Alternative Cleavage of Alzheimer-Associated Presenilins During Apoptosis by a Caspase-3 Family Protease

See allHide authors and affiliations

Science  18 Jul 1997:
Vol. 277, Issue 5324, pp. 373-376
DOI: 10.1126/science.277.5324.373


Most cases of early-onset familial Alzheimer's disease (FAD) are caused by mutations in the genes encoding the presenilin 1 (PS1) and PS2 proteins, both of which undergo regulated endoproteolytic processing. During apoptosis, PS1 and PS2 were shown to be cleaved at sites distal to their normal cleavage sites by a caspase-3 family protease. In cells expressing PS2 containing the asparagine-141 FAD mutant, the ratio of alternative to normal PS2 cleavage fragments was increased relative to wild-type PS2-expressing cells, suggesting a potential role for apoptosis-associated cleavage of presenilins in the pathogenesis of Alzheimer's disease.

Apoptosis (programmed cell death) is an evolutionary conserved form of cellular suicide that plays a beneficial role during development and homeostasis (1). Extensive evidence indicates that activation of apoptotic cell death is associated with a variety of neurodegenerative disorders (2), including Alzheimer's disease (AD) (3-5). Most cases of early-onset FAD are caused by mutations in the genes encoding the PS1 and PS2 proteins (6), both of which undergo regulated endoproteolytic processing to yield two stable fragments (7-9). Overexpression of the presenilins in transfected cells has been reported to increase susceptibility to apoptosis (10-12). Here we investigated whether the presenilins serve as substrates for apoptosis-associated cleavage in the programmed cell death pathway and assessed the effects of an FAD mutation on this process.

We initially identified the endogenous form of the PS2 endoproteolytic COOH-terminal fragment (CTF) using the COOH-terminal antibody to PS2, anti-PS2Loop (13). The endogenous PS2-CTF appeared as a 25-kD band in both human brain (14) and in uninduced H4 neuroglioma cells (Fig. 1). As reported for PS1 (7), no full-length PS2 was detected in brain (14) or in the uninduced cells (Fig. 1). In stably transfected H4 cells that were induced to overexpress PS2 containing a COOH-terminal FLAG epitope tag (9), the 54-kD full-length and high molecular weight forms of PS2 were observed along with multiple CTFs with apparent sizes of 26, 25, and 20 kD (Fig. 1, lanes 1 and 8), representing the transgene-derived normal cleavage product containing the eight–amino acid FLAG epitope, the endogenous CTF, and a smaller, alternative CTF, respectively (15).

Figure 1

Multiple PS2 COOH-terminal fragments (PS2-CTF) in the inducible H4 cells (9) overexpressing COOH-terminal epitope-tagged PS2. The 25-kD endogenous PS2 fragment (band e) was detected in both induced (lanes 1, 5 to 8) and uninduced (lanes 9 to 12) samples (13). Two additional fragments, including a 26-kD fragment representing the normal cleavage product containing an eight–amino acid FLAG epitope (band f), and the smaller, 20-kD fragment (band a), were detected only in the induced (lanes 1, 5 to 8) samples. The 20-kD CTF was selectively localized in the nonionic detergent (1% Triton X-100)–resistant cellular fraction (lane 2) (9, 13). Wild-type PS2 cells were induced for 48 hours, fractionated into detergent-insoluble and -resistant (lane 2) and detergent-soluble (lane 3) fractions and wash (lane 4), and then analyzed by protein immunoblotting (13). Total SDS lysates are shown in lane 1. Generation of the 20-kD PS2-CTF was blocked by zVAD (zVAD-fmk, a broad-spectrum caspase inhibitor) and zDEVD (zDEVD-fmk, a selective caspase-3 inhibitor) in cells overexpressing PS2 (lanes 5, 6, 9, and 10). zFA (cathepsin B inhibitor) was used as a control for the fluoromethylketone (fmk) group (lanes 7 and 11). Inducible PS2 cells were incubated with 100 μM zVAD (lane 5 and 9), zDEVD (lanes 6 and 10), and zFA (lanes 7 and 11), or solvent alone (dimethylsulfoxide; lanes 8 and 12) in the absence (uninduced, lanes 9 to 12) or presence (induced, lanes 5 to 8) of tetracycline for 24 hours.

We have previously shown that overexpression of PS2 in transfected cells results in the generation of a 20-kD CTF that predominantly localizes to the detergent (Triton)-resistant cellular fraction and that exhibits a very slow rate of turnover (9, 15). Whereas here the endogenous (Fig. 1, band e) and normal transgene-derived (Fig.1, band f) CTFs were primarily localized to the detergent-soluble cellular fraction (Fig. 1, lane 3), the smaller, alternative 20-kD CTF was enriched in the detergent-resistant fraction (Fig. 1, lane 2). Together, these data suggested that when overexpressed in stably transfected inducible H4 cells, PS2 is cleaved at a site located distal to the normal cleavage site, resulting in a smaller detergent-insoluble CTF.

Given recent reports that overexpression of PS2 induces apoptosis in transfected cells (10, 12), we examined the effects of zVAD, a broad-spectrum inhibitor for most mammalian interleukin-1β–converting enzyme ICE/ Ced-3 proteases (caspases), and zDEVD-fmk, a more selective inhibitor for caspase-3 (CPP32) family proteases (caspases-3, -6, and -7) (16, 17). Both zVAD and zDEVD blocked the generation of the 20-kD PS2-CTF in cells overexpressing PS2 (Fig. 1) but had no effect on levels of the endogenous 25-kD CTF (Fig. 1, band e), the 26-kD transgene-derived CTF (Fig. 1, band f), or other PS2 species. These findings indicated that in the absence of traditional apoptotic stimuli, overexpression of PS2 led to the activation of caspase-3 family proteases that alternatively cleaved PS2.

Caspase-3 is activated through a protease cascade in response to apoptotic stimuli and cleaves specific intracellular substrates during apoptosis (16, 17). To determine whether the alternative PS2-CTF was generated from full-length PS2 during apoptosis, we used a relatively low-expressing inducible H4 cell line in which only low to undetectable levels of the 20-kD CTF were observed at 12 hours after induction of PS2. When apoptosis was induced with 20 μM etoposide, the alternative 20-kD CTF as well as the corresponding alternative NH2-terminal fragment (NTF) were generated (Fig. 2). The alternative NTF was 34 kD, whereas the “normal” NTF, labeled as band f in the top panel of Fig. 2, was 30 kD. The alternative PS2 cleavage fragments were first observed at 3 hours after treatment with etoposide. Treatment with increasing concentrations of zVAD inhibited the generation of both alternative cleavage products (Fig. 2). These data indicated that full-length PS2 gave rise to the alternative NH2- and COOH-terminal cleavage fragments after the induction of apoptosis.

Figure 2

Cleavage of full-length 54-kD PS2 (fl) by a zVAD-sensitive protease activated during etoposide-induced apoptosis. Inducible H4 cells expressing PS2 containing the NH2-terminal FLAG epitope were induced for 12 hours and further incubated with media containing 20 μM etoposide for the time indicated, or first treated and incubated with the indicated concentrations of zVAD for 12 hours. Alternative (a′, 34 kD) and normal (f, 30 kD) NH2-terminal cleavage products were detected by anti-FLAG (top) (13), and the alternative (a, 20 kD) and endogenous plus normal COOH-terminal cleavage products (e/f, 25 kD) were detected by anti-PS2Loop (bottom).

We next examined whether endogenous PS2 is also cleaved by a caspase-3 family protease (or proteases). When native H4 cells were treated with staurosporine (Fig. 3A) or etoposide (14), apoptosis was induced as indicated by poly(ADP-ribose) polymerase (PARP) cleavage, proteolysis of inactive caspase-3 precursor (CASP-3), and decreased viability (assayed by trypan blue exclusion) (14). The progression of apoptosis correlated with the increased generation of the endogenous 20-kD PS2-CTF and decreased production of the 25-kD CTF (starting at 6 hours after treatment) (Fig.3A).

Figure 3

Cleavage of endogenous PS2 during apoptosis. (A) H4 human neuroglioma cells were treated with 1 μM staurosporine (STS) for timed intervals indicated on the top, and lysates were analyzed by protein immunoblotting (13) with antibodies to PS2loop, poly(ADP-ribose) polymerase (PARP), pro-caspase-3 (CASP-3), and HSP70. (B) Inhibition of STS-induced (left) or etoposide (ETO)-induced (right) cleavage of PS2, PARP, and CASP-3 in H4 cells by zVAD. Cells were first treated with inhibitors for 1 hour, then further incubated in the presence or absence of 2 μM STS (left) or 20 μM etoposide (right) for 18 hours, and lysates were analyzed by protein immunoblotting (13). (C) Inhibition of apoptosis-induced cleavage of PS2 by zDEVD (inhibitor for a caspase-3 family protease; 100 μM). Treatment of cells with STS and inhibitors was similar to that described for (B).

Alternative cleavage of endogenous PS2 during either staurosporine- or etoposide-induced apoptosis was specifically blocked by zVAD (Fig. 3B) and zDEVD (Fig 3C), both of which displayed no effect on normal processing of PS2 (generation of the 25-kD endogenous fragment). zVAD was a more potent inhibitor of apoptosis-induced cleavage of PS2 than zDEVD (18). This is most likely due to the fact that zVAD blocks the activation of the upstream protease cascade during the apoptotic pathway, whereas zDEVD specifically acts downstream as a competitive inhibitor for the caspase-3 cleavage site (17). These data indicated that in native H4 cells undergoing apoptosis, a caspase-3 family protease was responsible for the generation of the 20-kD PS2-CTF. Thus, like PARP (19) or lamins (20), PS2 appears to serve as an apoptotic death substrate that undergoes alternative proteolysis during apoptosis (21).

Differential cleavage by apoptotic and nonapoptotic proteases has also been demonstrated for the sterol-regulatory element binding proteins SREBP-1 and SREBP-2, endoplasmic reticulum proteins that control cellular cholesterol homeostasis (22). Both SREBP-1 and SREBP-2 are cleaved by caspase-3 (23) and caspase-3–related caspase-7 (24) at the site DSEPDSPVF for SREBP-1 and KDEPDSPPV for SREBP-2 (25). A potential consensus site (DSYDS, amino acids 326 to 330) (Fig. 4A) for cleavage by a caspase-3 family protease was localized in the large hydrophilic loop of PS2 (26). To test this site, we substituted either of the two Asp residues with Ala (Fig. 4A) (27). Both substitutions individually blocked the generation of the 20-kD CTF (Fig. 4B), indicating that PS2 is most likely cleaved after either Asp326 or Asp329. The presence of Asp326 might also be required for cleavage after Asp329, or vice versa. Meanwhile, the substitution Asp308→Ala (D308A) had no effect on cleavage. These data demonstrate that a caspase-3 family protease most likely cleaves PS2 at the predicted consensus cleavage site (Asp-X) that resides distal to the normal cleavage site to yield the apoptosis-associated PS2-CTF.

Figure 4

Abolition of zVAD/zDEVD-sensitive cleavage of PS2 by replacement of Asp326 or Asp329. (A) Consensus ICE/Ced-3 protease cleavage sites located after Asp326 or Asp329 (indicated by arrows A and B, respectively) at the PS2 loop domain encoded by exon 11. The predicted normal cleavage sites (8) are located in the distal region encoded by exon 10 (originally called exon 9) (36). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) Effect of D326A and D329A mutations on the generation of the 20-kD PS2 fragment (27). Inducible constructs encoding COOH-terminal FLAG epitope-tagged PS2 with indicated mutations (wild type, lane 1; D308A, lane 2; D326A, lane 3; and D329A, lane 4) were transiently transfected into tetracycline-responsive founder H4 cells (9). Cells were grown in the absence of tetracycline (induction) for 24 hours and further incubated in the presence of 20 μM etoposide for 9 hours and then analyzed by protein immunoblotting with either anti-PS2Loop (left) or anti-FLAG (right) (13).

To determine whether PS1 also undergoes alternative cleavage, we established inducible H4 cell lines expressing PS1 with a COOH-terminal FLAG epitope tag (27). In cells overexpressing high levels of PS1, in addition to the 24-kD “normal” transgene-derived CTF and the 23-kD endogenous CTF, a smaller, alternative 14-kD CTF was detected. However, in clonal cell lines expressing lower levels of PS1, the 14-kD CTF fragment was not detectable (14). As was observed for PS2, the generation of the alternative 14-kD PS1-CTF was blocked in the overexpressing cell lines by treatment with zVAD or zDEVD (14).

The 23-kD endogenous PS1-CTF was observed in native H4 cells as a doublet, presumably as a result of protein kinase C (PKC)–mediated phosphorylation of this fragment (28). Induction of apoptosis in native H4 cells with staurosporine led to the generation of the 14-kD PS1-CTF and also blocked the phosphorylation of the endogenous CTF (Fig.5). This is consistent with the fact that staurosporine inhibits PKC activity. Etoposide-induced apoptosis also led to the generation of the 14-kD CTF but, as expected, did not block phosphorylation of the endogenous PS1-CTF (Fig. 5). As with PS2, apoptosis-associated cleavage of endogenous PS1 was blocked by the treatment with zVAD (Fig. 5) or zDEVD (14). The size of the alternative PS1-CTF and a scan of the PS1 amino acid sequence suggested that the consensus sequence AQRDS (amino acids 343 to 346) is the most likely site for caspase-3–related cleavage in PS1, although this will require direct testing.

Figure 5

Cleavage of endogenous PS1-CTF during STS-induced apoptosis. Samples identical to those used for Fig. 3 were stained with anti-PS1Loop (7). PS1 was cleaved to generate the smaller 14-kD fragment during the progression of apoptosis (left). Inhibition of STS-induced (middle) and etoposide-induced (right) PS1 cleavage by zVAD is shown. Asterisk indicates nonspecific bands.

To begin to investigate a potential role for apoptotic PS2 cleavage products in AD pathogenesis, we examined the effect of the Asn141→Ile (N141I) Volga German FAD mutation (29) on the generation of the 20-kD PS2 CTF. Multiple wild-type or FAD mutant (N141I)-inducible cell lines expressing “low,” “middle,” or “high” amounts of PS2 were paired according to expression level and analyzed by protein immunoblotting (13). The ratio of the 20-kD CTF:26-kD CTF was increased approximately threefold in the PS2-N141I cells relative to the wild-type PS2 cells (Fig. 6) (30). Preliminary studies have also shown that the alternative 14-kD PS1-CTF accumulated in higher concentrations in lymphoblasts derived from patients with the Ala246→Glu (A246E) PS1 FAD mutation as compared to those from age-matched controls (31).

Figure 6

Relative increase of the 20-kD PS2 CTF in H4 cells expressing N141I mutant (Volga German) PS2. The CTFs generated from inducible H4 cells expressing “low” (lanes 1 and 2), “middle” (lanes 3 and 4) , and “high” (lanes 5 and 6) levels of wild-type (lanes 1, 3, and 5) and N141I FAD mutant (lanes 2, 4, and 6) PS2 (with COOH-terminal FLAG epitope tag) were detected by protein immunoblotting with anti-PS2Loop (13). In each case, the wild-type and PS2-N141I clonal cell lines were paired for similar levels of expression (30).

Apoptotic cell death has been reported to be a pathological feature of AD (3-5), although the exact contribution of apoptosis to the pathogenesis of AD remains unclear. Our studies indicate that PS1 and PS2 participate in the apoptotic pathway. It is possible that the alternative endoproteolytic PS1 and PS2 fragments generated by a caspase-3 family protease (or proteases) under apoptotic conditions may serve as pro-apoptotic effectors that alter the apoptotic threshold, making cells more vulnerable to apoptosis. The observed increase in apoptotic PS2 fragments in cells overexpressing the FAD mutant PS2-N141I raises the possibility that this fragment may contribute to enhanced susceptibility to apoptosis as a result of this mutation (10, 12). The N141I FAD mutation in PS2 and other FAD defects in PS1 and PS2 lead to increased production of amyloid-β42 (Aβ42) in fibroblasts and plasma of FAD patients as well as in transfected cells and transgenic mice expressing mutant PS genes (32). In addition, treatment of cells with Aβ induced apoptotic neuronal death in vitro (concentrations of 1 to 100 μM) and in vivo (3-5, 33) and down-regulated anti-apoptotic Bcl-2 expression in primary neurons (concentration of 100 nM) (34). Collectively, these observations suggest at least two potential pathogenic pathways: (i) alternative PS cleavage products directly participate in inducing pro-apoptotic conditions, which in turn lead to pathogenic changes associated with FAD, including neuronal and synaptic degeneration and subsequent generation of Aβ42; and (ii) alternative PS cleavage products induce the increased production or accumulation of Aβ42 that, in turn, effects pro-apoptotic changes and neuronal cell death.

Because the alternative cleavage of the presenilins occurs within 3 hours after an apoptotic stimulus (Fig. 2), our findings suggest that the alternative clip most likely precedes the effects of PS2-N141I on the processing of amyloid precursor protein, leading to an increased ratio of Aβ42:Aβ40 in transfected cells (32). The PS2-N141I mutation increases both the ratio of Aβ42:Aβ40 (32) and the ratio of the alternative:normal PS2 COOH-terminal fragments by roughly three- to fourfold. Thus, these molecular events may be intimately related consequences of the PS2-N141I mutation. Exactly how the activation of caspase-3, the enhanced generation of alternative presenilin cleavage fragments, and the increased generation of Aβ42 interact to contribute to AD neuropathogenesis will be an important topic for future studies.

  • * To whom correspondence should be addressed. E-mail: tanzi{at}


View Abstract

Stay Connected to Science

Navigate This Article