Endoproteolytic cleavage and proteasomal degradation of presenilin 2 in transfected cells.

Mutations in the presenilin genes, PS1 and PS2, cause a major portion of early onset familial Alzheimer's disease (FAD). The biological roles of the presenilins and how their pathological mutations confer FAD are unknown. In this study, we set out to examine the processing and degradation pathways of PS2. For regulated expression of PS2, we have established inducible cell lines expressing PS2 under the tight control of the tetracycline-responsive transactivator. Western blot analysis revealed that PS2 was detected as an approximately 53-55-kDa polypeptide (54-kDa PS2) as well as a high molecular mass form (HMW-PS2). Using a stably transfected, inducible cell system, we have found that PS2 is proteolytically cleaved into two stable cellular polypeptides including an approximately 20-kDa C-terminal fragment and an approximately 34-kDa N-terminal fragment. PS2 is polyubiquitinated in vivo, and the degradation of PS2 is inhibited by proteasome inhibitors, N-acetyl-L-leucinal-L-norleucinal and lactacystin. Our studies suggest that PS2 normally undergoes endoproteolytic cleavage and is degraded via the proteasome pathway.

A significant portion of Alzheimer's disease (AD) 1 is attributed to specific gene defects leading to familial Alzheimer's disease (FAD) (1)(2)(3)(4)(5). Two homologous genes, presenilin 1 (PS1) and presenilin 2 (PS2), are responsible for at least 50% of early onset (Ͼ60 years old) FAD (2,3). PS1 and PS2 are serpentine proteins consisting of six to nine predicted transmembrane domains interspersed with one large and multiple smaller hydrophilic loops (4,5). At the amino acid level, the two proteins are 67% identical and exhibit significant homology to two Caenorhabditis elegans gene products, sel-12 (approximately 50% identity) which has been predicted to facilitate Notch receptor function (6), and spe-4 (approximately 26% identity) which is involved in cytoplasmic trafficking of proteins during spermatogenesis (7). PS1 and PS2 are ubiquitously expressed (4,5) and in brain are expressed primarily in neurons, with similar regional distributions (8 -10). The presenilins are localized to the endoplasmic reticulum (ER) and the Golgi apparatus but not the plasma membrane suggesting a potential role in protein processing (8,11,43). To date, the PS1 and PS2 genes have been shown to contain 35 different mutations which are inherited in an autosomal dominant fashion in over 60 kindreds with early onset FAD (4, 5, 12; for summary, see Ref. 3). Recent studies suggest that the presenilins may directly or indirectly affect the processing of APP leading to increased production of A␤42 (13)(14)(15)(16). These results help to explain the relatively high degree of amyloid burden in the brains of FAD patients carrying PS1 and PS2 mutations. The pathogenic mechanism by which presenilin mutations lead to increased ␤-amyloid deposition and other neuropathological features of AD remains unclear. To begin understanding the role(s) of PS2 in normal cellular metabolism and AD pathogenesis, we investigated the processing and degradation pathways of PS2.

EXPERIMENTAL PROCEDURES
Materials-Tetracycline and ALLN (N-acetyl-L-leucinal-L-norleucinal) were purchased from Sigma. Lactacystin was supplied by Dr. E. J. Corey (Harvard University, Cambridge, MA). Other protease inhibitors and high quality Triton X-100 were purchased from Boehringer Mannheim. Brefeldin A (BFA) was purchased from Calbiochem.
Fusion Proteins and Antibodies-The glutathione S-transferase fusion protein encoding the large hydrophilic loop domain of PS2 was generated as described (17). Anti-PS2Loop is a polyclonal antiserum generated by immunizing rabbits with gel-purified human PS2 loop fusion protein. Specificity and characterization of the antiserum will be described elsewhere. Monoclonal (M2) and polyclonal (D-8) antibodies raised against FLAG peptide (DYKDDDDK) were purchased from IBI and Santa Cruz Biotechnology, respectively. Mouse monoclonal antiubiquitin antibody (Ubi-1) was purchased from Zymed Laboratories, Inc.. Rabbit polyclonal anti-ubiquitin antibodies were purchased from Sigma.
Generation of Founder Cell Line for an Inducible Expression System-Founder cell lines were generated by co-transfecting H4 neuroglioma cells in a 100-mm dish with 10 g of pUHD15-1, a plasmid encoding a tetracycline-repressible transactivator (18) and 1 g of pCMVneo. Individual G418-resistant colonies were isolated and characterized by transient transfection with the luciferase reporter plasmid pUHC13-3, whose promoter is induced by the transactivator. Luciferase induction in the presence and absence of tetracycline was measured by Western blot analysis using anti-luciferase antibody (Promega) to identify cells with maximal inducibility and tight regulation (data not shown).
Preparation of cDNA Constructs-The cDNAs for wild-type PS2 were subcloned from the pcDNA3 construct (8) into the tetracycline-inducible expression plasmid pUHD10-3 vector using polymerase chain reaction by Pfu polymerase (Stratagene). FLAG epitopes were added to either 5Ј or 3Ј ends of PS2 using polymerase chain reaction with the coding sequence for fusion to the sequence, DYKDDDDK. Resulting constructs, PS2s with either 3ЈFLAG or 5ЈFLAG peptides, were verified by DNA sequencing.
Generation of Stably Transformed Cell Lines with Inducible PS2 Constructs-The H4 founder cells were transfected with 10 g of each construct and 1 g of pCNH2hygro, conferring resistance to hygromycin. Hygromycin-resistant colonies were isolated in the presence of tetracycline and screened for PS2 expression by Western blot analysis using antibodies against the FLAG epitope-tag upon removal of tetracycline. For each PS2 construct, five clones demonstrating various induction levels with tight regulation by tetracycline were selected and used for further study. For induction of PS2, cells were washed five times with prewarmed phosphate-buffered saline to remove residual tetracycline and then incubated with complete media without tetracycline for the indicated hours.
Western Blotting and Immunoprecipitation-Protein samples were quantitated by the BCA protein assay kit (Pierce). SDS-PAGE was carried out using 4 -20% gradient Tris/glycine gels under reducing conditions. Proteins were transferred to polyvinylidene difluoride membrane (Bio-Rad) using a semi-dry electrotransfer system (Hoefer). The blots were blocked with 5% non-fat dry milk in TBST (25 mM Tris (pH 7.6), 137 mM NaCl, 0.15% Tween 20) for 1.5 h, incubated primary antibodies (M2, 3 g/ml; D-8, 1 to 1000; polyclonal anti-ubiquitin, 1 to 1000; Ubi-1, 1 to 1500) for 1.5 h, and secondary antibodies (horseradish peroxidase-conjugated anti-mouse or rabbit antibodies, 1 to 5000) in TBST. Between steps, the blots were washed with TBST for 30 min. For Western blotting with M2 antibodies, 5% (w/v) non-fat dry milk was included in the incubation steps for primary and secondary antibodies. The blot was visualized using the ECL Western blot detection system (Amersham). For immunoprecipitation, cells were lysed using IP buffer (10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 0.25% Nonidet P-40, 2 mM EDTA) plus protease inhibitors, and solubilized proteins were subjected to immunoprecipitation. The samples were precleared with protein A conjugated with magnetic beads (Perceptive Diagnostics) for 1 h in the cold room, incubated with either control (rabbit anti-mouse IgG) or anti-FLAG (D-8 at 10 g/ml) antibodies overnight, further incubated with protein A-magnetic beads (30 l/ sample) for 2 h in the cold room, and washed three times with IP buffer. Immunoprecipitates were collected using a magnetic bead collector, boiled in sample buffer, and subjected to SDS-PAGE and Western blotting using monoclonal anti-ubiquitin antibodies.

RESULTS AND DISCUSSION
Endoproteolytic Processing of PS2-To investigate the processing pathway of PS2, we established a regulated system for the expression of PS2. For this purpose, we developed inducible H4 cell lines expressing epitope-tagged versions of wild-type PS2 using a tetracycline-repressible transactivator (18). In this system, the presence of tetracycline in the culture medium suppresses PS2 expression, while its withdrawal results in induction of PS2 expression. PS2 production following induction was monitored by Western blot analysis using antibodies against either N-terminal or C-terminal FLAG epitope tags or using a polyclonal antisera specific for the large hydrophilic loop (HL-6) of PS2 (␣PS2Loop). Fig. 1 shows the results of Western blot analysis of a cellular lysate from representative, stably transformed cell lines expressing wild-type PS2 grown in the presence or absence of tetracycline for 48 h. No PS2 was observed for cell lines incubated in tetracycline-containing media, indicating the tight regulation of PS2 expression in this system. The SDS-extracted proteins detected by the monoclonal anti-FLAG antibody, M2, included full-length PS2 with an apparent molecular mass of 54 kDa and high molecular mass species of PS2 (HMW-PS2; Fig. 1) for both N-and Cterminal FLAG-tagged PS2. In addition, a C-terminal 20-kDa fragment (PS2-CTF) and an N-terminal 34-kDa fragment (PS2-NTF) were observed (Fig. 1), indicating that PS2 undergoes endoproteolytic cleavage like its homologue, PS1 (17). Both the N-terminal and C-terminal fragments appeared to be stable cellular products as opposed to degradation products since their presence was not affected by the absence of protease inhibitors in the lysis buffer, and they were not affected by prolonged incubation for up to 12 h at 37°C (data not shown). Our results reveal that a 54-kDa PS2 protein that undergoes endoproteolytic cleavage to generate 34-kDa N-terminal and 20-kDa C-terminal fragments. The generation of the PS2 endoproteolytic fragments could either be a step in the metabolic pathway of PS2 and/or a processing event necessary for the normal function of PS2.
Detergent Solubility of PS2 Fragments-We next compared the detergent solubility of the N-terminal and C-terminal endoproteolytic fragments. 48 h after induction, a comparison of equal amounts of proteins from the detergent (1% Triton X-100)-soluble and insoluble fractions revealed the PS2-NTF only in the soluble fraction, while the PS2-CTF was enriched in the detergent-resistant fraction (Fig. 2). Loading of equal volumes of the soluble and insoluble fractions, normalized for total cellular proteins, revealed a small amount of PS2-CTF in the soluble fraction, but the majority of the fragment localized to the insoluble fraction. The enrichment of the PS2-CTF with the detergent-resistant cellular fraction (Fig. 2) prepared by a procedure classically employed to isolate either detergent-resist- ant cytoskeletal structures (19 -21, 30, 31) or caveolae-like microdomains (32)(33)(34) suggests that the PS2-CTF may be a component of one of these cellular structures. Alternatively, this association could be due to the formation of detergentresistant HMW-PS2 complexes which localize to this fraction. The detergent insolubility of the PS2-CTF is unlikely to be due to the presence of the FLAG epitope since the PS2-CTF cleaved from the PS2 containing the N-terminal FLAG was also enriched in the detergent-resistant cellular fraction (data not shown).
Induction and Turnover of PS2-We next examined the time course of PS2 induction. Inducible PS2 cell lines were induced and harvested at time intervals up to 72 h. Equal amounts of protein were then analyzed from the detergent-resistant and detergent-soluble fractions (Fig. 3). PS2 carrying the C-terminal FLAG was first detected in the detergent-soluble fraction at 6 h post-induction as a full-length 54-kDa band. Thereafter, increasing amounts of HMW-PS2 were observed along with the appearance of a doublet owing to the presence of a band just below the 54-kDa band. By 72 h post-induction, the lower band of the doublet was not observed, and both the HMW-PS2 and the 54-kDa PS2 bands were less abundant relative to 48 h post-induction. In the detergent-resistant fraction, no obvious PS2 signal was detected by anti-FLAG antibody until 48 h post-induction at which point the PS2-CTF was observed. However, in "high"-expressing PS2 clonal cell lines (e.g. WF9), the PS2-CTF could be observed as early as 24 h post-induction (data not shown).
The inducible system was next employed to examine the turnover of PS2 containing the C-terminal FLAG (Fig. 4). PS2 was induced for 36 h, and PS2 turnover was assessed by adding tetracycline back to the media to repress further PS2 production (time point 0) and then testing detergent-soluble and detergent-resistant cellular fractions at timed intervals up to 24 h. Levels of HMW-PS2 and full-length 54 kDa PS2 were observed to progressively decrease over time and were largely undetectable after 24 h and 9 h post-repression, respectively, in the soluble fraction. In the detergent-resistant fraction, the PS2-CTF was observed as a highly stable fragment which remained relatively stable over the degradation time course. These data suggest that following cleavage the PS2-CTF is translocated to the detergent-resistant fraction where it exists as a relatively stable polypeptide.
Degradation of PS2 by the Proteasome Pathway-To further investigate the degradation of PS2, we tested the effects of a set of known cell-permeable protease inhibitors on the degradation of PS2, including pepstatin A, Pefabloc SC, E-64, leupeptin, and aprotinin. None of these protease inhibitors affected the turnover of the PS2-CTF (data not shown). Next, we tested whether PS2 is polyubiquitinated and degraded by the ubiquitin-proteasome pathway (22)(23)(24)(25). For this purpose, we used the proteasome inhibitors, ALLN and lactacystin (26), which are known to induce the accumulation of polyubiquitinated proteins by inhibiting the 20 S proteasome (the catalytic core of the 26 S complex) (27)(28)(29). Treatment with ALLN and lactacystin resulted in dramatically increased levels of HMW-PS2 while only full-length PS2 was observed in the absence of proteasome inhibitors (Fig. 5A).
To further explore the possibility that the HMW-PS2 con- tained polyubiquitinated PS2, PS2 containing the C-terminal FLAG was immunoprecipitated using polyclonal anti-FLAG antibodies and subjected to immunoblot analysis with a monoclonal anti-ubiquitin antibody (Fig. 5B). Ubiquitin-positive HMW-PS2 was detected and found to be significantly increased following treatment with 50 M ALLN. These findings were also confirmed using additional monoclonal anti-FLAG and polyclonal anti-ubiquitin antibodies (data not shown).
To determine whether polyubiquitinated HMW-PS2 serves as a degradation intermediate for the full-length PS2, constructs encoding wild-type (H6M-Ub) and dominant-negative ubiquitin (H6M-UbK48R), tagged with poly((6ϫ)-histidine) were transiently transfected into uninduced cells. Lysates from PS2 cells that were either uninduced or induced for 12 h (when the predominant species is full-length PS2 and not HMW-PS2; Fig. 3) in the presence or absence of ALLN were then subjected to nickel affinity chromatography, and bound products (ubiquitinated proteins) were subjected to immunoblot analysis with the ␣PS2Loop antibody (Fig. 5C). In the cells which were not treated with ALLN, no polyubiquitinated HMW-PS2 was observed with the exception of a small amount in one lane (Fig.  5C, lane 6) where dominant-negative ubiquitin was transfected into PS2-expressing cells. In contrast, induced cells which were treated with ALLN and were transiently transfected with H6M-Ub revealed abundant amounts of polyubiquitinated HMW-PS2 (Fig. 5C, lane 10). Meanwhile, the induced cells which were treated with ALLN and were transiently transfected with dominant-negative mutant H6M-UbK48R revealed a small amount of polyubiquitinated HMW-PS2 (Fig. 5C, lane 12) similar to the amount observed in lane 6. These findings indicate that full-length PS2 can be modified by epitope-tagged ubiquitin in vivo to form ubiquitinated HMW-PS2 and degraded through the ubiquitin-proteasome pathway. Thus, polyubiquitinated HMW-PS2 likely serves as an intermediate for full-length PS2 degradation in this system.
To determine the cellular site for polyubiquitination and proteasomal degradation of PS2, we tested the effects of BFA. BFA is known to induce the disassembly of the Golgi complex and led to rapid degradation of full-length PS2 (Fig. 6). However, treatment with both BFA and ALLN resulted in even greater accumulation of HMW-PS2 than with ALLN alone, while no significant increase was observed for PS2-CTF (Fig.  6). Similar results were obtained using lactacystin (data not shown). These results suggest that polyubiquitination and proteasomal degradation of PS2 occur in a pre-Golgi compartment (e.g. ER). These data also indicate that the generation and accumulation of the PS2-CTF may be regulated by additional proteolytic enzymes located in other subcellular sites.
The ubiquitin-proteasome pathway is known to play a role in the selective turnover of intracellular protein substrates via complete degradation. However, this pathway also participates in the functional alteration of specific proteins by limited proteolysis or endocytosis following polyubiquitination (for review, see Refs. [22][23][24][25]. The proteasome has been shown to degrade ER proteins (35). Thus, the ubiquitination and subsequent  2 and 3). Immunoprecipitates were separated on 4 -20% SDS-PAGE, and polyubiquitinated PS2 ([Ub] n -PS2) was detected by Western blot analysis using monoclonal (Ubi-1) ubiquitin antibody. C, binding of PS2 modified with epitope-tagged ubiquitin (H6M-Ub) to nickel affinity columns. Plasmid pCW7 encoding wild-type (H6M-Ub) or pCW8 encoding dominant-negative ubiquitin (H6M-UbK48R) was transiently transfected into uninduced PS2 cells (clone WF2). Transiently transfected PS2 cells were either uninduced (lanes 1, 3, 5, 7, 9, and 11) or induced for 12 h (lanes 2, 4, 6, 8, 10, and 12) in the absence (lanes 1-6) or presence of (lanes 7-12) ALLN. Lysates were subjected to nickel affinity chromatography as described under "Experimental Procedures," and bound materials were analyzed by Western blotting using anti-PS2Loop antibodies. degradation of PS2 by the proteasome pathway may serve as a means for regulating the turnover of PS2 in the ER. The ubiquitin-proteasome pathway may also be utilized in the transfected cells to dispose of excess PS2 thereby regulating the amount of PS2 that is available for endoproteolysis. Since there are extremely low levels of endogenous full-length PS1 (17) and PS2 (44), it is tempting to speculate that the proteasome may also play a role in regulating full-length presenilin degradation endogenously. It is unlikely, however, that the proteasome carries out the endoproteolytic cleavage of PS2 into the NTF and CTF since proteasome inhibitors did not block the generation of these fragments (Figs. 5A and 6; data not shown for the PS2-NTF). Two ER proteins that have been shown to be degraded by an ALLN/lactacystin-sensitive proteasomal pathway are 3-hydroxy-3-methylglutaryl-coenzyme A reductase (36,37) and the sterol regulatory element-binding protein (38,39). Interestingly, 3-hydroxy-3-methylglutaryl-coenzyme A reductase, a key regulatory enzyme of cholesterol biosynthesis, has been shown to span the membrane eight times (40), similar to the topological model recently proposed for PS1 (41,42). Thus, PS2 is now the second putative eight-transmembrane domain protein which has been localized to the ER (8,11,43) and is degraded by the proteasomal pathway.
These data show that PS2 expressed in transfected H4 cell lines undergoes endoproteolytic cleavage, is ubiquitinated, and degraded via an ALLN/lactacystin-sensitive proteasome pathway. Although it is unclear how aberrations in the metabolism of the presenilins ultimately lead to altered processing of APP and increased production of A␤42, conformational changes of presenilins due to FAD mutations could conceivably alter their metabolism and adversely affect the processing of APP.