Presenilin-1 mutations of leucine 166 equally affect the generation of the Notch and APP intracellular domains independent of their effect on Aβ₄₂ production

Antibodies.

The polyclonal and monoclonal antibodies against PS1 (3027, BI.3D7; ref. 27) and PS2 (3711, BI.HF5c; ref. 27), against the C terminus of APP (6687; ref. 11) and against Aβ_1–42 (3926; ref. 28) have been described. The monoclonal anti-myc antibody 9E10 was obtained from the Developmental Studies Hybridoma Bank, University of Iowa, or from Santa Cruz Biotechnology for use in immunofluorescence (1:300).

cDNA Constructs, Cell Culture, and Cell Lines.

cDNA constructs encoding PS1 L166 mutants were generated as described (11) and cloned into the pcDNA3.1/zeo(+) expression vector (Invitrogen). Human embryonic kidney 293 (K293) cells were transfected and cultured as described (11). K293/sw cells stably expressing Swedish mutant APP (swAPP) have been described (29).

Analysis of PSs, APP, and Notch Endoproteolysis.

PSs (27), APP, APP CTFs, secreted Aβ species (11), and NICD generation (7, 30) were analyzed as described. AICD generation was analyzed in vitro from membrane preparations of stably transfected K293/sw cells. Equal amounts of cells were resuspended (at least 0.5 ml per 10-cm dish) in homogenization buffer [10 mM Mops (pH 7.0)/10 mM KCl] containing 1× protease inhibitors (PI; Complete, Roche Molecular Biochemicals) and homogenized using a freeze–thaw procedure. Following swelling on ice for 10 min, cells were frozen in liquid N₂ for 5 min and allowed to thaw 28 min on ice and 2 min at RT. Equal volumes of the homogenate were centrifuged for 10 min at 1,000 × g at 4°C to prepare a post nuclear supernatant (PNS) fraction. In parallel, aliquots of the homogenate were solubilized in STEN-lysis buffer [50 mM Tris (pH 7.6)/150 mM NaCl/2 mM EDTA/1% Nonidet P-40] and, after a clarifying spin, subjected to protein determination. Membranes were pelleted from the PNS by centrifugation for 20 min at 16,000 × g at 4°C and resuspended (at least 75 μl per 10-cm dish) in assay buffer [150 mM sodium citrate (pH 6.4)/1 mM EDTA] according to the protein content determined in the aliquots of the homogenates above. To allow for AICD generation, aliquots of the membranes were incubated at 37°C for 1 h and analyzed for AICD as described (18). Aliquots of the membranes that were not used for AICD generation were analyzed for APP and APP CTF levels. To facilitate detection of recombinant AICD variants, cells were treated with 1.66 mM 1,10-phenanthroline for 90 min before the preparation of cell lysates.

PS1^−/− Rescue Assays.

PS1^−/− fibroblast cells were cotransfected with equal amount of either wild-type (wt) or L166 mutant PS1 cDNA and CBF1-luciferase reporter construct (31) by using Lipofectamine 2000 (GIBCO). The EGFP construct was also added at 1/25 molar ratio to each transfectant to normalize the transfection efficiency. Luciferase activity was measured from cell lysates 24 h post-transfection according to manufacturer's suggested method (Promega).

Confocal Microscopy.

Immunofluorescence was carried out as described (32) by using Alexa Fluor 488 (1:1,000, Molecular Probes) for detection. To detect cell nuclei of fixed cells, monomeric cyanine nucleic acid stain TO-PRO-3 iodide (1:1,000, Molecular Probes) was used. Confocal images were obtained with a Leica TCS SP confocal laser scanning microscope (Leica, Deerfield, IL). Single optical sections from each cell were chosen and processed using photoshop software (Adobe Systems, Mountain View, CA).

The PS1 L166P FAD Mutation Associated with Disease Onset in Adolescence Causes Greatly Increased Aβ₄₂ Generation and Impairs Notch Signaling.

Several FAD mutations have been identified in TM3 of PS1, suggesting that this TM is an important site for PS function. Here we report a previously uncharacterized FAD-associated mutation in TM3 of PS1 that changes the leucine residue at codon 166 to proline. This mutation is associated with an onset of AD in adolescence. In the female proband, secondary generalized seizures began at age 15, major depression occurred at 19, memory was clearly impaired by 24, ataxia and spastic paraparesis were recorded by 27, and moderate stage dementia by 28. Dementia, ataxia, and spasticity progressed until death at age 35. Numerous Aβ-immunopositive neuritic and cotton wool plaques were seen throughout the cerebral cortex and Aβ-immunopositive amyloid cores were abundant in the cerebellar cortex (J.M. and B.G., unpublished results).

Leucine 166 appears to be a rather important residue for PS function because it is conserved in all PSs, the only exceptions being the C. elegans PS homologue hop-1 and the Arabidopsis thaliana and Oryza sativa PSs. In addition, a FAD mutation changing leucine 166 to arginine has been identified in a Spanish family (33). To investigate this critical site of PS1 in more detail, we examined the effects of the L166 FAD mutations, as well as of several artificial mutations at this residue on APP and Notch endoproteolysis. We first investigated the pathological activity of the two FAD-associated PS1 L166 mutants on Aβ production by using a tissue culture system that has previously been proven to be very sensitive for the detection of abnormal Aβ₄₂ generation (ref. 34 and references therein). cDNAs encoding wt PS1, PS1 L166P, and PS1 L166R were stably transfected into K293/sw cells (29) and analyzed for PS expression. As shown in Fig. 1A, expression of wt PS1 and the L166 mutants allowed the generation of robust amounts of PS1 CTFs, demonstrating that the L166 mutants did not affect endoproteolysis of PS1. Functional expression of the PS1 variants was confirmed by the replacement of endogenous PS2 (ref. 35; Fig. 1A). We next analyzed the pathological function of the L166P and L166R mutation on total Aβ, Aβ₄₀, and Aβ₄₂ production. Conditioned media from metabolically labeled cells stably expressing wt PS1 or the two PS1 L166 FAD mutants were collected and analyzed for levels of total Aβ (Fig. 1B) or Aβ₄₀ and Aβ₄₂ species (Fig. 1C). Although the levels of total Aβ were similar (Fig. 1B), quantitation of Aβ₄₀ and Aβ₄₂ species revealed that the PS1 L166P caused a striking 5- to 6-fold increase of the Aβ₄₂/Aβ_total ratio (Fig. 1C and Table 1). Although not as high as observed for the L166P mutation, the L166R mutation showed a 3-fold increase of the Aβ₄₂/Aβ_total ratio (Fig. 1C and Table 1). Comparison of the PS1 L166P mutant with the PS1 M146L, L286V, Δexon9, and G384A FAD mutants underlined its exceptional Aβ₄₂ increase (Fig. 1D). This increase was comparable to that found for the PS1 G384A mutant (refs. 11 and 36; Fig. 1D) and considerably higher than that found for the large majority of the PS1 mutations (37).

We next investigated whether the L166P and L166R mutants also affect Notch signaling and assessed the activation of the Notch-dependent downstream transcription factor CBF1 (C-promoter binding factor 1), using luciferase as a reporter gene (31). PS1^−/− fibroblast cells were transiently transfected with wt PS1 or the PS1 L166P or L166R FAD mutants along with the CBF1-luciferase construct and activation of luciferase was measured. Robust luciferase activity was detected in PS1^−/− cells transfected with wt PS1 (Fig. 1E). In contrast, significantly reduced luciferase activity was observed with the L166P and L166R mutants. Thus, these data demonstrate that in addition to producing aberrant amounts of Aβ₄₂, the L166P and L166R mutants are also partially defective in Notch signaling.

Artificial L166 Mutants Cause Increased Aβ₄₂ Levels.

To examine the consequences of L166 mutations on γ-secretase cleavage in more detail, we changed leucine 166 to glutamate, alanine, serine, cysteine, and tryptophan. The respective cDNAs were stably transfected into K293/sw cells. All mutants were expressed at similar levels and replaced endogenous PSs (data not shown). We then analyzed whether the various artificial L166 mutants affect Aβ₄₂ production. Conditioned media were collected from metabolically labeled cells expressing wt PS1 or the artificial L166 mutants and analyzed for Aβ₄₀ and Aβ₄₂ production. Quantitation revealed that all mutations increased the Aβ₄₂/Aβ_total ratio to various extents, demonstrating that codon 166 is very sensitive to amino acid changes (Table 1). Introduction of the negatively charged amino acid glutamate led to the strongest elevation (4- to 5-fold) of the Aβ₄₂/Aβ_total ratio among the artificial L166 mutants. A rather strong increase in the Aβ₄₂/Aβ_total ratio was also observed for the serine and alanine substitutions. In contrast, the cysteine and tryptophan substitutions elevated the Aβ₄₂/Aβ_total ratio only slightly, similar to the majority of the known PS1 mutations (37). Thus, all artificial L166 mutants described here induce pathological Aβ₄₂ production and may therefore probably lead to AD if they occur in humans.

Differential Effects of PS1 L166 Mutants on Notch Endoproteolysis.

The results shown above suggest that the L166 mutants cause structural alterations of TM3 that result in an altered γ-secretase cleavage and lead to variable production of Aβ₄₂. We therefore next asked whether the various L166 mutants would also affect endoproteolysis of Notch. The above described cell lines expressing either wt PS1, FAD-linked, or artificial L166 mutations were transfected with the NotchΔE (NΔE) cDNA (38) and NICD formation was monitored in pulse–chase experiments as described (7). As a control, K293/sw cells stably co-expressing NΔE with the PS1 D385A mutant were used. NICD production was observed in all cell lines with the exception of cells expressing the PS1 D385A mutant that led to an almost complete inhibition of NICD generation, consistent with previous results (7, 11, 30). Consistent with impaired CBF1-mediated Notch signaling (Fig. 1E), the FAD-associated mutations L166P and L166R showed a significantly impaired conversion of NΔE to NICD (Fig. 2A). Marked impairment of NICD generation was also observed for the artificial L166W mutation, whereas the other L166 mutants displayed moderate or very little impairment in NICD formation. Quantitation confirmed the differential effects of the various L166 mutants on the conversion of NΔE to NICD (Fig. 2B).

To further confirm these results we followed the cellular distribution of Notch in confocal immunofluorescence experiments. Cells co-expressing NΔE and wt PS1, the two FAD-linked L166 mutants, and the artificial PS1 L166S and L166W mutants were permeabilized and co-stained with antibody 9E10 to the myc-tag of NΔE and with TO-PRO-3 iodide to visualize cell nuclei. Consistent with the results described above, cells expressing wt PS1 or the artificial PS1 L166S mutant showed a prominent nuclear Notch staining, suggesting efficient generation and nuclear translocation of NICD (Fig. 3). In contrast, cells expressing the PS1 L166P or L166R FAD mutants or the artificial L166W mutant showed much weaker staining of nuclear Notch, confirming the impaired generation and consequently reduced nuclear translocation of NICD (Fig. 3). Concomitant with reduced staining of nuclear Notch, increased staining of Notch on the cell surface was observed, indicating an inefficient conversion of cell-surface-located NΔE to NICD. In agreement with previous results (30), cell-surface staining of Notch, representing accumulated uncleaved NΔE precursor, was observed in cells expressing the PS1 D385A mutant (Fig. 3). In these cells, very little if any staining of nuclear Notch was observed (Fig. 3), consistent with an almost complete inhibition of NICD formation (see Fig. 2).

PS1 L166 Mutants Differentially Affect AICD Generation.

Because some L166 mutants, in particular the FAD-associated L166P mutant, drastically impaired NICD generation, we next investigated whether the S3-like γ-secretase cleavage of APP that generates AICD would also be affected by these mutants. K293/sw cells stably transfected with either wt PS1, the PS1 L166 mutants, or PS1 D385A were analyzed for AICD generation in in vitro assays that allow the production of robust amounts of AICD (16, 18, 39). We first analyzed AICD generation from cells expressing wt PS1. In agreement with previous findings, in vitro-generated AICD comigrated with a recombinant AICD variant starting at amino acid 50 of the Aβ domain (AICD₅₀) but not with a recombinant AICD variant starting at amino acid 43 (AICD₅₇) (Fig. 4A). Thus, these data confirm the efficient production of AICD that is generated by γ-secretase cleavage predominantly at position 49 of the Aβ domain. We next investigated whether the L166P mutant and the other L166 mutants affect AICD generation. As shown in Fig. 4B, all cell lines contained similar amounts of full-length APP and APP C-terminal fragments, CTFβ and CTFα, the substrates of γ-secretase, with the exception of the PS1 D385A mutant, which caused an expected strong accumulation of CTFβ and CTFα (6). AICD production was strongly reduced in cells expressing the L166P mutant. Interestingly, the L166W mutant also impaired AICD production consistent with its significant effect on S3 cleavage of Notch (compare Fig. 4 B and C with Fig. 2), whereas the other mutants displayed differential effects on AICD generation (Fig. 4B). Consistent with previous results (18, 20, 21), expression of the biologically inactive PS1 D385A mutant led to an almost complete block of AICD generation. To compare the effects of the L166 mutants on AICD production, we quantified AICD production relative to the amounts of CTFα, the predominant AICD precursor species in the K293/sw cells investigated (Fig. 4B). Quantitation of the AICD/CTFα ratios confirmed the differential effects of the L166 mutants on AICD production (Fig. 4C). Interestingly, comparison of the effects of the L166 mutants on NICD (Fig. 2B) and AICD production (Fig. 4C) indicated that Notch S3 and APP S3-like γ-secretase cleavage were affected by the L166 mutants in a similar manner. Indeed, statistical analysis (Spearman rank test) revealed that the effects of the L166 mutants on NICD and AICD production correlated (P < 0.02).