The Biological and Pathological Function of the Presenilin-1 ΔExon 9 Mutation Is Independent of Its Defect to Undergo Proteolytic Processing*

The two homologous presenilins are key factors for the generation of amyloid β-peptide (Aβ), since Alzheimer’s disease (AD)-associated mutations enhance the production of the pathologically relevant 42-amino acid Aβ (Aβ42), and a gene knockout of presenilin-1 (PS1) significantly inhibits total Aβ production. Presenilins undergo proteolytic processing within the domain encoded by exon 9, a process that may be closely related to their biological and pathological activity. An AD-associated mutation within the PS1 gene deletes exon 9 (PS1Δexon9) due to a splicing error and results in the accumulation of the uncleaved full-length protein. We now demonstrate the unexpected finding that the pathological activity of PS1Δexon9 is independent of its lack to undergo proteolytic processing, but is rather due to a point mutation (S290C) occurring at the aberrant exon 8/10 splice junction. Mutagenizing the cysteine residue at position 290 to the original serine residue completely inhibits the pathological activity in regard to the elevated production of Aβ42. Like PS1Δexon9, the resulting presenilin variant (PS1Δexon9 C290S) accumulates as an uncleaved protein and fully replaces endogenous presenilin fragments. Moreover, PS1Δexon9 C290S exhibits a significantly increased biological activity in a highly sensitive in vivo assay as compared with the AD-associated mutation. Therefore not only the increased Aβ42 production but also the decreased biological function of PS1Δexon9 is due to a point mutation and independent of the lack of proteolytic processing.

The two homologous presenilins are key factors for the generation of amyloid ␤-peptide (A␤), since Alzheimer's disease (AD)-associated mutations enhance the production of the pathologically relevant 42-amino acid A␤ (A␤42), and a gene knockout of presenilin-1 (PS1) significantly inhibits total A␤ production. Presenilins undergo proteolytic processing within the domain encoded by exon 9, a process that may be closely related to their biological and pathological activity. An AD-associated mutation within the PS1 gene deletes exon 9 (PS1⌬exon9) due to a splicing error and results in the accumulation of the uncleaved full-length protein. We now demonstrate the unexpected finding that the pathological activity of PS1⌬exon9 is independent of its lack to undergo proteolytic processing, but is rather due to a point mutation (S290C) occurring at the aberrant exon 8/10 splice junction. Mutagenizing the cysteine residue at position 290 to the original serine residue completely inhibits the pathological activity in regard to the elevated production of A␤42. Like PS1⌬exon9, the resulting presenilin variant (PS1⌬exon9 C290S) accumulates as an uncleaved protein and fully replaces endogenous presenilin fragments. Moreover, PS1⌬exon9 C290S exhibits a significantly increased biological activity in a highly sensitive in vivo assay as compared with the ADassociated mutation. Therefore not only the increased A␤42 production but also the decreased biological function of PS1⌬exon9 is due to a point mutation and independent of the lack of proteolytic processing.
Early onset Alzheimer's disease (AD) 1 can occur due to mutations within the genes encoding the ␤-amyloid precursor protein (␤APP) and the two presenilins, PS1 and PS2 (Refs. 1-3; summarized in Refs. 4 and 5). The mutations increase the generation of the 42-amino acid version of A␤ (A␤42), which aggregates more readily (6) and is therefore preferentially deposited in senile plaques (5). PS proteins are also involved in the physiological production of A␤, since the lack of PS1 expression in PS1 Ϫ/Ϫ mice results in a dramatically reduced A␤ production (7). All but one PS mutations are point mutations affecting conserved amino acids (4,5). However, due to a splicing error, the PS1⌬exon9 mutation results in the deletion of the domain encoded by exon 9 (8). This domain contains the cleavage site for proteolytic processing (9,10), and therefore PS1⌬exon9 accumulates as an uncleaved protein (9). Since proteolytic processing is highly regulated (9,11,12) and appears to be altered by PS mutations (Refs. 13-15; summarized in Ref. 16), the lack of proteolytic processing caused by the exon 9 deletion was expected to be responsible for its pathological activity. We now demonstrate the unexpected finding that the pathological function of PS1⌬exon9 as well as its reduced biological activity is independent of its lack to undergo proteolytic processing, but is rather due to a point mutation (S290C) that is the result of the aberrant exon 8/10 splice junction.
After gel purification the PCR products were mixed and subjected to a final PCR with primers PS1-187-F and PS1-STOP-R. The resulting PCR product was digested with EcoRI/XhoI and cloned into the pcDNA3.1/Zeo(ϩ) expression vector (Invitrogen). The cDNA was sequenced to verify successful mutagenesis.
Metabolic Labeling and Immunoprecipitation of PS-Stably transfected K293 cell lines were grown to confluence in 10-cm dishes. After starvation for 1 h in 4 ml of methionine-and serum-free MEM (MEM supplemented with 1% L-glutamine and 1% penicillin/streptomycin) cells were metabolically labeled with 700 Ci of [ 35 S]methionine (Promix, Amersham Pharmacia Biotech) in 4 ml of methionine-and serumfree MEM for 1 h. Cell extracts were prepared and subjected to immunoprecipitation of PS as described (12). PS immunoprecipitates were solubilized in sample buffer containing 4 M urea for 10 min at 65°C, * This work was supported by grants from the Verum Foundation (A␤42 generation) and the Deutsche Forschungsgemeinschaft (DFG) (to C. H. and R. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Combined Immunoprecipitation/Western Blotting-Stably transfected K293 cell lines were grown to confluence. Cell extracts were prepared and subjected to immunoprecipitation as described (12). Following gel electrophoresis, immunoprecipitated proteins were identified by immunoblotting (12). Bound antibodies were detected by enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech).
Analysis of A␤40 and A␤42-Culture medium (2 ml) was collected from confluent K293 cells grown in six-well dishes for 24 h. The medium was assayed for A␤40 and A␤42 using a previously described ELISA assay (12). A␤ peptides were immunoprecipitated using antibody 6E10 (Senetek) according to established protocols (21).
Transgenic Lines of Caenorhabditis elegans and Rescue Assays-Expression constructs were generated as described previously (22). Transgenic lines were established by microinjection of plasmid DNA mixtures into the C. elegans germ line to create extrachromosomal arrays as described previously (22). All plasmids used in this study were injected at a concentration of 20 ng/l into sel-12(ar171) or sel-12(ar171) unc-1(e538) hermaphrodites along with ttx-3:GFP as a cotransformation marker. Successful transformation was monitored by the expression of GFP in the AIY interneurons of F1 and F2 generation animals. Four independent lines from the progeny of F2 generation animals were established.
As the sel-12(ar171) animals never lay eggs (23), rescue of the sel-12 defect can be quantified by scoring egg laying behavior in transgenic animals (22,24). Consequently, for each transgenic line, we examined 50 transgenic animals for their ability to lay eggs and determined their brood size. The number of eggs laid by individual transgenic animals was counted and placed into four categories: "Eglϩϩϩ," robust egg laying, more than 30 eggs laid (wt phenotype); "Eglϩϩ," 15-30 eggs laid; "Eglϩ," 5-15 eggs laid; "EglϪ," no eggs laid.

RESULTS
The PS1⌬exon9 mutation changes a G to a T at the splice site for exon 9. It therefore destroys the minimal required consensus sequence for the splice acceptor site (8), which then results in an aberrant deletion of the domain encoded by exon 9 (Ref. 8; Fig. 1A). However, the mutation also changes codon 290 at the exon 8/10 splice junction from a serine to a cysteine (Ref. 8; Fig. 1A). We therefore have investigated if the single amino acid exchange or the deletion of the domain required for pro-teolytic processing is responsible for the pathological activity of the PS1⌬exon9 mutation in regard to A␤42 generation. For this purpose we mutagenized amino acid 290 of PS1⌬exon9 to a serine (PS1⌬exon9 C290S), thus correcting the point mutation (Fig. 1A). cDNAs encoding PS1⌬exon9 and PS1⌬exon9 C290S were stably transfected into kidney 293 cells overexpressing ␤APP containing the Swedish mutation. This cell line was used previously to determine the pathological effect of PS mutations on A␤ production (12,18). Untransfected K293 cells expressing endogenous PS, as well as cell lines overexpressing PS1⌬exon9 and PS1⌬exon9 C290S were pulse-labeled with [ 35 S]methionine for 1 h, and cell lysates were immunoprecipitated with antibody 3027 to the large loop of PS1. Both PS1⌬exon9 and PS1⌬exon9 C290S accumulated as full-length proteins (Fig.  1B). As reported previously (9, 10) very little endogenous fulllength PS1 could be detected in the untransfected cell line.
It has been shown before that overexpression of PS proteins results in the replacement of endogenous PS fragments and the accumulation of PS1⌬exon9 prevents the formation of stable PS fragments (9,11,12). To prove that overexpression of PS1⌬exon9 C290S still allows the reduction of endogenous PS fragments, cell lysates from unlabeled cells were immunoprecipitated with antibodies specific to the PS1 or PS2 loop domain (see "Experimental Procedures"), and PS fragments were identified with the PS1/PS2-specific monoclonal antibodies BI.3D7 and BI.HF5C (see "Experimental Procedures"). Whereas cell

FIG. 2. The pathological activity of PS1⌬exon9 is due to a point mutation at codon 290.
A, A␤ production of K293 cells expressing endogenous presenilins (control (Ctrl.)) and cell lines overexpressing PS1⌬exon9 or PS1⌬exon9 C290S. Conditioned medium from metabolically labeled kidney 293 cells expressing the indicated presenilin variants was immunoprecipitated with antibody 6E10. A␤40 and A␤42 were separated on a previously described gel system, which allows the specific resolution of both species (25). As reported before, A␤42 migrates faster as A␤40 (25). Note that cells expressing PS1⌬exon9 produced increased amounts of A␤42, whereas cells expressing PS1⌬exon9 C290S do not show such an increase. Two individual immunoprecipitations of each cell lysate are shown. B, quantitation of the A␤42 and A␤40 concentrations in conditioned medium of kidney 293 cells expressing the indicated presenilin variants using a previously described highly specific ELISA (12). Expression of PS1⌬exon9 but not the expression of PS1⌬exon9 C290S results in a 3-fold increase of A␤42 production. Identical results were observed with independent cell clones.
lines expressing endogenous presenilins produced PS1 as well as PS2 C-terminal fragments (Fig. 1, C and D), overexpression of PS1⌬exon9 and PS1⌬exon9 C290S inhibited the formation of endogenous PS1 and PS2 fragments (Fig. 1, C and D). This demonstrates that PS1⌬exon9 C290S, like PS1⌬exon9, fully replaced presenilins derived from the endogenous genes and also proves that PS1⌬exon9 C290S does not undergo proteolytic processing but rather accumulates as an uncleaved fulllength protein.
In order to investigate the pathological activity of the mutant PS derivatives on A␤ production, control cells as well as cells overexpressing PS1⌬exon9 and PS1⌬exon9 C290S were metabolically labeled with [ 35 S]methionine, and A␤40 and A␤42 peptides were immunoprecipitated from the conditioned medium using antibody 6E10. This antibody is raised to A␤1-17 and therefore recognizes A␤40 as well as A␤42 (see "Experimental Procedures"). Isolated A␤ peptides were then separated on a previously described gel system, which allows the specific resolution of A␤40 and A␤42 (25). As shown in Fig. 2A, cells overexpressing PS1⌬exon9 secreted elevated levels of A␤42 as compared with control cells. In contrast, the cell line stably transfected with PS1⌬exon9 C290S produced significantly lower amounts of A␤42 as cells expressing the Alzheimer's disease-associated PS1⌬exon9 mutation. This suggests that the pathological activity of the PS1⌬exon9 mutation is due to the point mutation generated at the aberrant splice junction. In order to quantitate A␤42 and A␤40 production, we used a previously described specific ELISA (12). In this assay, two highly specific monoclonal antibodies are used for the discrimination of both peptide species (12). As reported before (12,18,26), the PS1⌬exon9 mutation results in an approximately 3-fold increase of the A␤42/A␤ total ratio (Fig. 2B). In contrast, the cell line stably overexpressing PS1⌬exon9 C290S showed no increased A␤42/A␤ total ratio (Fig. 2B). Therefore using two different approaches we can show that the pathological activity of PS1⌬exon9 is independent of its lack of proteolytic processing but is rather caused by the single amino acid change at the aberrant splice junction.
Having demonstrated that PS1⌬exon9 C290S is pathologically inactive, we also wanted to test if this protein is sufficient to rescue a lin-12-mediated signaling defect that is a result of a mutant PS homologue (sel-12) in Caenorhabditis elegans (23). As reported previously, transgenic expression of human presenilins rescues the phenotype caused by mutations of sel-12 (22,24). In contrast, transgenic expression of the PS1⌬exon9 variant in C. elegans resulted in an incomplete rescue of the sel-12 mutant phenotype, as indicated by a significantly reduced brood size and reduced egg laying (Refs. 22 and 24; see also Table I). In order to assess the in vivo function of PS1⌬exon9 C290S, we tested its ability to rescue the putative sel-12 null allele ar171 (23). Four independent transgenic strains expressing PS1⌬exon9 C290S displayed robust egg laying (Table I).
Based on both numbers of laid progeny and numbers of eggs in utero, the phenotype of these strains is almost indistinguishable of that of wild type worms (Table I). These results demonstrate that PS1⌬exon9 C290S rescues the egg laying phenotype of the sel-12(ar171) mutant animals significantly better than PS1⌬exon9 (Table I). Therefore, the reduced biological activity of PS1⌬exon9 is also due to a single point mutation and completely independent of the lack of proteolytic processing. DISCUSSION Our results demonstrate that the pathological effect of the PS1⌬exon9 mutation is independent of its lack of proteolytic cleavage and the deletion of the exon 9-encoded domain. Reverting the cysteine residue generated by the exon 8/10 splice junction at codon 290 back to its wt residue (serine) still inhibits its proteolytic processing, causes an accumulation of the uncleaved protein, and prevents formation of endogenous PS fragments. Although the artificially generated PS1⌬exon9 C290S variant behaves like PS1⌬exon9 in regard to the characteristic biochemical features described above, it does not allow pathological overproduction of A␤42. Moreover, PS1⌬exon9 C290S regains full biological activity in a very sensitive in vivo assay system. This demonstrates that not only the pathological overproduction of A␤42 but also the reduced biological function is due to the single amino acid exchange. Therefore, similar to all other known PS mutations, the pathological effect of the PS1⌬exon9 mutation is due to a rather subtle amino acid exchange at a single highly conserved codon whereas the large deletion of the complete domain encoded by exon 9 does not affect the biological and pathological function. However, consistent with our previous results (27), we suggest that PS molecules lacking the domain encoded by exon 9 mimic a proteolytically processed and biologically active PS complex and therefore rescue the phenotype of the mutant nematode. Consequently, the rescuing activity of PS1⌬exon9 is significantly enhanced when the pathologically relevant point mutation at codon 290 is reverted to the wt residue.
It remains to be shown if the mutation at codon 290 is also pathologically active within the full-length protein like all other AD-associated point mutations. It may however be possible that the mutation only exhibits a pathological activity if it is aberrantly flanked by the domains encoded by exons 8 and 10. Structural changes that may specifically occur in the PS1⌬exon9 molecule (12,19) might be responsible for the pathological activity of this very unusual mutation.
Based on our results it would be interesting to investigate if the point mutations in the PS genes require the endoproteolytic cleavage for their pathological activity. Therefore artificial PS molecules should be generated containing an AD associated mutation as well as the smallest possible alteration of the sequence at the cleavage site which would inhibit PS processing. However, such mutations might be very difficult to gener-TABLE I Rescue of the sel-12 egg laying defect by human PS genes expressed from the sel-12 promoter For 50 transgenic animals each, the numbers of progeny were counted and grouped in the following categories: ϩϩϩ, over 30 progeny laid by individual animal; ϩϩ, 15-30 progeny laid; ϩ, 5-15 progeny laid; Ϫ, no progeny laid. The sel-12(ar171) strains carried an additional unc-1(e538) marker that did not affect egg laying or rescuing frequency. It was backcrossed several times from the strain originally published (23 ate since PS can be cleaved at multiple sites (10,12) and PS molecules containing larger deletions at the cleavage sites might mimic a proteolytically processed PS molecule (27).