Ubiquilin 1 Modulates Amyloid Precursor Protein Trafficking and Aβ Secretion*

Ubiquilin 1 (UBQLN1) is a ubiquitin-like protein, which has been shown to play a central role in regulating the proteasomal degradation of various proteins, including the presenilins. We recently reported that DNA variants in UBQLN1 increase the risk for Alzheimer disease, by influencing expression of this gene in brain. Here we present the first assessment of the effects of UBQLN1 on the metabolism of the amyloid precursor protein (APP). For this purpose, we employed RNA interference to down-regulate UBQLN1 in a variety of neuronal and non-neuronal cell lines. We demonstrate that down-regulation of UBQLN1 accelerates the maturation and intracellular trafficking of APP, while not interfering with α-, β-, or γ-secretase levels or activity. UBQLN1 knockdown increased the ratio of APP mature/immature, increased levels of full-length APP on the cell surface, and enhanced the secretion of sAPP (α- and β-forms). Moreover, UBQLN1 knockdown increased levels of secreted Aβ40 and Aβ42. Finally, employing a fluorescence resonance energy transfer-based assay, we show that UBQLN1 and APP come into close proximity in intact cells, independently of the presence of the presenilins. Collectively, our findings suggest that UBQLN1 may normally serve as a cytoplasmic “gatekeeper” that may control APP trafficking from intracellular compartments to the cell surface. These findings suggest that changes in UBQLN1 steady-state levels affect APP trafficking and processing, thereby influencing the generation of Aβ.

Alzheimer disease (AD) 3 is the most common cause of progressive neurological disorder leading to dementia. It is neuropathologically characterized by extracellular deposits of amyloid beta (A␤) peptide and by the generation of intracellular neurofibrillary tangles. Mutations in the amyloid precursor protein (APP), presenilin-1 (PSEN1), and presenilin-2 (PSEN2) genes are responsible for roughly half of the rare autosomal dominant, early-onset forms of the disease, which usually occur before the age of 60 (1)(2)(3)(4). Meanwhile, apolipoprotein E (APOE) is the only commonly accepted susceptibility factor for late-onset AD (5,6). Most mutations in APP, PSEN1, and PSEN2 genes lead to the increased production of A␤42 (relative to A␤40). A␤ is released from APP via sequential proteolytic cleavage by the ␤and ␥-secretases (7). In addition to APP, the presenilins, and APOE, it is evident that additional AD susceptibility genes exist; successfully identifying these novel risk genes is an extremely important task that will not only facilitate prediction and diagnosis of AD but can also elucidate novel therapeutic approaches to treating and preventing AD.
We have recently shown that genetic variants in the ubiquilin 1 (UBQLN1) gene, located on chromosome 9q22, increase the risk for AD, possibly by altering the expression and alternative splicing of this gene in brain (8). As is often the case with gene variants exerting modest effects on disease risk, subsequent genetic studies have both supported (9,10) and not supported (11,12) the initial genetic finding. Further support for the candidacy of UBQLN1 as an AD risk gene will require a comprehensive assessment of the functional role of UBQLN1 in the key pathways relevant to AD pathogenesis. UBQLN1 has previously been shown to interact with presenilin 1 (PS1) and presenilin 2 (PS2) proteins; overexpression of UBQLN1 was reported * This work was supported by the Extendicare Foundation and the Finnish to enhance the accumulation of presenilin holoproteins (13,14). More recently, down-regulation of UBQLN1 was reported to modulate PS1 endoproteolysis along with protein levels of nicastrin and PEN-2 in non-neuronal cell lines (15). These findings are particularly interesting given that the presenilins, nicastrin, and PEN-2 are all essential components of the ␥-secretase complex. Based on these data, it was anticipated that if the steady-state levels of these proteins are controlled by UBQLN1, APP processing and A␤ generation would be affected by changes in UBQLN1 expression. Immunohistochemical analyses have shown that anti-ubiquilin antibodies stain neurofibrillary tangles in AD brain as well as the Lewy bodies in Parkinson disease (13). Moreover, UBQLN1 was recently identified as part of purified polyglutamine aggregates and found to associate with the neuronal intranuclear inclusions in a mouse model of Huntington disease (16). These findings suggest that, beyond AD, UBQLN1 may play a more generalized role in neurodegenerative diseases characterized by abnormal protein accumulations, e.g. frontal lobe dementia, amyotrophic lateral sclerosis, and Parkinson disease.
UBQLN1 is an ubiquitin-like (UBL) protein that contains UBL and ubiquitin-associated (UBA) domains in its N and C termini, respectively. In addition to the presenilins, UBQLN1 plays a central role in regulating the proteasomal degradation of various proteins, including cyclin A, ␥-aminobutyric acid receptor, and hepatitis C virus RNA-dependent RNA polymerase proteins (17)(18)(19). Interestingly, UBQLN1 has been reported to exert differential effects on these proteins. Although overexpression of UBQLN1 has been reported to enhance presenilin and ␥-aminobutyric acid receptor expression, it has also been reported to promote the degradation of cyclin A and hepatitis C virus RNA-dependent RNA polymerase proteins. The UBL domain is known to be responsible for the binding of UBQLN1 to Rpn3, Rpn10a, and Rpn10e proteins in the 19 S subunit of the 26 S proteasome complex, whereas the UBA domain prefers to bind poly-, but not mono-ubiquitinated proteins (20). The UBL domain of UBQLN1 is also responsible for interaction with the ubiquitin-interacting motif of epidermal-growth factor receptor pathway substrate 15 (Esp15) (21). The UBL/ubiquitin-interacting motif-based interaction was proposed to be responsible for the sequestration of certain ubiquitin-interacting motif-containing endocytic proteins like epidermalgrowth factor receptor pathway substrate 15 into cytoplasmic ubiquitin-rich protein aggregates. Collectively, these data suggest that UBQLN1 is a key regulatory protein linking the ubiquitination machinery and the proteasome in mammalian cells. Therefore, even minor changes in the expression and/or function of this protein could potentially affect the steady-state levels of multiple protein targets.
Given the association of UBQLN1 gene variants with risk for AD and altered expression in brain (8), we set out to determine the as of yet unknown effects of UBQLN1 on APP synthesis, maturation, and metabolism. For this purpose, UBQLN1 levels were down-regulated in human H4 neuroglioma and human embryonic kidney (HEK293) cells using RNA interference (RNAi), and effects on APP holoprotein levels, APP maturation and turnover, APP ectodomain shedding (sAPP␣, -␤, and total), and A␤ secretion were assessed. Down-regulation of UBQLN1 dramatically increased the rate of APP maturation and trafficking through the secretory pathway leading to increased secretion of sAPP and A␤. Meanwhile, ␣-, ␤-, and ␥-secretase levels and activity were not affected by UBQLN1 RNAi treatment. We also employed a fluorescence resonance energy transfer (FRET)-based assay to show that UBQLN1 and APP come into close proximity in intact cells, independently of the presence of PS1 or PS2.

EXPERIMENTAL PROCEDURES
siRNAs and Plasmids-Silencer TM Pre-designed siRNA targeted to exon 5 of the human UBQLN1 gene (GGCGCATGTA-CACAGATAT) was used to knock down UBQLN1 using RNAi (Ambion). Silencer TM Negative control #1 siRNA was used as a control in RNAi experiments (Ambion). Human ADAM10 wild-type cDNA was cloned into peak 12 plasmid with HA tag fused at the C terminus. A human BACE1 wild-type cDNA was cloned into pcDNA4 with Myc-His tag fused at the C terminus of BACE1 (22).
Cell Cultures and Transfections-H4 human neuroglioma cells overexpressing APP751 or PS1 wild-type were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 200 g/ml G418. H4 naïve cells were grown in the same media, but without G418. HEK293-AP-APP cells overexpressing alkaline phosphatase (AP) and APP695 fusion protein and Bcl-XL/CrmA were grown as previously described (23). AP ectodomain was fused to the N terminus of full-length APP 695 lacking signal peptide (24). PS1/ PS2 double knock-out mouse embryonic fibroblasts (PSDKO-MEF) were maintained in OPTI-MEM media containing 10% fetal bovine serum. Human wild-type PS1 overexpressing PSDKO-MEFs were maintained in OPTI-MEM media containing 10% fetal bovine serum and 2.5 g/ml puromycin. All cells were grown in an incubator at 37°C containing 5% CO 2 .
Both the UBQLN1 and control siRNA were transfected into H4 cells (APP751/PS1 wild-type overexpressing and naïve) according to the manufacturer's instructions using a Nucleofector device (Amaxa). Transfection efficiency in each H4 cell line was found to be ϳ80 -90% based on counting of enhanced green fluorescent protein-positive cells. HEK293-AP-APP cells were transfected as previously described (23) using Lipofectamine 2000 (Invitrogen). Transfection efficiency was determined to be ϳ70 -80% according to enhanced green fluorescent protein.
RNA Extraction and Real-time PCR-Total RNA was extracted from H4-APP751 cells transfected with UBQLN1 and control siRNAs after 48 h of transfection using RNeasy Mini Kit (Qiagen). Equal quantities of DNase-treated RNA samples were subjected to cDNA synthesis using Superscript III Reverse Transcriptase (Invitrogen). Subsequently, SYBR Green Master PCR Mix (Applied Biosystems) and targetspecific PCR primers for APP (5Ј-TGAGCGCATGAATC-AGTCTC-3Ј and 5Ј-CCAGGCTGAACTCTCCATTC-3Ј), UBQLN1 (5Ј-GATCATTCAGCTCAGCAAACA-3Ј and 5Ј-GTATTCAAACCCAAGCTACTCAGA-3Ј), and GAPDH (5Ј-GGTCTCCTCTGACTTCAACA-3Ј and 5Ј-GTGAGG-GTCTCTCTCTTCCT-3Ј) were used for amplification of cDNA samples with iCycler real-time PCR machine (Bio-Rad). PCR primers were designed to amplify a region flanking two different exons, and the target specificity of PCR products was confirmed by sequencing. A standard curve method was used to obtain GAPDH-normalized APP and UBQLN1 values.
Secreted APP and A␤ Measurements in Conditioned Media-sAPP␣, sAPP␤, and sAPP total levels from conditioned media of transfected H4 and HEK293 cells were analyzed using Western blotting with 6E10 (Signet Laboratories), s␤APP-WT, and 22C11 (Chemicon International) antibodies, respectively. The C-terminal five amino acids of sAPP␤ were conjugated to ovalbumin and affinity-purified. Approximately 1 mg of the affinity-purified conjugate was mixed with Freund's complete adjuvant and injected intraperitoneally into rabbits, followed by four subsequent (weekly) injections of the suspension containing of 0.5 mg of the affinity-purified conjugate. Images were quantified using Quantity One software (Bio-Rad). In HEK293-AP-APP cells, sAPP total levels were also measured using AP assay (23). Briefly, aliquots of conditioned media were heatinactivated at 65°C for 30 min after which 20 l of the conditioned medium was added to 200 l of reaction solution with 5 mg of p-nitrophenyl phosphate (Sigma) as a substrate. Fluorescence emission was read at 405 nm. Each experiment was carried out in triplicate. A␤ ϫ40 and ϫ42 levels (picograms/ml) were quantified using sandwich enzyme-linked immunosorbent assay, and each experiment was carried out at least in triplicate from HEK293-AP-APP-conditioned media (26). All conditioned media values were normalized to total protein levels. In each case, down-regulation of UBQLN1 was confirmed from the total protein lysates.
In Vitro AICD Generation-For the in vitro AICD generation assay, a protocol similar to Pinnix et al. (27) and Tesco et al. (28) was used. H4-APP751 cells were grown on 100-mm plates after transfections with UBQLN1 and control siRNAs for 48 h. The cells were scraped in buffer A (50 mM HEPES, 150 mM NaCl, 5 mM 1,10-phenanthroline monohydrate, pH 7.4) and homogenized by passing through a 25-gauge 5/8 needle ten times. The homogenate was centrifuged at 10,000 ϫ g for 15 min. The membrane fraction (P10) obtained was washed once with buffer A and centrifuged at 10,000 ϫ g. Supernatants obtained after centrifugation (cytosolic protein fraction) were used to detect UBQLN1 (Zymed Laboratories Inc.) and ␤-tubulin (Sigma) protein levels by Western blotting. Total protein was measured in the P10 fraction, and the same amount of protein was incubated with 30 l of buffer B (50 mM HEPES, 150 mM NaCl, 5 mM 1,10-phenanthroline monohydrate, pH 7, and protease inhibitor mixture, Roche Applied Science) for 2 h on ice (as negative control) or at 37°C to induce the release of AICD. After incubation the samples were centrifuged at 10,000 ϫ g for 15 min. The supernatant was collected and analyzed by Western blotting using an anti-APP-CTF antibody (A8717, Sigma). The membrane fractions were lysed with radioimmune precipitation assay buffer (1% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 5 mM EDTA, 50 mM Tris, pH 8, 150 mM NaCl), and then equal volumes were analyzed by Western blotting. Images were quantified using Quantity One software (Bio-Rad), and C83-normalized AICD levels were reported.
Cycloheximide Degradation Time Course-A cycloheximide degradation time-course experiment was conducted with H4-APP751 cells transfected with UBQLN1 and control siR-NAs. 48 h after transfection, cells were treated with 30 g/ml cycloheximide for 0.5, 1.5, and 3.0 h after which cells were washed and lysed in radioimmune precipitation assay buffer. Equal amounts of total protein were used in Western blotting with APP (A8717, Sigma), UBQLN1 (Zymed Laboratories Inc.), and ␤-tubulin (Sigma)-specific antibodies. The half-life of APP im was calculated by plotting the APP im intensity values (Normalized to APP im 0-h value) with respect to different treatment times. Subsequently, logarithmic fitting was performed to obtain formula for half-life calculations.
Pulse-Chase Studies-UBQLN1 and control siRNA transfected H4-APP751 cells (48 h transfection) were split into two 100-mm plates (0-and 30-min pulse-chase samples) to perform pulse-chase experiment. Transfected cells were also split to 6-well plates to confirm UBQLN1 knockdown from protein lysates. Confluent (90%) pulse-chase plates were preincubated in methionine/cysteine-free (starve) medium for 30 min after which they were incubated in starve medium supplemented with 100 Ci/ml [ 35 S]methionine/cysteine per plate for 20 min (pulse). Then, cells were incubated in the presence of excess amounts of cold methionine/cysteine for 30 min (chase). The cells were then washed, lysed in radioimmune precipitation assay buffer, and immunoprecipitated with the A8717 antibody (Sigma). Samples were separated by SDS-PAGE using 4 -12% gels, fixed, dried, and exposed to a phosphorimaging screen (Bio-Rad). Images were read from a phosphorimaging screen using a Personal Molecular Imager FX, and the APP holoprotein and C83 bands were quantified using Quantity One software (Bio-Rad).
Biotinylation of Cell Surface Proteins-Biotinylation of cell surface proteins was performed according to Fournier et al. (29). H4-APP751 and HEK293-AP-APP cells were preincu-bated for 20 min in cold PBS, including Mg 2ϩ /Ca 2ϩ . Cells were then incubated with Sulfo-NHS-LC-Biotin (0.5 mg/ml in Mg 2ϩ /Ca 2ϩ containing PBS, Pierce) for 30 min at ϩ4°C. Excess biotin was quenched with 0.1 M glycine for 20 min, then washed with PBS, lysed in radioimmune precipitation assay buffer, and immunoprecipitated with streptavidin beads (Pierce) overnight. Samples were separated by SDS-PAGE using 4 -12% gels, and Western blotting with APP (A8717, Sigma), Transferrin receptor (Zymed Laboratories Inc.), and UBQLN1 (Zymed Laboratories Inc.) specific antibodies were performed. UBQLN1 down-regulation was confirmed from unbiotinylated protein fractions using Western blotting. To detect newly synthesized APP in cell surface, H4-APP751 cells were pulsed for 20 min with [ 35 S]Met and chased for 20 min. After 20-min chase, cell surface proteins were biotinylated and lysed as above. Equal amounts of protein were subjected to sequential immunoprecipitation with APP CTF antibody (A8717, Sigma) followed by streptavidinagarose (Pierce). Immunoprecipitated radioactive/biotinylated APP species were analyzed and quantified as in the pulse-chase experiment above.
Immunocytochemistry of MEF Cells-Prior to immunostaining, the cells were plated on poly-L-lysine-coated 96-well plates and grown to form a confluent monolayer. After two washing steps with PBS, cells were fixed in ice-cold methanol for 10 min, washed twice in PBS, and permeabilized in 1.5% normal donkey serum containing 0.1% Triton X-100 for 45 min. Application of primary antibodies mouse anti-APP CT (13G8, a gift from ELAN pharmaceuticals) and rabbit anti-UBQLN1 NT (Abcam) was performed in 1.5% normal donkey serum overnight at 4°C. After three washing steps in PBS, secondary antibodies conjugated to Alexa 430 (Invitrogen) or Cy3 (Jackson ImmunoResearch) were applied at room temperature for 60 -90 min.
Detection of FRET between APP and UBQLN1-FRET occurs between two fluorophores if they are within close proximity (ϳ10 nm) of each other. During fluorescent emission, some of the donor's excitation energy is non-radiatively transferred to the acceptor fluorophore, which leads to a characteristic shortening in donor fluorophore lifetime. After labeling the two epitopes of interest with the donor and acceptor fluorophores Alexa 430 and Cy3, respectively, the presence of a second lifetime indicates that a proportion of the donor-labeled epitopes is in close proximity to the acceptor-labeled epitopes. In this study, we employed an FLT Ultraevolution system (Tecan Trading AG) that allows for the characterization of the donor fluorescence lifetime on a high throughput level. Excitation of the donor fluorophore was carried out by a 440 nm laser head with a high repetition rate, and time-correlated single photon counting (TCSPC) was employed to reconstruct the donor fluorophore decay curve with a temporal resolution of 35 ps.
Data acquisition was performed using XFluor Software (Tecan Trading AG). Data analysis was carried out using a method for fitting fluorescence lifetime data. 4 In brief, the decay curve is assumed to follow the equation, A ϭ A I exp[(Ϫt/ I ) ␤ ] ϩ A 1 exp(Ϫt/ 1 ) ϩ A 2 exp(Ϫt/ 2 ), with A I , A 1 , and A 2 being the amplitude of the decay component associated with the instrument's background autofluorescence, the non-FRETing and FRETing donor fluorophores, respectively. I , 1 , and 2 are the characteristic decay constants of each of the components, and ␤ is the inverse heterogeneity or "degree of stretch" of the exponential (30,31). The fraction of interacting molecules can be calculated from the relative amplitudes of the two decay components (FRET strength, S), whereas the difference between the two lifetimes can be used as a measure of the intermolecular distance (FRET efficiency, E). To avoid cross-talk between decay components, the background and donor lifetimes are individually fit using a series of control conditions. As a measure of the number of molecules that are interacting in the respective sample, the FRET strength was calculated according to the formula, A 2 /(A 1 ϩ A 2 ). Each experiment was performed at least six times.
Statistical Analyses-Statistical analyses were performed using SPSS program version 11.0 or StatView for Windows program version 5.0.1. Independent-Samples t test (equal variances assumed) or Mann Whitney U test (equal variances not assumed) were used to test statistical significances between sample groups. Values are indicated as means Ϯ S.D. The level of statistical significance was set to p Ͻ 0.05.

Down-regulation of UBQLN1 Decreases Levels of APP
Holoprotein-Effects of UBQLN1 knockdown on APP holoprotein and APP CTF (C83/C99) levels were assessed in naïve human H4 neuroglioma cells as well as H4 cells overexpressing APP751 (H4-APP751 cells) and HEK293 cells overexpressing APP695 fused to alkaline phosphatase (HEK293-AP-APP cells) (Fig. 1). ␤-Tubulin was used to normalize UBQLN1 levels in all cell lines. UBQLN1 levels were specifically down-regulated an average 60 -70% in UBQLN1 siRNA (siUBQLN1) samples compared with control siRNA (siControl)-transfected samples. Levels of the C-terminal fragments, APP-C83 and APP-C99, were unchanged in the H4-APP751 cells, however, both APP immature (APP im ) and APP mature (APP m ) levels were decreased an average 30% after down-regulation of UBQLN1 (Fig. 1A). Similarly, levels of endogenous APP holoprotein were decreased in H4 naïve cells (Fig. 1B). In the HEK293-AP-APP cells, both overexpressed (AP-APP695) and endogenous APP holoprotein levels were decreased to a similar extent as that observed in the H4 cells (Fig. 1C). Because endogenous APP holoprotein levels were decreased to a similar extent as overexpressed APP in both cell lines, we could rule out artifactual effects of UBQLN1 knockdown on APP owing to overexpression of APP constructs. Interestingly, whereas we did not observe any differences in C83 or C99 levels in the APP overexpressing cells lines, in H4 naïve cells, C83 levels were significantly increased by an average 1.3-fold (p Ͻ 0.05). This result suggested that the observed effects of UBQLN1 knockdown effects on APP holoprotein were likely the result of altered APP processing.
To further investigate the observed effects of UBQLN1 knockdown on APP holoprotein levels, we assessed effects on APP mRNA levels. For this purpose, APP, UBQLN1, and GAPDH mRNA levels were determined in UBQLN siRNA-transfected and control siRNAtransfected H4-APP751 cells using real-time PCR. PCR primers used for APP mRNA quantification were designed to amplify the cDNA region from exons 16 and 17, thus capturing all three major mRNA variants of APP. In the UBQLN1 siRNA-transfected cells, GAPDH-normalized UBQLN1 mRNA levels were reduced to a similar extent (ϳ65%) as UBQLN1 protein in cell lysates. GAPDH-normalized APP mRNA levels did not differ between UBQLN1 siRNA-and control siRNA-transfected samples (Fig.  1D).
Down-regulation of UBQLN1 Increases Secretion of APP-We next assessed the effects of UBQLN1 knockdown on the levels of secreted APP (sAPP␣, sAPP␤, and total sAPP) in the conditioned media of H4 and HEK293 cells. Western blot analysis revealed that sAPP␣/sAP-APP␣ and sAPP total / sAP-APP total levels (normalized to total protein) were significantly increased 1.6-to 2.7-fold following UBQLN1 knockdown in conditioned media of H4-APP751 and HEK293-AP-APP cells (Fig. 2, A and  B). sAPP␤ and sAP-APP␤ levels were unchanged, although a slight increase was observed in the conditioned media of H4-APP751 cells following UBQLN1 knockdown. sAPP␣ levels in conditioned media of H4 naïve cells were significantly increased to that observed for the APP-overexpressing cells following UBQLN1 knockdown (p Ͻ 0.05, n ϭ 3, data not shown). The alkaline phosphatase moiety N-terminally fused to the APP construct in the HEK293-AP-APP cells allowed us to measure sAPP total levels using an AP protein assay (Fig. 2C). Consistent with the Western blotting results with the APP N-terminal antibody, sAPP total levels (normalized to total protein) were consistently increased an average 2.0-fold (p Ͻ 0.001, n ϭ 8). Thus, UBQLN1 knockdown also increased APP secretion in the HEK293-AP-APP cells. No differences in sAPP total levels were observed between control siRNA and mock transfected samples (Fig.  2C). UBQLN1 levels in protein lysates (normalized to ␤-tubu-FIGURE 1. Down-regulation of UBQLN1 decreases APP holoprotein levels. H4-APP751, H4 naïve, and HEK293-AP-APP cells were transfected with UBQLN1 (siUBQLN1) and control (siControl) siRNAs for 48 h, and total protein lysates were analyzed using Western blotting. APP immature (APP im ), APP mature (APP m ), as well as APP-C83 and APP-C99 bands were quantified, and ␤-tubulin-normalized levels are indicated. A, H4-APP751 cells, n ϭ 6; B, H4 naive cells; n ϭ 3; C, HEK293-AP-APP cells, n ϭ 4; D, H4-APP751 cells were transfected with UBQLN1 (siUBQLN1) and control (siControl) siRNAs for 48 h. APP, UBQLN1, and GAPDH mRNA levels were analyzed using real-time PCR. GAPDH-normalized APP mRNA levels were not affected by UBQLN1 RNAi. GAPDH-normalized UBQLN1 mRNA levels were significantly down-regulated by UBQLN1 knockdown, n ϭ 4. *, p Ͻ 0.05, ϮS.D.; **, p Ͻ 0.01. lin) were decreased an average 60 -70% after UBQLN1 RNAi in H4 and HEK293 cells.
Down-regulation of UBQLN1 Increases A␤ Secretion and Does Not Affect Steady-state Levels of ADAM10 or BACE1-We next set out to determine UBQLN1 knockdown affects A␤ secretion or ␣and ␤-secretase levels. First, we transfected HEK293-AP-APP cells with either ADAM10 or BACE1 cDNA constructs together with UBQLN1 siRNA or control siRNA. BACE1 levels were not affected by UBQLN1 knockdown (Fig.   3A). Consistent with previous studies (32,33), we observed an average 2.5-fold increase in APP-C99 (C89) levels and a decrease in C83 levels in the BACE1-transfected samples compared with mock samples (Fig.  3A). Both pro-and mature levels of ADAM10 were unchanged following UBQLN1 knockdown (Fig. 3B). In addition, endogenous ADAM10 (pro/mature) and BACE1 levels were not affected after UBQLN1 knockdown in HEK-293 AP-APP cells (Fig. 3C).
Next, we measured sAPP total levels from ADAM10-and BACE1transfected samples using the AP protein assay. In agreement with a previous study (23), sAPP total levels in ADAM10-and BACE1-transfected samples were increased an average of 2.5-and 12.1-fold, respectively, when compared with samples transfected with control siRNA alone (Fig. 3D). In samples co-transfected with ADAM10 plus UBQLN1 siRNA, sAPP total levels were increased to a similar extent (2.4-fold) as in samples transfected with UBQLN1 siRNA alone. Similar increases in sAP-APP␣/sAPP␣ levels were observed in samples co-transfected with ADAM10 plus UBQLN1 siRNA using 6E10 for Western blot analysis of conditioned media (Fig. 3B). In BACE1 plus UBQLN1 siRNA-transfected samples, the increase in sAPP total levels (1.5-fold) was not as pronounced as in the ADAM10 plus UBQLN1 siRNA-transfected samples (2.4-fold). However, Western blot analysis of conditioned media from BACE1 plus UBQLN1 siRNAtransfected samples revealed an average of 3.0-fold increase (p Ͻ 0.001, n ϭ 4) in sAP-APP␤ levels (Fig. 3A). These data suggest that UBQLN1 knockdown affects both sAPP␤ and sAPP␣ secretion to a similar extent.
Next, A␤ ϫ40 and ϫ42 levels were measured using A␤ enzyme-linked immunosorbent assay from the same HEK293 AP-APP-conditioned media samples used in the AP protein assay described above. Down-regulation of UBQLN1 significantly increased A␤ ϫ40 levels an average of 1.7-fold (p Ͻ 0.01). A similar increase (p ϭ 0.05), was also observed with the A␤ ϫ42 (Fig. 3E). Transfection of HEK293-AP-APP cells with BACE1 significantly increased A␤ levels as compared with FIGURE 2. Down-regulation of UBQLN1 increases secretion of APP. H4-APP751 and HEK293-AP-APP cells were transfected with UBQLN1 (siUBQLN1) and control (siControl) siRNAs for 48 h, and conditioned media were analyzed using Western blotting or alkaline phosphatase (AP) assay to detect secreted APP forms. A, UBQLN1 knockdown increased sAPP␣ and sAPP total levels (normalized to total protein) in conditioned media of H4-APP751 cells analyzed by Western blotting (n ϭ 4). sAPP␤ fragment is indicated as an arrow in the figure. B, UBQLN1 knockdown also increased sAP-APP␣ and sAP-APP total levels (normalized to total protein) in conditioned media of H4-APP751 cells analyzed by Western blotting (n ϭ 4). Endogenous sAPP␣ and sAPP total levels were also increased by UBQLN1 knockdown. C, sAPP total levels in HEK293-AP-APP-conditioned media were also measured using the AP assay. UBQLN1 knockdown increased sAPP total levels on average 2-fold, whereas no differences were observed between control siRNA and mock transfected condition media (n ϭ 8). *, p Ͻ 0.05, ϮS.D.; ***, p Ͻ 0.001. untransfected samples. Interestingly, BACE1 overexpression also elevated A␤ ϫ42 levels (4.0-fold increase) more profoundly than A␤ ϫ40 levels (2.0-fold increase). Co-transfection of BACE1 and UBQLN1 siRNA significantly increased A␤ ϫ40 levels (1.3-fold, p Ͻ 0.05), whereas A␤ ϫ42 levels were also increased, but not as significantly as those of A␤ ϫ40.

Down-regulation of UBQLN1 Does Not Affect Steady-state Levels of ␥-Secretase Components or in Vitro AICD Genera-
tion-We next asked whether increased A␤ secretion following UBQLN1 down-regulation was linked to effects on the ␥-secretase complex. For this purpose, we assessed steady-state levels of the ␥-secretase components PS1, nicastrin, PEN-2, and APH1aL in H4 cells stably overexpressing wild-type PS1. Levels of ␥-secretase components were not affected following 60 -70% siRNA-mediated down-regulation of UBQLN1 protein levels (Fig. 4A). UBQLN2 was almost undetectable in this cell line, and no differences in UBQLN2 protein levels were observed between the UBQLN1 siRNA-and control siRNA-transfected samples. Since down-regulation of UBQLN1 was previously reported to modulate PS1 endoproteolysis along with the protein levels of nicastrin and PEN-2 in HEK-293 cells (15), we measured the levels of ␥-secretase components in HEK-293 AP-APP cell line not overexpressing PS1 (Fig. 4B). Despite the 70% down-regulation of UBQLN1, endogenous PS1 CTF/NTF, nicastrin, PEN-2, and APH1aL levels were unaffected.
To determine the effects of UBQLN1 knockdown on ␥-secretase activity, an in vitro APP intracellular domain (AICD) generation assay was performed in the H4-APP751 cells. Membrane fractions of transfected samples were incubated at ϩ4°C and ϩ37°C for 2 h, after which both the membrane fractions and the supernatants were analyzed using APP C-terminal antibody in Western blots to detect C83 and AICD (Fig. 4C). Down-regulation of UBQLN1 was confirmed in the cytosolic protein fraction; as expected UBQLN1 was not detectable in the membrane fraction. Consistent with unchanged ␥-secretase component levels, down-regulation of UBQLN1 did not affect in vitro generation of AICD (normalized to APP C83 fragment).
Down-regulation of UBQLN1 Does Not Affect the Turnover of APP im -To further investigate the mechanism responsible for altered secretion of sAPP and A␤ following knockdown of UBQLN1, we performed a series of experiments, including a cycloheximide degradation time course to test whether UBQLN1 knockdown affects the turnover/degradation of APP im in H4-APP751 cells. We observed no significant differences in the mean half-lives of APP im between UBQLN1 knockdown (39 Ϯ 3 min) and control (42 Ϯ 4 min) samples (p ϭ 0.23, n ϭ 3) during the 3-h time course (Fig. 5A). In accordance with previous results (Fig. 1A), UBQLN1 knockdown significantly decreased both APP im and APP m levels (Ϫ30%, normalized to ␤-tubulin) at time point 0 h (p Ͻ 0.05, n ϭ 3), whereas APP C83 and C99 levels (normalized to ␤-tubulin) were comparable to control levels (Fig. 5A). After 30 min of cycloheximide treatment, however, APP im levels (normalized to ␤-tubulin) were significantly more affected (Ϫ39%) by UBQLN1 knockdown than were APP m levels (Ϫ2%) (p Ͻ 0.05, n ϭ 3) (Fig. 5B). Consistent with this observation, comparison of the 0-and 30-min time points revealed that the APP m /APP im ratio was significantly increased an average of 1.5-fold after 30 min of cycloheximide treatment (Fig. 5C). Similar increases were also observed at later time points (Fig. 5A). Although C83 and C99 levels (normalized to ␤-tubulin) were unchanged at both 0-and 30-min time points (Fig. 5A), levels of both CTFs were significantly increased, relative to APP total , following UBQLN1 knockdown at the 30-min time point (Fig. 5D). These data indicate that APP-CTF levels were not decreased simultaneously with APP holoprotein levels following down-regulation of UBQLN1. UBQLN1 levels were not affected by 3-h cycloheximide treatment, in agreement with previous study (13). Taken together, these data suggest that, although UBQLN1 knockdown does not affect APP im half-life, it does serve to accelerate APP maturation, consistent with increased APP secretion.
Down-regulation of UBQLN Increases the APP m /APP im Ratio-The cycloheximide degradation time course experiments suggested that knockdown of UBQLN1 may increase the APP m /APP im ratio owing to accelerated maturation of APP. However, to rule out the possibility that the effect of UBQLN1 down-regulation on APP maturation was not simply a consequence of cycloheximide blocking APP translation, we carried out a pulse-chase assays with [ 35 S]Met in H4-APP751 cells (Fig.  6A). Similar to the results of the cycloheximide time course, the APP m /APP im ratio was significantly increased an average 1.6fold (p Ͻ 0.01, n ϭ 4) after a 30-min chase in the UBQLN1 knockdown samples (Fig. 6B). Simultaneously, C83 levels were increased relative to APP total in the UBQLN1 knockdown samples, in agreement with the cycloheximide experiments (Fig.  5C). Meanwhile, total APP levels were decreased an average 21%, consistent with the notion that UBQLN1 knockdown leads to faster maturation and secretion of APP. The ratio of APP im between 0 and 30 min (30-min APP im /0-min APP im ) was not significantly different between UBQLN1 and control siRNA samples. This is also consistent with the cycloheximide time-course results in which the APP im half-life was not affected by UBQLN1 knockdown. UBQLN1 protein levels (normalized to ␤-tubulin) were decreased an average 70% following UBQLN1 RNAi treatment.
Down-regulation of UBQLN1 Increases Cell Surface Levels of APP-To further investigate the ability of UBQLN1 knockdown to enhance APP maturation, APP secretion, and A␤ generation, we next set out to test whether levels of cell FIGURE 3. Down-regulation of UBQLN1 increases A␤ secretion but does not affect steady-state levels of ADAM10 or BACE1. A, co-transfection of HEK293-AP-APP cells with Myc-tagged BACE1 and UBQLN1 siRNA (siUBQLN1) did not affect ␤-tubulin-normalized BACE1 levels (n ϭ 4). Overexpression of BACE1 increased APP-C99 (C89) levels and decreased APP-C83 levels. sAP-APP␤ levels (normalized to total protein) were increased in conditioned media on average 3-fold after co-transfection of BACE1 and UBQLN1 siRNA as compared with samples transfected with BACE1 and control (siControl) siRNA. B, co-transfection of HEK293-AP-APP cells with HA-tagged ADAM10 and UBQLN1 siRNA did not affect ␤-tubulin-normalized levels of pro-or mature ADAM10 (n ϭ 4). In addition, APP-C83 and APP-C99 levels were not affected in samples co-transfected with ADAM10 plus UBQLN1 siRNA compared with ADAM10 plus siControl samples. C, endogenous ADAM10 (pro/mature) and BACE1 levels (␤-tubulin normalized) were not affected by UBQLN1 knockdown in HEK293-AP-APP cells (n ϭ 4). D, sAPP total levels (normalized to total protein) were determined in conditioned media from HEK293-AP-APP cells using the AP assay after co-transfection with ADAM10 or BACE1 constructs together UBQLN1 (siUBQLN1) or control (siControl) siRNAs (n ϭ 4). sAPP total levels of co-transfected samples were calculated relative to control siRNA sample (normalized to 1). E, A␤ ϫ40 and ϫ42 levels normalized to total protein were measured using a sandwich enzyme-linked immunosorbent assay (n ϭ 4) in conditioned media of HEK293-AP-APP cells transfected with UBQLN1 (siUBQLN1) siRNA or control (siControl) siRNAs, alone or together with BACE1. UBQLN1 knockdown increased levels of both A␤ ϫ40 and A␤ ϫ42. *, p Ͻ 0.05, ϮS.D.; **, p Ͻ 0.01; ***, p Ͻ 0.001. OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 JOURNAL OF BIOLOGICAL CHEMISTRY 32247 surface holo-APP are increased following down-regulation of UBQLN1. For this purpose, H4-APP751 cell surface proteins were biotinylated with sulfo-NHS-LC-Biotin at ϩ4°C, immunoprecipitated with streptavidin-agarose beads, and then analyzed by Western blotting with an APP C-terminal antibody. APP m levels (normalized to transferrin receptor (TFR)) were significantly increased an average 1.5-fold in the UBQLN1 knockdown samples as compared with control samples (Fig. 7,  A and B). Also, comparison of raw values for levels of APP m and TFR in the UBQLN1 siRNA and control siRNA samples revealed that APP m levels were increased while TFR levels were unchanged. A faint APP im band was observed in the biotinylated samples suggesting that a minor portion of unbiotinylated APP im may have been pulled down together with biotinylated APP m . However, as expected, we did not observe any APP-CTFs or UBQLN1 in the biotinylated protein fraction. UBQLN1 was detected in the unbiotinylated protein supernatant obtained after streptavidin immunoprecipitation. To test whether greater amounts of newly synthesized APP is localized to the cell surface following down-regulation of UBQLN1, H4-APP751 cells were pulsed for 20 min with [ 35 S]Met and chased for 20 min. Subsequently, cell surface proteins were biotinylated, and the levels of radiolabeled APP molecules on the cell surface were determined using sequential immunoprecipitation with APP C-terminal antibody followed by streptavidin. Consistent with the non-radioactive biotinylation results (Fig. 7, A and B), more 35 S-Met-labeled/biotinylated APP m was detected at the cell surface (Fig. 7C) following knockdown of UBQLN1. TFR normalized AP-APP levels were also significantly increased (p Ͻ 0.05, n ϭ 4, data not shown) an average 1.3-fold in HEK293-AP-APP cells following UBQLN1 knockdown.

Ubiquilin 1 Regulates APP Secretion
UBQLN1 and APP Are in Close Proximity in Intact Cells-To determine whether close proximity between UBQLN1 and APP could be a determinant in APP maturation and trafficking, we developed a high throughput FRET-based assay using MEFs from double knock-out mice for PS1 and PS2 (PSDKO-MEF) that overexpress wild-type human PS1. In these cells, UBQLN1 and APP are expressed at endogenous levels that are sufficient for immunostaining and for the lifetime analysis. To assess changes in donor fluorophore lifetime, the UBQLN1 N terminus (UBQLN1-NT) was labeled with the donor fluorophore Alexa 430, and the APP C terminus (APP-CT) was labeled with the acceptor fluorophore Cy3. As is shown in Table 1, the lifetime of Alexa 430 (labeling the UBQLN1-NT) in the absence of an acceptor fluorophore was ϳ3800 ps. A second lifetime of ϳ1100 ps was observed when APP-CT was labeled with an acceptor fluorophore, indicating FRET, i.e. close proximity between a significant percentage of the two epitopes. The meas-ured FRET strength varied between 11 and 27%, suggesting that approximately one-fifth of the donor-fluorophore-labeled UBQLN1 molecules were in close proximity to a Cy3-labeled APP molecule. However, several attempts to co-immunoprecipitate APP and UBQLN1 from overexpressing cells or brain homogenates were unsuccessful. This could be due to the transient state of the interaction or might result from the small fraction of interacting molecules, as indicated by the low FRET strength in the high throughput FRET assay.
To test whether close proximity between UBQLN1 and APP occurs independently of the reported interaction of UBQLN1 with PS1 (13), the same FRET-based assay was performed in the PSDKO-MEF cells with no PS1 or PS2 present. As shown in Table 2, a second lifetime representing a similar FRET efficiency was observed with a FRET strength of between 14 and 25%, when the UBQLN1-NT and the APP-CT were both labeled with donor and acceptor fluorophores, respectively. Taken together, these data indicate that UBQLN1-NT and APP-CT are in close proximity to each other independently of the presence of PS1 and PS2 in the cell.
Down-regulation of UBQLN1 Decreases APLP2 Holoprotein Levels but Does Not Affect Steady-state Levels of FE65-To assess whether the effects of UBQLN1 knockdown are specific for APP, we next measured levels of the APP homologue amyloid ␤ (A4) precursor-like protein 2 (APLP2) following UBQLN1 RNAi treatment (Fig. 8A). APLP2 immature levels (normalized to ␤-tubulin) were decreased an average 50% following UBQLN1 knockdown in H4-APP751 cells, whereas APLP2 mature levels were unchanged (Fig. 8B). As observed with APP, the ratio of APLP2 mature relative to APLP2 immature was increased an average 2.2-fold following UBQLN1 RNAi, suggesting that both APP and APLP2 are similarly affected by UBQLN1 knockdown.
Because the effects of UBQLN1 knockdown in increasing APP maturation and secretion resembled the phenotype previously seen in FE65/FE65L-overexpressing cells (34,35), we next determined whether UBQLN1 knockdown affects FE65 levels in H4-APP751 cells (Fig. 8A). Steady-state levels of FE65 remained unchanged following UBQLN1 knockdown (Fig. 8B), ruling out the possibility that UBQLN1 knockdown directly affects FE65 levels.

DISCUSSION
We have shown that down-regulation of UBQLN1 increases the secretion of both sAPP and A␤ in both neuronal and nonneuronal cells. Subsequent analyses revealed this to be the result of accelerated maturation and trafficking of APP through the secretory pathway. Consistent with the notion that knockdown of UBQLN1 enhances the trafficking of APP, elevated FIGURE 4. Down-regulation of UBQLN1 does not affect steady-state levels of ␥-secretase complex proteins or in vitro AICD generation. Effects of UBQLN1 down-regulation on steady-state levels of ␥-secretase complex proteins and ␥-secretase activity were assessed in an H4 cell line overexpressing wild-type PS1 by Western blot analysis and an in vitro AICD generation assay. A, no differences were observed in PS1 C-and N-terminal (CTF/NTF) fragments or full-length (FL)-PS1 levels normalized to ␤-tubulin. PEN-2, mature nicastrin (NCTmat), and APH-1aL levels were also unchanged. UBQLN1 was specifically down-regulated by RNAi on average 65% (n ϭ 8). UBQLN2 levels were almost undetectable in this cell line, and no differences in UBQLN2 levels were observed between UBQLN1 and control siRNA transfected samples. B, down-regulation of UBQLN1 did not affect endogenous PS1 CTF/NTF, nicastrin, PEN-2, and APH1aL levels in HEK293-AP-APP cells (n ϭ 3). C, UBQLN1 knockdown did not affect in vitro AICD generation in H4-APP751 cells. Membrane fractions (P10) of transfected samples were incubated at ϩ4°C and ϩ37°C for 2 h in which after both the membrane fractions (M) and the supernatants (S) were analyzed on Western blots using an APP C-terminal antibody to detect C83 and AICD, respectively (n ϭ 4). UBQLN1 knockdown was confirmed in the cytoplasmic protein fraction. **, p Ͻ 0.01, ϮS.D.
amounts of newly synthesized, radiolabeled holo-APP accumulate at the cell surface. Additionally, biotinylation of cell surface proteins revealed that more holo-APP was located on the cell surface following knockdown of UBQLN1. UBQLN1 knockdown also significantly increased the APP m /APP im ratio as assayed by cycloheximide degradation time-course and pulse-chase experiments. These findings indicate that down-regulation of UBQLN1 affects the APP maturation process in the early secretory pathway. The effects of UBQLN1 knockdown on APP maturation and trafficking appear to be independent of ␥-secretase, because steadystate levels of ␥-secretase complex components and ␥-secretase activity (based on in vitro generation of AICD) were not affected. In addition, UBQLN1 knockdown did not affect the levels of BACE1 and ␣-secretase candidate, ADAM10, or their functions. It is still possible that UBQLN1 knockdown may affect other ␣-secretase candidates such as TACE (tumor necrosis factor ␣-converting enzyme) and ADAM9. However, this possibility seems unlikely in light of the combined effects of UBQLN1 knockdown on APP maturation, trafficking, cell surface localization, and sAPP/A␤ secretion. Consistent with the notion of accelerated trafficking and processing of APP, UBQLN1 knockdown also significantly decreased steady-state levels of APP holoprotein, particularly immature APP. In contrast, levels of APP-CTFs were either unchanged (in APP-overexpressing cell lines) or increased (C83 levels in H4 naïve cells). UBQLN1 knockdown increased both C83 and C99 levels relative to APP im in H4-APP751 cells after 30 min of cycloheximide treatment or pulse chase. Collectively, our data suggest that UBQLN1 knockdown enhances the flux of APP through the secretory pathway leading to an increased portion of mature APP reaching the cell surface and increased secretion of sAPP and A␤. Because we also observed that UBQLN1 knockdown also increased levels of secreted A␤ in HEK293 cells, it is unlikely that accumulation of APP on the cell surface is a consequence of the decreased rate of endocytosis. In support of this, it has recently been shown that the overexpression of dynamin I dominant negative mutant (K44A) increases shedding of the APP ectodomain (sAPP␣) while significantly reducing the release of A␤ in HEK293 cells consistent with the premise that APP internalization is necessary for A␤ generation in HEK293 cells (36). In this context, however, it should be noted that, in HeLa cells, the dynamin I mutant (K44A) not only increased sAPP␣ levels, but also increased endogenous A␤ secretion suggesting that A␤ can be produced directly at the plasma membrane (37).
UBQLN1 has previously been shown to interact with PS1 and PS2, and overexpression of UBQLN1 enhanced the accumulation of PS holoprotein in non-neuronal cells (13). More recently, it was reported that UBQLN1 knockdown increases levels of the presenilin N-terminal and C-terminal endoproteolytic fragments, along with levels of nicastrin and PEN-2 in HEK293 cells (15). In the present study, down-regulation of UBQLN1 did not affect levels of PS1 holoprotein or its endoproteolytic fragment in H4 cells line stably overexpressing wildtype PS1. In agreement with Massey et al. (15), we did observe a slight, but statistically insignificant increase in steady-state levels of PEN-2 (1.1-fold) and nicastrin (1.2-fold) following downregulation of UBQLN1; APH1aL levels were unchanged. Similar to results obtained from H4 cells, we did not detect any changes in ␥-secretase component levels in HEK-293 AP-APP cells. Consistent with our observation of unchanged ␥-secretase component levels in both neuronal and non-neuronal cell lines, UBQLN1 knockdown did not affect in vitro generation of AICD, indicating that down-regulation of UBQLN1 does not interfere with ␥-secretase activity in human H4 neuroglioma cells. Thus, the discrepancy between our study and that of Massey et al. (15) regarding the effects of UBQLN1 knockdown on ␥-secretase complex component levels cannot be explained simply by use of different cell lines (non-neuronal versus neuronal). Because down-regulation of UBQLN1 altered levels of PS1, nicastrin, and PEN-2 by ϳ1.2to 1.4-fold in normal HEK-293 cells (15), it is possible that such subtle changes may have been undetectable in the present study.
Although UBQLN1 knockdown did not affect PS1 levels in the current study, it is interesting to note that UBQLN1 knockdown and PS1 deficiency or the expression of PS1 loss-of-function mutants (38,39) exerted similar effects on at least the maturation and trafficking of APP. In contrast to our study, A␤ FIGURE 6. Down-regulation of UBQLN1 increases APP m /APP im ratio in pulse-chase assays. A, H4-APP751 cells were transfected with UBQLNI (siUBQLN1) or control (siControl) siRNAs for 48 h after which cells were pulsed for 20 min with [ 35 S]Met and chased in the presence of excess amounts of cold methionine/cysteine for 30 min. Total protein lysates were immunoprecipitated with the APP C-terminal antibody and exposed to a phosphorimaging screen for quantification. B, quantified pulse-chase data. APP m /APP im ratio was significantly increased after 30-min chase (n ϭ 4). APP total ϭ APP im ϩ APP m . **, p Ͻ 0.01, ϮS.D. . Down-regulation of UBQLN1 increases cell surface levels of APP m . A, H4-APP751 cells were transfected with UBQLN1 (siUBQLN1) or control (siControl) siRNAs for 48 h. Subsequently, cell surface proteins were biotinylated with Sulfo-NHS-LC-Biotin at ϩ4°C, immunoprecipitated with streptavidin-agarose beads and analyzed using Western blotting with anti-APP C-terminal, anti-transferrin receptor (TFR), and anti-UBQLN1 antibodies. Down-regulation of UBQLN1 was confirmed in the unbiotinylated (ϭcytoplasmic) protein fraction. B, quantified APP m and TFR raw levels as well as TFR-normalized APP m levels (n ϭ 4). UBQLN1 knockdown significantly increased APP m levels on the cell surface on average 1.5-fold. C, more newly synthesized APP m is localized to cell surface of the H4-APP751 cells following knockdown of UBQLN1. Cells were pulsed for 20 min with [ 35 S]Met and chased for 20 min in which after cell surface proteins were biotinylated and the levels of radiolabeled APP molecules on the cell surface were determined using sequential immunoprecipitation with APP CTF antibody followed by streptavidin. *, p Ͻ 0.05, ϮS.D. levels were not increased in these two PS1 loss-of-function studies owing to lack of ␥-secretase activity. It has previously been suggested that PS1 regulates the recruitment or association of trafficking factors with cytoplasmic sorting signals within APP, thus controlling the sorting of APP to the cell surface (39). In this context, it is tempting to speculate whether UBQLN1 is one of those suggested trafficking factors, a "gatekeeper," which together with PS1 and other trafficking factors co-operatively modulates APP trafficking to the cell surface in the secretory pathway. In this scenario, down-regulation of UBQLN1 might be expected to potentiate APP trafficking from the trans-Golgi network to the cell surface. On the other hand, it was recently shown that APP and PS1 are packed differently into COPII vesicles early in the secretory pathway suggesting that APP and PS1 trafficking from the ER are normally uncoupled (40).
UBQLN1 has been reported to regulate nicotine-induced upregulation of neuronal nicotinic acetylcholine receptors (41). Overexpression of UBQLN1, together with neuronal nicotinic acetylcholine receptors, dramatically reduced the expression of receptors on the cell surface. Consistent with the hypothesis that UBQLN1 and PS1 co-operatively regulate trafficking of proteins to the cell surface, Leem et al. (38) showed that a PS1 deletion mutant increased assembly and surface expression of nicotinic acetylcholine receptors, in addition to APP. These data suggest that other cell surface proteins, beyond APP, may also be affected by UBQLN1 knockdown. Because UBQLN1 knockdown did not affect trafficking of TFR to the cell surface, protein trafficking was not globally affected. However, the fact that APLP2 holoprotein levels and maturation were changed in a similar manner to that observed for APP suggests that UBQLN1 down-regulation also affects trafficking of other APP family members. Because UBQLN1 has previously been shown to interact with the cytoplasmic loop and C-terminal domain of PS1 (13), we considered the possibility that UBQLN1 modulates APP trafficking via PS1. Given that we could observe FRET between fluorescently labeled APP and UBQLN1 proteins in intact cells independently of the presence of PS1 and PS2, the presenilins do not seem to be required for bringing UBQLN1 and APP into close proximity. It has been previously shown that overexpression of FE65/FE65L leads to similar effects on APP maturation/ secretion as those observed here following knockdown of UBQLN1 (34,35). Based on these data, we speculated that UBQLN1 might regulate the binding of adapter proteins such as FE65 to APP or, alternatively, that down-regulation of UBQLN1 might modulate FE65 levels through an unknown mechanism. In the current study, we did not observe any changes in FE65 levels following UBQLN1 knockdown, ruling out the possibility that UBQLN1 directly regulates FE65 levels. However, keeping in mind that UBQLN1 is in close proximity to APP, it remains to be determined whether the down-regulation of UBQLN1 modulates binding of other adapter proteins to APP, e.g. FE65L, X11, and others.  4). B, quantification of APLP immature (im) and mature (m) levels (␤-tubulin normalized) revealed that APLP2 im levels were significantly decreased similar to APP while APLP2 m levels were unchanged. The APLP2 m /APLP2 im ratio was significantly increased on average 2.2-fold. Steady-state levels of FE65 remained unchanged following UBQLN1 knockdown. **, p Ͻ 0.01, ϮS.D.  In summary, we have observed that down-regulation of UBQLN1 accelerates APP maturation and secretion while not interfering with ␣-, ␤-, or ␥-secretase levels or function in both neuronal and non-neuronal cell lines. We also show that knockdown of UBQLN1 increases cell surface levels of APP, and secretion of sAPP, A␤40, and A␤42. Finally, we showed that UBQLN1 comes into close proximity to APP in the absence of PS1 in intact cells. Collectively, our findings suggest that UBQLN1 may normally serve as a cytoplasmic gatekeeper, which together with PS and other factors may control APP trafficking from intracellular compartments to the cell surface. These findings also suggest that changes in UBQLN1 steadystate levels in the brain may affect APP trafficking and processing, thereby influencing the generation of sAPP and A␤, and thus risk for AD. These data, together with previous data implicating DNA variants in UBQLN1 gene as minor risk factors conferring for late-onset AD, warrant further investigation of the potential role of UBQLN1 in the etiology and pathogenesis of AD.