To identify molecules that specifically bind to apoE4, various fractions obtained from mouse brain were loaded onto a FLAG-apoE4(1–272) or FLAG-apoE4(1–299) affinity column. The proteins bound to these columns were eluted with an excess amount of FLAG peptide and the extracts were subjected to SDS-PAGE and silver staining. When each mouse brain fraction was loaded onto a FLAG-apoE4(1–272) or FLAG-apoE4(1–299) affinity column and eluted samples were subjected to SDS-PAGE, many protein bands were detected as candidate apoE4-associated proteins (Fig. 1). These bands were detected only when the mouse brain fractions and FLAG-apoE4 recombinant proteins coexisted.

To identify these proteins, the protein bands were subjected to LC-MS/MS analysis. As a result, we identified sixteen apoE4-associated proteins, which are shown in Table 1. These candidate proteins include the ATP synthase protein α and β subunits, which were previously reported 30. Interestingly, ten among the candidate proteins are shown to be associated with mitochondria.

Next, we examined whether these proteins really associate with apoE in living cells. We cotransfected each candidate protein and FLAG-apoE4 into Neuro2a cells. Twenty-four hours following the transfection, the cells were harvested and treated with Triton X-100 solubilization buffer to obtain cell lysate as described in Materials and Methods. We carried out immunoprecipitation using these samples with an anti-FLAG antibody. The immunoprecipitate was then subjected to western blotting analysis using antibody specific for each protein or anti-HA antibody. This is because, in case we could not find antibodies specific for some candidate proteins, we could still generate HA-tagged proteins. Among the proteins identified by LC-MS/MS analysis, we found that three proteins, ubiquinol cytochrome c reductase core protein 2 (UQCRC2), cytochrome C1, and cytochrome c oxidase subunit 4 isoform 1 (COX IV 1) were associated with apoE4(1–299) and apoE4(1–272).

To examine whether mouse UQCRC2 is really associated with apoE proteins in cells, Neuro2a cells were cotransfected with mammalian expression plasmids encoding FLAG-apoE4(1–272) or FLAG-apoE4(1–299) and plasmids encoding mouse UQCRC2. Western blot analysis showed that the signal representing mouse UQCRC2 was clearly detected in the immunoprecipitate from the cells cotransfected with apoE4(1–272), while a very faint signal for UQCRC2 was detected in that from the cells cotransfected with apoE4(1–299). ApoE proteins were similarly immunoprecipitated in both samples (Fig. 2A). These results suggest that UQCRC2 prefers to associate with apoE4(1–272) than with apoE4(1–299).

To examine whether human UQCRC2 also associates with apoE4, we performed an experiment using human HA-tagged UQCRC2 expression vector. Similar to the results using mouse UQCRC2-transfected cells, the signals representing human HA-tagged UQCRC2 were clearly detected in the immunoprecipitate from the cells cotransfected with apoE4(1–272), while a very faint signal for HA-UQCRC2 was detected in that from the cells cotransfected with apoE4(1–299). Under these conditions, apoE4(1–272) and apoE4(1–299) proteins were similarly immunoprecipitated in both samples (Fig. 2B).

Next, we examined the association of human cytochrome C1 and apoE proteins using Neuro2a cells cotransfected with HA-tagged human cytochrome C1 expression vector and plasmids encoding FLAG-apoE4(1–272) or FLAG-apoE4(1–299). A strong signal for HA-tagged human cytochrome C1 was detected in the immunoprecipitate from the cell samples expressing apoE4(1–272), while a very faint signal for cytochrome C1 was detected in those transfected with apoE4(1–299) plasmid. ApoE4(1–272) and apoE4(1–299) proteins were similarly detected in both samples (Fig. 2C).

In addition, whether COX IV 1 is associated with apoE was determined. Human COX IV 1 was immunoprecipitated in the samples from apoE4(1–272)-expressing cells and a very faint signal for COX IV 1 was detected in the samples from apoE4(1–299)-expressing cells, whereas apoE4(1–272) and apoE4(1–299) proteins were similarly immunoprecipitated in both samples (Fig. 3). Interestingly, these proteins, UQCRC2, cytochrome C1, and COX IV 1, are associated more strongly with apoE4(1–272) than with apoE4(1–299). Concerning other proteins, we carried out similar experiments; however, no association of these proteins with apoE4(1–272) and apoE4(1–299) was found in cultured cells (data not shown).

The results indicate that apoE4(1–272) binds to mitochondrial proteins; therefore, we next determined whether the levels of apoE4(1–272) and apoE4(1–299) are also associated with the mitochondria. We, thus, determined the levels of apoE4(1–272) and apoE4(1–299) in mitochondrion-rich fractions isolated from Neuro2a cells transfected with apoE4(1–272) or apoE4(1–299). The level of apoE4(1–272) in the pellet, a mitochondrion-rich fraction, was greater than that of apoE4(1–299) (Fig. 4). VDAC, a mitochondrion marker, was recovered in the pellet fraction (Fig. 4).

It is known that UQCRC2 and cytochrome C1 are subunits of mitochondrial respiratory complex III and COX IV 1 is a subunit of mitochondrial respiratory complex IV. It was reported that apoE4(1–272) induces mitochondrial dysfunction 25262729. Because these proteins are associated more strongly with apoE4(1–272) than with apoE4(1–299), we investigated whether the enzymatic activities of mitochondrial respiratory complexes III and IV change, when apoE4(1–272) is overexpressed in cultured cells. Complex III activity was expressed as the difference in the reduction of cytochrome c with or without antimycin A and myxothiazol, both of which are complex III inhibitors. Expectedly, the complex III activity of apoE4(1–272)-overexpressing Neuro2a cells was lower than that of apoE4(1–299)-overexpressing cells (Fig. 5A). Because apoE4(1–272) associates with UQCRC2 and cytochrome C1, there was a possibility that the decrease in complex III activity in Neuro2a cells expressing apoE4(1–272) was due to the interaction between apoE4 and complex III. Complex IV activity was expressed as the difference in the oxidation of ferrocytochrome C with or without KCN and Na₃N, both of which are complex IV inhibitors. The complex IV activity of apoE4(1–272)-overexpressing cells was significantly lower than that of apoE4(1–299)-overexpressing cells (Fig. 5B). The levels of the mitochondrial proteins UQCRC2 and cytochrome C1 in apoE4(1–272)- and apoE4(1–299)-overexpressing cells were similar, as demonstrated by western blot analysis using anti-UQCRC2 and anti-cytochrome C1 antibodies (Fig. 5C).

Because apoE4(1–272) decreases the activities of mitochondrial complexes III and IV, we next examined whether the overexpression of apoE4(1–272) also affects ATP synthase activity and mitochondrial membrane potential. Unexpectedly, apoE4(1–272) and apoE4(1–299) showed no effect on ATP synthase activity (Fig. 6A). We further examined the effect of the overexpression of apoE4(1–272) and apoE4(1–299) on mitochondrial membrane potential. Neuro2a cells transfected with these apoE4 species were stained with JC-1, a fluorescent dye that has been shown to be a reliable indicator of mitochondrial membrane potential changes in intact cells. After three hours of treatment of Neuro2a cells with 1 μM valinomycin, a K⁺ ionophore that disrupts the transmembrane electrical gradient, the intensity of red fluorescence (FL2) markedly decreased, whereas that of green fluorescence (FL1) slightly increased, as demonstrated by JC-1 staining (Fig. 6B), indicating the dissipation of mitochondrial membrane potential. On the other hand, there were no significant differences in FL1 and FL2 intensities among Neuro2a cells transfected with mock, apoE4(1–272) and apoE4(1–299) (Figs. 6B, C, and 6D), indicating that the overexpression of apoE4(1–272) and apoE4(1–299) has no effect on mitochondrial membrane potential.

Brains obtained from C57BL6 male mice were homogenized with a homogenizing buffer (10 mM Tris-HCl, pH 7.4 1 mM EDTA 0.25 M sucrose), and the homogenate was centrifuged at 1,000 g for 10 min at 4°C. The supernatant was recentrifuged at 10,000 g for 20 min at 4°C, and the resulting precipitate was suspended in a homogenizing buffer. The supernatant was recentrifuged at 100,000 g for 1 min at 4°C, and the resulting precipitate was suspended in a homogenizing buffer. The supernatant was used as the cytosolic fraction. The proteins in the 10,000 g and 100,000 g pellet fraction were extracted with homogenizing buffer containing 1 M KCl. The resulting pellet was solubilized with 2% Triton X-100 for 1 h at 4°C and then centrifuged at 100,000 g of 1 h at 4°C. The supernatants were used as 10,000 g or 100,000 g membrane extracts.

Recombinant FLAG-apoE4 fusion proteins encoding WT (1–299) or C-terminal truncated apoE4 (1–272) were prepared and purified as follows. PCR products encoding apoE4 (1–272) and apoE4 (1–299) were subcloned into pFLAG-MAC expression vector (Sigma). These plasmids were transformed into the BL21 strain of Escherichia coli and induced with isopropyl-1-thio-b-_D-galactopyranoside to produce FLAG fusion proteins. The bacteria were suspended in PBS, and vigorous sonication was performed before centrifugation at 10,000 g for 20 min. The resulting supernatants were applied to anti-FLAG M2-agarose affinity resin column (Sigma) and then eluted with an elution buffer (TBS containing 100 μg/ml FLAG peptide (Sigma)). Purified FLAG fusion proteins were dialyzed against TBS.

Recombinant FLAG-apoE4(1–299) or FLAG-apoE4(1–272) fusion protein coupled to anti-FLAG M2-agarose affinity resin was used to identify affinity-purified apoE4-binding proteins. The fractionated mouse brain samples were applied to the FLAG-apoE4(1–299)- and FLAG-apoE4(1–272)-anti-FLAG M2-agarose affinity resin column. The proteins bound to the resin column were then eluted with the elution buffer. The eluted proteins were dialyzed against the dialysis buffer, concentrated, and subjected to SDS-PAGE. The gels were stained with SilverQuest Silver Staining kit (Invitrogen). The proteins specifically associated with apoEs demonstrated as silver-stained bands were cut out, digested with trypsin, and subjected to LC-MS/MS analysis.

The proteins in the silver-stained bands were reduced with 10 mM dithiothreitol at room temperature for 2 h and alkylated with 40 mM iodoacetamide in the dark at room temperature for 30 min. Each sample was digested with trypsin (4 μg/ml; Trypsin Gold, Promega) in 40 mM NH₄HCO₃/10%ACN at 37°C overnight. The extracted peptides were then separated via nano liquid chromatography (LC) (Paradigm MS4, Michrom BioResources, Inc., Auburn, CA) using a Magic C18 column (0.2 × 50 mm; Michrom BioResources, Inc.; Auburn, CA). The LC eluent was analyzed using an LCQ Advantage MAX mass spectrometer (Thermo Fisher Scientific) equipped with an ion-spray source. All MS/MS spectra were searched by SEQUEST algorithm from BioWorks software (Thermo Fisher Scientific).

Neuro2a cells were grown in DMEM medium, supplemented with 10% FBS, 50 units/ml penicillin, 50 mg/ml streptomycin, and 2 mM glutamine at 37°C in a humidified 5% CO₂ 95% air incubator.

PCR products encoding apoE4 (1–272) and apoE4 (1–299) were subcloned into pFLAG-CMV-2 expression vector (Sigma). Neuro2a cells were co-transfected with FLAG-apoE4 (1–272 or 1–299) plasmids and mammalian expression plasmids coding the candidate of apoE4 associating proteins (Toyobo, Japan) using Lipofectamine 2000 (Invitrogen, CA, USA). As necessary, mammalian expression plasmids coding the candidate apoE4-associated proteins were fused with HA-tag of their C-terminus using KOD-PLUS Mutagenesis Kit (Toyobo). The plasmids were transfected to Neuro2a cells as follows. Neuro2a cells were plated on a 6-cm plate at a cell density of 1.5 × 10⁶ and cultured in the culture medium described above. The next day, the ells were transfected with the plasmid employing Lipofectamine 2000 reagent. On culture day 3, the cells were lysed with 500 μl of Triton X-100 solubilization buffer (10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 10 mg/ml leupeptin, 1 mM PMSF, 0.5% Triton X-100). The cell lysate was incubated with anti-FLAG M2-agarose affinity resin. After an overnight incubation at 4°C, the beads were washed three times with Triton X-100 solubilization buffer and then eluted with the elution buffer (PBS containing FLAG peptide at a concentration of 100 μg/ml). The immunoprecipitates were then analyzed by western blotting with anti-HA monoclonal antibody (Sigma), anti-UQCRC2 monoclonal antibody (Abcam), and anti-COX IV 1 monoclonal antibody (Cell Signaling).

Mitochondria isolated from homogenates of Neuro2a cells transfected with ApoE4 (1–299 or 1–272) were used for enzymatic analysis.

Neuro2a cells transfected with apoE4(1–272) and apoE4(1–299) plasmids were harvested and homogenized with a homogenizing buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.25 M sucrose), and the homogenate was centrifuged at 1,000 g for 10 min at 4°C. The resulting supernatant was further centrifuged at 10,000 × g for 20 min at 4°C, and the resulting pellet (ppt) fractions were obtained. The ppt fractions were resuspended in 100 μl of assay buffer (1 mM ADP and 5 mM sodium succinate) and incubated for 5 min at 37°C. 100 mM Tris-HCl buffer (pH 7.6) containing 4 mM EDTA was added and the fractions were further incubated for 2 min at 100°C. Then the fractions were plated on ice and ATP level in each fraction was determined using an ATP bioluminescence assay kit CLS II (Roche Diagnosis GmbH, Mannheim, Germany).

ApoE4(1–272 or 1–299)-transfected Neuro2a cells were stained with 2 μM JC-1 (Molecular Probes, Eugene, OR) at 37°C for 20 min. Then, the cells were analyzed using a flow cytometer FACSCalibur (Becton Dickinson, Franklin Lakes, NJ) with FlowJo software (Tree Star Inc., Ashland, OR).

StatView computer software (Windows) was used for statistical analysis. Statistical significance of differences between samples was evaluated by multiple pairwise comparisons among the sets of data using ANOVA and the Bonferoni t-test.