Apolipoprotein E4 Domain Interaction Occurs in Living Neuronal Cells As Determined by Fluorescence Resonance Energy Transfer*

Apolipoprotein (apo) E4 is a major risk factor for Alzheimer disease. Although the mechanisms remain to be determined, the detrimental effects of apoE4 in neurobiology must be based on its unique structural and biophysical properties. One such property is domain interaction mediated by a salt bridge between Arg-61 in the N-terminal domain and Glu-255 in the C-terminal domain of apoE4. This interaction, which does not occur in apoE3 or apoE2, causes apoE4 to bind preferentially to certain lipoprotein particles in vitro and in vivo. Here we used fluorescence resonance energy transfer (FRET) to determine whether apoE4 domain interaction occurs in living neuronal cells. Neuro-2a cells were transfected with constructs encoding apoE3 or apoE4 in which yellow fluorescent protein (YFP) was fused to the N terminus, and cyan fluorescent protein (CFP) was fused to the C terminus. To generate a FRET signal that can be detected by spectrum confocal microscopy, the labeled N and C termini must be in close proximity (<100 A). FRET signals occurred in cells transfected with YFP-apoE4-CFP but not in those transfected with YFP-apoE3-CFP, suggesting that the N and C termini of apoE4 are in close proximity in living cells and that those of apoE3 are not. FRET signals did not occur in cells cotransfected with YFP-apoE4 and apoE4-CFP, suggesting that the FRET in YFP-apoE4-CFP-transfected cells was intramolecular. Mutation of Arg-61 to Thr or Glu-255 to Ala in apoE4, which disrupts domain interaction, abolished FRET in Neuro-2a cells, strongly suggesting that the FRET in YFP-apoE4-CFP cells was caused by domain interaction. ApoE4-producing cells secreted less phospholipid than apoE3-producing cells, but after disruption of domain interaction in apoE4, phospholipid secretion increased to the levels seen with apoE3, suggesting that domain interaction decreases the phospholipid-binding capacity of apoE4. Thus, apoE4 domain interaction occurs in living neuronal cells and may be a molecular basis for apoE4-related neurodegeneration.

similar expression of apoE3. Domain interaction may contribute to the neurodegenerative effects of apoE4 in those transgenic mice (31,34,59,60). However, domain interaction has not been demonstrated in living neuronal cells.
To explore this fundamental question, we used fluorescence resonance energy transfer (FRET) to analyze living neuronal cells expressing apoE3 or apoE4. FRET-the nonradiative transfer of photon energy from an excited fluorophore (donor) to another fluorophore (acceptor)-occurs only when the donor and acceptor are in close proximity (<100 Å). Thus, this approach can be used to measure nanometer scale distances. It has been used to study domain reorganization induced by lipid association in human apoE3 (61,62). Recent advances, such as new fluorescent probes and new imaging methods, have made it possible to use this technique to detect protein-protein interactions and changes in protein conformation in living cells (63-69).
For example, cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) can be used to label two proteins to detect intermolecular interactions or to label two termini of a protein to detect intramolecular domain interaction in living cells by measuring FRET (63-74).
In the current study, we used FRET to examine apoE4 domain interaction in living neuronal cells expressing apoE3 or apoE4 fused to fluorescent proteins (YFP-apoE3-CFP and YFP-apoE4-CFP). Our results demonstrate that domain interaction of apoE4 occurs in living neuronal cells, which might be a molecular basis for apoE4-related neurodegeneration. 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate, and Roche complete protease inhibitors) for 30 min. After centrifugation at 13,000 rpm for 15 min in an Eppendorf centrifuge (Westbury, NY), apoE labeled with both YFP and CFP was detected in the supernatant by Western blotting with M2 monoclonal antibodies against the FLAG tag (Sigma) (75).

Imaging FRET by Multiphoton Confocal Microscopy-Neuro-2a cells transiently transfected
with the YFP-apoE3-CFP or YFP-apoE4-CFP cDNA construct were grown in serum-free minimum essential medium for 18-20 h, illuminated with a Zeiss two-photon laser, and examined with a Zeiss confocal microscope. To visualize CFP fluorescence, transfected cells were excited at 458 nm, and images were acquired at 465-512 nm. FRET images were acquired at the emission wavelength of YFP (543-618 nm) when CFP was excited (66, 67, 76, 77).
Although YFP and CFP have broad absorption and emission peaks that partially overlap, the Meta Detector of the Zeiss multiphoton confocal system can deconvolve the contribution of each fluorochrome to each pixel by spectral analysis using information obtained from the reference spectra (YFP or CFP alone). Thus, we could image YFP and CFP fluorescence simultaneously in neurons expressing YFP-apoE3-CFP or YFP-apoE4-CFP at similar levels for FRET analysis. anti-apoE immunoprecipitation from that obtained before immunoprecipitation. The levels of apoE3, apoE4, YFP-apoE3-CFP, YFP-apoE4-CFP, or YFP-apoE4-Thr-61-CFP in the conditioned medium were measured semiquantitatively by anti-apoE Western blotting (81) and used to normalize the lipids bound with various forms of apoE.
Statistical Analysis-Results are reported as mean ± SD. Differences were evaluated by t test or analysis of variance.

RESULTS AND DISCUSSION
To determine if apoE4 domain interaction occurs in living neuronal cells, we prepared cDNA constructs encoding YFP-apoE3-CFP or YFP-apoE4-CFP (Fig. 1A). Control cDNA constructs encoding YFP-apoE3, apoE3-CFP, YFP-apoE4, or apoE4-CFP were also prepared. All of these fusion proteins have a FLAG tag at the N-terminus. After transient transfection of mouse neuroblastoma Neuro-2a cells with these constructs, all of the singly and doubly labeled forms of apoE were detected at similar levels in cell lysates (Fig. 1A) and medium (data not shown) by Western blotting with monoclonal anti-FLAG.
Since the emission spectrum of CFP (460-520 nm) overlaps with the excitation spectrum of YFP (480-520 nm), spectrum confocal microscopy can be used to measure FRET in living cells (70,71). If CFP and YFP are in close proximity (<100 Å), part of the emission energy under CFP excitation is transferred from CFP to YFP, thereby increasing YFP emission and decreasing CFP emission. Thus, the occurrence of FRET in transfected Neuro-2a cells expressing doublelabeled apoE indicates close proximity of the N-and C-termini, consistent with domain interaction in apoE4; the absence of FRET or a very weak FRET signal suggests that the two termini are farther apart, consistent with the lack of domain interaction in apoE3 (Fig. 1B). decreases, reflecting energy transfer from CFP to YFP (Fig. 3A). Therefore, the ratio of FRET (YFP) to CFP fluorescence can be used as a measure of FRET intensity. The higher the ratio, the stronger the FRET signal. The YFP-apoE4-CFP-transfected cells had a much higher ratio of FRET to CFP than the YFP-apoE3-CFP-transfected cells (Fig. 3B), suggesting that FRET occurs in YFP-apoE4-CFP cells but not in YFP-apoE3-CFP cells.
Finally, we measured the change of the donor (CFP) fluorescence before and after laser photobleaching of the acceptor (YFP) fluorescence. If FRET occurs, this photobleaching will increase the donor (CFP) fluorescence because the energy cannot be transferred from CFP to YFP (66, 73, 80). After photobleaching of YFP, CFP fluorescence increased significantly (~25%) in YFP-apoE4-CFP cells (Fig. 4A), indicating the occurrence of FRET, but decreased slightly in YFP-apoE3-CFP cells, indicating the absence of FRET (Fig. 4B). The slight decrease in CFP fluorescence probably reflects photobleaching of CFP due to the partial overlap of the excitation spectra of YFP and CFP (73,80).   (Fig. 5A). The ratios of FRET (YFP) to CFP fluorescence were significantly lower in Neuro-2a cells expressing the mutant proteins than in cells expressing YFP-apoE4-CFP at similar levels (Fig. 5B), suggesting that the N-and the C-termini of these two apoE4 mutants are separated by more than 100 Å.
Interestingly, the ratios of YFP to CFP fluorescence in cells expressing the mutant proteins were similar to those of cells expressing YFP-apoE3-CFP. Thus, the two apoE4 mutants appear to be structurally similar to apoE3, as reported (35,36). These data strongly suggest that the FRET in YFP-apoE4-CFP cells is caused by apoE4 domain interaction.

Fusion of Fluorescent Proteins with ApoE3 and ApoE4 Does Not Alter Their Lipid Binding
Properties-To determine whether the fusion of fluorescent proteins with apoE3 and apoE4 alter their lipid binding properties, which is one of the major functions of apoE, we measured lipid content of secreted apoE-containing lipoproteins. Both apoE4 and YFP-apoE4-CFP bound approximately 25% less phospholipid (p < 0.05) than apoE3 and YFP-apoE3-CFP (Fig. 6A).
Low levels of cholesterol and undetectable levels of triglyceride were found in the medium (data not shown). These results indicate that fusion of fluorescent proteins with apoE3 and apoE4 does not alter their lipid binding properties, suggesting that the presence of YFP and CFP does not alter significantly the folding or structural properties of apoE. These results also indicate that the cells secreted less phospholipid with apoE4 than with apoE3, suggesting decreased phospholipidbinding capacity of apoE4. Furthermore, a strong FRET signal was detected in the medium containing YFP-apoE4-CFP (Fig. 6B), consistent with apoE4 domain interaction occurring in secreted phospholipid-rich lipoprotein particles.

Domain Interaction Decreases Phospholipid-binding Capacity of ApoE4-To determine
whether the domain interaction is responsible for the decreased phospholipid-binding capacity of apoE4, we compared the amount of phospholipid bound by YFP-apoE4-Thr-61-CFP, in which the domain interaction had been disrupted (Fig. 5), with that bound by YFP-apoE4-CFP or YFP-apoE3-CFP. The YFP-apoE4-Thr-61-CFP bound significantly more phospholipid than YFP- apoE4-CFP (p < 0.05). Importantly, YFP-apoE4-Thr-61-CFP and YFP-apoE3-CFP bound similar amounts of phospholipid (Fig. 6C). Thus, disruption of domain interaction restores phospholipid-binding capacity of apoE4 to a level similar to that of apoE3. These results strongly suggest that domain interaction decreases phospholipid-binding capacity of apoE4, leading to secretion of less phospholipid from apoE4-producing cells than from apoE3-producing cells.
In summary, the current study demonstrates that apoE4 domain interaction occurs in living neuronal cells as determined by FRET and confirms that the FRET technique is a powerful tool to detect intramolecular domain interaction of proteins in living cells. This study also reveals that domain interaction decreases phospholipid-binding capacity of apoE4, leading to decreased secretion of phospholipid from apoE4-producing cells. Consistent with this finding, human apoE-producing primary astrocytes from apoE4 knock-in mice secrete less phospholipid than those from apoE3 knock-in mice (82). Morphological and behavioral analyses of transgenic mice expressing apoE3 or apoE4 specifically in central nervous system neurons revealed significant age-dependent and excitotoxin-induced neurodegeneration and behavioral deficits in apoE4, but not apoE3, mice (55,58). Since the structural and biophysical properties of a protein determine its physiological or pathological functions, the intraneuronal apoE4 domain interaction observed in this study might be a molecular basis for apoE4-related neurodegeneration in apoE4 transgenic mice (31,34,59,60)