Soluble amorphous aggregates were prepared as previously described [21] . Briefly, 0.11 mM Aβ₄₂ (Peptide Institute Inc., Osaka, Japan) was dissolved in 1,1,1,3,3,3-hexafluo- ro-2-propanol, and incubated at 4˚C for 16 h followed by 37˚C for 3 h, after which the solution was lyophilized. The steps including dissolution and lyophilization were repeated twice, and lyophilized Aβ₄₂ was dissolved in deionized water. The concentration of Aβ₄₂ in the solution was determined by its absorbance at 280 nm using the molar extinction coefficient of 1450 M⁻¹cm⁻¹ of Aβ₄₂, which has one tyrosine residue and no tryptophan residues.

Direct ELISA was performed as previously described [12] . The soluble aggregates (100 μl) were used as solid-phase antigens, and were stained on 96-well plates (F96 MAXISORP NUNC-IMMUNO PLATE; Thermo Fisher Scientific Inc., Rochester, NY, USA). The monoclonal antibody 37-11 or 77-3, previously conjugated to HRP using a Peroxidase Labeling Kit-SH (Dojindo Molecular Technologies Inc., Kumamoto, Japan), was used as a primary antibody. The substrate SIGMAFAST OPD (Sigma-Aldrich, St. Louis, MO) was used for color development according to the manufacturer’s instructions. For sandwich ELISA, the antibody 77-3 was used as the primary antibody, and HRP-conjugated antibody 37-11 was used as the secondary antibody as described previously [13] .

The sizes and shapes of the aggregates were observed by atomic force microscopy (AFM, JSPM-5200, JEOL Ltd, Tokyo, Japan) as previously described [12] . The aggregates (10 μl) were dropped on fresh mica and dried by desiccation, and measured using the altering current mode at room temperature, with a frequency of approximately 190 kHz, typical of resonances with a 4.5 N/m spring constant.

Metal ions are known to bind to Aβ and induce the formation of the amorphous aggregates. Metal ions were hypothesized to change the conformation of the epitopes, recognized by antibodies 37-11 and 77-3, and used for the highly sensitive detection of soluble aggregates. Various metal ions (50 μM MnCl₂, FeCl₂, NiCl₂, CuCl₂, or ZnCl₂) were added to the soluble aggregates and signal intensities were observed after sandwich ELISA using the two antibodies. The addition of CuSO₄ increased the signal intensity approximately 9.1-fold; however, the effect of other metal ions had little effect on signal intensities (data not shown). This result is consistent with the previous reports suggesting that binding of Cu²⁺ to Aβ is stronger than that of other metal ions. Thus, the increased signal intensity was considered to be caused by Cu²⁺ binding to Aβ The effect of different concentrations of CuSO₄ on the signal intensity of soluble aggregates in the sandwich ELISA is shown in Figure 1. The signal increased in the presence of 10 μM CuSO₄, and reached a maximum in the presence of CuSO₄ ranging in concentration from 25 to 75 μM. These concentrations of Cu²⁺ were much higher than those of Aβ (130 nM in 100 μl), used for the ELISA, and may have induced the formation of amorphous aggregates. Some studies have shown that Cu²⁺ at sub-equimolar metal ion/Aβ₄₂ ratios induces the formation of fibrils, whereas supra-molar ratios induce the formation of non-fibrillar forms [15] . In addition, the Cu²⁺ concentrations in this report were similar to those found in the serum (7 - 41 μM) [22] . Figure 2 shows the effects of CuSO₄ and CuCl₂ on the signal intensities of the sandwich ELISA. The effect of CuCl₂ was greater than that of CuSO₄ suggesting that the counter anion also affects signal intensity.

It is possible that the epitopes, recognized by antibodies 37-11 and 77-3, are formed on the surface of soluble aggregates in the presence of Cu²⁺. To evaluate the formation of new epitopes, direct ELISA was performed. Figure 3 shows the effect of 50 μM CuCl₂ on signal intensities. Unlike the results of the sandwich ELISA (Figure 2), Cu²⁺ only minimally enhanced the signal intensity. These results suggest that Cu²⁺ increases the intensity by gathering distinct aggregates with two distinct epitopes recognized by antibodies 37-11 and 77-3, rather than generating these epitopes. The AFM images (Figure 4), showing the formation of larger aggregates through the addition of Cu²⁺, were

consistent with this theory. Based on this hypothesis, the epitopes recognized by antibodies 37-11 and 77-3 might be separately located on the surface of distinct aggregates, and few aggregates have both epitopes on their surface.

This report shows that the addition of Cu²⁺ increases the reactivity of the sandwich ELISA for soluble aggregates. This finding could contribute to the development of a precise and highly sensitive ELISA for Aβ aggregates by enhancing the reactivity of the antibodies.