In order to explore the potential catalytic effect of the aspartic acid residues in PS1 TM 6 and 7 domains, we generated mutant PS1 including those containing either a single DTG triplet or both triplets after residue D257 or/and D385 (see Table 1). The transient transfection experiments were carried out with equal amount of each mutated PS1 construct and secreted Aβ₄₀ and Aβ₄₂ were measured by ELISA. Two stable cell lines, named H125.3-16 and H167-11 cell lines that express human APP carrying Swedish or London mutation respectively, were used for assessing changes of secreted Aβ₄₀ and Aβ₄₂. The values of Aβ from the mock (empty pcDNA3 vector) transfected cells were used for normalization to calculate the % change in Aβ values obtained after transfection with the various plasmids shown in column 2. When the H125.3-16 cells were transfected with wild-type PS1 (wt-PS1), the ratio of Aβ₄₂ over total Aβ was unchanged (Table 1). As expected, PS1 mutant carrying the familial mutation (PS1-M146V) 4041 cause a preferential production of Aβ₄₂ (Table 1). While mutant PS1 carrying a single DTG triplet at D257 (PS1-D257TG) behaved similar to wt PS1, mutant PS1 carrying a single DGT triplet at D385 (PS1-D385TG) caused a remarkably increased production of Aβ₄₂ but not Aβ₄₀, and this increase was even more dramatic than the above mutants (Table 1). A mutant PS1 containing both the M146V and D385TG mutations was not any more productive in Aβ₄₂ secretion than either mutation separately (data not shown), suggesting no synergistic effect for these two mutations. Transient transfection of these PS1 mutant constructs into the human IMR-32 neuroblastoma cells or a mouse neuroblastoma cell line N2A-APP, which expresses human Swedish APP as previously described 4243, also produced similar patterns of Aβ levels (data not shown). We also generated stable cells expressing both Swedish APP and various PS1 mutants as shown in Table 2. PS1 mutant harboring D385TG produced about 15-fold higher levels of Aβ₄₂ than control while Aβ₄₀ was only increased by 1.5-fold than control (Table 2). Similarly, stable cells expressing PS1-D257TG/D385TG also produced significantly higher levels of Aβ₄₂ than control cells.

A Western blot of protein extracts from transfected cells showed that transfected PS1 proteins in various cell lines are comparable (Figure 1), suggesting that the above shift of Aβ₄₂ in cells expressing mutant PS1 was not due to the dramatically altered expression of PS1 variants. Thus, generation of artificial DTG motif at D385 appears to dramatically favor the production of Aβ₄₂.

Previously studies have shown that mutation of either D257 to A257 or D385 to A385 causes significant reduction of total Aβ production 26. To determine whether the enhanced Aβ₄₂ production seen in PS1-D385TG mutant was truly due to the introduction of an aspartyl protease DTG triplet, we then made substitutions to disrupt DTG motif in PS1-D385TG or PS1-D257TG/D385TG template. Interestingly, disruption of DTG motif in PS1-D257T

A

A

A

The effects on the secretion of Aβ, caused by the PS1 mutants harboring the DTG triplet were explored by treating transfected cells with two well characterized γ-secretase inhibitors: L-685,458, an aspartyl protease transition state analog 46, and fenchylamine, a sulfonamide derivative 47. Production of Aβ₄₀ was initially increased in APA2 cells (expressing both Swedish APP and wt PS1, △), APD3 (expressing both Swedish APP and PS1-M146V, ▲) and their parental line H125.3-16 (expressing only Swedish APP, ▼) upon addition of L-685,458 up to 0.3 μM, but became reduced with the increased dose of L-685,458 while reaching a complete inhibition at 3 μM (Figure 2A). The cell line APE12, expressing both Swedish APP and PS1-D257TG, showed a similar biphasic curve but in a lower dose range (0.1 μM and 1 μM, respectively; data not shown). Contrary to the above cases, cell lines expressing PS1 mutants PS1-D385TG (APB10,) and PS1-D257TG/D385TG (APC5, +) did not display biphasic curves and showed inhibition of Aβ₄₀ production at all concentrations of the inhibitor used in the experiments. Interestingly, this non-biphasic curve was also seen in the cell line H167-11 expressing the London mutation APPV642F (□). This suggests that the conformational changes seen in APP London mutation are likely compatible with the D385TG mutation in PS1 during the interaction between the enzyme and its substrate; both facilitate γ-secretase to produce Aβ₄₂.

The biphasic effects of L-685,458 on Aβ₄₂ secretion from the cell lines expressing wt PS1 or familial PS1 mutation were even more robust than that on Aβ₄₀ secretion (Figure 2B). Again, cell lines expressing either APP London mutation or D385TG triplet displayed no obvious biphasic effects. It appeared that L-685,458 has less inhibitory potency on Aβ₄₂ production than on Aβ₄₀ production as it required a higher dose to inhibit Aβ₄₂ than Aβ₄₀ in cells expressing PS1 carrying D385TG triplet. Cell toxicity was not obvious during the treatments with this drug up to 30 μM for any of the cell lines tested (data not shown), excluding the possibility that the Aβ reduction seen in the above experiments was due to drug toxicity.

A biphasic production of Aβ₄₀ was also seen in H125.3-16 (+), H143.3 (▽), and APA2 (△) cell lines treated with the less potent γ-secretase inhibitor fenchylamine sulfonamide: it took 12 μM to reach the peak of enzymatic activity and 30 μM to achieve complete inhibition (Figure 3A). Cells expressing PS1-D257TG (APE12) were also more sensitive to the drug treatment with 3 μM for the full stimulation (data not shown). Again, cells expressing PS1 D385TG mutants (APB10, ○) showed highest sensitivity to the treatment with non-biphasic responses (Figure 3A). Similarly, there appeared no biphasic Aβ₄₀ production in cells expressing APP London mutation when treated with fenchylamine sulfonamide.

For the production of Aβ₄₂, fenchylamine sulfonamide produced a large stimulus response on cells expressing endogenous PS1 or transfected wt PS1, but had a weak stimulating effect or no inhibition on the other cell lines (Figure 3B). The stimulating effect was probably due to the low potency of this drug on the inhibition of γ-secretase activity in producing Aβ₄₂, and the biphasic effect reflects a low inhibitory profile.

The nonselective COX inhibitors NSAIDs Ibuprofen, Indomethacin and Sulindac sulfide (an active metabolite of the pro-drug Sulindac) have been shown to reduce Aβ₄₂ levels through increasing Aβ₃₈ production 48, suggesting that NSAIDs can modulate γ-secretase activity. We found that all of the above compounds lowered Aβ₄₂ secretion in a dose dependent manner in APB10 cell line expressing both Swedish APP and PS1-D385TG or APD cell line expressing both Swedish APP and the clinical mutation of PS1 M146V (Table 4). Sulindac sulfide was the most potent, followed by Indomethacin and Ibuprofen, showing 50% inhibition at about 30–50 μM for Sulindac Sulfide, 50–150 μM for Indomethacin and 300 μM for Ibuprofen. The nonselective COX inhibitors NSAIDs Aspirin (Acetylsalicylic acid) and Naproxen did not have any inhibitory effect (data not shown). The COX-2 NSAID inhibitor Meloxicam showed about 50% inhibition at 300 μM for both Aβ₄₀ and Aβ₄₂ which was similar to Ibuprofen for Aβ₄₂, while Ibuprofen showed no effect on Aβ₄₀ production.

To examine the Aβ species under the treated conditions, media from the three tested cell lines expressing PS1, PS1-D385TG and PS1-M146V were collected, and Aβ peptides were immunoprecipitated with monoclonal antibody 4G8. Sulindac Sulfide at a concentration of 100 μM was chosen for treating the cell lines for its potent inhibitory effects seen in the above experiments. The immunoprecipitates were examined on bicine-urea gels to have better resolution of various Aβ species 48. As shown in Figure 4, Aβ₄₀ was a predominant species in this wt-PS1 expressing H125.3-16 cell line (Figure 4, lane 1). Aβ₄₂ accounted for only 13% of total Aβ species that is defined as the sum of the Aβ 38, 39, 40, and 42 bands (as determined by optical density measurements). As expected, the H125.3-16 cells treated with Sulindac Sulfide produced no measurable Aβ₄₂ while Aβ₃₈ was increased from 15% to 33% of total Aβ species (Figure 4 lanes 2). Prior to any treatment, substantially more Aβ₄₂ was produced in PS1-D385TG expressing cells (Figure 4 lanes 3) than in cells expressing either endogenous levels of wt PS1 (lanes 1) or PS1-M146V (lanes 5), and this was in consistent with the ELISA results. The Western blot results also indicated that PS1-D385TG elevated Aβ₃₈ and Aβ₃₉ production, and the increased Aβ₄₂ species accounted for only 25% of the total Aβ in PS1-D385TG-expressing cells instead of 33% in PS1-M146V-expressing APC5 cells. After Sulindac Sulfide treatment, Aβ₄₂ formation was completely shifted to Aβ₃₈ in both H125.3-16 and APC5 cell lines (Figure 4, lanes 2 and 6). However, Aβ₄₂ production was only partially reduced in PS1-D385TG-expressing cells under the same treatment conditions (Figure 4, lane 4). As noted, while the levels of Aβ₃₈ were increased from 25% of total Aβ species to 33% in the PS1-D385TG expressing cells treated with 100 μM Sulindac Sulfide, it was the Aβ₃₉ which had a more significant decline from 20% to undetectable in respect to the total Aβ species (Figure 4, lane 4). Apparently, NSAID has differential affects on the selective cleavage of APP-CTFs by the γ-secretase activity when PS1 is mutated.

Notch receptors undergo three distinct proteolytic cleavages during maturation and activation, and the third cleavage of Notch (S3 site) occurs within the plasma membrane by the PS1-containing γ-secretase and results in the release and translocation of the intracellular domain into the nucleus to execute Notch signaling 49. To determine whether the PS1 mutations carrying the DTG triplet would affect Notch processing, we developed a protocol based on the translocation of the Notch intracellular domain (NICD) to the nucleus after S3 cleavage 50. For each of the 5 cell lines examined, cleavage of Notch by the γ-secretase activity was determined based on luciferase activity that measures binding of NICD to the reporter. Figure 5 summarized results from three independent experiments. While the clinical mutation PS1-M146V showed similar processing activity at the S3 site to the wild type PS1, a significant reduction of S3 cleavage was found in cells expressing PS1 mutants containing DTG motif at D385 (Figure 5). Specifically, the Notch cleavage activities of PS1-D385TG, PS1-DTG/DTG, PS1-D257TG and PS1-M146V have decreased by about 62%, 46%, 22% and 7%, respectively, compared to PS1wt activity (Figure 5). Obviously, the mutations in PS1 containing a DTG motif caused a significant change in the conformation in the γ-secretase complex that is not favoring the cleavage of the Notch substrate.

HEK-293 cells were initially transfected with sixteen μg of DNA (pcDNA3.1/Hygro- vector inserted with either APP Swedish or London mutant. Two selected stable cell lines expressing Swedish APP were designated to be H143.3-23 and H125.3-16 while H167-11 for the cell line APP London mutant. A similar procedure was followed for the establishment of 125.3-16 cells that express PS1, PS1-D385TG, PS1-D257TG/D385TG, PS1-M146V, or PS1-D257TG cDNA inserts. The table below shows the nomenclature for the APP/PS1 stable cell lines and the clone used for follow-up studies. Three to ten clones were picked for each cell line and examined for the levels of Aβ₄₀ and Aβ₄₂ in conditioned media. The average ratio of Aβ₄₂/total-Aβ for ten wt PS1-clones was 0.139 ± 0.22; 0.644 ± 0.089 for nine PS1-D385TG clones; 0.564 ± 0.133 for five PS1-D257TG/D385TG clones, 0.240 ± 0.033 for three PS1-M146V clones, and 0.087 ± 0.044 for ten PS1-D257TG clones. One clone for each PS1 DNA construct was selected from the set of clones for follow-up.

H125.3-16 APP-Sw

APA clone #2 APP-Sw + PS1 wt

APB clone #10 APP-Sw + PS1-D385TG

APC clone #5 APP-Sw + PS1-D257TG/D385TG

APD clone #3 APP-Sw + PS1-M146V

APE clone #12 APP-Sw + PS1-D257TG

Notch undergoes cleavage by the γ-secretase to release NICD that will translocate into nucleus to result in expression of target gene. To examine this cleavage, cell lines were seeded in 6 well dishes at 6 × 10⁵ cells per ml (2 ml/well) and each well was transfected with 6.25 ng Notch ΔE-GVP 61, 1.55 μg pFR-Luc (UAS-firefly luciferase, Stratagene), and 62.5 ng pRL-CMV (renilla luciferase, Promega) for the luciferase assay or 2 μg Notch1ΔE for the Notch the next day. After incubation for 3 hrs, transfection media was replaced with growth media and cells were allowed to grow for an additional 48 hrs. Luciferase assay were performed according to the protocols from manufacturer (Dual-Luciferase Reporter Assay System, Promega). Briefly, each well was washed twice with PBS and harvested with 500 μl PLB buffer Lysates were transferred to eppendorf tubes and allowed to freeze at -80° for up to one week before being assayed. Lysates (2 or 20 μl) was transferred to each well of a 96 well plate and luciferase activity was measured using the Promega Dual-Luciferase reagents and a Lumiskan Ascent luminometer (ThermoLabsystems).

The analysis of Aβ levels from conditioned media under specified conditions was performed as described previously 62. Statistical analysis of the Aβ₄₀ and Aβ₄₂ levels was performed using the Student's t-test.

Cells were first grown in DMEM media for 24 hours in 6 well dishes to near confluence and then treated with drugs such as 100 μM of sulindac sulfide in 1% DMSO. After incubation for 24 hrs, one ml of conditioned media was used for immunoprecipitation with monoclonal antibody 4G8 under standard overnight immunoprecipitation conditions as described previously 63. The extensively washed immunoprecipitates were resolved on a Tricine-Urea gel. For Western with cell lysates, cell extracts were prepared in TENT buffer (50 mM Tris pH 8.0, 150 mM NaCl, 2 mM EDTA and 1% Triton X-100) with protease inhibitor cocktails. Equal amount of protein extracts were resolved on a 4–12% NuPage Bis Tris gel from Invitrogen (Carlsbad, CA). Monoclonal antibody 6E10 was used to detect Aβ species.

Cells were plated at 50 to 100 thousand per well. After 48 hours, when cells were confluent, medium was replaced by cell medium containing the drug at each dilution. Each drug dilution was run in triplicate wells. After 24 hours incubation, half the volume of the conditioned cell medium was collected for measuring Aβ₄₀ and Aβ₄₂ by ELISA, while the remaining was saved for replication. The plate with the remaining cells was used for the MTS reduction assay to assess drug toxicity to the cells. All drugs were dissolved in DMSO at a concentration 1000 fold higher than the final drug concentration in the cell media for a final concentration of DMSO of 0.1%. The drugs used were Fenchylamine, L-685,458 (from Bachem), Sulindac sulfide and sulfone (a second metabolite of Sulindac) (all from Biomol Research Labs Inc.), Acetylsalicylic acid (from ICN), (S)-Naproxen (from Cayman Chemical Co), Meloxicam (from Calbiochem). The experiment with the selective γ-secretase inhibitors was replicated. The comparison was made between Aβ₄₀ and Aβ₄₂ levels after drug treatment versus mock treated control. Cellular toxicity in treated cells was evaluated according to the procedures as previously described 64.