Proteolytic processing of low density lipoprotein receptor-related protein mediates regulated release of its intracellular domain.

The low density lipoprotein (LDL) receptor-related protein (LRP) is a multifunctional cell surface receptor that interacts through its cytoplasmic tail with adaptor and scaffold proteins that participate in cellular signaling. Its extracellular domain, like that of the signaling receptor Notch and of amyloid precursor protein (APP), is proteolytically processed at multiple positions. This similarity led us to investigate whether LRP, like APP and Notch, might also be cleaved at a third, intramembranous or cytoplasmic site, resulting in the release of its intracellular domain. Using independent experimental approaches we demonstrate that the cytoplasmic domain is released by a gamma-secretase-like activity and that this event is modulated by protein kinase C. Furthermore, cytoplasmic adaptor proteins that bind to the LRP tail affect the subcellular localization of the free intracellular domain and may regulate putative signaling functions. Finally, we show that the degradation of the free tail fragment is mediated by the proteasome. These findings suggest a novel role for the intracellular domain of LRP that may involve the subcellular translocation of preassembled signaling complexes from the plasma membrane.

The low density lipoprotein (LDL) receptor-related protein (LRP) is a multifunctional cell surface receptor that interacts through its cytoplasmic tail with adaptor and scaffold proteins that participate in cellular signaling. Its extracellular domain, like that of the signaling receptor Notch and of amyloid precursor protein (APP), is proteolytically processed at multiple positions. This similarity led us to investigate whether LRP, like APP and Notch, might also be cleaved at a third, intramembranous or cytoplasmic site, resulting in the release of its intracellular domain. Using independent experimental approaches we demonstrate that the cytoplasmic domain is released by a ␥-secretase-like activity and that this event is modulated by protein kinase C. Furthermore, cytoplasmic adaptor proteins that bind to the LRP tail affect the subcellular localization of the free intracellular domain and may regulate putative signaling functions. Finally, we show that the degradation of the free tail fragment is mediated by the proteasome. These findings suggest a novel role for the intracellular domain of LRP that may involve the subcellular translocation of preassembled signaling complexes from the plasma membrane.
Seven structurally closely related cell surface receptors constitute the core of the low density lipoprotein (LDL) 1 receptor gene family. They include the LDL receptor, the LDL receptorrelated protein (LRP or LRP1), LRP1b, megalin, very low density lipoprotein (VLDL) receptor, apolipoprotein E receptor 2 (apo-ER2), and MEGF7 (multiple epidermal growth factor-like domains containing protein 7) (1). Although the members of this evolutionarily ancient gene family all share the same typ-ical arrangement of protein modules in their extracellular domains, the biological functions of the individual members of the family are highly diverse and include roles in cellular ligand uptake and endocytosis, the transmission of extracellular signals, vitamin and cholesterol homeostasis, brain development, the modulation of neurotransmission, and protection from neurodegeneration (1,2).
Of this entire receptor family, LRP interacts with by far the largest number of proteins (3). Ligands that bind to its extracellular domain include, for instance, ␣2-macroglobulin, plasminogen activators, clotting factors, lipases, and the amyloid precursor protein (APP). The cytoplasmic tail of LRP also interacts with an extended set of intracellular adaptor and scaffold proteins, e.g. Dab1, c-Jun amino-terminal kinase interacting proteins, the postsynaptic density protein PSD-95 (4), and the trivalent scaffold protein FE65 (5). The latter protein contains two phosphotyrosine binding domains, of which the first binds to the LRP tail, whereas the second domain interacts with an NPXY motif in the cytoplasmic tail of APP (6 -9).
Mice in which the LRP gene has been disrupted by homologous recombination die early during embryonic development (10). The phenotype of the few malformed LRP-deficient embryos that survive until around E9 -E10 is complex and difficult to explain by defects in ligand endocytosis alone, suggesting that LRP, like the VLDL receptor and the apo-ER2 (11), could have essential developmental signaling functions that involve interactions of its cytoplasmic tail with the intracellular signal transduction machinery.
LRP is a type I integral membrane protein, like the other members of the family. However, only LRP and its closest relative, LRP1b, are cleaved at a site that matches the consensus sequence recognized by furin (12,13), a processing proteinase that frequently serves to activate or modify cell surface receptors and secreted proteins upon exit from the secretory pathway (14). Furin cleavage of the 600-kDa LRP precursor protein produces the mature cell-surface receptor, which consists of a carboxyl-terminal 85-kDa fragment and a non-covalently attached 515-kDa amino-terminal subunit.
Notably, furin also cleaves members of the Notch family of cell surface signaling proteins, which, like LRP, contain multiple EGF repeats in their extracellular domain (15,16). Additionally, Notch proteins undergo another metalloproteinasemediated processing step at the cell surface (17,18), which is followed by the presenilin/␥-secretase-dependent release of their intracellular domains (ICD) (19,20,47). Notch-ICD translocates to the nucleus where it stimulates expression of target genes through interaction with transcription factors of the CSL family (21).
Interestingly, shedding of LRP from the cell surface has also been reported (22), and it, too, involves a metalloproteinase (23). Furthermore, LRP can form a complex with APP by interactions of both the extracellular (24) and the intracellular domains (5,25). APP also undergoes several proteolytic processing events that, as in Notch, lead to the presenilin/␥-secretasedependent release of the cytoplasmic tail. Recently it was shown that FE65 binding to the APP tail not only greatly enhances nuclear translocation of the complex but that this also serves to recruit transcriptional activators, which could potentially stimulate the expression of target genes that are regulated by APP cleavage events (26).
The purpose of the current study was to investigate whether LRP, in a manner analogous to Notch and APP, may also be subject to intramembranous or cytoplasmic proteolytic processing that results in the release of its intracellular domain from the membrane. Our results demonstrate that such a cleavage event occurs, that a protease with ␥-secretase like properties is involved, and that intracellular proteins such as FE65 under certain conditions can stimulate the translocation of the tail fragment into the nucleus. The amount of biologically active cleaved tail fragment can be regulated by activation of protein kinase C (PKC) by phorbol esters and by inhibition of the proteasome.
Plasmids and Construction of the LRP-Gal4/VP16 Vector-For construction of LRP-Gal4/VP16 the 12,659 -14,098 HindIII fragment of the human LRP cDNA (28) was amplified by PCR using the following primers, which introduced an NheI site 5Ј to the endogenous HindIII site at 12,659, eliminated the stop codon at nucleotide 14,099 and introduced an NheI site at this position: 5Ј-GCTAGCCAAGCTTTCAGT-CATCGGCAGCATCCGGCTC-3Ј (sense-LRP), 5Ј-GCTAGCGCCAAGG-GGTCCCCTATCTCGTCCTCAGG-3Ј (antisense-LRP). The PCR product was cloned into the plasmid pMstGV via its NheI sites. pMstGV is based on the commercially available vector pM (CLONTECH, Palo Alto, CA). It codes for a VP16 transactivation domain 3Ј-terminal to the Gal4 binding domain of pM and does no longer carry a stop codon 5Ј-terminal to the coding sequence for Gal4. The coding sequences for the LRP fragment plus the Gal4 and VP16 domains were then excised by Hind-III and ligated into the HindIII cut pcDNA3.1-LRP vector (28,29). Orientation and sequences of inserts were verified by DNA sequencing.
For the construction of pcDNA3.1-LRP-Gal4, the sequence coding for the VP16 domain was excised from the pMstGV-LRP fragment intermediate product with EcoRI and BamHI. Then the newly generated ends of the plasmid were made blunt in a fill-in reaction with the Klenow enzyme (Roche Molecular Biochemicals, Mannheim, Germany). The vector was then re-ligated and sequenced. Its small HindIII fragment was cloned into the HindIII-precut pcDNA3.1-LRP plasmid as described.
Primers 5Ј-GAGATTGGAAACCCCGCCTACAAGATG-TACGAA-3Ј (sense-T/A) and 5Ј-TTCGTACATCTTGTAGGCGGGGTT-TCCAATCTC-3Ј (antisense-T/A) were used to introduce a point mutation into the first NPXY motif NPTY of the cytoplasmic tail of LRP thereby changing its threonine residue to alanine. The same fragment of LRP that was used for the construction of pcDNA3.1-LRP-Gal4/V16 was amplified by PCR in two overlapping pieces using the sense-LRP and antisense-LRP-T/A primers for the first part and sense-LRP-T/A and antisense-LRP for the second part. These two overlapping fragments, together with the LRP-sense and LRP-antisense primers, were used in a second PCR to generate the whole LRP-HindIII fragment bearing the T/A mutation in the sequence coding for the cytoplasmic tail of LRP. The subsequent cloning steps were the same as described for pcDNA3.1-LRP-Gal4/VP16. pMstGV (pGal4/VP16) was generously provided by X.-W. Cao and T. Sü dhof, University of Texas Southwestern Medical Center, Dallas, TX, as were the vectors pMstGV-APP695 (pAPP-Gal4/VP16) and pMstGV-LDLR (pLDLR-Gal4/VP16) and the pG5E1B-Luc plasmid, which contains five Gal4 binding sites and the E1B minimal promoter in front of the luciferase firefly gene. pRK5-FE65 (9) and pcDNA3.1-mDab1 (5) have been described previously. pCMV-␤-Gal was obtained from Stratagene (La Jolla, CA).
Cell Culture and Transfection of 293 Cells-The human embryonic kidney cell line HEK 293 (CRL-1573, ATCC) was maintained in monolayer culture at 37°C in an 8% CO 2 atmosphere. On day 0 cells were set up at a density of 5 ϫ 10 5 per 60-mm dish in Dulbecco's modified Eagle's medium (DMEM, glucose, 1 g/liter, Cellgro, Herndon, VA) containing 100 units/ml penicillin, 100 g/ml streptomycin sulfate (Cellgro) and 10% (v/v) fetal calf serum (FCS) (Atlanta Biologicals, Norcross, GA). On day 2 cells were transfected by calcium phosphate precipitation (MBS kit, Stratagene, La Jolla, CA) with 1.5 g of the pG5E1B-Luc plasmid per 60-mm dish, 0.05 g of pCMV-␤-Gal for internal control of transfection efficiency, the respective receptor construct (1 g/60-mm dish), and expression vectors for different adapter proteins or other components as indicated in the text. On day 3 treatment with PMA or other reagents was initiated. PMA was solubilized in Me 2 SO at a concentration of 1 mM. For stimulation of transfected 293 cells the cell culture medium (DMEM-low glucose ϩ 10% (v/v) FCS ϩ 100 units/ml penicillin/ 100 g/ml streptomycin) was exchanged for medium containing 100 nM PMA, and the cells were cultured for another 24 h before harvesting. Culture of N2aSW10 and N2a385. 16 Cells-N2a cells stably expressing the APP "Swedish mutant" (N2aSW10) (30) or the APP Swedish mutant and a dominant negative presenilin-1 (D385A-PS1) (31) were kindly provided by Sangram S. Sisodia, Chicago, IL. Cells were grown in 50% DMEM/50% Opti-MEM (Invitrogen, Rockville, MD) supplemented with 5% (v/v) FCS, 100 units/ml penicillin, 100 g/ml streptomycin sulfate, and 200 g/ml G418 (Invitrogen) for N2aSW10 cells or 200 g/ml G418 and 200 g of hygromycin (Invitrogen) for N2a385.16 cells, in a 8% CO 2 atmosphere at 37°C.
Reporter Gene Assays-293 and N2a cells were harvested for reporter gene assays 48 h after transfection. The culture medium was removed, and the cells were washed once with PBS. After addition of 400 l of Reporter Lysis Buffer (Promega, Madison, WI) per 60-mm dish, cells were processed according to the manufacturers' instructions. Luciferase gene expression was analyzed using the Luciferase Assay System (Promega). A 10-s integral measurement of light emission from each reaction mixture was performed in a tube luminometer (Optocomp II, MOM Instruments).
␤-Galactosidase assays for internal control of transcription efficiency were carried out with an aliquot of the cell lysates prepared for the luciferase assay using the Chemiluminescent ␤-Gal Detection Kit (CLONTECH, Palo Alto, CA). The normalization of luciferase activity to ␤-galactosidase activity was done by dividing the relative light units values obtained with the first assay by those from the latter reaction. All transfections for reporter gene assays were done in duplicate and repeated in at least two independent experiments.
Preparation of Cell Membranes for Western Blot Analysis-Cells were grown to confluency in 100-mm culture dishes. For the preparation of membranes, culture medium was aspirated and cells were washed twice with ice-cold PBS. Cells were scraped into 1 ml of phosphate-buffered saline (PBS) per dish and centrifuged at 850 ϫ g at 4°C. Cells were resuspended in hypotonic buffer A (10 mM HEPES-KOH, pH 7.8, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 1 mM NaVO 3 ), left on ice for 20 min, and then passed 10 times through a 22-G needle. After centrifugation at 850 ϫ g for 5 min at 4°C, the supernatant was recentrifuged at 100,000 ϫ g for 30 min. Finally, the membrane pellet was resuspended in SDS buffer (10 mM Tris-HCl, pH 6.8, 100 mM NaCl, 1% SDS, 1 mM EDTA).
In Vitro Generation of Free LRP Tail-The assay was adapted from the protocol described by Sastre et al. (32) for the in vitro generation of the APP carboxyl-terminal fragment. Briefly, 293 cells were transfected with 5 g of pcDNA3.1-LRP expression plasmid per 100-mm dish by calcium phosphate precipitation. After 2 days the cells were harvested and washed with ice-cold PBS as described above. The cell pellet was resuspended in 0.5 ml of homogenization buffer per 100-mm dish (10 mM MOPS, pH 7.0, 10 mM KCl, 1 pellet per 10 ml of Complete mini proteinase inhibitor mixture from Roche Molecular Biochemicals, Mannheim, Germany) and incubated on ice for 10 min. Cells were broken by passing through a 22-G needle and centrifuged at 1000 ϫ g for 15 min at 4°C. The supernatant was recentrifuged at 16,000 ϫ g for 20 min at 4°C. The membrane pellet was then washed once with ice-cold homogenization buffer and was resuspended in 50 l of assay buffer (150 mM sodium citrate, pH 6.4, proteinase inhibitors as above). Membrane samples were divided into two 25-l aliquots and incubated in the absence or presence of 500 nM DAPT at 0°C or at 37°C for 1 h. Samples were then centrifuged at 100,000 ϫ g for 1 h at 4°C, and pellets and supernatants were analyzed by 10 -20% SDS-gel electrophoresis.
SDS-PAGE and Immunoblot Analysis-SDS-gel electrophoresis and immunoblot analysis of whole cell or membrane lysates were performed according to standard procedures on 4 -12% or 10 -20% SDS-polyacrylamide gels. After electrophoresis and protein transfer to polyvinylidene difluoride membranes, immunoblot analysis was carried out with rabbit polyclonal antibodies against a carboxyl-terminal epitope of LRP (28). Bound antibodies were visualized by enhanced chemiluminescence (ECL) using SuperSignal CL-HRP Substrate (Pierce, Rockford, IL) according to the manufacturer's instructions.

RESULTS
The ␥-secretase-mediated intramembranous cleavage events that in Notch and APP lead to the release of the proteins' intracellular domains into the cytoplasm occur at a site preceding a valine located four or three residues, respectively, from the cytoplasmic face of the membrane (32). The transmembrane segment of LRP contains two valine residues at a corresponding position (Fig. 1) raising the possibility that LRP might be cleaved at the same relative position and that this cleavage might also involve ␥-secretase.
To test whether the cytoplasmic tail of LRP can be released by intramembranous or intracellular proteolysis, we constructed an expression vector encoding a chimeric protein, in which a yeast Gal4 DNA binding domain and a VP16 transactivation domain from herpes simplex virus were fused to the carboxyl-terminal end of the cytoplasmic tail of full-length LRP (pLRP-Gal4/VP16; Fig. 2A). Proteolytic release of the tail-Gal4/ VP16 fusion protein from the membrane was detected by measuring Gal4/VP16-dependent luciferase gene expression from a reporter construct containing multiple Gal4 binding motifs and the adenoviral E1B minimal promoter, followed by the luciferase gene (pG5E1B-Luc). A similar experimental strategy had previously been employed to detect the release of Notch ICD and for identifying the carboxyl-terminal fragment of APP as part of a transcriptionally active complex (19,20,26).
Proteolytic Release of the LRP Cytoplasmic Domain-To determine whether the LRP tail can be proteolytically released from the membrane, we transfected HEK 293 cells with pLRP-Gal4/VP16, pG5E1B-Luc, and with a ␤-galactosidase expression vector (pCMV-␤Gal) to control for transfection efficiency. Transcriptional stimulation by a soluble Gal4/VP16 fusion protein or by an LDL receptor-Gal4/VP16 (LDLR-Gal4/VP16) chi-meric protein, which is not expected to undergo processing, served as positive and negative controls, respectively. LRP-Gal4/VP16 stimulated luciferase activity by ϳ80 fold, whereas LDLR-Gal4/VP16 had no effect (Fig. 2B), even though expression of the shorter LDLR-Gal4/VP16 fusion protein was considerably higher than that of the almost five times larger LRP-Gal4/VP16 construct (data not shown). These findings indicate that the cytoplasmic tail of LRP can be proteolytically released from the membrane.
Reporter Gene Activity in LRP-Gal4/VP16-expressing Cells Is Enhanced by PMA Treatment-We next sought to determine, whether the rate at which the cytoplasmic tail of LRP is released from the membrane can be modulated. Incubation of LRP-Gal4/VP16-transfected cells with the LRP ligands ␣2macroglobulin and thrombospondin, and with a fusion protein of the LRP binding chaperone RAP with the constant region of human IgG (33), which can dimerize receptors, had no effect on the transcriptional activation of the Luc reporter (data not shown), suggesting that binding of ligands to the extracellular domain does not generally activate the cleavage. However, incubation of the transfected cells with the phorbol ester PMA robustly increased reporter gene activity from the LRP-Gal4/ VP16 construct by ϳ8 fold but not from the APP-Gal4/VP16 or LDLR-Gal4/VP16 control constructs (Fig. 3A).
Because phorbol esters are activators of PKC, we determined whether the stimulatory effect of PMA could be blocked by simultaneous treatment of cells with Calphostin C, a nonisozyme-specific PKC inhibitor. Incubation of PMA-treated LRP-Gal4/VP16-transfected cells with 1 M Calphostin C completely abolished the PMA-mediated increase in reporter gene activity (Fig. 3B), suggesting that PKC can regulate LRP tail release.
The cytoplasmic tail of LRP contains a possible PKC phosphorylation site as part of the NPTYK sequence motif. To test whether this PKC phosphorylation site was directly involved in the regulation of tail release, nuclear import, or stimulation of reporter gene transcription, we mutated the putatively phosphorylated threonine of the PKC consensus site to an alanine residue. This mutation did not significantly affect the ability of PMA to stimulate LRP tail release (Fig. 4A), suggesting that PKC-mediated phosphorylation of the LRP tail is not the primary mechanism by which PMA increases reporter gene activation. Also, deletion of the carboxyl-terminal 25 or 46 amino acids of the LRP tail did not abolish basal or PMA-induced reporter gene transcription (data not shown).
To determine whether accelerated turnover of LRP at the plasma membrane, resulting in increased release of the tail from the membrane, might be the reason for the PMA effect, we incubated untransfected HEK 293 cells in the absence (Fig. 4B,  lane 1) or presence (lane 2) of 100 nM PMA for 24 h and then prepared lysates from the treated and non-treated cells. Immunoblot analysis revealed greatly decreased levels of the 85-kDa LRP fragment and of the 600-kDa precursor after incubation with PMA, whereas a shorter fragment of ϳ25kDa that reacted with an antibody directed against the carboxyl-terminal epitope of LRP was increased in the membrane fractions of PMA-treated 293 cells compared with untreated cells (Fig. 4C,  lanes 2 and 4). This truncated carboxyl-terminal fragment of LRP is likely generated by proteolytic processing, or shedding, of the LRP ectodomain (23). Thus, PMA appears to increase LRP shedding, which would result in increased availability of substrate for the subsequent intramembranous or cytoplasmic cleavage event. ␥-Secretase performs such an intramembranous proteolytic cleavage step in Notch and APP and thus is a candidate protease that might mediate the release of the LRP cytoplasmic domain from the 25-kDa precursor. In a preliminary experiment the ␥-secretase inhibitor DAPT (27)

did indeed increase the levels of the 25-kDa fragment in the absence (lane 3) or presence (lane 1) of PMA.
A ␥-Secretase-like Activity Participates in the Release of the LRP Intracellular Domain-To further examine whether a ␥-secretase-like activity might be involved in the release of the cytoplasmic tail of LRP, we treated pAPP-Gal4/VP16-or pLRP-Gal4/VP16-transfected cells (Fig. 5A)  Activation of reporter gene activity is an indirect measure of LRP tail release from cellular membranes. To directly demonstrate this release of the cytoplasmic domain of LRP, we transfected 293 cells with an LRP expression plasmid and incubated isolated membranes for 1 h in vitro in the presence or absence of DAPT at 0°C or at 37°C in the absence of cytoplasm (Fig.  5B). Incubation at 37°C, but not at 0°C, caused the release of an ϳ12-kDa fragment that could be detected with an antibody directed against the extreme carboxyl terminus of the LRP tail from the membranes (indicated by the arrow). The appearance of this fragment was completely blocked by DAPT. Furthermore, inhibition of ␥-secretase activity by culturing 293 cells in the presence of increasing concentrations of DAPT was accompanied by the accumulation of the carboxyl-terminal 25-kDa fragment in the membrane fraction (Fig. 5C). Increased levels of this 25-kDa fragment were also observed in membrane preparations of N2a cells that stably express a dominant negative presenilin-1 mutant (PS1-D385A) (31), compared with control cells (Fig. 5D). LRP tail cleavage-dependent reporter gene activity was reduced by ϳ75% in PS1-D385A-expressing N2a cells compared with PS1-wt cells (Fig. 5E). Taken together, these results suggest that ␥-secretase is involved in the processing of LRP but that one or more additional ␥-secretaseinsensitive, and presumably cytoplasmic, proteases can also participate.
Modulation of LRP Cytoplasmic Domain-dependent Reporter Activity by Tail Binding Proteins-The cytoplasmic tails of LRP and APP contain binding sites for a variety of adaptor and scaffold proteins, including FE65, a scaffolding protein that stimulates the nuclear import of the APP cytoplasmic domain and is likely also involved in the transcription of APP target genes (26), and Dab1, an adaptor protein that mediates signaling through the VLDL receptor and the apo-ER2 (11,35). To determine whether FE65 can modulate reporter gene activation, we cotransfected 293 cells with pLRP-Gal4/VP16 and with an expression vector for FE65. APP-Gal4/VP16 and pcDNA3.1-Dab1 plasmids were used as controls. Cells were maintained in the presence of fetal calf serum at all times. Fig.  6A shows that FE65 cotransfection robustly enhanced APP-Gal4/VP16-dependent reporter gene activation, whereas it had no effect in cells expressing LRP-Gal4/VP16 under these conditions (gray boxes). Cotransfection with the Dab1 expression plasmid strongly reduced reporter gene activation in both cases (open boxes) suggesting that Dab1 may either interfere with cleavage directly or may prevent the binding of cytoplasmic factors that promote nuclear translocation of the cytoplasmic domain-Gal4/VP16 fusion protein. However, if the transfected cells were cultured in serum-free medium for 24 h before harvesting, coexpression of FE65 was now capable of stimulating reporter gene activity in LRP-Gal4/VP16-expressing cells by ϳ8-fold (Fig. 6B). Because the absolute levels of basal activity in the absence of FE65 were comparable, whether or not serum was present, these results suggest that either the interaction of FE65 with the LRP tail may be enhanced by post-transcrip- Cell lysates were assayed for reporter gene activity as described above. APP-Gal4/VP16 served as a control for the efficacy of DAPT. B, release of the cytoplasmic domain of LRP from membranes in vitro and is inhibited by DAPT. 293 cells were transfected with 5 g of pcDNA3.1-LRP (expression plasmid for LRP) in 100-mm dishes. 48 h later the cells were lysed, and membranes were isolated as described under "Experimental Procedures." Resuspended membranes were incubated at 0°C or at 37°C for 1 h in the absence or presence of 500 nM DAPT. Pellet (P100) and supernatant (S100) fractions were prepared by centrifugation at 100,000 ϫ g for 1 h and analyzed by immunoblotting with an antibody directed against the carboxyl terminus of LRP. C, treatment of 293 cells with DAPT leads to accumulation of a membrane-bound 25-kDa fragment of LRP. 293 cells were transfected with 5 g of pcDNA3.1-LRP in 100-mm dishes. 12 h after transfection the cells were treated with the indicated concentrations of DAPT for 24 h. Membranes were isolated and analyzed by immunoblotting with an antibody directed against the carboxyl terminus of LRP. D, a 25-kDa LRP carboxyl-terminal fragment accumulates in N2a cells stably transfected with a dominant negative PS1-mutant cDNA. Membranes were prepared from N2a cells that expressed (PS1-D385A) or did not express (PS1-wt) a dominant negative acting PS1 cDNA (31) and analyzed as described in C. E, PS1-wt-and PS1-D385Aexpressing N2a cells were transfected with 1.5 g of pG5E1B-Luc, 0.05 g of pCMV-␤-Gal, and 1 g of pcDNA3.1-LRP-Gal4/VP16 (LRP-Gal4/ VP16) using the MBS kit (Stratagene, La Jolla, CA). After 2 days cell lysates were assayed for reporter gene activity. tional modification in the absence of serum or that the FE65/ LRP tail complex is stabilized under these conditions. When FE65 binds to the cytoplasmic tail of APP, it can recruit the transcriptional activator Tip60 into the complex, obviating the need for the presence of the VP16 transactivator domain for reporter gene activation (26). Because Tip60 interacts with the same PTB domain in FE65 by which FE65 binds to the LRP tail, we sought to investigate whether binding of FE65 to LRP and recruitment of Tip60 might be mutually exclusive. Fig. 6C shows that an LRP-Gal4 fusion protein that lacks the VP16 transactivator domain is incapable of stimulating the luciferase reporter gene expression, whether FE65, PMA, or fetal calf serum (data not shown) were present or not.
The Intracellular Domain of LRP Is Degraded by the Proteasome-To determine whether the proteasome may be involved in the regulation of the potential biological functions of the LRP cytoplasmic domain, we exposed cells transfected with either the LRP-Gal4/VP16 construct or with the soluble Gal4/VP16 control plasmid to the proteasome inhibitor lactacystin (Fig. 7). Inhibition of proteasomal degradation strongly enhanced LRP-Gal4/VP16-stimulated reporter gene activity (Fig. 7A) but had only a modest effect on the reporter gene activity in the control cells (Fig. 7B).

DISCUSSION
In this study we have reported on a novel proteolytic processing event that mediates the release of the cytoplasmic domain of LRP from the membrane. The intracellular domain of LRP contains sequence motifs for a variety of cytoplasmic adaptor and scaffold proteins (4,5). Proteolytic cleavage of the receptor, either within the membrane by a ␥-secretase-like activity or close to the cytoplasmic leaflet, results in the release of this domain into the cytoplasm, where it may modulate signaling, or further translocate to other subcellular compartments.
A characteristic feature that the Notch family members, the amyloid precursor protein APP, and, as most recently recognized, the epidermal growth factor receptor family member ErbB4 have in common is the sequential proteolytic processing at multiple extracellular and intramembranous sites that ultimately results in the release of their cytoplasmic domains from the membrane (18, 26, 36 -39). Before intramembranous cleavage can proceed, the bulk of the extracellular domains of the protein apparently must be shed from the cell surface. In each case, metalloproteinases, e.g. tumor necrosis factor-alpha converting enzyme, have been shown to play a role (17,40), although in the case of APP, ␤-secretase, i.e. the aspartyl proteinase beta-site APP cleaving enzyme, can also mediate this event (41). LRP, like the Notch family members, undergoes cleavage by furin in a late secretory compartment (12) and is cleaved by a metalloproteinase (23). Furthermore, the amino acid sequence of the transmembrane segment of LRP close to the cytoplasmic leaflet resembles that of Notch and APP (Fig.  1), which prompted us to investigate whether cleavage at a third site in the membrane might also mediate release of the cytoplasmic domain. Using a luciferase reporter (Fig. 2A) and a biochemical in vitro assay, we could readily detect this cyto- plasmic tail release (Figs. 2B and 5B), raising the question whether ␥-secretase, which mediates this step in the other proteins, may also act on LRP. This hypothesis was supported by earlier work from Van Uden and colleagues (42), who had shown that overexpression of wild type or of a mutant dominant active form of presenilin 1 (PS1) greatly decreased the expression of LRP at the surface of cultured cells. In the reporter gene assay (Fig. 5A) as well as in the in vitro cleavage assay (Fig. 5B) the ␥-secretase inhibitors DAPT (27) and L-685,458 (34) strongly inhibited LRP tail cleavage. Furthermore, incubation of the cells with the inhibitor resulted in the accumulation of a membrane-bound carboxyl-terminal fragment of LRP of ϳ25 kDa (Fig. 5C). The same fragment also accumulated in N2a cells that stably express a dominant negative presenilin-1 mutant (PS1-D385A) (Fig. 5D), suggesting that ␥-secretase may indeed mediate this event. However, because neither ␥-secretase inhibitors nor dominant negative PS1 completely inhibited release of the cytoplasmic tail of LRP, another, potentially cytoplasmic, protease may also participate.
As had been noted during the studies on Notch cleavage (43), sequences in the cytoplasmic tail of LRP itself were not required for the regulation of this cleavage step, because truncations of the LRP tail (data not shown) and point mutations that alter the PKC consensus phosphorylation site in the LRP tail ( Fig. 4A) had no significant effect on reporter gene activation. However, exposure of cultured cells to phorbol esters, which activate PKC, strongly increased the activity of the luciferase reporter gene (Fig. 3A). This was accompanied by a profound reduction in cellular LRP with a concomitant augmentation of the levels of a membrane-associated carboxyl-terminal fragment of LRP of ϳ25kDa (Fig. 4, B and C), suggesting increased extracellular cleavage and ectodomain shedding of the protein and, consequently, increased substrate availability for the one or more enzymes that mediate the ultimate intramembranous or cytosolic cleavage step. Consistent with this interpretation is the finding by Ni et al. (37) who showed that induction of PKC by phorbol esters stimulates tumor necrosis factor-␣ converting enzyme mediated turnover of ErbB4 and accumulation of the membrane-associated carboxyl-terminal fragment. That the increase in LRP turnover was caused by stimulation of PKC was further supported by the finding that Calphostin C, an inhibitor of PKC that is not specific for any particular isozyme, prevented the cytoplasmic release of the tail in the luciferase reporter gene assay (Fig. 3B).
LRP, like APP, binds the scaffold protein FE65, which was originally identified as a transcriptional activator (44), on its cytoplasmic tail (5). Cao (26) recently reported on the FE65mediated stimulation of the transcription of a luciferase reporter gene in an assay that was dependent on the release of the APP cytoplasmic domain from the membrane. In these experiments FE65 mediated this transcriptional stimulation in two ways: First, FE65 directly promotes nuclear translocation (45), and second, FE65 binds the transcriptional activator and histone acetyltransferase Tip60 via its amino-terminal PTB domain (26). We tested, therefore, whether FE65 was also capable of enhancing reporter gene activity in our assay, which was dependent on the release of the LRP intracellular domain. Under normal cell culture conditions, i.e. in the presence of fetal calf serum, which contains a broad range of mitogens and growth factors, FE65 did not significantly alter LRP-Gal4/VP16dependent reporter gene transcription (Fig. 6A). In contrast, cotransfection of the cells with a Dab1 expression plasmid potently suppressed reporter gene transcription. This finding suggests that Dab1 binding to the tail may interfere with another endogenous factor that, like FE65, can promote nuclear import, or that can directly stimulate gene transcription.
In the absence of serum, reporter gene activity was greatly enhanced by FE65 cotransfection (Fig. 6B) suggesting that post-translational modification of LRP, FE65, or another protein that interacts with FE65 underlies this effect. In contrast to the findings for APP, FE65 was incapable of stimulating reporter gene activation when the VP16 transactivation domain was omitted from the LRP tail fusion complex, in the presence or absence of PMA (Fig. 6C), suggesting that interaction of the amino-terminal PTB domain of FE65 with the LRP tail and with Tip60 are mutually exclusive.
How could LRP tail cleavage modulate cellular signaling? The LRP intracellular domain contains numerous binding sites for adaptor and scaffold proteins that constitutively, or under specific conditions, may recruit other biologically active proteins, e.g. kinases such as mitogen-activated protein kinase kinases, JNK, or Src family members into the complex. Release of the LRP tail from the membrane by intramembranous or cytosolic cleavage could translocate this complex to target sites anywhere in the cell, including the nucleus, where it may modulate intracellular signaling events. Regulated intramembranous proteolysis and release of a transcriptional activator has first been shown for the sterol regulatory element binding proteins (46). At present no specific proteins or genes have been identified that are regulated by the release of the ErbB4, APP, or the LRP tail from the plasma membrane, whereas in the case of Notch family members CSL-regulated genes have been identified as targets for the Notch ICD (21).
The proteolytic release of intracellular domains has turned out to be another distinct mechanism by which cells can relay signals from the extracellular space to their interior. The question now arises how universal this mechanism is. Shedding has long been recognized as a feature by which the extracellular domains of many cell surface proteins are released into the FIG. 7. Lactacystin increases reporter gene activity. 293 cells were transfected with 1.5 g of pG5E1B-Luc, 0.05 g of pCMV-␤-Gal, and with 1 g of pcDNA3.1-LRP-Gal4/VP16 (LRP-Gal4/ VP16) or 0.1 g of pMstGV (Gal4/VP16). 24 h after transfection cells were treated with 10 M lactacystin for 24 h. Cell lysates were assayed for reporter gene activity. extracellular space or into the circulation. We now have to wonder whether in many, if not all, of these instances shedding of the extracellular domains is followed by an intramembranous cleavage. This may involve ␥-secretase or a similar enzyme with relaxed sequence specificity that requires a substrate with a short extracellular domain. The subsequent release of the cytoplasmic domain may be a general mechanism by which these proteins transmit a final message to their host cell.