Abbreviations used : Aβ, β-amyloid peptide ; AD, Alzheimer's disease ; AEBSF, 4-(2-aminoethyl)benzenesulfonyl fluoride • HCl ; Cal I, calpain inhibitor I ; Cal II, calpain inhibitor II ; DCI, 3,4-dichloroisocoumarin ; DMSO, dimethyl sulfoxide ; FAD, familial Alzheimer's disease ; MALDI-TOF-MS, matrix-assisted laser desorption/ionization-time-of-flight-mass spectroscopy ; PAGE, polyacrylamide gel electrophoresis ; PS, presenilin ; RIPA, radioimmunoprecipitation assay ; SDS, sodium dodecyl sulfate.
Address correspondence and reprint requests to Dr. T. Honda at Laboratory for Alzheimer's Disease, Brain Science Institute, RIKEN, Saitama 351-0198, Japan.
Abstract : Presenilin 1 (PS1) has been identified as a causative gene for most early-onset familial Alzheimer's disease. Biochemical studies revealed that PS1 exists predominantly as two processed fragments in cells and brain tissues. We prepared stably transfected cells expressing the wild-type and familial Alzheimer's disease-associated mutants of PS1 and investigated the enzyme that participates in the metabolism of PS1. After treatment of the cells with proteasome inhibitors, the full-length PS1 was significantly accumulated. The levels of N- and C-terminal fragments were also increased. The accumulation of PS1 with a deletion of exon 10, which is unable to be processed, on treatment of the transfected cells with lactacystin indicated that proteasome can degrade full-length PS1. A synthetic peptide that includes the processing region of PS1 was cleaved by 20S proteasome at the putative processing sites after Met288 and Glu299. Metabolic labeling experiments showed that the appearance of the N-terminal fragment was attenuated by the inhibitor. Finally, 28-kDa N- and 20-kDa C-terminal fragments were generated by purified PS1 in vitro. These data indicated that the proteasome pathway is involved in PS1 processing. These results demonstrate that the proteasome pathway plays dual roles in processing and degradation of PS1.
Alzheimer's disease (AD) is a progressive neurode-generative disorder with devastating memory loss. Senile plaques composed of extracellular accumulations of β-amyloid, intracellular deposits of neurofibrillary tangles, and neuronal loss are three well-characterized pathological features of this disease. There are sporadic and familial forms of the disease, which are pathologically almost indistinguishable.
We and others previously reported that PS1 was synthesized as a ≈48-kDa protein and processed into ≈28-kDa N-terminal and ≈20-kDa C-terminal fragments in cultured cells and human brain (Mercken et al., 1996 ; Thinakaran et al., 1996). That one-third of identified mutations are located near this processing site suggests the biological and pathological importance of these processed fragments. We previously showed that at least three mutations, M146V, A246E, and C410Y, affect this processing (Mercken et al., 1996 ; Murayama et al., 1997). To understand this processing further, we prepared cell lines stably expressing wild-type and mutant human PS1 and studied the effects of protease inhibitors on the metabolism of PS1. Among several inhibitors, proteasome inhibitors resulted in significant accumulation of full-length PS1. The N- and C-terminal fragments were also accumulated. Kinetic analysis, determination of cleaved sites using a synthetic peptide and proteasome, and in vitro digestion of PS1 by proteasome revealed that proteasome is involved not only in the degradation, but also in the processing of PS1.
Calpain inhibitor I (Cal I), calpain inhibitor II (Cal II), and leupeptin were purchased from Nacalai Tesque. 4-(2-Aminoethyl) benzenesulfonyl fluoride • HCl (AEBSF) was from Pentapharm. 3,4-Dichloroisocoumarin (DCI) was from Sigma. Calpastatin was from Calbiochem. MG132 and E64d were from Peptide Institute. Lactacystin was a kind gift from Dr. Omura, Kitasato Institute, Japan (Omura et al., 1991).
Construction of wild-type and mutant human PS1 expression vectors
A full-length human PS1 was cloned from a human brain cDNA library. FAD-associated mutations were introduced by PCR-based site-directed mutagenesis and confirmed by sequencing. cDNAs encoding wild-type and mutant PS1 were subcloned into pCI-neo (Promega) or pEF321 (a gift from Dr. S. Sugano, Institute of Medical Science, University of Tokyo, Tokyo, Japan). The latter is an expression vector with the human elongation factor 1α promoter (Kim et al., 1990). None of the constructs used here contains a VRSQ motif.
Cell cultures and transfection
Both human neuroblastoma SH-SY5Y and human embryonic kidney 293 cells were cultured in Dulbecco's minimal essential medium supplemented with 10% fetal bovine serum at 37°C in an atmosphere of 95% air and 5% CO2. For stable transfection, 5 × 105 cells were transfected with 2 μg of plasmid DNA with the use of LipofectAMINE (GibcoBRL) according to the manufacturer's instructions. Transfectants were selected with 500 μg/ml G418. After the second cloning, the concentration of G418 was reduced to 200 μg/ml for maintenance.
Polyclonal antibodies M5 and hL312 were raised in rabbits against the hydrophilic loop domains, amino acids 299-313 and 312-343 of human PS1, respectively. Monoclonal antibodies MKAD3.3 and 3.4 to the N-terminal region of human PS1 have been described previously (Mercken et al., 1996). MKAD3.3 and 3.4 were used as ascites fluid and hybridoma culture supernatant, respectively. All polyclonal antisera used here were affinity purified with corresponding peptide-coupled agarose gel columns (Pierce).
Protease inhibitor treatment
The cells (7 × 105) in 6-cm dishes were treated for 6 h with the indicated concentrations of protease inhibitors or vehicle [dimethyl sulfoxide (DMSO)] alone. The cells were washed twice with cold phosphate-buffered saline and harvested. The cells were lysed with 50 mM Tris-HCl containing 1% Triton X-100, 2mM dithiothreitol, 1 mM AEBSF, 10 μg/ml pepstatin, 1 mM Cal I, and 10 μg/ml leupeptin, then briefly sonicated. The lysates were centrifuged at 15,000 g for 20 min, and the resultant supernatants were subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE)/western blot analyses.
Metabolic labeling and immunoprecipitations
PS1-transfected cells (1 × 106) in 6-cm dishes were pretreated with 50 μM Cal I or vehicle (DMSO) alone in 2 ml of methionine- and cysteine-free Dulbecco's minimal essential medium (GIBCO) with 2% dialyzed fetal bovine serum for 1 h and labeled with 3.7 MBq [35S]methionine and [35S]cysteine (Pro-mix, Amersham) for 15 min in 2 ml of the above medium. The cells were then washed with phosphate-buffered saline and lysed with 450 μl of radioimmunoprecipitation assay (RIPA) buffer [50 mM Tris-HCL (pH 7.4) containing 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 1 mM AEBSF, 10 μg/ml pepstatin, 1 mM Cal I, and 10 μg/ml leupeptin]. After incubation for 30 min on ice, the lysates were centrifuged at 15,000 g for 20 min at 4°C. The supernatants were incubated with protein G-Sepharose (Pharmacia) for 30 min and centrifuged for 5 min at 15,000 g. The precleared supernatants were used for immunoprecipitation. For immunoprecipitation, monoclonal antibody MKAD3.3 precoupled with protein G-Sepharose in RIPA buffer was added to cell supernatants. After incubation at 4°C for 17 h, samples were centrifuged at 3,000 g for 2 min and pellets were washed five times with RIPA buffer. The pellets were then resuspended in 25 μl of SDS sample buffer, incubated for 5 min at 55°C, and subjected to SDS-PAGE. Dried gels were exposed to x-ray film (BioMax MR, Kodak) and densitometry was performed or directly quantified with a BAS 5000 Phosphor Imaging System.
SDS-PAGE and western blot
Protein samples were separated in 13% T SDS-PAGE. The gels were then electrically blotted onto polyvinylidene difluoride membranes (Immobilon-P, Millipore) and blocked with 5% nonfat dried milk in TBST [20 mM Tris-HC] (pH 7.4) containing 150 mM NaCl and 0.05% Tween 20] for 1 h. The first and second antibodies were also diluted with 5% milk/TBST. Horseradish peroxidase-labeled secondary antibodies (Jackson Immuno Research) were used and the signals were detected with a chemiluminescence substrate (Pierce).
In vitro cleavage of synthetic peptide L281 by proteasome
Ten micrograms of a synthetic peptide, L281(281-311), was incubated with 0.2 μg of 20S bovine proteasome at 30°C for 30 min in 50 mM Tris-HCl (pH 8.0) with 0.04% SDS. Digests were subjected to matrix-assisted laser desorption/ionization (MALDI)-time-of-flight (TOF)-mass spectroscopy (MS) or to reversed-phase HPLC, to separate cleaved peptides. Each peak was subjected to protein sequencing (Applied Biosystems 477A) or MALDI-TOF-MS analysis. MALDI-TOF-MS was measured on a PerSeptive Biosystems Voyager Linear mass spectrometer by using α-cyano-4-hydroxycinnamic acid as a matrix.
In vitro processing of immunoprecipitated PS1 by proteasome
Semiconfluent PS1-transfected cells in a 10-cm dish were starved in 4 ml of methionine- and cysteine-free Dulbecco's minimal essential medium with 2% dialyzed fetal bovine serum for 1 h and labeled with 14.8 MBq [35S]methionine and [35S]cysteine (Pro-mix, Amersham) for 15 min in 4 ml of the above medium. PS1 was immunoprecipitated as described above. After final wash with RIPA buffer, immunoprecipitates were washed two times with 50 mM Tris-HCl (pH 7.4) containing 0.04% SDS and used as substrate. Bovine 20S proteasome was added and incubated at 37°C for 6 h with agitation. Reaction was stopped by addition of SDS sample buffer.
Proteasome inhibitors block the processing of PS1 in cells
We prepared the transfected human neuroblastoma SH-SY5Y cells stably expressing human wild-type PS1 under the control of cytomegalovirus early promoter. As we reported previously (Mercken et al., 1996), monoclonal antibody MKAD3.4 to the N-terminal region of human PS1 stained the full-length PS1 with the molecular mass of ≈48 kDa and also more intensely stained the processed N-terminal half of the ≈28 kDa, predominantly in the membrane fraction from nontransfected SH-SY5Y cells (Fig. 1). An antibody, hL312, to the hydrophilic loop domain recognized both ≈48-kDa full-length PS1 and two or three bands of 20-24 kDa, which represented unphosphorylated and phosphorylated forms of C-terminal fragments, mainly in the membrane fraction (Seeger et al., 1997 ; Walter et al., 1997). In transfected cells, the amounts of full-length PS1 were significantly increased, as were those of both N- and C-terminal fragments. It is interesting that M5, a polyclonal antibody to 299-314 of human PS1, reacted with the C-terminal fragment like hL312, but did not react with full-length PS1.
To identify the proteases that are involved in the metabolism of PS1, transfected cells were exposed to several protease inhibitors and the cell lysates were analyzed by means of western blotting. As shown in Fig. 2, peptide aldehyde Cal I at 10 and 30 μM caused significant accumulation of full-length PS1. This peptide aldehyde is known to inhibit cysteine proteases, such as calpains and lysosomal cathepsins, and proteasome. To identify the protease involved in the metabolism of PS1, other protease inhibitors were then tested. To exclude processing and degradation of PS1 during lysate preparation, lysis buffer containing 1 mM Cal I was used in all experiments. Although Cal II inhibits calpains as effectively as Cal I, little accumulation was observed. Leupeptin at up to 100 μM did not cause any accumulation. The calpain inhibitor calpeptin and the structurally unrelated, membrane-permeable, potent cysteine protease inhibitor E64d at concentrations up to 100 μM also failed to cause accumulation (data not shown). We then tried MG132 and DCI as proteasome inhibitors. Treatment with these proteasome inhibitors caused marked accumulation of full-length PS1, as shown in Fig. 2, in the order of Cal I > MG132 > DCI. Although DCI also inhibits serine proteases, AEBSF, a nonspecific potent serine protease inhibitor, at 100 μM did not show any effect. The possibility that the target protease of DCI is a serine protease can therefore be ruled out. Finally, to confirm the involvement of proteasome in the metabolism of PS1, lactacystin, a highly specific, potent, irreversible proteasome inhibitor was used (Fig. 3). Lactacystin caused significant accumulation of ≈48-kDa full-length PS1 at concentrations from 1 μM. The amounts of N- and C-terminal fragments also increased in a dosedependent manner, suggesting that both fragments are also digested by proteasome.
To confirm that the effect of proteasome inhibitor is a general but not a cell-specific event, another transfected cell line, i.e., human embryonic kidney 293 cells stably transfected with wild-type human PS1, and nontransfected SH-SY5Y were treated with Cal I under the same conditions. Western blots of transfected 293 cells showed that the processing pattern of PS1 was almost the same as that of transfected SH-SY5Y cells and when transfected 293 cells were treated with Cal I, full-length PS1 was significantly accumulated (data not shown). Moreover, the accumulation of full-length PS1 was also seen in nontransfected SH-SY5Y, although the increase was small (Fig. 2). These results indicated that the proteasome pathway is actually involved in the metabolism of PS1 under normal conditions.
Peptide digestion by 20S proteasome
The above results indicated that proteasome may be involved in degradation or processing of full-length PS1, or both. To clarify this question, we conducted the following three experiments.
First, to determine whether proteasome can cleave the exact processing sites of PS1, a synthetic peptide, L281 (amino acids 281-311), which contains the entire processing region of human PS1, was digested with bovine 20S proteasome. After incubation of L281 with 20S proteasome for 30 min, the aliquot of the reaction mixture was analyzed directly by MALDI-TOF-MS. These conditions were decided to determine the primary cleavage sites, as proteasome is a multicatalytic protease and longer incubation results in complete digestion. Two fragments with molecular weights of 1,373.0 and 1,434.2 were predominant as cleaved products, demonstrating that two peptides, 300-311 (Mr = 1,371) and 299-311 (Mr = 1,442), were produced. After reversed-phase HPLC separation of the digest, all collected peaks were subjected to MALDI-TOF-MS and direct sequence analysis. As shown in Fig. 4, in addition to the above two peptides, two minor peptides corresponding to amino acids 281-288 and 281-289 were identified. Although the peaks were higher than the former two peptides, amounts of peptides C and D were less than one-half of those of peptides A and B. These data indicate that two of the reported sites of PS1, after Met298 and Glu299, can be processed by proteasome (Podlisny et al., 1997 ; Wisniewski et al., 1997). We could not find the middle part of L281, amino acids 290-298, in which the other two reported cleavage sites are located. Presumably, this region is very sensitive to proteasome, and was digested completely.
To confirm further the involvement of proteasome in the processing, we examined whether the processing of PS1 is actually inhibited by proteasome inhibitor. As N- and C-terminal fragments are also protected by proteasome inhibitor as shown in Fig. 3, we conducted a short-term pulse-labeling experiment, to analyze the effect of the inhibitor only on the processing of full-length PS1. Transfected human kidney 293 cells expressing human wild-type PS1 under the human elongation factor 1α promoter were pretreated with and without 50 μM Cal I for 1 h. Labeling was conducted for 15 min, then PS1 was immunoprecipitated with antibody MKAD3.3 directed against the N-terminal (Fig. 5A). The N-terminal fragment appeared within 15 min. When cells were treated with Cal I, the level of the N-terminal fragment decreased significantly (p < 0.05) (Fig. 5B). We could not see the protection of full-length PS1. Instead, the level of full-length PS1 was slightly decreased by Cal I treatment, although it was not statistically significant. These data may indicate that accumulation of the full-length PS1 by pretreatment with Cal I inhibits the new synthesis of PS1. This slightly decreased synthesis might mask the protection of full-length PS1. Considering these effects, we introduced a new parameter, the ratio of the N-terminal fragment to the sum of the N-terminal fragment plus full-length PS1. It indicates the degree of processing more properly. As indicated in Fig. 5C, it was significantly reduced (p < 0.01). These data showed that processing of PS1 is fast and that proteasome inhibitor can attenuate this event.
In vitro processing of immunoprecipitated PS1 by 20S proteasome
Pulse-labeled PS1 was immunoprecipitated and incubated with bovine 20S proteasome in vitro. After 6 h, both 28-kDa N-terminal and 20-kDa C-terminal fragments were produced (Fig. 6). These results, however, showed inefficient degradation and processing of full-length PS1 by proteasome compared with those in intact cells. Proper recognition of PS1 by protreasome might need membrane integration of PS1 and/or 26 proteasome system instead of 20S.
Relationship between FAD-associated mutations and metabolism of PS1 by proteasome
We then examined the effects of FAD-associated mutations on the metabolism of PS1 by proteasome. As shown in the wild-type PS1 transfected cells, Cal I treatment significantly accumulated full-length PS1 in transfected cells expressing PS1 with Leu392Val mutation, Cys410Tyr mutation, or a deletion of exon 10 (Fig. 7). Given that PS1 with a deletion of exon 10 is unable to be processed because of the lack of the processing region, the accumulation of this mutant PS1 clearly suggests that full-length PS1 can be degraded via the proteasome pathway.
The proteins PS1 and PS2 are mostly detected as cleaved products in cultured cells and brain tissues (Mercken et al., 1996 ; Thinakaran et al., 1996 ; Kim et al., 1997). One-third of missense mutations of PS1 are located near the cleavage region. Moreover, deletion of exon 10 of PS1, which is also associated with FAD, results in the accumulation of uncleaved PS1 and a concomitant decrease in cleaved forms of PS1. These data suggest that these processed forms of PS1 and/or the processing event have a physiological role. We further investigated this processing event by using PS1-transfected cells.
When cells were treated with proteasome inhibitors, the full-length PS1 was significantly accumulated. The accumulation of full-length PS1 can be explained by inhibition of degradation and/or processing into two fragments. In accordance with previous reports on PS1 (Fraser et al., 1998 ; Marambaud et al., 1998) and PS2 (Kim et al., 1997), we found that the full-length PS1 was degraded by proteasome. The accumulation, lactacystin treatment, of the full-length PS1 with a deletion of exon 10, which is unable to be processed, strongly supported this idea. The N- and C-terminal fragments were increased by the treatment with proteasome inhibitor, suggesting that they are also degraded via the proteasome pathway. Although the proteasome pathway is a major cellular protein degradation system located mainly in cytoplasm, recent data show that proteasome can also degrade proteins within the endoplasmic reticulum (Brodsky and McCracken, 1997 ; Kopito, 1997). The recovery of accumulated PS1 in the membrane fraction (data not shown) confirmed that proteasome acts on PS1 in the endoplasmic reticulum.
Our results demonstrate that the proteasome pathway has a role in the processing of PS1, too. The involvement of the proteasome pathway is supported by three lines of evidence. First, two of four previously reported processing sites, after Met298 and Glu299, were cleaved by bovine 20S proteasome in an in vitro experiment using a synthetic peptide. Although we could not identify the cleavages that Thr291 and Met292, our failure to detect peptides containing the region from Ser290 and Met298 suggests that this region might be very sensitive to proteasome and might have been completely digested. Second, the short-term metabolic labeling experiment focused only on the processing of PS1 and showed that treatment with proteasome inhibitor attenuated the processing of PS1. Third, N- and C-terminal fragments were produced from the immunoprecipitated PS1 by proteasome in vitro.
As we have shown here, a part of newly synthesized PS1 is processed into two fragments and the rest is rapidly degraded. The degradation of full-length PS1 and processed two fragments are degraded via the proteasome pathway. We demonstrated that the Cal I-sensitive proteasome pathway plays a part in the processing of PS1. However, the fact that the N- and C-terminal fragments were not decreased by proteasome inhibitor but actually increased suggests the existence of another unidentified processing enzyme. As proteasome is a multicatalytic protease and is regulated by several subunits, some proteasome systems may be involved in the metabolism of PS1, and the processing activity of proteasome, compared with degradation, might not be as sensitive to the inhibitors that we used.
We used transfected cells, as other groups did. It is possible that overexpression of membrane proteins causes activation of the proteasome system, resulting in rapid degradation of the proteins. However, the accumulation of full-length PS1 was also observed when non-transfected cells were treated with lactacystin. This observation supports that full-length PS1 is a natural substrate of proteasome under physiological conditions.
Rapid degradation of cellular proteins via the proteasome pathway has been well studied. However, little is known about site-specific cleavage by proteasome. During NFκB activation, not only degradation of IκB, but also processing of p105 precursor protein to generate p50 subunit is dependent on the proteasome pathway (Fan and Maniatis, 1991 ; Palombella et al., 1994). Our finding on the processing of PS1 is another example of participation of proteasome in proteolytic processing.
The ubiquitin-proteasome pathway has been suggested to be dysfunctional in AD (McDermott and Gibson, 1991 ; Muller and Schwartz, 1995). It has been reported that the inhibition of proteasome increased the secretion of both Aβ40 and Aβ42 without changing the Aβ42 ratio (Marambaud et al., 1997a,b ; Yamazaki et al., 1997). Our present results suggest that dysfunction of the proteasome system results in impairment of PS1 metabolism. Together with our previous finding that some PS1 mutants are hardly processed (Mercken et al., 1996 ; Murayama et al., 1997), this study suggests an important role of PS1 metabolism in the mechanism of Aβ processing. The combined effect of PS1 mutations, which increase the Aβ42 ratio, and proteasome dysfunction, which causes mismetabolism of PS1 resulting in the elevated total Aβ level, may affect the age at onset of AD.