LRP1 Downregulates the Alzheimer’s β-Secretase BACE1 by Modulating Its Intraneuronal Trafficking,,

The β-secretase called BACE1 is a membrane-associated protease that initiates the generation of amyloid β-protein, a key event in Alzheimer’s disease. However, the mechanism of inraneuronal regulation of BACE1 is poorly understood.


Introduction
Abnormal accumulation of amyloid ␤-protein (A␤) within specific brain regions is thought to play a primary role in the pathogenetic mechanism of Alzheimer's disease (AD) (Hardy and Selkoe, 2002). Recent evidence also supports the view that soluble A␤ oligomers constitute initiator culprits of AD (Larson and Lesné, 2012). The ␤-secretase called ␤-site APP-cleaving enzyme 1 (BACE1) is a membrane-bound aspartyl protease that initiates the generation of A␤ by cleaving the amyloid precursor protein (APP) (Vassar et al., 1999). BACE1 is primarily expressed in neurons in the brain (Vassar et al., 1999) and is possibly involved in AD pathology . Since BACE1 inhibition is highly effective in reducing A␤ production, BACE1 is an important therapeutic target in AD (Stockley and O'Neill, 2008;. BACE1 is known to be regulated at transcriptional, post-transcriptional, translational, and post-translational levels (Rossner et al., 2006;Stockley and O'Neill, 2008;, but the mechanisms underlying BACE1 regulation in neurons are only partly understood. Posttranslational regulation appears critical, because BACE1 regulation occurs after the protein is matured and transported through neuronal processes. Low-density lipoprotein receptor-related protein 1 (LRP1) is a transmembrane receptor that belongs to the LDL receptor gene family (Cam and Bu, 2006;Lillis et al., 2008;Zlokovic et al., 2010). LRP1 is a very large molecule (ϳ600 kDa) that is cleaved by furin to generate a non-covalently associated heterodimer consisting of an ␣-chain (515 kDa) containing the ligand-binding domains, and a ␤-chain (85 kDa) containing the transmembrane domain and the cytoplasmic tail (Cam and Bu, 2006;Lillis et al., 2008;Zlokovic et al., 2010). LRP1 is highly expressed in the brain and exerts multiple functions, including endocytosis of specific proteins. Many ligands for LRP1 exist, including apolipoprotein E and ␣2-macroglobulin, which are also risk factors for AD (Ulery and Strickland, 2000;Cam and Bu, 2006;Lillis et al., 2008;Zlokovic et al., 2010). LRP1 is known to act through scaffolding proteins such as Fe65 to play a role in the endocytosis of APP , and possibly serves as a substrate of BACE1 (von Arnim et al., 2005). Interestingly, the LRP1 intracellular domain has been reported to be important for targeting APP and BACE1 to lipid rafts (Yoon et al., 2007), considered important sites for the generation and accumulation of A␤ (Vetrivel and Thinakaran 2010;Hicks et al., 2012). Thus, LRP1 may have a role in A␤ production, possibly through interaction with APP and/or BACE1; however, the exact relationship between LRP1 and BACE1 remains unclear. LRP1 is also known to play active roles in A␤ clearance such that it mediates brain-to-blood A␤ clearance at the bloodϪbrain barrier (Cam and Bu, 2006;Lillis et al., 2008;Zlokovic et al., 2010;Kanekiyo and Bu, 2014). In the current study, we focused on the molecular relationship between BACE1 and LRP1, and discovered that LRP1 downregulates BACE1 by modulating its intracellular trafficking and stability through protein interaction in neurons.

Cell culture
The mouse embryonic fibroblast lines, MEF-1 (LRP1-wildtype (WT)) and PEA-13 (LRP1-knockout (KO)), obtained from American Type Culture Collection, and human embryonic kidney (HEK) 293 cells were cultured in DMEM supplemented with 10% fetal bovine serum. HEK293 cells were cultured on collagen I-coated dishes or plates (Iwaki). Primary neuronal cultures were prepared from cerebral cortices of rat embryos at embryonic day 17, as described previously Motoki et al., 2012). Cells were plated on poly-L-lysine-coated dishes or plates and maintained in Neurobasal medium containing B27 supplements (Invitrogen).

Western blot analysis
Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer containing protease inhibitors. Western blotting of cell lysates was performed using a standard procedure as described previously . Protein band densities were quantified using an image analyzer LAS-1000 (Fuji Film).

Coimmunoprecipitation
Membrane proteins were extracted from HEK293 cells coexpressing BACE1 and LRP-L4 and immunoprecipitated with 1D4 antibody as described previously . Immunoprecipitated proteins were analyzed by Western blotting with anti-LRP1 antibody. Anti-BACE1 (MAB9311) was used for immunoprecipitation in coimmunoprecipitation experiments of endogenous BACE1 and LRP1.

A␤ measurement
The amounts of A␤40 in conditioned media were measured using sandwich ELISA kits (Wako), as described previously (Motoki et al., 2012).

Immunocytochemistry
HEK293 cells or primary neurons cultured on cover slips were fixed with 4% paraformaldehyde in PBS. Fixed cells were permeabilized and blocked with 0.3% Triton X-100 and 1% horse serum in PBS, and incubated with primary antibody for 1 h, followed by incubation with Alexa488-conjugated anti-mouse or anti-rabbit IgG secondary antibody (Molecular Probes) for 1 h. For double immunolabeling, cells were subsequently stained with a second primary antibody, followed by incubation with Alexa568-conjugated anti-goat or anti-mouse IgG, Cy5conjugated anti-mouse IgG, or DyLight649-conjugated anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Laboratories), as appropriate. For triple immunofluorescence staining with antibodies against rhodopsin (1D4) and HA tags, and an antibody against organelle markers (EEA1, rab7a, ␤-COP, or ␥1-adaptin), cells were first incubated with primary antibodies against the organelle marker followed by incubation with the appropriate secondary antibody, then with goat anti-HA followed by incubation with Alexa568-conjugated anti-goat IgG, and finally with 1D4 followed by incubation with Cy5-conjugated anti-mouse IgG. For triple immunostaining of neurons, goat anti-EEA1 and rabbit anti-HA antibodies were used. In some cases, CanGet Signal Immunostain Immunoreaction Enhancer Solution (Toyobo) was used to increase the sensitivity of the reaction with primary antibodies.

Lipid raft isolation
Lipid rafts were isolated using sucrose density gradient ultracentrifugation, as described previously (Motoki et al., 2012). Generally, the lipid raft marker flotillin-1 fractionated into fraction 4 of the 10 fractions collected.

Cycloheximide chase experiments
Cells were plated on six-well plates, incubated in the presence of cycloheximide (100 M) for up to 12 h, and analyzed by Western blotting, as described previously . In cotreatment experiments, cells were coincubated with cycloheximide and chloroquine (50 M).

Statistics
All results are presented as means Ϯ SEMs. Data were statistically analyzed using one-way ANOVA followed by a Tukey multiple comparison test or Student's t test with a significance threshold of p Ͻ 0.05.

BACE1 protein expression is increased in LRP1-knockout cells
We first compared the protein expression level of BACE1 in WT and LRP1-KO cells. Western blot analyses of cell lysates showed that the protein expression levels of endogenous BACE1 and APP in LRP1-KO cells were significantly higher than those in WT cells (Fig. 1A,B). The increase in APP level in LRP1-KO cells is consistent with a previous report . The specificity of anti-BACE1 antibody (AB5832) was confirmed by comparison with another anti-BACE1 antibody (D10E5), as presented below (see Fig. 3E). Additionally, sandwich ELISAs revealed that the amount of A␤40 in conditioned media of LRP1-KO cells was higher than that in media conditioned by LRP1-WT cells (Fig. 1C). Overexpression of BACE1 or APP in LRP1-KO and LRP1-WT cells using recombinant adenoviruses resulted in much higher levels of BACE1 or APP in LRP1-KO than in LRP1-WT cells (Fig. 1D). It was previously reported that exogenously expressed intracellular domain of LRP1 facilitates lipid raft targeting of BACE1 and APP (Yoon et al., 2007). We compared the Figure 1 BACE1 protein expression is increased in LRP1-KO cells relative to WT cells. A, Cell lysates of LRP1-KO and WT cells were analyzed by Western blotting with antibodies against APP, BACE1, and LRP1. B, Quantification of relative protein levels in A (n ϭ 3, ‫ء‬p Ͻ 0.05). C, The amounts of A␤40 in 24-h-conditioned media from LRP1-KO and WT cells were measured by sandwich ELISA (n ϭ 3, ‫ء‬p Ͻ 0.05). D, LRP1-KO and WT cells infected with recombinant adenoviruses expressing LacZ, WT APP, or rho-tagged BACE1 were maintained for 2 d, and cell lysates were analyzed by Western blotting as above. APP CTFs were analyzed by immunoprecipi-tationϪWestern blot analysis with anti-APP antibody, as described previously (Motoki et al., 2012). Images from the same blots were merged in the left panel. E؊G, LRP1-KO and WT cells infected with recombinant adenoviruses expressing LacZ (E), BACE1 (F), or WT APP (G) were subjected to sucrose density gradient fractionation, as described in Materials and Methods. Each fraction was analyzed by Western blotting with antibodies against the rhodopsin tag (1D4), APP, BACE1, or flotillin-1. The level of BACE1 or APP in each fraction was quantified and expressed as a percentage of the total level in all fractions. For detection of APP CTFs in G, each fraction was additionally analyzed by immunoprecipitationϪWestern blot analysis with an anti-APP antibody, and the levels of ␣-CTF and ␤'-CTF in each fraction were quantified as above (n ϭ 3).

Figure 2
LRP-L4 down-regulates BACE1 in HEK293 cells and primary neurons. A, LRP1 constructs. LRP-L4 is a functional LRP1 mini-receptor; an HA epitope was inserted into LRP-L4 after the signal peptide (SP) at the N-terminus. LRP1 is cleaved by furin to generate an N-terminal ␣-chain and a C-terminal ␤-chain. TM, Transmembrane domain. B, HEK293 cells were cotransfected with BACE1 and either LRP-L4 or empty vector. Similarly, cells were cotransfected with APP and LRP-L4 or BACE2 and LRP-L4. Cell lysates were analyzed by Western blotting with the indicated antibodies. C, HEK293 cells were cotransfected with BACE1 and the indicated amounts of LRP-L4 and/or empty vector. Cell lysates were analyzed by Western blotting with 1D4 or anti-LRP1 antibody. Relative BACE1 levels were quantified and graphed. D, Primary cultured neurons were infected with adenoviruses expressing BACE1 (5 moi) plus those expressing the indicated amounts of recombinant LRP-L4 and/or LacZ, and cell lysates were analyzed by Western blotting after 2 d. The graph indicates relative mature BACE1 levels. M, Mature; im, immature. E, Primary neurons were coinfected with the indicated adenoviruses expressing swAPP, BACE1, and/or LRP-L4 (3 moi each), and maintained for 2 d. The total amount of adenovirus infected was equalized by addition of LacZ adenovirus. Cell lysates were analyzed by Western (continued in page 6).
lipid raft distribution of endogenous and overexpressed BACE1 and APP between LRP1-KO and LRP1-WT cells by sucrose density gradient fractionation. Although the levels of APP or BACE1 in raft (mainly fraction 4) and non-raft (fractions 7-10) fractions appeared relatively higher in LRP1-KO than LPR1-WT cells in the blots, there were no appreciable differences in the distribution pattern of endogenous (Fig. 1E) or exogenous BACE1 (Fig. 1F) or APP (Fig. 1G), as assessed by the relative level in each fraction of the total level, between the two cell types. These findings suggest that endogenous LRP1 does not affect lipid raft association of BACE1 or APP. Interestingly, consistent with an increased levels of A␤40 in LRP1-KO cells, an analysis of APP C-terminal fragments (CTFs) indicated that ␤'-CTF, a metabolite derived from ␤-secretase processing of APP Motoki et al., 2012), predominated in LRP1-KO cells compared with ␣-CTF, a product of ␣-secretase processing of APP, whereas ␣-CTF was more prominent than ␤'-CTF in LRP1-WT cells (Fig. 1D,G), implying that APP processing by BACE1 is promoted in LRP1-KO cells. These data suggest that the steady-state levels, but not lipid raft association, of BACE1 may be regulated by LRP1.

LRP1 downregulates BACE1 protein expression in HEK293 cells and primary neurons
Having confirmed an essential role of endogenous LRP1 in regulating BACE1 levels, we next used LRP-L4, a functional LRP1 mini-receptor containing an N-terminal HA tag ( Fig. 2A), to investigate whether exogenously expressed LRP1 can influence BACE1. Coexpression of BACE1 with the C-terminal rhodopsin (rho) tag and LRP-L4 in HEK293 cells induced a remarkable decrease in the level of BACE1 protein compared with that in cells expressing BACE1 alone (Fig. 2B), an effect that was dependent on the expression level of LRP-L4 (Fig. 2C). In contrast, the levels of BACE2 and APP proteins were not affected by LRP-L4 (Fig. 2B), suggesting a specific effect of LRP-L4 on BACE1. Similarly, primary cultured neurons coexpressing rho-tagged BACE1 and LRP-L4 via recombinant adenoviruses showed reduced levels of BACE1 compared with neurons expressing BACE1 and LacZ. Again, this effect was found to be dependent on the expression level of LRP-L4, as indicated by a clear inverse relationship between the amount of LRP-L4 and BACE1 levels (Fig. 2D). These data suggest that LRP1 negatively regulates BACE1 protein expression in both HEK293 cells and primary cultured neurons.
Furthermore, primary cultured neurons coexpressing swAPP, BACE1, and LRP-L4 secreted lower amounts of A␤40 than those coexpressing swAPP, BACE1, and LacZ; this effect was attributable to the remarkable reduction in BACE1 levels (Fig. 2E). In contrast, LRP-L4 and ␤-chain levels were comparable between neurons coexpressing BACE1 and swAPP and those expressing swAPP only. In addition, the levels of both cellular APP and secreted A␤40 were unaltered in neurons coexpressing swAPP plus LRP-L4 compared with those expressing swAPP only (Fig. 2E). These data suggest that LRP1 may inhibit A␤ production through downregulation of BACE1 in primary cultured neurons.

Physical interaction and colocalization of BACE1 and LRP1
We next investigated the mechanism by which LRP-L4 downregulates BACE1. We analyzed interactions between the two molecules by coimmunoprecipitation, using HEK293 cells transfected with BACE1 and LRP-L4 or LRP-L4 alone. These experiments revealed that LRP-L4 was coimmunoprecipitated with BACE1, suggesting that the BACE1 interacts with LRP-L4 (Fig. 3A). Notably, full-length LRP-L4 rather than the ␤-chain primarily immunoprecipitated with BACE1, implying that the N-terminal extracellular region of LRP-L4 is important for the interaction.
We then performed immunocytochemical analyses to examine whether LRP-L4 influences the intracellular localization of BACE1 in HEK293 cells and primary neurons. In these experiments, comparable levels of BACE1 were expressed in cells coexpressing BACE1 and LRP-L4 and those expressing BACE1 alone by adjusting the amounts of BACE1 cDNA or recombinant BACE1 adenoviruses used for transfection or transduction, respectively (Figs. 3C, 4A). BACE1 and LRP-L4 were visualized using antibodies to the rhodopsin tag (1D4) and hemagglutinin (HA) tag, respectively. Immunostaining for 1D4 revealed positive granular staining in the cytosol in BACE1-expressing HEK293 cells. Interestingly, the size of 1D4-positive granules appeared to be larger in cells coexpressing BACE1 and LRP-L4 compared to those expressing BACE1 alone (Fig. 3B). Double-immunofluorescence staining with anti-1D4 and anti-HA antibodies revealed positive HA immunostaining in the cytosol and nuclear membrane, and showed that HA immunoreactivity partially overlapped with that of 1D4, indicating colocalization of BACE1 and LRP-L4 (Fig. 3B). The HA-immunostaining pattern was comparable between cells expressing BACE1 alone and those expressing BACE1 and LRP-L4. Similar findings were obtained with primary neurons expressing BACE1 and/or LRP-L4 as well. Immunostaining for 1D4 showed that immunopositive granules localized in perikarya and neuritic processes. Immunostaining for HA showed positive reticular immunoreactivity in perikarya and neurites. In neurons coexpressing BACE1 and LRP-L4, 1D4 and HA immunoreactive signals partially overlapped, indicating colocalization of BACE1 and LRP-L4 (Fig. 3D).
Furthermore, coimmunoprecipitation experiments showed that endogenous LRP1 ␤-chain was coprecipitated with BACE1 by anti-BACE1 antibody, but not by negative continued blotting as above. The graph indicates relative mature BACE1 levels. A␤40 levels in 24-h-conditioned media were determined with sandwich ELISA. C؊E, n ϭ 3; ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01.

Figure 3
Physical association and colocalization of BACE1 and LRP-L4. A, HEK293 cells were transfected with LRP-L4 and either BACE1 or empty vector. Protein extracts of membrane fractions were immunoprecipitated with 1D4 antibody and the precipitated proteins were analyzed by Western blotting, as described in Materials and Methods. B, HEK293 cells transfected with BACE1 plus LRP-L4, BACE1 only, or LRP-L4 only were analyzed by double-immunofluorescence staining with 1D4 (green) and anti-HA (magenta) antibodies. Overlapping 1D4 and HA immunoreactive signals were observed in cells coexpressing BACE1 and LRP-L4. Scale bar, 10 m. C, Primary neurons were infected with the indicated amounts of recombinant adenoviruses expressing BACE1, LRP-L4, and/or LacZ. Two days after infection, cell lysates were analyzed by Western blotting with 1D4, anti-LRP1, or anti-␤-galactosidase. Comparable BACE1 levels were observed in neurons infected with BACE1 adenoviruses (2.5 moi) only and those infected with BACE1 (5 moi) and LRP-L4 (1.25 moi) adenoviruses. D, Primary neurons grown on coverslips were infected with BACE1 plus LacZ adenoviruses, BACE1 plus LRP-L4 adenoviruses, or LRP-L4 plus LacZ adenoviruses. Cells were analyzed as in B. Neurons coexpressing BACE1 and LRP-L4 exhibited overlapping 1D4 and HA immunoreactive signals in soma and (continued in page 8).
control antibody (Fig. 3E), suggesting that endogenous BACE1 and LRP1 interact physically with each other. Since native LRP1 ␣-chain and ␤-chain are connected through noncovalent interaction, it is possible that BACE1 physically associated with such a native form of LRP1. Double immunofluorescence staining using Alexa568-labeled anti-BACE1 antibody also clearly showed that the immunoreactivities of endogenous BACE1 and LRP1 overlapped in soma and neurites, demonstrating their colocalization in these subcellular regions (Fig. 3F).

LRP-L4 decreases BACE1 stability and reduces the cell-surface expression of BACE1
To gain a more in-depth understanding of the inhibitory effect of LRP-L4 on BACE1, we performed cell-surface biotinylation experiments in which biotinylated proteins were avidin-agarose precipitated and subjected to Western blotting. In this experiment, levels of BACE1 in cells expressing BACE1 only were comparable to those in cells expressing BACE1 plus LRP-L4. The level of mature, cell-surface BACE1 in cells expressing BACE1 plus LRP-L4 was significantly lower than that in cells expressing BACE1 only (Fig. 4A). This finding suggests that LRP-L4 affects BACE1 protein levels at the plasma membrane, possibly by altering BACE1 intracellular trafficking.
We surmised that LRP-L4 might affect the stability of BACE1 proteins, and therefore performed cycloheximide chase experiments. During the 12 h chase period, BACE1 levels were only slightly reduced in cells expressing BACE1 only, whereas the levels were markedly reduced (to ϳ60% of the control level) in cells expressing BACE1 plus LRP-L4 (Fig. 4B). These data suggest that LRP-L4 decreases the stability of BACE1.
To further examine whether BACE1 downregulation by LRP-L4 is associated with endocytosis from the cell surface, we coexpressed mutant dynamin 2 (Dyn2 K44A), which inhibits endocytosis (McNiven et al., 2000), with BACE1 and LRP-L4. LRP-L4 caused downregulation of BACE1 even with endocytosis inhibited by expression of myc-tagged Dyn2 K44A (Fig. 4C). The protein level of BACE1 as well as that of LRP-L4, especially the ␤-chain, was higher in cells transfected with Dyn2 K44A than in cells without Dyn2 K44A, suggesting that inhibition of endocytosis results in the accumulation of both proteins. These data suggest that downregulation of BACE1 caused by LRP-L4 may be independent of endocytosis.

LRP1 facilitates BACE1 transport from early to late endosomes, promoting BACE1 lysosomal degradation
We hypothesized that LRP-L4 may alter BACE1 subcellular localization to promote its degradation, and therefore investigated the subcellular location of BACE1 using HEK293 cells and primary neurons. To this end, we performed triple immunostaining with anti-HA, 1D4, and antibodies against early endosomal antigen 1 (EEA1), rab7a, ␤-COP, or ␥1-adaptin, markers of early endosomes, late endosomes, the Golgi apparatus, and the trans-Golgi network (TGN), respectively. In HEK293 cells coexpressing BACE1 and LRP-L4, 1D4 and HA immunoreactive signals partially overlapped with those of EEA1 and rab7a (Fig.  5A). In contrast, in HEK293 cells expressing only BACE1, 1D4 immunoreactivity clearly overlapped with that of EEA1, but only marginally with that of rab7a (Fig. 5A). We additionally observed only limited colocalization of 1D4 immunoreactive signals with those of ␤-COP or ␥1-adaptin in both cells expressing only BACE1 and those expressing BACE1 and LRP-L4 (Fig. 5B). The finding that 1D4positive granules in cells coexpressing BACE1 and LRP-L4 were larger than those in cells expressing only BACE1 seemed to reflect the different subcellular locations of BACE1 in these cells.
Moreover, 1D4 and HA immunoreactivity colocalized with that of EEA1 and rab7a in both soma and neurites of primary neurons coexpressing BACE1 and LRP-L4, whereas 1D4 was substantially colocalized with EEA1, but not with rab7a, in neurons expressing BACE1 only (Fig.  5C). Colocalization of 1D4, HA, and ␤-COP or ␥1-adaptin immunoreactive signals was marginal in neurons coexpressing BACE1 and LRP-L4 as well as those expressing BACE1 only (Fig. 5D). The presence of EEA1 and rab7a immunoreactivity in neuronal processes is consistent with the notion that endosomal organelles are distributed throughout the soma, dendrites, and axons (Lasiecka and Winckler, 2011). These data suggest that LRP1 facilitates BACE1 sorting from early to late endosomes in both HEK293 cells and primary neurons.
We reasoned that LRP-L4 promotes trafficking of BACE1 to lysosomes, where BACE1 degradation occurs. To clarify this point, we tested the effect of chloroquine, a lysosomotropic agent, in cycloheximide chase experiments, as described above. Treatment of HEK293 cells coexpressing BACE1 and LRP-L4 with cycloheximide plus chloroquine almost completely rescued the reduction in BACE1 levels in cells treated with cycloheximide only, suggesting that LRP-L4 likely destabilizes BACE1 through increased lysosomal degradation (Fig. 5E). In addition, the reduced level of full-length LRP1 in cycloheximide-treated continued neurites. Scale bar, 20 m. E, Protein extracts of membrane fractions of primary neurons were immunoprecipiated with anti-BACE1 (MAB9311) or 1D4 (negative control), followed by immunoblotting with anti-LRP1. The blots were reprobed with anti-BACE1 antibodies (AB5832 and D10E5). Images from the same blots were merged in this figure. F, Primary neurons grown on coverslips were analyzed by double-immunofluorescence staining with anti-LRP1 and Alexa568-labeled anti-BACE1. BACE1 and LRP1 immunoreactivities were clearly overlapped in both soma and neurites of neurons. Scale bar, 20 m. cells was only partially recovered by cotreatment with cycloheximide and chloroquine (Fig. 5E).
Finally, we used triple immunofluorescence staining to analyze the subcellular localization of endogenous BACE1 Figure 4 LRP1 decreases BACE1 stability and reduces the cell surface expression of BACE1. A, HEK293 cells were transfected with the indicated amounts of BACE1 and either LRP-L4 or vector. Cell-surface biotinylation experiments were performed as described in Materials and Methods. Western blots of total cell lysates and avidin-agarose-precipitated material are shown. Relative BACE1 levels were quantified and graphed. B, HEK293 cells transfected with BACE1 plus LRP-L4 or vector as in A were subjected to cycloheximide chase experiments, as described in Materials and Methods. After incubation with cycloheximide (CHX) for the indicated times, cells were lysed and analyzed by Western blotting. Relative BACE1 levels at 0 and 12 h were quantified and graphed. (A, B: n ϭ 3, ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01). C, HEK293 cells were cotransfected with BACE1, LRP-L4, and/or Dyn2 K44A as indicated. The total amount of DNA was equalized by the addition of vector. Cell lysates were analyzed by Western blotting with appropriate antibodies. Relative BACE1 levels in blots were quantified and graphed (n ϭ 3, ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01). were analyzed by triple-immunofluorescence staining with anti-EEA1/anti-rab7a (green), anti-HA (red), and 1D4 (blue) antibodies, or double-immunofluorescence staining with anti-EEA1/anti-rab7a (green) and 1D4 (red) antibodies. Note that colocalization of HA, 1D4, and EEA1/rab7a immunoreactive signals was observed in cells expressing BACE1 and LRP-L4, whereas colocalization of signals for 1D4 and EEA1, but not rab7a, was observed in cells expressing BACE1 only. Scale bars, 20 m. B, HEK293 cells transfected with BACE1 plus LRP-L4 or BACE1 only were analyzed by triple-immunofluorescence staining with anti-␤-COP/anti-␥1-adaptin (green), anti-HA (red), and 1D4 (blue) antibodies, or double-immunofluorescence staining with anti-␤-COP/anti-␥1-(continued in page 11). and LRP1 in neurons. The results revealed that both endogenous BACE1 and LRP1 immmunoreactivities were clearly colocalized with EEA1 in soma and neurites and partially with rab7 mainly in soma (Fig. 6A,B). In contrast, only limited colocalization of endogenous BACE1, LRP1, and GM130 (a Golgi marker) was observed in soma (Fig.  6C). These data clearly suggest the association of endogenous BACE1 and LRP1 in endosomal compartments, particularly early endosomes, of neurons.

Discussion
The mechanisms underlying the regulation of BACE1 in neurons have not yet been fully elucidated. It was recently shown that BACE1 is transported in axons and dendrites of neurons, where it likely functions to generate A␤ (Sannerud et al., 2011;Das et al., 2013;Deng et al., 2013;Buggia-Prévot et al., 2013;Ye and Cai, 2014). Therefore, it is critical to understand how BACE1 is regulated in axons and dendrites. BACE1 can be regulated posttranslationally by modulation of intracellular trafficking and degradation (Zhi et al., 2011;Tan and Evin, 2012;Zhang and Song, 2013). It is known that BACE1 is recycled between endosomes and the TGN or the plasma membrane (Zhi et al., 2011;Tan and Evin, 2012;Zhang and Song, 2013) and cleaves APP mainly in acidic endosomal compartments (Stockley and O'Neill, 2008;Zhi et al., 2011;Tan and Evin, 2012;Zhang and Song, 2013;. BACE1 degradation appears to occur predominantly through the lysosomal pathway, although the involvement of the proteasomal pathway has been reported (Qing et al., 2004;Koh et al., 2005;, Kandalepas et al., 2013. However, the mechanisms by which BACE1 is directed to degradation pathways are poorly understood. Here, we found that LRP1 negatively regulates BACE1 through proteinϪprotein interactions and showed that this negative regulation is mediated by modulation of BACE1 intracellular trafficking and stability. Consistent with the finding that the expression level of BACE1 is increased in LRP1-KO cells relative to LRP1-WT cells, we observed that LRP-L4 exerts a specific suppressive effect on BACE1 in both HEK293 and primary continued adaptin (green) and 1D4 (red) antibodies, as in A. C, Primary neurons grown on coverslips were infected with BACE1 plus LRP-L4 adenoviruses or BACE1 plus LacZ adenoviruses. Cells were analyzed as in A, except that goat anti-EEA1 and rabbit anti-HA antibodies were used. Neurons coexpressing BACE1 and LRP-L4 exhibited overlapping 1D4, HA, and EEA1/rab7a immunoreactive signals in soma and neurites, whereas neurons expressing only BACE1 exhibiting overlapping signals of 1D4 and EEA1, but not rab7a. Scale bars, 20 m. D, Primary neurons infected with BACE1 plus LRP-L4 adenoviruses or BACE1 plus LacZ adenoviruses were analyzed by triple-or double-immunofluorescence analysis performed with anti-␤-COP/anti-␥1-adaptin, anti-HA, and 1D4 antibodies, as in B. Scale bars, 20 m. E, HEK293 cells transfected with BACE1 plus LRP-L4 were subjected to cycloheximide (CHX) chase experiments, in which cells were coincubated with or without chloroquine (CQ; 50 M). After 12 h, cells were lysed and analyzed by Western blotting. Relative levels of BACE1 and LRP-L4 were quantified and graphed (n ϭ 3, ‫ء‬p Ͻ 0.05). neurons. Focusing on the mechanism of BACE1 downregulation by LRP-L4, we established that the two proteins physically interact and colocalize, and found that LRP-L4 affects the stability of BACE1, possibly through their interaction. The inhibitory effect of LRP-L4 on BACE1 expression may be independent of endocytosis. Notably, LRP-L4 influences the intracellular trafficking of BACE1; it decreases cell-surface expression of BACE1 and facilitates BACE1 transport from early to late endosomes, thereby promoting lysosomal degradation. Furthermore, we obtained evidence that endogenous BACE1 and LRP1 interact with each other and are colocalized in endosomal compartments of neurons.
On the basis of these data, we propose the following hypothetical model (Fig. 7): BACE1 and LRP1 interact with each other and traffic from the TGN to early endosomes; this trafficking may be independent of endocytosis. BACE1ϪLRP1 complexes in early endosomes are preferentially sorted into late endosomes in both soma and neuronal processes of neurons. BACE1 is then degraded in lysosomes, and LRP1 may possibly be recycled back to the cell surface. Interestingly, it has been shown that RAP, one of the ligands of LRP1, is transported from early to late endosomes with LRP1 for lysosomal targeting and LRP1 is recycled back to the plasma membrane (Laatsch et al., 2012). LRP1 may thus mediate targeting of BACE1 and RAP to lysosomes by a similar trafficking mechanism. Although it is plausible that BACE1 is intimately associated with and regulated by LRP1 endogenously, additional evidence will be required to firmly establish this. Further research is also required to elucidate the molecular basis of the interaction between BACE1 and LRP1.
It has previously been shown that GGA3, one of the GGA proteins involved in the transport of cargo proteins from the TGN to endosomes, plays a significant role in the sorting of BACE1 from endosomes to lysosomes . However, it is not yet clear how GGA3 participates in BACE1 regulation in neurons, especially in dendrites and axons. Another recent study also reported that snapin, a dynein motor adaptor, mediates BACE1 retrograde trafficking to lysosomes for degradation (Ye and Cai, 2014). Other factors also appear to be involved in BACE1 sorting in endocytic and recycling pathways, including sorting nexin 6, sorting nexin 12, and sortilin (He et al, 2005;Wahle et al., 2005;Finan et al., 2011;. Further research is also needed to fully understand the mechanisms by which BACE1 sorting to lysosomes is coordinately regulated by interacting proteins, including LRP1 and GGA3, in neurons. LRP1 has previously been indicated to be one of the substrates of BACE1 (von Arnim et al., 2005;von Einem et al., 2010). However, we observed that BACE1 overexpression affects neither LRP-L4 and ␤-chain levels nor the intracellular expression pattern of LRP-L4. Furthermore, recent searches for BACE1 substrates in neurons using a proteomic approach did not detect LRP1 (Kuhn et al., 2012;. Considering these discrepant findings, LRP1 may have relatively low affinity with BACE1 as a substrate under physiological conditions. Whether LRP1 has a role in the negative regulation of BACE1 and A␤ production in vivo is an important issue to be clarified. A recent study by Kanekiyo and colleagues (2013) showed that LRP1 deletion in forebrain neurons increases insoluble A␤ levels as well as amyloid burden in the cortex of APP/PS1 mice. They indicated that a disturbance of LRP1-mediated neuronal A␤ uptake and clearance is responsible for the exacerbation of A␤ deposition in neuronal LRP1-KO mice. However, their data appear to be consistent with the idea that neuronal LRP1 negatively Figure 7 Hypothetical model. In neurons, BACE1 is recycled through the TGN, the plasma membrane and early endosomes, maintaining relative stable BACE1 levels. LRP1 complexes with BACE1 and facilitates the transit of BACE1 from early to late endosomes, promoting lysosomal targeting and degradation of BACE1. BACE1-LRP1 complexes may be sorted directly from the TGN to early endosomes. LRP1 may possibly be recycled back from late endosomes to the plasma membrane, as in the case of RAP-LRP1 complexes (Laatsch et al., 2012). regulates BACE1 and A␤ production, insofar as the exacerbated A␤ deposition in LRP1-KO mice may be due, at least in part, to increased neuronal A␤ production. A similar study, however, showed that lowering LRP1 levels in hippocampal neurons did not significantly alter A␤ levels and amyloid plaque numbers (Xu et al., 2012). In another study, transgenic mice expressing an LRP1 minigene exhibited small increases in the levels of both soluble and insoluble A␤ in the cortex . The reason for these discordant data is currently unknown.
Lipid rafts are thought to be involved in the generation and accumulation of A␤ (Vetrivel and Thinakaran 2010;Hicks et al., 2012). Previous studies have pointed to a possible significant role for exogenously expressed LRP1 C-terminal domain in the lipid raft targeting of APP as well as BACE1 (Yoon et al., 2007). However, our data using LRP1-KO cells do not support such a role for endogenous LRP1. A recent report has indicated that BACE1 cleavage of APP mainly occurs in non-raft fractions in primary neurons (Motoki et al., 2012). Whether BACE1 association with LRP1 occurs within or outside lipid rafts of endosomal compartments in neurons would be an interesting question for future research.
There have been conflicting reports on the expression levels of LRP1 protein in AD brains.  reported reduced LRP1 expression in AD samples compared to age-matched controls; however, other studies have reported increased or unchanged LRP1 expression (Qiu et al., 2001;Causevic et al., 2003;Donahue et al., 2006). Although there does not appear to be a clear correlation between LRP1 expression and AD pathology, LRP1 plays significant roles in processes that are directly associated with the pathological progression of AD. Importantly, LRP1 expressed in brain capillaries (endothelial cells and pericytes) plays a critical role in clearance of A␤ across the bloodϪbrain barrier (Lillis et al., 2008;Zlokovic et al., 2010;Kanekiyo and Bu, 2014). LRP1 expressed in other cell types (vascular smooth muscle cells, glial cells, and neurons) appears to mediate cellular A␤ clearance (Lillis et al., 2008;Kanekiyo and Bu, 2014). LRP1 also mediates anti-apoptotic function in neurons (Fuentealba et al., 2009) and has an essential role in synaptic function (May et al., 2004;. Moreover, LRP1 is capable of influencing neuronal A␤ generation through BACE1 downregulation, as revealed in the current study. Thus, LRP1 has various beneficial functions that protect against AD pathology, including lowering A␤ levels in the brain. Thus, future research into LRP1 biology may aid in developing novel strategies to treat or prevent AD. For instance, any agent that is capable of upregulating LRP1 in the brain may be of therapeutic value to AD.