A Loss of Function Mutant of the Presenilin Homologue SEL-12 Undergoes Aberrant Endoproteolysis in Caenorhabditis elegans and Increases Aβ42 Generation in Human Cells*

The familial Alzheimer's disease-associated presenilins (PSs) occur as a dimeric complex of proteolytically generated fragments, which functionally supports endoproteolysis of Notch and the β-amyloid precursor protein (βAPP). A homologous gene, sel-12, has been identified in Caenorhabditis elegans. We now demonstrate that wild-type (wt) SEL-12 undergoes endoproteolytic cleavage in C. elegans similar to the PSs in human tissue. In contrast, SEL-12 C60S protein expressed from thesel-12(ar131) allele is miscleaved in C. elegans, resulting in a larger mutant N-terminal fragment. Neither SEL-12 wt nor C60S undergo endoproteolytic processing upon expression in human cells, suggesting that SEL-12 is cleaved by aC. elegans-specific endoproteolytic activity. The loss of function of sel-12 in C. elegans is not associated with a dominant negative activity in human cells, because SEL-12 C60S and the corresponding PS1 C92S mutation do not interfere with Notch1 cleavage. Moreover, both mutant variants increase the aberrant production of the highly amyloidogenic 42-amino acid version of amyloid β-peptide similar to familial Alzheimer's disease-associated human PS mutants. Our data therefore demonstrate that the C60S mutation in SEL-12 is associated with aberrant endoproteolysis and a loss of function in C. elegans, whereas a gain of misfunction is observed upon expression in human cells.

FAD-associated PS mutations are thought to gain a pathological misfunction in the endoproteolytic processing of ␤APP. Similar to the ␤APP mutations, PS mutations result in the enhanced production of the highly amyloidogenic 42-amino acid variant of amyloid ␤-peptide (A␤42) (6).
PSs are not only involved in the aberrant A␤ production in rare FAD cases but are also required for physiological A␤ production. A PS1 gene knock-out inhibits A␤ production and results in the accumulation of C-terminal ␤APP fragments, which are thought to be the immediate precursors for the ␥-secretase cleavage (19). The inhibition of ␥-secretase cleavage indicates that PSs are directly involved in endoproteolysis of ␤APP. This is supported by the finding that two aspartate residues located within the putative TM domains 6 and/or 7 of PS1 and PS2 are critically required for A␤ production (20 -24). Moreover, Wolfe et al. (25) hypothesized that PSs are aspartyl proteases and therefore identical to the ␥-secretase, which is strongly supported by the photoaffinity labeling of PS1 and PS2 (26,27).
PSs do not only support the ␥-secretase cleavage of ␤APP but also a similar cleavage of Notch (28 -30) and Ire1 (31). FADassociated mutations (32) as well as mutagenesis of residue 286 of PS1 to charged amino acids (33) and mutations of the aspartates located in TM6 and TM7 severely reduce endoproteolysis of Notch (21,34,35). Further evidence for a function of PSs in Notch signaling is also provided by a knock-out of the PS1 gene (28,36,37), which results in a developmental phenotype similar to the Notch knock-out. Moreover, genetic evidence indicates that the PS homologous gene sel-12 of the nematode Caenorhabditis elegans is also directly involved in Notch signaling, because two mutations in sel-12 reduce the activity of a hyperactive allele of lin-12, the C. elegans Notch homologue (38). These sel-12 alleles result in an egg-laying defect (38) and a functional defect of two neurons involved in the animal's temperature memory (39). The additional deletion of the second C. elegans PS homologue, hop-1, strongly enhances the sel-12 phenotype and results in sterility or early embryonic lethality, depending on the maternal contributions of either hop-1 or sel-12 (39 -41). This is similar to the finding in mice where the additional ablation of the PS2 gene leads to a full Notch phenotype (42,43). Both the egg-laying phenotype and the neuronal defects in C. elegans can be fully rescued by overexpression of wild-type (wt) PS1 and PS2 (44 -47). However, FAD-associated PS1 or PS2 mutants exhibit only a weak activity in this genetic background, suggesting that human FAD mutations exhibit a reduced function rather than a dominant negative function (39,44,45). On the other hand, FADassociated PS1 mutations fully rescue the developmental deficits of the PS1 Ϫ/Ϫ mice (48,49).
Although sel-12 function is genetically well understood (38,39,41,44,45,50), little is known about the biochemical abnormalities, which on the molecular level interfere with sel-12 activity. We therefore analyzed SEL-12 expression and investigated its endoproteolysis as well as its function in proteolytic processing of ␤APP and Notch. Our data indicate that a loss of sel-12 function is associated either with a severe truncation of the resulting protein (sel-12(ar171)) or a defect in endoproteolysis (sel-12(ar131)). Moreover, the sel-12 loss of function mutation sel-12 C605 causes a gain of misfunction upon expression in human cells by increasing aberrant A␤42 generation.
Construction of cDNAs-The cDNAs encoding PS1 C92S and sel-12 C60S were constructed by polymerase chain reaction mediated mutagenesis of codon 92 of PS1 cDNA and codon 60 of sel-12 cDNA (GeneBank AF171064) using appropriate primers (51). The mutant cDNAs were cloned into the expression vector pcDNA3.1/zeo(ϩ) (Invitrogen) and sequenced to verify successful mutagenesis.
Preparation of Protein Extracts from Cultured Cells and C. elegans Hermaphrodites-Cells were lysed in 10ϫ radioimmune precipitation buffer containing 10% Triton X-100, 5% deoxycholic acid (Sigma), and 1% SDS, and diluted 10-fold by PBS containing a protease inhibitor mix (Sigma). 150 mg of frozen worms were resuspended in 300 l of 10ϫ radioimmune precipitation buffer. The suspension was sonicated for 15 s ϫ 3 times on ice. Upon sonication, 600 l of PBS supplemented with protease inhibitors and 600 mg of glass beads (B. Braun, Melsungen, Germany) were added, and the suspension was incubated for 30 min under constant agitation at 1000 rpm/4°C. After removal of the glass beads, the lysate was sonicated again for 10 s ϫ 2 times on ice. Insoluble material was removed by centrifugation (15,000 ϫ g at 4°C). Then, 2.1 ml of PBS containing protease inhibitors was added to the lysate, and SEL-12 was immunoprecipitated as described (51).
Combined Immunoprecipitation/Western Blotting-Cell lysates or worm extracts were immunoprecipitated using the indicated antibodies. Following gel electrophoresis, immunoprecipitated SEL-12 deriva-tives were identified by immunoblotting using antibody SA6848. Human PSs were identified by immunoblotting using the monoclonal antibody PS1N (to detect the PS1 NTF), BI.3D7 (to detect the PS1 CTF), or BI.HF5c (to detect the PS2 CTF). Bound antibodies were detected by enhanced chemiluminescence using standard procedures (ECL, Amersham Pharmacia Biotech).
CNBr Digest-Proteolytic fragments of PS1 or SEL-12 were purified by immunoprecipitation. Immunoprecipitated SEL-12 and PS derivatives were identified by immunoblotting. Adjacent bands were excised from the gel and incubated in 70% (v/v) formic acid with or without 500 l of 80 mg/ml cyanogen bromide (CNBr) overnight at 4°C (54). The solution was dried in a Speed-Vac, redissolved in 50 l of H 2 O, and dried again. The dried peptides were dissolved in SDS sample buffer and separated by SDS-polyacrylamide gel electrophoresis. SEL-12 fragments were identified by immunoblotting.
Analysis of ␤APP Metabolites-Stably transfected HEK 293 cell lines were grown to confluence. For the analysis of A␤ and p3 in conditioned media, cells were metabolically labeled with 300 Ci of [ 35 S]methionine (Promix, Amersham Pharmacia Biotech) as indicated. A␤ species and p3 were immunoprecipitated from conditioned media with antibody 3926 (52). A␤1-40 and A␤1-42 were separated on previously described Tris-Bicine gels (56).

RESULTS
Aberrant Endoproteolysis of SEL-12 C60S in C. elegans-So far, the expression and endoproteolysis of SEL-12 in the nematode C. elegans has not been studied in detail. We therefore first analyzed endoproteolysis of wt SEL-12 in the nematode. Protein extracts from wt worms (N2) were prepared and SEL-12 derivatives were isolated by immunoprecipitation using two independent polyclonal antibodies (SA6848 and 322), which were raised against the large cytoplasmic loop of SEL-12 ( Fig. 1a). Precipitated SEL-12 derivatives were identified by immunoblotting using antibody SA6848. As a negative control we also analyzed protein extracts from HEK 293 cells transfected with human PS1 as well as worms expressing the sel-12(ar171) allele, which results in a premature stop of sel-12 translation (W225stop) and should therefore not give rise to any specific translation product detected by the antibodies used. Although, indeed, no specific sel-12 translation products were identified in C. elegans expressing the sel-12(ar171) allele ( Fig. 1b), we detected an approximately 24-kDa CTF in the wt worms (N2) (Fig. 1b). Endoproteolysis of SEL-12 is consistent with the findings that PSs from all other species analyzed, including mice (11), zebrafish (18), and Drosophila (57,58), are proteolytically processed as well. When we analyzed protein extracts derived from worms expressing the sel-12(ar131) allele (SEL-12 C60S), we surprisingly did not detect SEL-12 C-terminal fragments of similar molecular weight as observed in the wt worms. Instead of the ϳ24-kDa CTF obtained in the wt N2 worms, a novel fragment of ϳ32 kDa was observed (Fig. 1b). In parallel the holoprotein appeared to accumulate as well. This indicates that SEL-12 C60S is not efficiently processed in the worm and that an alternative processing activity leads to an aberrant cleavage.
Alternative Cleavage of SEL-12 C60S Occurs Close to TM7-Based on the findings described in Fig. 1b, it was difficult to predict the site of alternative cleavage. Alternative cleavage could either occur further C-terminal producing a larger NTF or further N-terminal producing a larger CTF. In both cases the epitopes for antibodies SA6848 and 322 would be retained in the novel proteolytic product. To discriminate between these two possibilities, we performed a cyanogen bromide (CNBr) digest of isolated SEL-12 fragments. Because no methionine residues are observed within the cytoplasmic loop, we predicted that CNBr should not further digest the wt SEL-12 CTF but may result in additional cleavage products upon incubation with the larger SEL-12 C60S fragment (Fig. 2a). If the miscleavage would produce a larger NTF due to a C-terminal cleavage close to TM7, CNBr treatment would create a ϳ17-kDa fragment (corresponding to a peptide between Ala-203 and TM7). In contrast, a larger fragment would be produced by CNBr digestion (between Ala-203 and the C terminus of SEL-12) if endoproteolysis of SEL-12 C60S would occur further N-terminal (Fig. 2a).
SEL-12 loop fragments derived from the wt gene as well as from the sel-12(ar131) allele were isolated by immunoprecipitation (Fig. 2b), separated on Tris-Tricine gels and visualized by immunoblotting using antibody SA6848. Bands of interest were cut out of the gel and subsequently digested with CNBr. Similar experiments were carried out with isolated NTF and CTF of human PS1 (Fig. 2b). As shown in Fig. 2b, CNBr treatment did not reveal additional fragments, when SEL-12 fragment from the wt N2 worm was digested. This was fully confirmed by CNBr digestion of the human PS1 CTF, whereas the human PS1 NTF was sensitive to CNBr as expected (Fig.  2b). In contrast, the alternative SEL-12 proteolytic fragment gave rise to an ϳ17-kDa species (Fig. 2b), which was completely absent upon digestion of the SEL-12 fragment derived from the wt gene (Fig. 2b). Based on its molecular mass, the 17-kDa peptide may correspond to a CNBr fragment starting at Ala-203 within the predicted TM5 (Fig. 2a). These results demonstrate that the alternative SEL-12 fragment in mutant sel-12(ar131) worms is generated by a cleavage further C-terminal to the physiological cleavage site thus giving rise to a larger N-terminal fragment (Fig. 2c).
The Alternative Cleavage Is Not Produced by Caspases-It was previously shown that caspases can mediate alternative cleavage of PS1 and PS2 (59 -62). SEL-12 indeed contains three potential caspase cleavage sites after the aspartate residues 276, 284, and 345 in the large cytoplasmic loop (Fig. 3a). Furthermore, FAD-associated mutations may induce apoptosis (63), and the SEL-12 C60S mutation behaves very similar to FAD mutations (see below). We therefore investigated whether caspases could generate the alternative fragment observed in the SEL-12 C60S worms. If SEL-12 is cleaved at Asp-345, an in vitro digest of isolated SEL-12 holoprotein could generate a fragment co-migrating with the alternative fragments observed in the sel-12(ar131) worms (Fig. 3a). On the other hand, if SEL-12 C60S is cleaved at Asp-276 or Asp-284, the SEL-12 loop antibodies should recognize a caspase-generated CTF, which migrates considerably faster than the regular CTF of SEL-12 (Fig. 3a). Based on previous findings (55, 59 -61) caspase-3 was chosen for the in vitro assay. As substrates we used the wt SEL-12 and the SEL-12 C60S holoproteins isolated from HEK 293 stably transfected with the sel-12 cDNA (see next paragraph). As shown in Fig. 3 (b and c), cleavage of wt SEL-12 as well as SEL-12 C60S by caspase-3 resulted in an alternative CTF with a lower molecular mass than the CTF derived from wt worms. Therefore, these data suggest that caspase-3 cleavage of SEL-12 occurs after either Asp-276 or Asp-284 (see Fig.  3e). To confirm this finding under in vivo conditions, we induced apoptosis in HEK 293 expressing PS1 or SEL-12 with 1 M staurosporine (STS). To control STS-induced apoptosis we also monitored the production of the previously described (62) FIG. 1. Endoproteolysis of SEL-12. a, schematic diagram of SEL-12 and the antigens (bars) used to generate the worm-specific antibodies. The arrow denotes the expected conventional cleavage site (8,9). b, expression and endoproteolysis of SEL-12 in C. elegans. Lysates from wt PS1-transfected HEK 293 cells or worms (N2, Bristol Strain N2; ar131, sel-12 C60S; ar171, sel-12 W225stop) (38) were subjected to immunoprecipitation/ Western blotting with the indicated antibodies. Lysates from wt PS1-overexpressing cells did not show any specific bands, demonstrating no cross-reaction of SEL-12-specific antibodies with human PS1. Lysates from the sel-12(ar171) worms showed no significant SEL-12 derivatives consistent with a premature translational stop at W225. The corresponding NTF cannot be recognized by the antibody SA6848 (see "Discussion"). Note that immunoprecipitation with two independent antibodies, SA6848 and 332, obtained very similar results. The asterisk corresponds to IgG light chains.
alternative CTF of PS1 (Fig. 3d, lower panel). Under conditions where STS-induced apoptosis led to the generation of the caspase-generated alternative PS1 CTF, a prominent approximately 20-kDa CTF occurred in HEK 293 cells expressing SEL-12. This fragment co-migrated with a SEL-12-derived CTF, which was generated by caspase cleavage in vitro (Fig.  3d, upper panel). These results are consistent with previous results on caspase-generated CTFs of human PSs (59,(62)(63)(64). Because the caspase-generated fragment identified by antibody SA6848 exhibits a considerably different molecular mass as the alternative fragments observed in the sel-12(ar131) worms (Fig. 3c), these data demonstrate that the alternative cleavage in the worm is not mediated by caspases induced during apoptosis ( Fig. 3e; see also Fig. 2c). (Note that only the caspase cleavage of SEL-12 is observed in human cells; see next paragraph.) Lack of Endoproteolysis of SEL-12 in Human Cells-To test if the miscleavage of SEL-12 can be observed upon expression in human cells, the wt sel-12 cDNA as well as the cDNA encoding sel-12 C60S were stably transfected into HEK 293 cells. This cell line was previously used in many studies (i.e. Refs. 21,33,35,65) to investigate endoproteolysis of PSs as well as their function in Notch and ␤APP endoproteolysis. To mimic the SEL-12 C60S mutation in human PS1, we introduced the corresponding mutation at the conserved codon 92 (PS1 C92S; Fig. 4a) and generated cell lines stably expressing this cDNA construct.
PS1 and SEL-12 derivatives were identified by a combined immunoprecipitation/Western blotting protocol. Surprisingly, this revealed no detectable endoproteolysis of SEL-12 (wt or C60S) in human cells, although high levels of the holoprotein were observed (Fig. 4b, first panel). Because worms are grown at 15-25°C, the lack of SEL-12 endoproteolysis may be due to aberrant folding of SEL-12 at 37°C. To test if that could be the case, HEK 293 cells were grown at 25°C for 24 h. However, under these conditions endoproteolysis of SEL-12 was still not M and X denote methionine residues and the predicted conventional cleavage site, respectively. Note that no methionine residues occur in the SEL-12 CTF. b, CNBr digestion of PS1 and SEL-12 fragments. Lysates from the indicated cells/worms were immunoprecipitated with the indicated antibodies (upper panel). Bands of interest were recovered and incubated in the absence (Ϫ) or presence (ϩ) of CNBr. CNBr-sensitive/resistant bands were recovered by Western blotting using the indicated antibodies (lower panel). As expected, the PS1 NTF was sensitive to CNBr. However, only limited digestion was observed, which may be due to aggregation of the highly hydrophobic cleavage products. In contrast, the PS1 CTF is largely resistant to CNBr digestion. Note that the PS1 CTF contains 1 to 3 methionine residues depending on its cleavage site at Met-292, Met-298, or Met-457 (69,73). Because these residues are either very close to the N or C termini, CNBr cleavage at these sites will not substantially affect the molecular mass of the remaining peptide. The SEL-12 CTF is resistant to CNBr digestion as expected. Note that a ϳ15-kDa fragment was generated by CNBr treatment from the mutant SEL-12 C60S peptide, which is not observed upon digestion of the wt SEL-12 CTF. Arrowheads denote the bands which co-migrate with the original band cut out from polyacrylamide gels. c, schematic diagram of endoproteolysis of wt SEL-12 and SEL-12 C60S. The asterisk indicates the C60S mutation. observed (Fig. 4b, second panel). In contrast, human PS1 did undergo sufficient endoproteolysis under these conditions (Fig.  4b, third panel). This suggests that SEL-12 undergoes wormspecific endoproteolysis and indicates that human cells lack a component required for SEL-12 cleavage. Furthermore, introduction of the corresponding C92S mutation into human PS1 led to the generation of a C-terminal fragment co-migrating with PS1 CTFs derived from the wt cDNA (Fig. 4c). This also indicates that the failure in the endoproteolysis of mutant SEL-12 is worm-specific and can not be observed upon expression of the corresponding PS1 mutation in human cells.
We also analyzed whether ectopic expression of SEL-12 results in the replacement of endogenous PS fragments. Replacement of endogenous PS fragments is a widely observed phenomenon and thought to be an indication of a stable expression of the ectopic PS variant (11,16). We observed that SEL-12 could at least partially replace endogenous PS1 and PS2 fragments, because derivatives of both endogenous PSs are reduced ( Fig. 4c; see also Fig. 4b, lower panel). However, sel-12-mediated replacement of endogenous PSs is not as efficient as the replacement caused by ectopic PS expression (Fig. 4c).
SEL-12 C60S and PS1 C92S Facilitate Notch1 Endoproteolysis in Human Cells-Because SEL-12 C60S results in reduced Notch signaling in the sel-12(ar131) allele, we next analyzed whether SEL-12 C60S or the corresponding human PS1 C92S mutation interfere with Notch1 endoproteolysis in hu-man cells in a dominant negative manner. Cells expressing wt SEL-12, SEL-12 C60S, endogenous PSs, PS1 C92S, or the FAD-associated PS1 L286V were co-transfected with the Notch1 ⌬E derivative described previously (21,28,33,35,53). In pulse-chase experiments we then followed the production of the Notch1 intracellular cytoplasmic domain (NICD). Cells were metabolically labeled for 15 min with [ 35 S]methionine and chased for 60 min in the presence of excess amounts of unlabeled methionine. A 60-min cold chase was chosen, because we and others previously found efficient NICD formation at this time point (21,33,66). Interestingly, SEL-12 C60S as well as PS1 C92S efficiently supported NICD formation like all other PS derivatives (Fig. 5; left panel). From these results we conclude that, in contrast to the loss of function caused by the active site mutation (Fig. 5; right panel), the cysteine to serine mutations in TM1 do not interfere with Notch1 endoproteolysis in human cells. Moreover, neither accumulation of ␤APP CTFs nor decreased secretion of A␤ was observed in cells expressing SEL-12 C60S or PS1 C92S (data not shown). Therefore, these mutations do not interfere with Notch1 and ␤APP endoproteolysis in a dominant negative manner.
A Gain of Misfunction in Human Cells-The C60S mutation occurs at a highly conserved amino acid residue (Fig. 4a). This point mutation is therefore very similar to almost all FADassociated point mutations, which also occur at evolutionary conserved residues and involve chemically rather subtle amino FIG. 3. Caspase cleavage of wt SEL-12 and SEL-12 C60S. a, schematic diagram of SEL-12 and the aspartic residues within and close to the hydrophilic loop. Three candidate aspartic residues for caspase cleavage were identified, which may give rise to the alternative cleavage product. b, in vitro digestion of wt SEL-12 and SEL-12 C60S by caspase-3. Lysates of cells expressing wt SEL-12 or SEL-12 C60S (see also Fig. 4) were immunoprecipitated with SA6848 to isolate SEL-12 holoproteins. Immunoprecipitates were incubated with (ϩ) or without (Ϫ) caspase-3. Following proteolytic digestion, the fragments were detected by immunoblotting using the same antibody. Successful digestion by caspase-3 was monitored by the generation of PS1 caspase fragment from PS1 CTF (55) (data not shown). c, the SEL-12 caspase fragment does not co-migrate with the alternative fragment observed in sel-12(ar131) worms. Caspase-3-digested SEL-12 as well as lysates from sel-12(ar131) mutant or wt worms were immunoprecipitated with antibody SA6848. Precipitated SEL-12 derivatives were identified by immunoblotting. d, in vivo and in vitro produced SEL-12 fragments co-migrated. HEK 293 cells expressing wt PS1 or SEL-12 were treated with (ϩ) and without (Ϫ) 1 M STS to induce apoptosis. SEL-12 derivatives were identified by immunoprecipitation/immunoblotting. As a control, SEL-12 was digested with caspase-3 in vitro. Under the same conditions STS treatment resulted in the production of the previously described (62) alternative PS1 CTF (lower panel). e, schematic diagram of caspase-mediated cleavage of SEL-12. Compare with Fig. 2c. Asterisks indicate degradation products stabilized by ALLN treatment. acid exchanges (67). We therefore investigated whether the PS1 C92S mutation, which corresponds to the C60S mutation of the worm (see above) exhibits a pathological function in terms of increased A␤42 generation. Conditioned media from metabolically labeled cells stably transfected with wt PS1, the FAD-associated PS1 L286V mutation, and the PS1 C92S mutation were collected and immunoprecipitated with antibody 3926. This antibody immunoprecipitates all A␤ species, including A␤40 and A␤42 (47). Immunoprecipitates were separated on a previously described gel system, which allowed the specific detection and quantitation of both A␤ species (56). Consistent with previous results (33,68) this revealed that the FADassociated PS1 L286V mutation increased the A␤42/A␤40 levels (Fig. 6a). Interestingly, the PS1 C92S mutation caused the production of very high levels of A␤42. Quantitation revealed an increase of approximately 3-fold (Fig. 6a). The data were further confirmed by a previously described enzyme-linked immunosorbent assay (17,18,21,33,35,47,69) (data not shown). Therefore, the C60S homologous mutation in human PS1 behaves like a FAD-associated mutation. We next analyzed whether wt SEL-12 or the SEL-12 C60S mutation affects A␤42 generation in human cells. Interestingly, an increased level of A␤42 was observed upon expression of the wt cDNA. This was further elevated upon the expression of the C60S mutation (Fig. 6a). Increased A␤42 production driven by wt SEL-12 indicated that wt SEL-12 has a pathological activity in human cells. Similar to our work on zebrafish PS1 (18), this appears to be due to several different amino acids within SEL-12 protein at positions corresponding to previously identified FAD-associated point mutations (Fig. 6b). DISCUSSION Experiments to rescue the egg-laying phenotype of C. elegans (sel-12(ar131), sel-12(ar171)) are now frequently used to test the biological activity of human PSs (44,45,47,50). However, very little is known about the molecular mechanisms of the sel-12 mutant alleles. Specifically, endoproteolysis of SEL-12, the PS homologue in the worm has so far not been investigated. Because endoproteolysis of human PS proteins appears to be an indication for functional expression and complex formation (11,13,14,17,70), we now studied the expression of wt SEL-12, SEL-12 C60S (sel-12(ar131)) and SEL-12 W225stop (sel-12(ar171)) in C. elegans.
SEL-12 undergoes endoproteolysis in C. elegans very similar to presenilin homologues of other species (18,57,58). As observed in other species, we found high levels of a SEL-12 CTF and only low amounts of the corresponding holoprotein. Expression of the sel-12(ar171) allele, which results in a premature translational stop at W225 in TM6, produces a truncated derivative that corresponds to a truncated NTF as suggested before (38). Because we and others have shown previously that such a fragment is unstable and biologically inactive, the W225stop mutation may correspond to a functional knock-out (17,71,72). In contrast the sel-12(ar131) allele is robustly expressed. However, we surprisingly found that the C60S mutation inhibited endoproteolysis at the wt cleavage site. We identified a larger NTF, which suggests that the aberrant endoproteolytic cleavage occurs much further C-terminal to the conventional cleavage site, most likely close to TM7. We had sequenced the sel-12 coding region in sel-12(ar131) worms and tested the correct mRNA length by reverse transcription-polymerase chain reaction (data not shown). Therefore, the aberrant fragment is not due to additional mutations that may result in alternative mRNA splicing. In addition, the sel-12(ar131) worms had been backcrossed extensively to remove FIG. 4. Expression and endoproteolysis of PS1 C92S, wt SEL-12 and SEL-12 C60S in human cells. a, sequence alignment of the TM1 region of SEL-12 and PS1/PS2. Based on the significant sequence similarity, Cys-60 of SEL-12 corresponds to Cys-92 of human PS1. Boldface letters denote conserved sequences. b, expression and endoproteolysis of wt SEL-12 and SEL-12 C60S in HEK 293 cells. Cells were grown at 37°C (first panel) or 25°C (second and third panel). Aliquots of the lysates were subjected to immunoprecipitation/Western blotting with SA6848 to detect SEL-12 derivatives (first and second panel). To identify human PS1 derivatives, cell lysates were immunoprecipitated with antibody 3027 and immunoblotted with BI.3D7 (third panel). c, replacement of endogenous PSs by overexpressed SEL-12. PS1 derivatives were identified by a combined immunoprecipitation/immunoblotting protocol using antibodies 3027/BI.3D7 (upper panel). PS2 derivatives were identified by a combined immunoprecipitation/immunoblotting protocol using antibodies 3711/BI.HF5c (lower panel). Note that overexpression of SEL-12 reduces expression of endogenous PS1 and PS2. background mutations in other genes.
The alternative cleavage is remarkable, because the SEL-12 C60S mutation occurs far away from the site of endoproteolytic processing. This suggests that the mutation affects the structure of SEL-12, thus blocking the conventional cleavage site but making a secondary alternative site available for aberrant proteolytic processing. Interestingly, a very similar phenomenon was observed for an envelope protein of spleen necrosis virus (73). Several independent loss-of-function mutations located N-terminal from the conserved retroviral cleavage site of the protein induce aberrant endoproteolysis at a secondary site.
When sel-12 cDNAs were expressed in human cells, we did not observe any detectable endoproteolysis. It should be emphasized that heterologous overexpression of zebrafish PS1 in human cells allowed normal endoproteolysis (18). Interestingly, the SEL-12 holoprotein was active in A␤ generation, because wt SEL-12 as well as SEL-12 C60S increased the levels of A␤42 generation very similar to the FAD-associated point mutation PS1 L286V. This is remarkable, because recently it was claimed that the uncleaved holoprotein of PSs is a proteolytically inactive zymogene (26). However, together with previous findings (69,76) this appears to be unlikely, because at least some uncleaved PS derivatives can support aberrant A␤42 generation and do not inhibit Notch endoproteolysis in a dominant negative manner like the active site mutations (35,69,23). The failure of SEL-12 to undergo endoproteolysis in human cells may be due to the lack of sequence conservation at the endoproteolytic cleavage site (Fig. 7; Refs. 69,74,75). This suggests that the endoproteolytic activity of human cells does not recognize SEL-12. However, because SEL-12 induces A␤42 generation in human cells (and is therefore pathologically active), our data suggest that endoproteolysis of PSs and ␥-secretase activity are not necessarily correlated.
The C60S mutation occurs at a highly conserved position very similar to the FAD-associated PS1 mutation. Indeed, the introduction of the serine to cysteine mutation at the homologous position of human PS1 resulted in a significant increase of A␤42 generation, as it is observed in all FAD-associated PS mutations. One may therefore speculate that the PS1 C92S mutation could be found at some point in a family with early onset FAD. Strikingly, during the time this manuscript was under consideration, this mutation has been reported to occur in an Italian family (76).
FIG. 6. The pathological function of PS1 C92S, wt SEL-12, and SEL-12 C60S. a, quantitation of the A␤1-42/A␤1-40 ratio. Conditioned media from metabolically labeled cells were immunoprecipitated with antibody 3926 to detect all species of A␤. A␤1-40 and A␤1-42 were identified on a previously described Tris-Bicine gel system (56). Quantitation of the A␤ 1-42/A␤ 1-40 ratio was performed by phosphorimaging. Bars represent the mean Ϯ S.E. of three independent experiments. Asterisks (*) and (**) correspond to statistical significance (*p Ͻ 0.01; Student's t test) and (**p Ͻ 0.001; Student's t test), respectively. b, amino acid exchanges of SEL-12 at sites corresponding to FAD mutations in human PS1 and PS2. Boldface letters indicate SEL-12specific amino acid exchanges at positions, which are conserved in human PS1 and PS2. Cleavage of human PS1 and PS2 is heterogeneous and appears to occur at three different sites. Arrows denote reported cleavage sites (69,74,75,77). Boldface letters indicate conserved amino acids. All three cleavage sites are not conserved in SEL-12.