was collected from the cells (with and without Triton X-100 treatment). Samples were further diluted and processed according to manufacturers’ protocol. The increase in amount of enzyme activity in the supernatant directly correlates with the amount of formazan formed, which is proportional to the number of lysed cells. Absorbance was measured at 492 nm with reference wavelength at 620 nm on a plate reader spectrophotometer.
Total RNA from cells was isolated using TRI Reagent® (Sigma-Aldrich) following manufacturer’s protocol. 1 μg of total RNA was used for reverse transcription with iScript™ cDNA synthesis kit (Bio-Rad) according to the manufacturer’s protocol. Real-time PCR was performed using iTaq™ Universal SYBR® Green supermix (Bio-Rad) following manufacturer’s instructions. Relative gene expression levels were calculated with the ΔΔCt method using GAPDH for normalization.
L. R acknowledges the financial support from the Velux Foundation, the Swiss National Science Foundation grant, the Bangerter Stiftung, the Baugarten Stiftung and the Synapsis Foundation. L. R and V. U acknowledge the funding support from the European Neuroscience Campus of the Erasmus Mundus Program.
We thank G. Yu for the HeLa-swAPP cells.
All the experiments were conducted according to the standard ethical guidelines.