Characterization of Oligomer Formation of Amyloid-β Peptide Using a Split-luciferase Complementation Assay*

Amyloid-β peptide (Aβ) is the amyloid component of senile plaques in Alzheimer disease (AD) brains. Recently a soluble oliomeric form of Aβ in Aβ precursor protein transgenic mouse brains and AD brains was identified as a potential causative molecule for memory impairment, suggesting that soluble Aβ oligomers cause neurodegeneration in AD. Further characterization of this species has been hampered, however, because the concentrations are quite small and it is difficult to monitor Aβ oligomers specifically. Here we developed a novel method for monitoring Aβ oligomers using a split-luciferase complementation assay. In this assay, the N- and C-terminal fragments of Gaussia luciferase (Gluc) are fused separately to Aβ. We found that conditioned media from both N- and C-terminal fragments of Gluc-tagged Aβ1–42 doubly transfected HEK293 cells showed strong luminescence. We used gel filtration analyses to analyze the size of oligomers formed by the luciferase complementation assay, and found that it matched closely with oligomers formed by endogenous Aβ in Tg2576 neurons. Large oligomers (24–36-mers), 8-mers, trimers, and dimers predominate. In both systems, Aβ formed oligomers intracellularly, which then appear to be secreted as oligomers. We then evaluated several factors that might impact oligomer formation. The level of oligomerization of Aβ1–40 was similar to that of Aβ1–42. Homodimers formed more readily than heterodimers. The level of oligomerization of murine Aβ1–42 was similar to that of human Aβ1–42. As expected, the familial AD-linked Arctic mutation (E22G) significantly enhanced oligomer formation. These data suggest that Gluc-tagged Aβ enables the analysis of Aβ oligomers.

Alzheimer disease (AD) 2 is common progressive neurodegenerative disorder causing dementia. In AD patient's brains, loss of neocortical synapses correlates with cognitive impairment (1). A␤ is the amyloid component of senile plaques in AD brains (2,3) and is derived from the A␤ precursor protein (APP) through sequential proteolytic cleavage by ␤-secretase and ␥-secretase (4). Based on several genetic and biochemical studies, the amyloid hypothesis is now widely accepted as the pathogenesis of AD (4). It has recently been suggested that picomolar concentrations of dimeric, or oligomeric, rather than monomeric or fibrillar forms of A␤, are the most neurotoxic, but characterization of these forms of soluble A␤ is technically difficult.
In previous in vitro A␤ fibrillization studies using synthetic A␤, a variety of soluble prefibrillar species, including A␤ protofibrils (5,6), A␤-derived diffusible ligands (7), and amylospheroids (8) were reported as neuronal toxic molecules. Recently, soluble A␤ oligomers, A␤*56 (9) and A␤ dimers (10) were isolated from APP transgenic mouse brain and AD brain and these A␤ oligomers caused synaptic dysfunction and memory impairment. Moreover A␤ oligomers were observed to associate with senile plaques and correlate with synaptic loss in AD brains (11). These data suggest that prefibrillar A␤ oligomers may be a causative species for synaptic loss and neuronal death in AD brains (12).
To understand how A␤ forms oligomers, it is important to monitor A␤ oligomers specifically and quantitatively. However isolation of A␤ oligomers from brains requires biochemical extraction and separation by size exclusion chromatography (SEC) (9,10). It is thus difficult to monitor the low concentrations of A␤ oligomers that appear to have biological activity, and analyze their dynamics. In this study, we developed a novel method for monitoring A␤ oligomers using a split-luciferase complementation assay. Using this method, we found that A␤ oligomers can be built within cells and are then secreted. We also found that A␤ preferred to form homogenous oligomers and familial AD-linked E22G mutation enhances oligomer formation of A␤. These results suggest that the assay may be useful to look for molecules that impact A␤ oligomerization.
Cell Culture and cDNA Transfection-HEK293 cells were cultured in Opti-MEM (Invitrogen) with 10% fetal bovine serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin at 37°in 5% CO 2 atmosphere. Transient and stable cell lines were generated by transfecting cDNA plasmids using Lipo-fectamine2000 (Invitrogen) as suggested by the manufacturer and selection in Opti-MEM with 100 g/ml of Zeocin (Invitrogen). For luciferase assays of the CM, we incubated HEK293 cells until 80% confluency, changed the media to Opti-MEM without fetal bovine serum, and further incubation for 3 days.
Luciferase Assay-Conditioned media (CM) from cells was collected and centrifuged at 1,200 rpm for 5 min to remove cell debris. Cells were lysed by cell lysis reagent (Promega). After adding 17 g/ml of coelenterazine (NanoLight technology) diluted by Opti-MEM into samples, luciferase activity was immediately measured using a Wallac 1420 (PerkinElmer). Statistical analysis was performed by one-way analysis of variance (ANOVA) using Prism 5 for Mac OSX (GraphPad). Following ANOVA, the Bonferroni post hoc test was applied.
Immunoblotting and ELISA-Equal amounts of CM or cell lysates were electrophoresed on 10 -20% Novex Tris-Glycine gels (Invitrogen) in Tris-Glycine SDS running buffer for SDS-PAGE (Invitrogen). Gels were transferred to PVDF membrane (PolyScreen, PerkinElmer), and blocked for 30 min at room temperature in 5% nonfat skim milk/TBST (Tris-buffer saline with 0.1% Tween20). Membranes were probed with 1 g/ml of anti-A␤ monoclonal antibody 6E10 (Signet), or anti-A␤42 specific monoclonal antibody 21F12 (Elan Pharmaceuticals, South San Francisco, CA) in TBST for 2 h at room temperature or for 12 h at 4°C. Following incubation with horseradish peroxidaseconjugated secondary antibody (Bio-Rad) for 1 h at room temperature, immunoreactive proteins were developed using an ECL kit (Western Lightning, PerkinElmer) and detected on Hyperfilm ECL (GE Healthcare). For ELISA-based A␤ assays, sandwich ELISA BNT77/BA27 or BNT77/BC05 (Waco Chemicals, Richmond, VA) were used as suggested by the manufacturer.
Size Exclusion Chromatography-750 l of CM were separated by size exclusion chromatography on a Superdex200 column (GE Healthcare) in 50 mM ammonium acetate pH 8.5 with an AKTA purifier 10 (GE Healthcare).

Detection of A␤ Oligomers
Using the Split-luciferase Complementation Assay-To monitor A␤ oligomers, we developed a novel method using a split-luciferase complementation assay. In this assay, the amino (N)-and carboxyl (C)-terminal fragments of Gluc, which emits luminescence 200-fold higher compared with firefly or Renilla luciferase (15), are fused separately to the N terminus of A␤1-42. We constructed the N-terminal fragment of Gluc-tagged A␤ (luci-A␤) or C-terminal fragment of Gluc-tagged A␤ (ferase-A␤). To enable the secretion of the split Gluc-tagged A␤ from cells, we inserted a secretory signal sequence of Immunoglobulin kappa upstream of the sequence encoding the split Gluc tags. Luci-A␤ and ferase-A␤ do not support luminescence by themselves. We hypothesized that, if luci-A␤ and ferase-A␤ interact together as an A␤ dimer/oligomer, the N-and C-terminal fragments of Gluc would reconstitute a functional molecule and exhibit luminescence (Fig. 1A). This technique is also known as a bimolecular luminescence complementation assay (BiLC) (16). We transiently transfected mock plasmid, full-length Gluc-A␤ (luciferase-A␤), luci-A␤, ferase-A␤, or both luci-A␤ and ferase-A␤ (luci-A␤/ferase-A␤) into HEK293 cells, incubated 48 h and collected the CM. Upon immunoblotting analysis of the CM of transfectants, luci-A␤ exhibited 20 and 22 kDa doublet bands, and ferase-A␤ exhibited 14 and 16 kDa doublet bands (Fig. 1B). In addition, greater than 50 kDa smear bands were observed in CM from luci-A␤/ ferase-A␤ transfected (double expression) HEK293 cells (Fig.  1B). We measured the luminescence and found that CM from double expression cells showed strong luminescence. On the other hand, the CM from luci-A␤ or ferase-A␤ separately expressing HEK293 cells did not show any luminescence (Fig.  1C). As a control, the CM from luciferase-A␤ overexpressing HEK293 exhibited 30 and 32 kDa doublet bands (Fig. 1B) and, as expected, showed substantially higher luminescence compared with double expression HEK293 cells (Fig. 1C). To rule out the possibility that overexpression of both N-terminal and C-terminal fragments of Gluc reconstitute the activity by themselves without A␤ sequence, we transfected both luci and ferase into HEK293 cells. However the CM form, both luci-and ferase-overexpressing HEK293 cells did not show strong luminescence (Fig. 1C). These data suggested that the split Gluctagged A␤ formed dimers/oligomers and reconstituted active luciferase in the CM.
To examine the size distribution of oligomers formed by Gluc-tagged A␤, we subjected CM from HEK293 cells stably expressing luci-A␤/ferase-A␤ to SEC. We applied CM onto Superdex200 SEC column, collected fractions, and measured the luminescence of each fraction. We found that the greater than 300 kDa fractions strongly exhibited luminescence and that less than 40 kDa fractions also weakly exhibited luminescence, suggesting that the majority of the split Gluc-tagged A␤ formed oligomers (estimated as 36ϳ24-mer, assuming that the complex is made of luci-A␤ and ferase-A␤ and does not reflect luciferase activity associated with complexes of amyloid oligomers with amyloid-binding proteins) and the minority of split Gluc-tagged A␤ formed dimers (Fig. 1D). Immunoblot analysis also detected both luci-A␤ and ferase-A␤ most prominently in the 8 -10 ml fractions (Fig. 1E). Interestingly, denaturing Western blot preparations revealed stable dimers/trimers even in the high molecular weight fraction.
Next, we examined the correlation between the direction of split luciferase tags and luciferase activity. We fused the N-terminal fragment of Gluc (A␤-luci) or the C-terminal fragment of Gluc (A␤-ferase) to the C terminus of A␤1-42 ( Fig. 2A). We transiently transfected A␤-luci, A␤-ferase, or A␤-luci/A␤-fe-rase into HEK293 cells. On immunoblotting analysis of the CM, A␤-luci was detected as 16 and 18 kDa doublet band proteins, and A␤-ferase was detected as 10 and 12 kDa doublet band proteins (Fig. 2B). We found that the CM from A␤-luci/A␤ferase expressing HEK293 cells showed luminescence and that the CM from A␤-luci or A␤-ferase separately expressing HEK293 cells did not show any luminescence (Fig. 2C). To compare luciferase activities between the direction of split-luciferase tag, we transfected luci-A␤/ferase-A␤ (N-N), A␤-luci/ A␤-ferase (C-C), luci-A␤/A␤-ferase (N-C), and A␤-luci/ferase-A␤ (C-N) into HEK293 cells. All N-N, C-C, N-C, and C-N conformations supported luminescence, although the luci-A␤/ ferase-A␤ (N-N) and luci-A␤/A␤-ferase (N-C) appeared to produce a somewhat stronger signal. The luci-A␤/ferase-A␤ double pair was used for further studies.
Split Gluc-tagged A␤ Oligomers Are Built Intracellularly-To understand where split Gluc-tagged A␤ forms oligomers, we measured luminescence within the cells overexpressing luci-A␤/ferase-A␤. We found the cell lysate from doubly expressing HEK293 cells also showed luminescence (Fig. 3A). On the other hand, cell lysate from luci-A␤ or ferase-A␤ separately expressing HEK293 cells did not show any luminescence. On immunoblot analysis of cell lysates, luci-A␤ was detected as a 22 kDa protein and ferase-A␤ expressed as a 16 kDa protein (Fig. 3B). To examine the possibility that A␤ oligomer formation occurs extracellularly, we co-cultured luci-A␤ stably expressing HEK293 cells and ferase-A␤ stably expressing HEK293 cell. However we did not detect luminescence from the CM of co-cultured cells (Fig. 3C, co-culture). We also made a mixture between the CM from luci-A␤ stably expressing HEK293 cells and ferase-A␤ stably expressing HEK293 cell. However we also did not detect luminescence from the mixture of CM (Fig. 3C, luci-A␤ CM ϩ ferase-A␤ CM). These results suggested Gluc-tagged A␤ oligomers are built within the cells and secreted. Once secreted, we suggest that they are relatively stable, and therefore secreted non-luminant luci-A␤ oligomers mixed with secreted ferase-A␤ oligomers do not readily exchange to form luminescent pairs.
To examine whether A␤ oligomer formation occurs intracellulary, we subjected cell lysate from HEK293 cells stably expressing luci-A␤/ferase-A␤ to SEC. We applied cell lysate on to a Superdex200 SEC column, collected fractions and measured the luminescence of each fraction. We found that, similar to CM (Fig. 1D), the greater than 300 kDa fractions strongly exhibited luciferase activity, suggesting that the split Gluctagged A␤ formed oligomers (estimated as 36ϳ24-mer) intracellulary and is then secreted from cells (Fig. 3D). Immunoblot analysis also detected both luci-A␤ and ferase-A␤ most prominently in the 8 -9 ml fractions (Fig. 3E).
We previously published that A␤ oligomers are rich in the CM from 14-day-cultured primary culture neurons of Tg2576 APP transgenic mouse (19). To determine whether native A␤ also forms oligomers intracellulary, we applied cell lysate of primary culture neurons of Tg2576 mouse to SEC, collected fractions and measured the A␤ concentration using a human A␤ specific ELISA. We found A␤ eluted as 36ϳ24-mer, 9ϳ8mer, 3-mer, or 1-mer size (Fig. 3F), suggesting that native A␤ also forms oligomers intraneuronally. With regard to the largest fraction, we cannot exclude the possibility that Gluc-tagged A␤ or native A␤ interacts with unidentified factors and is thus eluted into larger molecular size fractions. A difference is apparent between Fig. 3D and Fig. 3F in the exact distributions; we postulate that in the HEK293 cells, the concentration of Gluc-tagged A␤ is very high, and they may form HMW oligomers very rapidly. On the other hand, in the Tg2576 neurons, the concentration of A␤ is low, and they may form oligomers more slowly and also show some intermediate size oligomers.
Murine A␤ Forms Oligomers Similar to Human A␤-Between human A␤ and murine A␤, there are 3-amino acid differences (R5G, Y10F, and H13R, Fig. 5A). We recently found that human A␤ secreted from 14-day-cultured primary neurons of Tg2576 mice formed 10 -30 kDa oligomers and murine A␤ secreted from 14-day-cultured primary culture neurons of wild-type mouse also formed the same sized oligomers (19). However the levels of murine A␤ oligomers in the CM of wildtype mouse neurons were substantially lower than those of human A␤ in the CM of Tg2576 mouse neurons. To determine whether murine A␤ forms oligomers like human A␤, we constructed split Gluc-tagged murine A␤1-42 (R5G, Y10F, and H13R, Fig. 5A). We also constructed a human/murine hybrid (Hu/Ms) A␤1-42 with split Gluc tag (Y10F and H13R, Fig. 5A). We transiently transfected these plasmids into HEK293 cells and collected their CMs. Because the epitopes of anti-human A␤ mAb 6E10 are residues 3-8 of human A␤, murine A␤ is negative for 6E10 whereas Hu/Ms A␤ is positive for mAb 6E10. On the other hand, mAb 21F12 is A␤42 C-terminal end specific antibody and recognizes human, murine, and Hu/Ms A␤42. Similar to split Gluc-tagged human A␤, we found that murine luci-A␤ and Hu/Ms luci-A␤ exhibited 20 and 22 doublet bands and that murine ferase-A␤ and Hu/Ms ferase-A␤ exhibited 14 and 16 doublet bands in the CM (Fig. 5B, right panel). As expected, mAb 21F12 revealed split Gluc-tagged human, murine and Hu/Ms A␤ and mAb 6E10 revealed split Gluc-tagged human and Hu/Ms A␤ (Fig. 5B). The CM from both Hu/Ms luci-A␤ and Hu/Ms ferase-A␤, or from both murine luci-A␤ and murine ferase-A␤ stably expressing HEK293 cells showed strong luminescence. On the other hand, as expected, the CM from Hu/Ms luci-A␤, Hu/Ms ferase-A␤, murine luci-A␤, or murine ferase-A␤ separately expressing HEK293 cells did not show any luminescence (Fig. 5C). We next compared the luminescence of reconstituted Gluc from human, murine, and Hu/Ms A␤ oligomers. We collected the CM from both human luci-A␤ and ferase-A␤, from both murine luci-A␤ and murine ferase-A␤, or from both Hu/Ms luci-A␤ and Hu/Ms ferase-A␤ transiently expressing HEK293 cells, and measured the luminescence, standardized by the expression levels of A␤ (Fig. 5D). Compared with human A␤, murine A␤ showed 97.9% luminescence and Hu/Ms A␤ showed 107.2% luminescence. Among them, there were no significant differences, suggesting that murine A␤ forms oligomers similar to human A␤.
Familial AD-linked Arctic Mutation Enhanced Split Gluctagged A␤ Oligomer Formation-The E22G Arctic mutation is one of the familial AD-linked mutations within the A␤ region in the APP gene and it has been reported to enhance protofibril formation (20). To determine whether the E22G mutation affects A␤ oligomerization, we constructed E22G mutant split Gluc-tagged A␤1-42 plasmids (Fig. 6A, luci-A␤E22G and ferase-A␤E22G) with which we generated luci-A␤E22G, ferase-A␤E22G, and both luci-A␤E22G and ferase-A␤E22G transiently expressing HEK293 cells. Luci-A␤E22G exhibited 20 and 22 kDa doublet bands, and ferase-A␤E22G exhibited 14 and 16 kDa doublet bands in CM (Fig. 6B). The CM from luci-A␤E22G/ferase-A␤E22G doubly expressing HEK293 cells showed strong luminescence. On the other hand, as expected, the CM from luci-A␤E22G or ferase-A␤E22G separately expressing HEK293 cells did not show any luminescence (Fig.  6C). We next compared luminescence of the CM from luci-A␤/ ferase-A␤ expressing cells with the CM from luci-A␤E22G/ ferase-A␤E22G expressing cells after standardization by the expression levels of A␤. We found that E22G mutant Gluctagged A␤ oligomers exhibited 178.0% higher luminescence compared with wild-type Gluc-tagged A␤ oligomers, suggesting that E22G may enhance not only protofibril formation but also oligomer formation (Fig. 6D).
Split Gluc-tagged A␤ Reduced Oligomer Formation by Incubation with Synthetic A␤-A␤ oligomer is one of the intermediate species in the process of amyloid fibril formation. We next examined the relationship between oligomer formation and fibril formation. To enhance fibril formation, we incubated split Gluc-tagged A␤ oligomers with 50 g/ml of fresh synthetic A␤1-40 or A␤1-42 at 37°C for 24 h. Synthetic A␤1-42 fibrilizes faster than synthetic A␤1-40 (17). We found that split Gluc-tagged A␤ oligomer luminescence was significantly reduced to about 24.2% by incubation with synthetic A␤ (Fig.  7A). Synthetic A␤1-40 and A␤1-42 significantly decreased the luminescence of split Gluc-tagged A␤ oligomers about 40.2 and 56.8%, respectively (Fig. 7A). We saw amyloid fibrils in the split Gluc-tagged A␤ oligomers that had been incubated with synthetic A␤, but we did not see any amyloid fibrils in the equivalent preparation of split Gluc-tagged A␤ oligomers alone by electron microscopy (data not shown), arguing that one effect of the synthetic A␤ might be to enhance fibrillization and precipitate split luciferase A␤. On the other hand, we found that whole Gluc-tagged A␤ oligomers did not show any significant reduction of the luminescence after incubation with synthetic A␤ and that, in fact, synthetic A␤1-40 and A␤1-42 significantly increased luminescence of whole Gluc-tagged A␤ (Fig.  7B). Although it is not clear why synthetic A␤1-40 or A␤1-42 increases the luminescence of whole Gluc-tagged A␤, this result shows that A␤ fibrils do not quench the luminescence of Gluc. We therefore favor the interpretation that the addition of excess "cold" A␤ competes for dimeric split luciferase molecules, leading to a decrease in the likelihood that split luciferase dimers form functionally active pairs, consistent with the possibility that incubation with synthetic A␤ reduces the specific oligomer conformation needed to support luminescence in the BiLC format when incorporated with A␤ fibrils.

DISCUSSION
In this study we generated a novel method for monitoring A␤ oligomers using a bimolecular luminescence complementation assay. Gluc-tagged A␤ oligomers, secreted from both luci-A␤ and ferase-A␤ stably expressing HEK293 cells, formed highmolecular-weight A␤ oligomers. Although Gluc-tagged A␤ is an artificial protein, it appears that it forms oligomers similar to natural A␤ oligomers. We demonstrate that the bimolecular complementation assay oligomers are biochemically similar to native A␤ oligomers produced by Tg2576 neurons in terms of their behavior on SEC columns. They also behave as anticipated based on previous literature using native A␤ in a variety of experimental assays, with expected changes in complementation efficiency produced by manipulating hetero versus homodimer formation, the effects of the E22G mutation, and the detection of both intracellular and secreted forms of oligomeric A␤. Specifically, using the new Gluc-tagged A␤ oligomer assay, we found that 1) A␤ formed oligomers within the HEK293 cells which were secreted, 2) split Gluc-tagged A␤ tended to form A␤40 or A␤42 homogeneous oligomers, compared with A␤40 and A␤42 heterogeneous oligomers, 3) murine A␤ formed oligomers similar to human A␤, 4) the E22G Arctic mutation enhanced oligomer formation of split Gluctagged A␤. Importantly, the high sensitivity of the assay allowed us to readily detect and characterize the presence of oligomers within the intracellular compartment, strongly supporting the idea that oligomers can form intracellularly, and provide an ultrasensitive tool for their further study.
To monitor A␤ oligomers specifically, several methods, including SEC (10, 21), A␤ oligomer specific antibodies (22,23,24), chemical crosslinking method (25), mass spectrometry (26), or single-antibody sandwich enzyme-linked immunosorbent assay (27,28), have been reported. Compared with these methods, BiLC method using Gluc-tagged A␤ is a highly-sensitivity, quantitative, and simple method for monitoring A␤ oligomers. The assay has the further advantage that it does not require concentration of the sample or other manipulations that may impact dimer/oligomer formation, and it is ideal for high-throughput type assays.
It is possible that the addition of tags (even the relatively small tags involved in the Gaussia luciferase system) could impact A␤ characteristics. We therefore examined a variety of systems in which the biological characterization of oligomer or protofibril formation has been previously evaluated, and find quite analogous results with the bimolecular complementation assay. We found that Gluc-tagged A␤ formed oligomers within cells. Although it is likely that the secretion characteristics of the Gluc-tagged A␤ differ from native A␤, the observation that A␤ can form oligomers intracellularly is in accord with several studies reporting detection of intraneuronal A␤ (29,30). In FAD mutant APP overexpressing Chinese hamster ovary cells (clone 7PA2), SDS-stable A␤ oligomers were detected intracellularly (31). Moreover, in APP transgenic mouse brains, A␤ oligomer specific antibody detected intraneuronal A␤ oligomers (23). In this study, we also found that A␤ may formed oligomers in Tg2576 APP transgenic mouse neurons (Fig. 3F).
We found that split Gluc-tagged A␤ tended to form A␤40 or A␤42 homogeneous oligomers, compared with A␤40 and A␤42 heterogeneous oligomers (Fig. 4). Shankar et al. (10) found similar results, with the majority of dimers appearing to be homodimers, when they examined A␤ dimers isolated from AD brains using A␤40 or A␤42 C-terminal end specific antibodies. Although the level of fibrillization of A␤42 is much higher than A␤40 (17), A␤40 is also known to form dimers, trimers, and tetramers (25), or amylospheroid (8) in vitro. Why does A␤ prefer to form homo-oligomers rather than hetero-oligomers? We cannot rule out the possibility that this result is secondary to the structure of the split Gluc constructs, but Shanker et al. (10) report, using endogenous A␤, that heterodimers are rare. Recently A␤ oligomers made by a mixture of synthetic A␤40 and A␤42 (ratio 7:3) was found to be more stable and neurotoxic compared with A␤40 or A␤42 solo oligomer (32). A␤40 and A␤42 is also reported to coassemble into amyloid fibrils (33,34), suggesting that A␤40 or A␤42 homo oligomer may be mixed and may form further aggregated structures.
We also found that murine A␤ formed oligomers similar to human A␤ (Fig. 5). We detected 10 -30 kDa murine A␤ oligomers in the CM from 14-day-cultured primary culture neurons of wild-type mice (19). Recently endogenous murine A␤ was reported to have a crucial role in synaptic activity (35), and endogenous murine A␤ oligomers suggested a role regulating synaptic activity. Moreover, by the prolonged inhibition of A␤-degrading enzyme neprilysin in the brain of wild-type mice, it has been reported that murine A␤ is deposited in the brain (36), suggesting that murine A␤ may undergo a process of oligomerization and fibrillization similar to human A␤.
The familial AD-linked E22G Arctic mutation has been reported to enhance protofibril formation (20) and oligomer formation (37) in vitro, and amyloid deposition in vivo (38). In this study, using split Gluc-tagged A␤ oligomers, we confirmed that E22G mutation enhanced the oligomer formation in the CM of HEK293 cells. Because A␤ N-terminal portion mutations (R5G, Y10F, and H13R in murine A␤) did not affect A␤ oligomer formation, we suggest that the mid portion of A␤ including the Glu-22 residue may be important for A␤ oligomerization.
In summary, we developed a novel technique for specifically monitoring and quantifying A␤ oligomers. The split luciferase complementation assay may be useful for the general question of oligomeric protein behavior, and we have recently demonstrated that ␣-synuclein oligomers can be similarly studied using this technique (39). A␤ oligomerization is a key process in AD pathogenesis, and monitoring A␤ oligomers is necessary not only for elucidation of the molecular mechanism of AD pathogenesis but also for development and assessment of disease modifying therapies for AD. Using this technique we will further elucidate A␤ oligomer functions and its kinetics.