FABP3 Protein Promotes α-Synuclein Oligomerization Associated with 1-Methyl-1,2,3,6-tetrahydropiridine-induced Neurotoxicity*

Background: αSyn toxicity is triggered by oligomerization of αSyn, and its formation is partly regulated by PUFAs. Results: MPTP-induced neurotoxicity and αSyn oligomerization are attenuated in Fabp3−/− mice. Conclusion: FABP3 is implicated in arachidonic acid-induced αSyn oligomerization and promotes dopaminergic cell death. Significance: FABP3 aggravates MPTP-induced neuronal toxicity and αSyn accumulation. α-Synuclein (αSyn) accumulation in dopaminergic (DA) neurons is partly regulated by long-chain polyunsaturated fatty acids. We found that fatty acid-binding protein 3 (FABP3, H-FABP), a factor critical for arachidonic acid (AA) transport and metabolism in brain, is highly expressed in DA neurons. Fabp3 knock-out (Fabp3−/−) mice were resistant to 1-methyl-1,2,3,6-tetrahydropiridine-induced DA neurodegeneration in the substantia nigra pars compacta and showed improved motor function. Interestingly, FABP3 interacted with αSyn in the substantia nigra pars compacta, and αSyn accumulation following 1-methyl-1,2,3,6-tetrahydropiridine treatment was attenuated in Fabp3−/− compared with wild-type mice. We confirmed that FABP3 overexpression aggravates AA-induced αSyn oligomerization and promotes cell death in PC12 cells, whereas overexpression of a mutant form of FABP3 lacking fatty-acid binding capacity did not. Taken together, αSyn oligomerization in DA neurons is likely aggravated by AA through FABP3 in Parkinson disease pathology.

Parkinson disease (PD) 2 is a common motor disorder affecting Ͼ1% of the population over 65 years of age worldwide (1). Histopathologic features of PD are the loss of dopaminergic (DA) neurons in the substantia nigra pars compacta (SNpc) and the presence of cytoplasmic protein aggregates, known as Lewy bodies (LBs) (2). ␣-Synuclein (␣Syn), a 140-amino acid protein, is associated with synaptic vesicles in presynaptic nerve terminals (3), and ␤-sheet fibrillar aggregates, including ␣Syn, are major components of LBs. ␣Syn accumulation is associated with progressive loss of DA neurons, implicating that activity in PD pathogenesis (4). In addition, duplication/triplication (5)(6)(7) and missense mutations (A53T, A30P, E46K, H50Q, and G51D) (8 -12) in the ␣Syn gene SNCA are linked to familial early onset PD, suggesting that the mutations accelerate ␣Syn aggregation and disease progression.
␣Syn toxicity is triggered by oligomerization of ␣Syn in vitro (13) and in vivo (14), indicating that oligomerization underlies cytotoxic events in PD. However, mechanisms underlying ␣Syn oligomerization in DA neurons are unclear. Previous reports suggested that ␣Syn binds fatty acids, particularly long-chain polyunsaturated fatty acids (PUFAs) (15,16), and that ␣Syn oligomerization and the appearance of LB-like inclusions in cultured mesencephalic neuronal cells are enhanced by exposure to PUFAs (17)(18)(19). In addition, abnormally high PUFA levels are observed in ␣Syn-transfected mesencephalic neuronal cells and in PD brains, whereas lower levels are seen in mice lacking ␣Syn (17,18), suggesting that PUFA binding to ␣Syn is a key event in generating pathogenic ␣Syn oligomers.
Because PUFAs are insoluble in an aqueous cellular environment, fatty acid-binding proteins (FABPs) acting as cellular shuttles are essential to transport them to appropriate intracellular compartments (20). Among the FABPs, FABP3, which is expressed in neurons (21), shows a preference for binding to n-6 fatty acids (22). Indeed, Fabp3 knock-out (Fabp3 Ϫ/Ϫ ) mice exhibit a 24% reduction in incorporation of arachidonic acid (AA) into brain membranes and reduced levels of total n-6 fatty acids in major phospholipid classes in membranes (23), suggesting that FABP3 is critical for neuronal AA uptake and metabolism. We report herein that FABP3 is highly expressed in DA neurons and accelerates ␣Syn oligomerization, thereby aggravating AA-induced ␣Syn oligomerization and its toxicity.

EXPERIMENTAL PROCEDURES
Animals-Generation of Fabp3 Ϫ/Ϫ mice was described previously (24). Adult 12-week-old mice were used in all experiments. Mice were housed under climate-controlled conditions with a 12-h light/dark cycle and provided standard food and water ad libitum. Experiments were approved by the Institutional Animal Care and Use Committee at Tohoku University.
Behavioral Tests-In training sessions for the rotarod task, mice were placed on a drum (ENV-576M; Med Associates, St. Albans, VT) rotating at 20 rpm until the latency to fall from the drum exceeded 200 s. For test sessions, mice were placed on the rotating rod and latency to fall was recorded for up to 5 min. The beam-walking task was performed as described previously (25).
Cell Culture and Viability Assay-PC12 cells were maintained in Dulbecco's minimal essential medium (DMEM) supplemented with 10% horse serum, 5% fetal bovine serum (FBS), and penicillin/streptomycin (100 units/100 g/ml) at 37°C under 5% CO 2 . Cells were transfected using Lipofectamine 2000 (Invitrogen) as described previously (26). Conditioning living PC12 cells with AA was carried out as described previously (18). Briefly, at 32 h post-transfection in serum-free DMEM, fatty acid-free bovine serum albumin (BSA, Sigma)-AA (Sigma) complexes were added to the medium. These complexes were prepared by mixing BSA with AA (at a 1:5 molar ratio) in binding buffer containing 10 mM Tris-HCl (pH 8.0), 150 mM NaCl at 37°C for 30 min. After treatment of cells with AA for 16 h, a final concentration of 500 M 1-methyl-4-phenylpyridinium (MPP ϩ , Sigma) was added for an additional 24 h. Survival experiments were performed as described previously (28). The appearance of condensed nuclear staining with DAPI (Vector Laboratories) served as an indicator of cell death. Triplicate cultures were used for each condition, and each experiment was performed at least three times.
In Vitro ␣Syn Oligimerization-Recombinant human ␣Syn (Enzo Life Sciences, Farmingdale, NY) was incubated with AA and recombinant human His-tagged FABP3 (Cayman Chemical, Ann Arbor, MI) at the indicated concentrations in binding buffer containing 10 mM Tris-HCl (pH 8.0), 150 mM NaCl at 37°C for 30 min. For detection by immunoblotting, samples were mixed with SDS buffer without boiling.
Chemical Cross-linking Reactions-We performed chemical cross-linking reactions to identify ␣Syn oligomerization (29). For in vitro cross-linking of recombinant proteins, dithiobis-(succinimidylpropionate) (DSP) (Pierce) was added to the incubation mixture with a final concentration of 30 M, and the recombinant proteins were incubated at 37°C for 30 min in PBS containing protease inhibitors. The cross-linking reactions were terminated by incubation with Tris-HCl (pH 7.5) at 50 mM final concentration for 15 min at room temperature. Samples were mixed with SDS buffer without boiling. For in situ crosslinking in PC12 cells, transfected cells in 60-mm dishes were washed with PBS and incubated with DSP (1 mM) at 37°C for 30 min. The cross-linking reactions were terminated in the dishes by incubation with Tris-HCl (pH 7.5) at 50 mM final concentration for 15 min at room temperature. After chemical cross-linking, cells were collected by scraping and homogenized in buffer containing PBS with 1% Triton X-100 and protease inhibitors. Samples were mixed with SDS buffer without boiling. We also detected ␣Syn oligomerization in mouse brain samples using the native method without chemical cross-linking.
Statistical Evaluation-All values were expressed as means Ϯ S.E. Comparison between two experimental groups was made using the unpaired Student's t test for immunoblot and immunohistochemical analyses. Behavioral tests were analyzed using two-way analysis of variance, followed by one-way analysis of variance for each group and Dunnett's tests. p Ͻ 0.05 was considered significant.

RESULTS
Fabp3 Ϫ/Ϫ Mice Are Resistant to MPTP-induced DA Neurodegeneration in the SNpc-15-kDa cytoplasmic FABPs occur as 13 different isoforms that are widely distributed in various tissues. Among FABPs, FABP3, FABP5, and FABP7 are expressed in brain (30). FABP5 is predominantly expressed in immature neurons and glial cells and FABP7 is in glial cells, whereas FABP3 is highly expressed in mature neurons (21). In the substantia nigra, strong FABP3 immunoreactivity was observed in the SNpc but not in the substantia nigra reticular. That immunoreactivity was totally abolished in Fabp3 Ϫ/Ϫ mice (Fig. 1A) (26). In addition, analysis of TH immunoreactivity indicated that most FABP3-positive neurons were dopaminergic (Fig.  1B).
To address the role of FABP3 in the pathogenesis of PD, mice were treated with MPTP ( Fig. 1, C-E). We observed no significant difference in the number of TH-positive neurons between saline-treated wild-type (WT) and Fabp3 Ϫ/Ϫ mice. In WT mice, 4 weeks after the last MPTP injection, the number of TH-positive neurons in SNpc was markedly reduced compared with numbers seen in saline-treated WT mice (t ϭ 5.39, p Ͻ 0.01, n ϭ 6 each). By contrast, the number of TH-positive neurons was unchanged by MPTP treatment in Fabp3 Ϫ/Ϫ mice (t ϭ 0.84, p ϭ 0.231, n ϭ 6 each). The number of TH-positive neurons was significantly rescued in MPTP-treated Fabp3 Ϫ/Ϫ mice compared with MPTP-treated WT mice (t ϭ 4.38, p ϭ 0.011). These results suggest that endogenous FABP3 aggravates DA neurotoxin-induced cell death.
MPTP-induced Neurotoxicity Is Attenuated in the Striatum of Fabp3 Ϫ/Ϫ Mouse Brain-Next we confirmed that DA terminals in the striatum are less damaged by MPTP in an Fabp3 Ϫ/Ϫ background. TH immunoreactivity in the striatum was reduced in WT mice following MPTP treatment but was unchanged in MPTP-treated Fabp3 Ϫ/Ϫ mice ( Fig. 2A). To quantify TH immunoreactivity in striatal regions, we performed immunoblotting analysis and found that TH protein levels in the striatum of MPTP-treated animals were significantly higher in Fabp3 Ϫ/Ϫ (66.6 Ϯ 7.4%) compared with WT (46.3 Ϯ 5.0%) mice (Fig. 2B) (t ϭ 2.27, p ϭ 0.018, n ϭ 8 -12).
FABP3 Deficiency Attenuates MPTP-induced ␣Syn Accumulation in the SNpc-We next asked whether the resistance to MPTP-induced DA neurodegeneration in Fabp3 Ϫ/Ϫ mice is associated with reduced ␣Syn oligomerization. Immunoblot analysis using an ␣Syn-specific antibody showed that although the levels of ␣Syn 15-kDa monomer and oligomers were unchanged in both denatured and nondenatured extracts of substantia nigra from saline-treated WT and Fabp3 Ϫ/Ϫ mice, significantly higher levels of ␣Syn oligomers were seen in nondenatured substantia nigra extracts in WT compared with Fabp3 Ϫ/Ϫ mice following MPTP treatment (p Ͻ 0.01, n ϭ 5) (Fig. 4A). In denatured samples, levels of ␣Syn monomer and FABP3 were significantly up-regulated in MPTP-treated WT mice but not in Fabp3 Ϫ/Ϫ mice (Fig. 4B) (␣Syn (t ϭ 4.1, p Ͻ 0.01) and FABP3 (t ϭ 2.44, p Ͻ 0.05)). In confocal microscopic analysis, consistent with our result in Fig. 1, the TH immunoreactivity in the substantia nigra was markedly reduced in MPTP-treated WT mice. In addition, immunolabeling with ␣Syn antibody showed significant ␣Syn accumulation in DA cell bodies in the substantia nigra of MPTP-treated WT mice. In contrast, only mild ␣Syn immunoreactivity was detected in DA cell bodies of MPTP-treated Fabp3 Ϫ/Ϫ mice (saline-treated WT, 34 cells; MPTP-treated WT, 248 cells; saline-treated Fabp3 Ϫ/Ϫ , 42 cells; MPTP-treated Fabp3 Ϫ/Ϫ , 89 cells; n ϭ 5 each) (Fig. 4C).
FABP3 Makes Complexes with ␣Syn Oligomers-In previous reports, purified recombinant human ␣Syn could bind radiolabeled oleic acid ( 14 C-18:1) and decosahexaenoic acid (22:6) in vitro, and these fatty acids promote ␣Syn oligomerization (16,18). We now asked whether the direct binding of AA to ␣Syn promotes its oligomerization. After 30 min of incubation without AA, the ␣Syn was only detected as monomeric form (15 kDa). In contrast, incubation with AA clearly promoted the oligomerization of recombinant ␣Syn (60 -100 kDa). Importantly, the ␣Syn oligomerization was enhanced by adding recombinant human FABP3 (50 M AA p ϭ 0.0016 and 100 M AA p ϭ 0.024, in FABP3 absence versus presence, n ϭ 3 each) (Fig. 5A). To confirm the interaction between ␣Syn and FABP3 in vitro, we performed chemical cross-linking, a well established biochemical method to identify ␣Syn oligomerization (29). Consistent with the previous report, we detected crosslinker-induced ␣Syn oligomers with 60 and 90 kDa. The 60-kDa formation was FABP3 concentration-dependent manner (Fig. 5B, left, arrow). In addition, we found a few minor immunoreactive bands with 70 and 95 kDa, which correspond to FABP3 immunoreactive bands with 70 and 95 kDa (Fig. 5B,  right, arrowhead). Using FABP3 antibody, we also observed a few minor ␣Syn-FABP3 complexes with 60 and 90 kDa, which likely contain ␣Syn oligomers (Fig. 5B, left and right, arrow). To further confirm an interaction between ␣Syn and FABP3 in vivo, we performed immunoprecipitation of ␣Syn from substantia nigra extracts using ␣Syn antibody. The immunoprecipitates were then immunoblotted with FABP3 antibody. ␣Syn-FABP3 oligomeric complexes with 65 and 90 kDa were observed in MPTP-treated WT mice but not in Fabp3 Ϫ/Ϫ mice (Fig. 5C). In addition, most FABP3 immunoreactivity colocalized with ␣Syn accumulation in DA cell bodies of MPTPtreated WT mice (Fig. 5D, arrow), suggesting that FABP3 makes complexes with ␣Syn oligomers and promotes ␣Syn oligomerization.
Finally, we addressed whether AA promotes FABP3-induced ␣Syn aggregation. Interestingly, 100 M AA treatment enhanced levels of ␣Syn oligomerization in FABP3-transfected cells (202.6 Ϯ 10.2% of ␣Synand FABP3-cotransfected cells with MPP ϩ , p Ͻ 0.01, n ϭ 3 each), and oligomerization was markedly attenuated in cells transfected with the FABP3(F16S) construct, a mutant lacking fatty acid binding capacity (Fig. 7A) (27). These results indicate that AA-bound FABP3 increases ␣Syn oligomerization. More importantly, exposure of FABP3overexpressing cells to AA significantly promoted cell death in response to MPP ϩ compared with cells expressing ␣Syn alone, and FABP3(F16S) overexpression significantly rescued cells from AA-potentiated FABP3-induced cell death (

DISCUSSION
In this study, we report that FABP3 is implicated in the MPTP-induced neuronal toxicity and ␣Syn accumulation. We first observed that Fabp3 Ϫ/Ϫ mice were more resistant to neu- rotoxin-induced DA neurodegeneration and motor deficits in the murine PD model. The ameliorating effects seen in Fabp3 Ϫ/Ϫ mice were highly correlated with a reduction in ␣Syn oligomerization in the SNpc. We then confirmed enhanced ␣Syn oligomerization in response to up-regulated FABP3 expression and FABP3-mediated AA incorporation following neurotoxin exposure. Based on these observations, we suggest that FABP3 up-regulation by MPTP accelerates ␣Syn oligomerization and accumulation, leading to DA neurodegeneration.
Interestingly, others have proposed that ␣Syn could function as an FABP, as it exhibits an ␣-helical lipid-binding motif similar to class A2 lipid-binding domains seen in apolipoproteins and which accounts for binding to membrane phospholipids (16). However, titration microcalorimetry analysis indicates that ␣Syn binds monomeric AA and decosahexaenoic acid with only low affinity (K d ϭ 1-4 M) (31), which is about 2 orders of magnitude less affinity than classical FABPs, including FABP3 (32). In addition, unlike the case with classical FABPs, NMR spectroscopy has not identified specific fatty acid-binding sites or similarities in tertiary structure between ␣Syn and FABP (33). Thus, regulation of ␣Syn by PUFAs may require a specific lipid composition or the presence of neuron-specific lipidbinding partners. FABP3 would interact with ␣Syn and AA to promote ␣Syn oligomerization.
More importantly, FABP3 is highly expressed in DA neurons in SNpc and plays critical roles in DA neurotoxicity in vivo. Because Fabp3 Ϫ/Ϫ mice exhibit markedly reduced incorporation of AA into brain tissue plasma membranes (23), we hypothesized that DA neuroprotection in Fabp3 Ϫ/Ϫ mice is elicited by AA-dependent production of prostaglandin E2 (PGE 2 ), which others have shown to be responsible for cyclooxygenase-2 (COX-2)-mediated neurotoxicity in neuroinflammatory events (34). To investigate potential roles for PGE 2 production in Fabp3 Ϫ/Ϫ mice, we determined levels of released PGE 2 in mesencephalic cultures treated with MPP ϩ . Unexpectedly, we observed no significant difference in PGE 2 production between WT and Fabp3 Ϫ/Ϫ mesencephalic cells. 3 This observation indicates that neurotoxin-induced PGE 2 production does not account for inhibition of DA neuronal death seen in Fabp3 Ϫ/Ϫ mice.
MPP ϩ , a toxic metabolite of MPTP, is an inhibitor of complex I in the mitochondrial electron transport chain and a substrate for the dopamine transporter, therefore accumulating in DA neurons and eliciting neurodegeneration (35). Interestingly, the N-terminal 32 amino acids of human ␣Syn contain a cryptic mitochondrial targeting signal (36), and ␣Syn is accumulated in the mitochondria of post-mortem PD brains (36). FABP3 overexpression causes mitochondrial dysfunction and induces apoptosis in the P19 mouse teratocarcinoma cell line (37); FABP3 may also induce a mitochondrial dysfunction and is implicated in oxidative stress induced by MPTP toxicity.
In our study, we found the significant reduction of the number of ␣Syn-accumulated cells in DA cell bodies of MPTPtreated Fabp3 Ϫ/Ϫ mice compared with MPTP-treated WT mice (Fig. 4C). However, MPTP-treated mouse models do not induce ␣Syn-containing inclusions, similar to LBs (2). Further study will be required to investigate some differences in formation of ␣Syn-containing inclusions between the MPTP treatment model and the PD model by rotenone treatment (38) or ubiquitin-proteasome inhibitor treatment (39).
Increased AA intake is reportedly correlated with PD risk (40), and higher levels of AA and total n-6 PUFAs have been observed in post-mortem PD brains than healthy controls (41). Proteomic analysis of human substantia nigra indicated higher levels of FABP3 protein in PD patients than in control subjects (42). Higher FABP3 levels have been reported in the sera of patients with dementia accompanied by LBs (43) and of PD patients (44) compared with Alzheimer disease patients and nondemented controls. Although further studies are war-3 N. Shioda and K. Fukunaga, manuscript in preparation. FIGURE 5. FABP3 makes complexes with ␣Syn oligomers. A, left, recombinant human ␣Syn (r␣Syn) was incubated with or without AA at the indicated concentrations at 37°C for 30 min. Samples were immunoblotted with anti-␣Syn antibody. Of note, ␣Syn oligomerization was enhanced by adding recombinant human FABP3 (rFABP3). Right, quantitative densitometry analysis. *, p Ͻ 0.05; **, p Ͻ 0.01 in the presence versus absence of rFABP3. B, recombinant human ␣Syn at a concentration of 10 g/ml was cross-linked using 30 M DSP in the presence or absence of recombinant human FABP3 (10 or 100 g/ml). Samples were immunoblotted with anti-␣Syn (left) or with anti-FABP3 (right) antibody. C, coimmunoprecipitation of ␣Syn and FABP3 in total lysates from the substantia nigra. Extracts were immunoprecipitated (IP) with anti-␣Syn antibody, and immunoprecipitates were then immunoblotted (WB) with anti-FABP3 antibody. D, confocal images showing localization of ␣Syn (red) and FABP3 (green) in the substantia nigra. ␣Syn and FABP3 are colocalized in DA cell body (arrow). Scale bar, D, 20 m. WT, wild-type mice; KO, Fabp3 Ϫ/Ϫ mice. ranted, our findings suggest that up-regulation of FABP3 protein and increased AA/PUFA incorporation likely function in LB formation in PD. These results also provide an intriguing clue with respect to a potential molecular target for neurodegeneration in human ␣-synucleinopathies, including PD.