The construction of APP as bait, and screening method was described elsewhere 20.

Cell lines, transfection methods, mammalian expression constructs of APP, APP-Ncas, APP Swedish, APP London, APP-YFP, myc-tagged Notch, APLP2 were described 2027. CD74 is cloned from the two-hybrid screening, and its mouse counterparts, Ii p31 and Ii p41 were generous gifts from Dr. Norbert Koch 28. All CD74 constructs, both human and mouse, were cloned into pcDNA3.1 vector (Invitrogen) with a N-terminal FLAG tag. The Dutch and Australian mutations of APP were introduced into APP by QuikChange XL (Stratagene).

The following antibodies and antibody beads were used: αFLAG (M2, Sigma F1804); αmyc (Cell Signaling, 2276); αAPP (22C11, Chemicon MAB348); αsAPPα (IBL 11088); αsAPPβ (IBL 18957); αAPP C-terminal fragments (CTF) (αAPPct, Invitrogen/Zymed 36-6900); αAPLP2 (EMD/Calbiochem 171616); α transferrin receptor; Flag M2 beads (Sigma A2220). Rabbit polyclonal antibody and horse radish peroxidase conjugated secondary antibodies are from Southern Biotechnology.

All Western blot samples were separated by 4-12% Criterion Gels (Bio-Rad 3450125) with NuPAGE MES running buffer (Invitrogen NP0002). The gels were transferred to nitrocellulose membrane (Whatman BA85 Protran) with NuPAGE transfer buffer (Invitrogen NP0006) in a Trans-Blot cell (Bio-Rad). After the transfer, the membrane was blocked in PBS containing 5% skim milk, and incubated with indicated primary antibodies over night in the same buffer. The membrane was washed in PBS containing 0.05% Tween-20, and incubated in corresponding secondary antibodies for 1 hour. The membrane was washed in PBS containing 0.05% Tween-20 again, and the bands were visualized with Super Signal West Pico Chemiluminescent Substrate (Pierce) and Hyblot X-ray films (Denville Scientific).

HeLa cells were transfected with indicated combinations of plasmids. The transfected cells were lysed in the Hepes-Triton buffer (20 mM Hepes/NaOH pH 7.4, 1 mM EDTA, 150 mM NaCl, 0.5% Triton-X 100) on ice. The lysates were cleared by spinning at 20,000 g for 10 min. To precipitate the immune complex, the cleared lysates were mixed with FLAG beads, or with indicated antibodies and protein A beads (Pierce). A rabbit polyclonal antibody was used as a negative control of αAPPct precipitation. After being washed three times with the same buffer, the precipitated beads were boiled in 2 × SDS Buffer and analyzed by Westerm blot. The immunoprecipitants analyzed correspond to the four times of the inputs.

HeLa cells were plated on coverslips coated with poly L-lysine (Sigma), and were transfected with indicated plasmids. The cells on coverslips were fixed, permeabilized with 0.2% Triton X-100, stained as described 29. APP-YFP and Notch-GFP were detected with YFP/GFP fluorescence. Anti-FLAG and αAPPct stainings were visualized with Alexa Fluor 594 goat anti-mouse IgG (H+L) (Invitrogen) and Alexa Fluor 488 goat anti-rabbit IgG (H+L) (Invitrogen), respectively. Cells were imaged sequentially with the Albert Einstein Analytical Imaging Facility BioRad Radiance 2000 confocal microscope using a 63× NA1.4 oil objective and 488 nm and 568 nm laser lines. The images shown are representative of the phenotypes observed in at least three independent experiments.

Regions of interest were analyzed with ImageJ software (National Institute of Health) and Colocalization Indices plug-in 30. Colocalization coefficient (CC) 31 and intensity correlation quotient (ICQ) 32 were calculated with the software by setting the threshold of CD74 staining so that the calculation is limited to CD74-positive vacuoles. CC ranges from -1 to 1: 1 indicates complete colocalization; 0, ramdom colocalization, -1, complete negative colocalization. Positive ICQ indicates positive colocalizatioin.

HEK293APP cells were transfected with pcDNA3 or with FLAG-CD74. One day after the transfection, the cells were further incubated in fresh media for 24 hours. The conditioned media were collected and cleared by spinning at 20,000 g for 10 min. Cells were lysed in the Hepes-Triton buffer, and protein concentration was determined by Bradford method using bovine serum albumin as a standard. Aβ in the media and the cell lysates were measured using the ELISA kits for Aβ40 and Aβ42 (IBL 17713 and 17711, respectively).

Equal amounts of total lysates and the cleared media were analyzed by Westerm blot.

Synthetic Aβ40 and Aβ42 (American Peptide Company) were dissolved in 50 mM Tris/HCl pH 8.0 at 1 mg/ml and stored at -70°C until use. HeLa cells were transfected with pcDNA3 or FLAG-CD74. One day after transfection, the media was replaced with fresh media containing synthetic Aβ40 and Aβ42 at approximately 100 ng/ml. The media were collected at indicated time points and Aβ was measured as above.

HeLa cells were transfected with pcDNA3 or FLAG-CD74 together with APP. One day after the transfection, the cells were starved for 90 minutes by incubating in DMEM without cysteine and methionine (Invitrogen). The cells were labelled at 500 Ci/ml of ³⁵S cysteine and methionine for 30 minutes and chased in the complete DMEM for indicated times. After the chase, the cells were lysed and immunoprecipitated with αAPPct as above. The precipitants were analyzed by SDS-PAGE and autoradiography.

To identify APP ligands, we screened human brain cDNA library with split-ubiquitin system using APP as bait 20. This screening is designed so that the each bait and prey are fused to each half of modified ubiquitin, which upon interaction releases the DNA binding domain and activation domain. They are translocated to the nuclei and transcriptionally activate the downstream reporter gene 33. This screening enabled us to clone transmembrane proteins, which are very difficult to clone in regular yeast two hybrid screening because the transmembrane proteins do not translocate to the nucleus to activate the downstream signal.

One of the clones isolated contained the entire CD74 cDNA (Figure 1A). CD74 serves as an invariant chain of Class II major histocompatibility complex (MIIC) 34, and it also serves as a receptor for macrophage migration inhibitory factor 24. CD74 is a type II transmembrane protein, which is processed by protease during its maturation as a chaperon of Class II MHC molecules 23.

In the same split-ubiquitin yeast screening using APP as bait, we cloned BRI2 and BRI3 202122. Since BRI2 and BRI3 are also single pass, type II transmembrane proteins themselves, and interact with APP and inhibit the production of Aβ 202122 we decided to test if the same is true for CD74 in mammalian cell context.

Full length CD74 with an N-terminal FLAG tag were transfected to HeLa cells. As shown in Figure 1B, the immunoprecipitation of CD74 from the total lysates indicated that predominantly mature APP (mAPP) interacted with CD74. There is little binding to the two APP C-terminal fragments (APP CTFs, indicated as C99 and C83). No consistent change was observed for the amount of APP CTFs. This CD74-APP interaction was further confirmed by the reciprocal immunoprecipitation. The total lysates were prepared from the HeLa cells cotransfected with FLAG-CD74 and APP, and the CD74 fragments bound to the precipitated APP were analyzed. The Westerm blot shows that CD74 indeed bound to APP. Interestingly, the 33 kDa fragment (CD74 NTF33), but not the 14 kDa fragment (CD74 NTF14) of CD74 was associated with APP, even though NTF14 was clearly the major CD74 NTF in the total lysates (Figure 1B). The preferred binding of NTF33 over NTF14 indicates that NTF 14 is not sufficient for the CD74-APP interaction, and hints that the trimeric region of CD74 is required for the interaction.

Next, we attempted to determine the region of CD74 required for this interaction. cDNA obtained in the two-hybrid screening, full length CD74, two fragments of CD74, and the two mouse CD74 isoforms (p31 and p41) were cloned into a FLAG-tagged mammalian expression vector (Figure 1A), and cotransfected into HeLa cells together with APP. Immunoprecipitation with FLAG beads showed that both human and mouse CD74 bound mature APP. The lack of interaction found in the deletion construct CD74 (1-115) confirms that the luminal portion containing the trimeric domain is required for the CD74 binding to APP. The deletion of first 23 amino acids (CD74 24-216) also disrupted the interaction, hinting to the possibility that the endosomal-targeting signal 35 might be involved in this interaction.

To investigate the specificity of CD74-APP binding, myc-tagged Notch and APLP2 were transfected into HeLa cells together with pcDNA3 or FLAG-CD74. They lysates from the transfected cells were precipitated with FLAG beads to analyze if Notch or APLP2 co purified with CD74. As indicated in Figure 2A, CD74 did not precipitate Notch. Nor did it precipitate APLP2, a close homolog of APP (Figure 2B). The lack of interaction with Notch or APLP2 further attests the specificity of the CD74-APP interaction.

Next, we tested how other APP metabolites were affected by the overexpression of CD74. HeLa cells were transfected with APP together with pcDNA3, FLAG-CD74, FLAG-CD74 (24-216), and mouse CD74 (Ii p31), and one day later, the cells were further incubated in fresh media for 24 hours. APP fragments in the media and in the cell were analyzed by Westerm blot. As shown in Figure 2C, CD74 over expression did not show significant change in APP, APP CTFs, sAPPα, or sAPPβ (left panels: Western blots, right panels: quantification of the blots). Student's t test showed no significant difference between any combinations of the values.

Next, we asked if CD74 changes the production of Aβ. HEK293APP cells, which stably express APP, were transfected with pcDNA3 or with FLAG-CD74. The transfectants were further incubated in fresh media as above and the secreted Aβ40 and Aβ42 were measured with Aβ ELISA. As shown in Figure 2D, exogenous expression of CD74 decreased both the secreted Aβ40 and Aβ42. On the contrary, there is no significant change of cytoplasmic Aβ40 (Figure 2E). The levels of cytoplasmic Aβ42 were too low to be detected (data not shown).

Next, we questioned if the overexpression of CD74 affects the stability of Aβ added to the media. HeLa cells were transfected with pcDNA3 or FLAG-CD74. On the next day, the media was replaced with fresh media containing of Aβ40 and Aβ42. The ELISA of the media taken at the indicated time points showed there was no significant difference in the stability of Aβ40 (p = 0.890) and Aβ42 (p = 0.141) by paired Student's t test analysis (Figure 2E).

In order to determine whether metabolism of APP is modified by CD74, we performed pulse-chase labeling of APP. HeLa cells were transfected with pcDNA3 or FLAG-CD74 together with APP. The cells were pulse labeled with ³⁵S methionine and cysteine and chased for indicated hours. The autoradiography of immunoprecipitated APP indicates that CD74 did not change the conversion of cell-bound APP significantly (Figure 2F).

Next, we tested if various mutants of APP interact with CD74. HeLa cells were transfected with APP, APP-Ncas, APP Swedish, APP Dutch, APP London, and APP Australian mutants. These are all APP mutations that in humans cause familial forms of Alzheimer's disease 36, except for APP-Ncas that is an APP construct missing the COOH-terminal 31 amino acids 37. The lysates were prepared from the transfected cells, and CD74 was precipitated. The FLAG immunoprecipitations show that CD74 fully interacts with mature form of all APP variants (Figure 2G).

Next, we sought to identify how CD74 and APP are distributed inside the cell. As shown in Figure 3A for Hela cells transfected with APP-YFP or FLAG-CD74, APP is localized in intracellular reticular structures, while CD74 is targeted to endosomes 35. Overexpression of CD74 and has been reported to create vacuoles 35 that we also can appreciate (Figure 3A).

When these two were co expressed in the same cell, a significant portion of APP co localized with CD74 in these vacuoles (Figure 3B, upper panels). This CD74-dependent redistribution of APP is specific since, for example, CD74 did not co-localize in these vacuolar structures with Notch-GFP, reinforcing the above mentioned biochemical observation that CD74 does not bind Notch (Figure 2A).

Next, we tested if the major APP CTF fragments, C99 and C83, changed intracellular localization if expressed together with CD74. For this, HeLa cells were cotransfected with these APP variants and FLAG-CD74, and stained with αAPPct to visualize APP, C99 and C83, and αFLAG to visualize CD74. As expected, APP co localized with CD74 nicely in the vacuolar structures. However, C99 and C83 did not show detectable accumulation to the CD74 induced vacuolar structures. This observation could explain the initial immunoprecipitations results that CD74 barely interact with APP CTFs (Figure 1B).