The presenilins (PS1 or PS2, for a review, see Sisodia and St George-Hyslop 2002) are 6–8 pass trans-membrane proteins (Doan et al. 1996; Li and Greenwald 1996) that localize to the endoplasmic reticulum (ER), Golgi complex, cell surface, as well as nuclear envelope, kinetochore and centrosome (De Strooper et al. 1997; Li et al. 1997; Georgakopoulos et al. 1999; Ray et al. 1999). The presenilins play a role in the production of amyloid β-peptide (Aβ) from the amyloid protein precursor (APP) (Borchelt et al. 1996; Citron et al. 1997). They have been shown to be essential for the gamma-cleavage of APP (De Strooper et al. 1998) and have been thought to contain the active site for γ-secretase and suggested to be part of a complex with γ-secretase activity (Wolfe et al. 1999; Esler et al. 2000; Li et al. 2000a,b; Seiffert et al. 2000). However, the exact function of presenilins is not clear. Functional orthologues of the human presenilins have been identified in both Caenorhabditis elegans and Drosophila. Sel-12 is one of three C. elegans presenilin homologs (sel-12, hop-1 and spe-4) (Levitan and Greenwald 1995; Li and Greenwald 1997; Westlund et al. 1999). Mutations in sel-12 cause a defect in egg laying by altering signaling through the Notch/lin-12 pathway (Levitan et al. 1995). The sel-12 mutant phenotype can be rescued by human PS1 or PS2 (Levitan et al. 1996), indicating that PS1, PS2 and sel-12 are functional homologues. The sel-12 mutant phenotype can be suppressed by loss of function mutations in a second gene, SEL-10 (Wu et al. 1998), which probably results in rescue of the egg laying defect by increasing the activity of a functionally redundant presenilin, hop-1. SEL-10 protein is a homologue of yeast Cdc4, a member of the SCF (Skp1-Cdc53/CUL1-F-box protein) E2-E3 ubiquitin ligase family (Hubbard et al. 1997). The SCF E2-E3 ubiquitin ligasescontain a catalytic core consisting of Skp1, Rbx1, Cdc53/CUL-1 and an E2 ubiquitin transferase, Cdc34 (Tyers and Willems 1999). These are targeted to substrates for ubiquitination by adapter proteins (e.g. Cdc4, Grr1, Met30, β-TrCP) containing an F-box motif and WD40 repeats (Patton et al. 1998; Winston et al. 1999). Human SEL-10 (also named as hCdc4 as for human Cdc4) was recently reported to mediate theubiquitination and turnover of phosphorylated cyclin E andthe gene encoding hCdc4 was found to be mutated in a cellline derived from breast cancer that expressed extremely high levels of cyclin E (Strohmaier et al. 2001). Human SEL-10 has also recently been shown to negatively regulate Notch signaling by mediating ubiquitination of Notch throughinteraction (Öberg et al. 2001; Wu et al. 2001). There is evidence that presenilinsare ubiquitinated and undergo degradation through the ubiquitin-proteasome pathway (Kim et al. 1997; Marambaud et al. 1998; Steiner et al. 1998). Physical interaction between C. elegans SEL-10 and SEL-12 has been shown previously (Wu et al. 1998). Thus, SEL-10 may be the F-box adapter protein that recruits presenilins for ubiquitination and subsequent degradation. In this study, we show that human SEL-10 interacts with PS1 and enhances PS1ubiquitination, thus altering cellular levels of unprocessed PS1 and its N- and C-terminal fragments. This leads to an alteration in the metabolism of APP and to an increase in the production of amyloid β-peptide, the principal component of amyloid plaque in Alzheimer's disease (AD) (Selkoe 1993).
Mutations in the human presenilin genes (PS1 or PS2) have been linked to autosomal dominant, early onset Alzheimer's disease (AD). Presenilins, probably as an essential part of gamma-secretase, modulate gamma-cleavage of the amyloid protein precursor (APP) to the amyloid β-peptide (Aβ). Mutations in sel-12, a Caenorhabditis elegans presenilin homologue, cause a defect in egg laying that can be suppressed by loss of function mutations in a second gene, SEL-10. SEL-10 protein is a homologue of yeast Cdc4, a member of the SCF (Skp1-Cdc53/CUL1-F-box protein) E2-E3 ubiquitin ligase family. In this study, we show that human SEL-10 interacts with PS1 and enhances PS1 ubiquitination, thus altering cellular levels of unprocessed PS1 and its N- and C-terminal fragments. Co-transfection of sel-10 and APP cDNAs in HEK293 cells leads to an alteration in the metabolism of APP and to an increase in the production of amyloid β-peptide, the principal component of amyloid plaque in Alzheimer's disease.
amyloid precursor protein
the cystic fibrosis trans-membrane conductance regulator
expression sequence tag
fluorescence-activated cell sorter
human embryonic kidney cells
presenilin 1 C-terminal endoproteolytic fragment
presenilin 2 N-terminal endoproteolytic fragment
Materials and methods
Cloning and northern analysis
Incyte clone (028971) was identified as the human homologue ofC. elegans sel-10 and its sequence was used to design fourantisense oligonucleotide primers (TCACTTCATGTCCACATCAAAGTCC, GGTAATTACAAAGTTCTTGTTGAACTG, CCCTGCAACGTGTGTAGACAGG, and CCAGTCTCTGCATTCCACACTTTG) to amplify the remainder of the human SEL-10 sequence. ‘Electronic northern’ analysis revealed expression of SEL-10 in hippocampus and mammary gland so these tissues werechosen for 5′ rapid amplification of cDNA ends (RACE) cloning using Marathon kit (CloneTech, Palo Alto, CA, USA). Marathon-ready cDNAs from hippocampus and mammary gland were prepared as directed in the kit. PCR products were cloned into the TA vector pCR3.1 (Invitrogen, Carlsbad, CA, USA), and isolates were sequenced. An alternate 5′ oligonucleotide primer was also designed based on Incyte clones that have 5′ ends that differ from the hippocampal SEL-10 sequence (CTCAGACAGGTCAGGACATTTGG). Blastn was used to search the Incyte databases LifeSeq and LifeSeqFL. Gap alignments and translations were performed with GCG programs (University of Wisconsin Genetic Computer Group, Madison, WI, USA). Northern analysis was conducted with standard procedure using 32P-labeled cDNA or PCR product from a specific region of the cDNA.
Plasmids and transfections
The human SEL-10 cDNA was inserted into the EcoR1 site of the vector pCS2 + MT (gift of Jan Kitajewski, Columbia University College of Physicians and Surgeons). This fused a 5′ 6-myc epitope tag in-frame to the fifth methionine of the hippocampal SEL-10 cDNA. The hippocampal and mammary (as referred to ‘hippocampus form’ and ‘common form’) SEL-10 cDNA diverge upstream of this methionine. A PS1 cDNA with a 3′-FLAG tag (PS1-C-FLAG) was inserted into the pcDNA3.1 vector. An APP cDNA containing the Swedish KM→NL mutation and an attenuated ER retention sequence consisting of a C-terminal di-lysine motif (APP695Sw-KK) was inserted into the pIRES-EGFP vector (Clontech). Human embryonic kidney (HEK293) cells were grown to 80% confluence in Dulbecco's modified Eagle's medium (DMEM) with 10% (v/v) fetal bovine serum (FBS) and transfected with the above cDNAs. A total of 10 µg DNA/6 × 106 cells was used for transfection with a single plasmid. For co-transfection of multiple plasmids, an equal amount of each plasmid was used for a total of 10 µg DNA using LipofectAmine (Gibco BRL, Rockville, MA, USA). Single cells were sorted into each well of one 96-well plate containing growth medium without G418 by fluorescence-activated cell sorter (FACS) using an EPICS Elite ESP flow cytometer (Coulter, Hialeah, FL, USA) equipped with a 488-nm excitation line supplied by an air-cooled argon laser. After a 4-day recovery period, G418 was added to the medium to a final concentration of 400 µg/mL for the selection of stably transfected cell lines. Stable expression of cDNA was confirmed by detection of the specific protein or sequence tag by western blot.
Immunoprecipitation, western blot and ELISA
PS1 and APP antibodies have been characterized (Citron et al. 1997; Li et al. 1997; Mehta et al. 1998) and ubiquitin antibody was from Dako, Carpinteria, CA, USA. Cell lysates with equal amount of protein were precipitated with antibody and protein G-Sepharose at 4°C for 2 h and the precipitated beads were washed with TENT (50 nm Tris-HCl pH 8.0, 2 mm EDTA, 150 mm NaCl, 1% Triton X-100) buffer (Wu et al. 1998) then analyzed by western blot using 4–12% NuPage Bis-Tris gel and PVDF membrane (Novex, San Diego, CA, USA), peroxidase-conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA) and SuperSignal West Pico Luminol/Enhancer (Pierce, Rockford, IL, USA) following the procedures by the manufacturers. The ELISA for Aβ 1–40 and Aβ 1–42 in conditioned media from the cell cultures was performed as described (Mehta et al. 1998). For each experimental condition, each medium sample of the three identical cultures was measured twice and the data were analyzed with standard t-test.
Cloning of human and mouse SEL-10 cDNA
We have reported the human SEL-10 protein sequence recently in a collaborative publication (Wu et al. 2001). Both human and mouse orthologs of C. elegans SEL-10 have been identified in expression sequence tag (EST) databases, although the sequence information is incomplete (Hubbard et al. 1997). We used rapid amplification of cDNA ends (RACE) to clone amplification products containing the full coding sequences for human and mouse SEL-10 (GenBank Accession Number). The human SEL-10 gene was localized to chromosome 4q31.2–31.3 by in situ hybridization (data not shown). The predicted protein sequences of human and C. elegans SEL-10 have 47.6% amino acid identity. Human SEL-10 contains an F-box domain as found in other SCF family members (Fig. 1). It also contains seven WD40/β-transducin repeats (Neer et al. 1994) as seen in yeast Cdc4p and C. elegans SEL-10, suggesting that the protein forms a seven-bladed propeller structure (Sondek et al. 1996) (Fig. 1). There are two variants of human SEL-10 cDNA isolated from mammary gland and hippocampus libraries, named ‘common form’ and ‘hippocampus form’ cDNAs, respectively, which differ in 5′ sequence upstream of a common translation initiation site (Fig. 2a). Analysis of the human SEL-10 gene sequence showed that there are 11 exons in SEL-10 gene. Exons 3–11 encode the common region of the two forms of SEL-10, while exon 1 encodes the N-terminal part of common form and exon 2 encodes the N-terminal part of hippocampal form SEL-10. This indicates that the human SEL-10 transcript undergoes differential splicing in a tissue-specific fashion or that the gene contains multiple, tissue specific promoters. The common form contains three in-frame methionines upstream of the common initiation site and the hippocampal form contains four. Whether or not these encode proteins with different N-termini is not known. The common form transcript is ubiquitously expressed as two isoforms of 4 and 5.5 kb (Fig. 2b). The 5.5-kb isoform was detected in all tissues tested except kidney, while the 4-kb isoform was only detected in heart, brain and muscle. The hippocampal form of 4-kb can be detected only in brain (Fig. 2b).
Interaction of human SEL-10 and PS1
We first assessed the physical interaction of human SEL-10 and PS1. SEL-10, tagged with an N-terminal 6-Myc epitope (SEL-10-myc), and PS1 were transiently coexpressed in human embryonic kidney cells (HEK293) and their interaction was assessed by immunoprecipitation. Complexes between SEL-10 and PS1 could only be detected in the presence of a proteasome inhibitor, lactacystin, which was added to the cultures at the time of transfection (Fig. 3a). When immunoprecipitated with anti-PS1 loop antibody, which was raised against the loop between trans-membrane domains six and seven and recognizes the N-terminal and C-terminal domains of PS1 as well as the full-length protein (Li et al. 1997), only the immunoprecipitate from co-transfected cells contained SEL-10-myc, indicating that SEL-10 can interact with PS1. As co-immunoprecipitation was not observed in cells without lactacystin treatment, this suggests that the complex can only be captured by blocking the entry of PS1 into the proteasome degradation pathway. The interaction between SEL-10 and PS1 was confirmed in the reverse experiment by using anti-myc antibody for immunoprecipitation of SEL-10. PS1 is cleaved within the cytoplasmic loop between trans-membrane domains six and seven by an unknown protease that generates N- and C-terminal fragments (PS1-NTF and PS1-CTF, respectively) (Thinakaran et al. 1996). The SEL-10-myc immunoprecipitates contained primarily full length and high molecular weight forms of PS1 (Fig. 3b), but very low amounts of PS1-NTF and undetectable PS1-CTF (data not shown). This was probably due to the accumulated high level of full-length PS1 and low level of PS1-NTF and -CTF in transiently transfected cells. This observation was similar to the previous report that co-immunoprecipitation of C. elegans SEL-10 with SEL-12 also brings down unprocessed SEL-12, but not detectable N- and C-terminal fragments (Wu et al. 1998).
Effect of SEL-10 on ubiquitination and protein level of PS1
Next, we examined the effect of SEL-10 transfection on PS1 ubiquitination. PS1 immunoprecipitated from co-transfected cells contains a higher level of ubiquitination compared to cells transfected with PS1 alone, as shown by probing with anti-ubiquitin antibody (Fig. 4a). This result suggests that complex formation between SEL-10 and PS1 facilitates ubiquitination as implied previously by the need for lactacystin to demonstrate SEL-10/PS1 complex accumulation.
We then investigated how ubiquitination affects PS1 protein level. HEK293 cells were co-transfected with either SEL-10 or PS1, alone or in combination, and immunoprecipitates were probed with anti-PS1 loop antibody (Fig. 4b). In cells co-transfected with PS1 and SEL-10 in comparison with cells transfected with PS1 alone, PS1-NTF and PS1-CTF were decreased. These observations suggest that SEL-10 mediated ubiquitination of PS1-CTF and PS1-NTF leads to their degradation. No obvious change in the amount of unprocessed PS1 was observed.
It is notable that Fig. 4a shows more high molecular weight (HMW) PS1 in cells transfected with PS1 and SEL-10 than in the cells transfected with PS1 alone, while Fig. 4b shows similarly high level of HMW PS1 in both transfection conditions. The HMW signals in Fig. 4a, detected by anti-PS1L immunoprecipitation followed by western blot using anti-ubiquitin antibody, represent only ubiquitinated PS1. Therefore, there is a clear difference in the HMW PS1 level between transfections with and without SEL-10. Meanwhile, the HMW signals in Fig. 4b, detected by anti-PS1 immunoprecipitation followed by anti-PS1 western blot, represent all forms of HMW PS1 including ubiquitinated PS1 as well as detergent-resistant PS1 complexes as previously reported for PS2 (Kim et al. 1997).
Effect of SEL-10 on Aβ production
In order to examine the impact of SEL-10 mediated ubiquitination of PS1 on amyloid β-peptide (Aβ) production, SEL-10 with c-myc tag was expressed by either transient or stable transfection in HEK293 cells, with or without co-expression of wild-type PS1 or APP with Swedish mutation. Aβ peptides in the conditioned media of the cell cultures were measured by enzyme immunoassays that could distinguish the 1–40 and 1–42 forms of the peptide (Aβ1–40 and Aβ1–42). In transient expression experiments (Fig. 5a), the endogenous Aβ peptides production was not changed by transient single or co-expression of SEL-10 and PS1. Transient expression of PS1 alone did not change Aβ peptides production as well, as constantly reported (for a review, see Kovacs and Tanzi 1998). However, transient expression of SEL-10 with APP elevated the exogenous Aβ1–40 and 1–42 peptides production by more than two-fold in comparison with APP expression alone. Transient co-expression of PS1 with APP also increased Αβ1–40 and Αβ1–42 levels by more than three-fold, similar to previous reports (Ancolio et al. 1997) although this is not a consistent finding (Li and Greenwald 1996). Co-expression of SEL-10 and PS1 with APP had an additive effect with an increase in Αβ production of approximately seven-fold. No effect on the ratio of Αβ1–42/total Αβ was observed with all values falling in the range 8.5% to 13.5%. The stimulation of Αβ production upon expression of SEL-10 was consistently observed across experiments; however, the degree of stimulation did vary. To confirm and extend the result, a series of HEK293 cell lines were derived with stable expression of SEL-10 (Fig. 5b). Endogenous PS1-CTF level in the two stable SEL-10 cell lines was decreased compared with the stable cell line transfected with control vector pcDNA3.1. The two SEL-10 cell lines showed four- and seven-fold increases in endogenous Aβ1–40 peptide production compared with HEK293 cells transfected with just the pcDNA3.1 vector. No significant change in Aβ1–42 peptide was observed in these two stable SEL-10 cell lines, indicating the elevated Aβ1–40 peptide was not due to a general variation of the cell lines. Similar to the endogenous Aβ level, the exogenous Aβ1–40 and Aβ1–42 levels, upon transient transfection with APP, were higher in the SEL-10 stable cell lines then the control pcDNA3 cell line. Transient expression of PS1 has a similar effect. It increases the cellular level of unprocessed PS1 while having relatively little effect on the levels of PS1-NTF and PS1-CTF fragments (data not shown).
Our data indicate that SEL-10 interacts with PS1, stimulates PS1 ubiquitination, and probably recruits the PS1 into the proteasome pathway for protein degradation. SEL-10 is likely to function as an adaptor protein that assembles the core catalytic complex of an SCF E2-E3 ubiquitin ligase (Patton et al. 1998). Recognition of most SCF substrates by F-box/WD40 repeat adaptor proteins is phosphorylation dependent (Steiner et al. 1998), suggesting that this may be an additional level of cellular regulation of presenilin levels. Indeed, phosphorylation of presenilins has been well reported (Walter et al. 1996; De Strooper et al. 1997). It required inhibition of proteasome function by lactacystin to demonstrate complex formation between human SEL-10 and PS1 in HEK293 cells. In contrast, the C. elegans SEL-10/SEL-12 complex accumulates in human HEK293 cells in the absence of proteasome inhibitors, suggesting that nematode SEL-10 is unable to assemble the human core catalytic complex (Wuet al. 1998). Also, co-immunoprecipitation brings down unprocessed SEL-12 with C. elegans SEL-10, but notubiquitinated, high molecular weight forms of SEL-12.
It is interesting that SEL-10 transient and stable transfections increase Aβ level while the cellular level of PS1-NTF and PS1-CTF decreases in the transfected cells, because a decreased PS1-NTF and -CTF level would be expected to decrease Aβ level. This observation is unexpected but not surprising. It has been reported that down-regulation of PS1 expression by stably inducible antisense transfection increased Aβ1–42 secretion by as much as fivefold, which correlated with a 60% decrease in PS1-NTF (Refolo et al. 1999). How decrease of PS1-NTF level is related to the increase of Aβ level remains an interesting mechanism to resolve. Another possible explanation for the increase of Aβ level by SEL-10 transfection is that SEL-10-mediated ubiquitination might modulate PS1 function in APP processing. Although the best-characterized task of ubiquitin is targeting proteins for destruction by the proteasome, a growing body of evidence is indicating other functions of ubiquitin. Such proteolysis-independent functions include ubiquitin-dependent endocytosis, transcriptional control and DNA repair (for reviews, see Jentsch and Schlenker 1995; Pickart 2001; Weissman 2001). It is important to explore how SEL-10 mediates ubiquitination of PS1 and modulates APP processing.
Recently it was reported that SEL-10 interacts with Notch and mediate its ubiquitination ( Öberg et al. 2001; Wu et al. 2001). Interestingly SEL-10 showed various effects on Notch function. Inhibition of SEL-10 activity by expression of dominant negative WD40 domains of SEL-10 caused increase of Notch4 signaling, but not Notch1 (Wu et al. 2001), suggesting a different fate of a protein upon ubiquitination other than degradation. Also, ubiquitinated Notch accumulated in the presence of proteasome inhibitor were not as potent in signaling as those not ubiquitinated (Öberg et al. 2001), suggesting a modification role of ubiquitination by SEL-10, which parallels to our prediction that SEL-10 may modify the cleavage of PS1 and modulate Aβ production. In light of these observations, the observed increase of Aβ production upon over expression of SEL-10 may also reflect an ubiquitination-mediated modification of the function of presenilins in γ-cleavage of APP.
Human SEL-10 was also recently reported to mediate the ubiquitination and turnover of phosphorylated cyclin E (Strohmaier et al. 2001). Cyclin E is one of the activators of the cyclin-dependent kinase Cdk2 and is thought to be critical for the initiation of DNA replication and therefore controls S phase progression. Interestingly, the involvement of presenilins in cell division has also been suggested by previous studies. PS1 has been reported to be associated with nuclear membrane, centrosomes and kinetochores (Li et al. 1997). Antisense against PS1 and PS2 decreased the transition to blastocyst stage in mouse embryos and PS2 over-expression in HEK 293 cell-arrested cell cycle at S phase (Jeong et al. 2000). These data suggest that presenilins play key roles in cell division and differentiation. Therefore the common role for SEL-10/hCdc4 in ubiquitination of PS1, notch and cyclin E may suggest a common feature of these proteins in the control of cell division and differentiation. It is notable that there are two splicing isoforms of SEL-10 mRNA (Fig. 2). The dominant expression of the ‘common form’ SEL-10 in brain and skeletal muscle suggest its role in ubiquitination of PS1 that is alsodominantly expressed in brain and skeletal muscle (Levy-Lahad et al. 1995; Li et al. 1995; Sherrington et al. 1995). It will be interesting to determine functional difference between the two isoforms.
Degradation of an eight-pass trans-membrane protein such as presenilin presents a topological hurdle, as the presenilin protein must be extracted from the membrane and delivered to the proteasome. Like presenilins, a number of other multipass integral membrane proteins such as the cystic fibrosis trans-membrane conductance regulator (CFTR) are degraded through the ubiquitin-proteasome pathway (Ward et al. 1995; Johnston et al. 1998). Over-expression or inhibition of proteasome activity allows accumulation of ubiquitinated, high molecular weight forms of CFTR and PS1 in a distinct pericentriolar structure that has been termed the aggresome. Pericentriolar localization of PS1 is seen in normal cells, suggesting that a portion of PS1 may be targeted to the aggresome pathway (Citron et al. 1997). Perhaps the more than 80 different missense mutations in PS1 associated with early onset Alzheimer's disease cause protein mis-folding and enhance the propensity of mutant PS1 to enter the aggresome pathway. This conceivably could have impact on intracellular production of Αβ peptide and could explain the enhancement of Aβ production caused by increased expression of SEL-10.
The authors wish to thank Iva Greenwald, Howard Hughes Medical Institute, and Jan Kitajewski, Columbia University College of Physicians and Surgeons for their numerous intellectual contributions to this work.