Improvement of a low pH antigen-antibody dissociation procedure for ELISA measurement of circulating anti-Abeta antibodies

Abstract

Background: Prior work from our group found that acid dissociation (pH 2.5 incubation) of serum from APP transgenic mice vaccinated against Abeta increased the apparent anti-Abeta titers, suggesting antibody masking by antigen in the ELISA assay. Subsequently, we found that pH 2.5 incubation of serum from unvaccinated non-transgenic mice showed antibody binding to Abeta1-42, but no increase when other proteins, including shorter Abeta peptides, coated the ELISA plate. To investigate further the effects of low pH incubation on apparent anti-Abeta1-42 signals, we examined normal sera from nonTg unvaccinated mice, nonTg mice vaccinated with Abeta peptide (to produce authentic anti-Abeta antibodies) or a monoclonal antibody against Abeta (6E10) using competitive-inhibition ELISA and Abeta epitope mapping assays. In addition, we examined use of a less stringent low pH procedure at pH 3.5, to ascertain if it had the same effects as the pH 2.5 procedure.

Results: We believe there are three distinct effects of pH 2.5 incubation.; A) an artifactual increase in binding to full length Abeta by mouse immunoglobulin which has low affinity for Abeta, B) an inactivation of anti-Abeta antibodies that is time dependent and C) unmasking of high affinity anti-Abeta antibodies when high levels of circulating Abeta is present in APP transgenic mice. All three reactions can interact to produce the final ELISA signal. Incubation of sera from unvaccinated nonTg mice at pH 2.5 enhanced ELISA signals by process A. Conversely, pH 2.5 incubation of sera from vaccinated nonTg mice with caused a time dependent reduction of antibody signal by process B (overcoming the increase caused by A). The artifactual anti-Abeta ELISA signal enhanced by pH 2.5 incubation of normal mouse sera could not be effectively competed by low to moderate concentrations of Abeta, nor bind to shorter Abeta peptides in a manner similar to authentic anti-Abeta antibodies. Incubation of mouse sera at pH 3.5 caused neither an apparent increase in anti-Abeta ELISA signal, nor an inactivation of the ELISA signals resulting from either vaccination or monoclonal antibodies. However, incubation at pH 3.5 was able to completely reverse the reduction in ELISA signal caused by Abeta complexing with antibodies in sera from vaccinated mice or monoclonal anti-Abeta antibodies.

Conclusion: Incubation at pH 3.5 is sufficient to dissociate Abeta bound to anti-Abeta antibodies without producing artifactual increases in the signal, or inactivating authentic antibody binding. Thus, use of pH 3.5 is a considerable improvement over pH 2.5 incubation for unmasking anti-Abeta antibodies in ELISA assays to measure antibodies in APP transgenic mouse sera.

Figure 1

Incubation at pH 2.5 induced increase in anti-Aβ ELISA signal is present not only in sera from vaccinated mice, but also untreated mice. Sera were collected from APP transgenic mice vaccinated (Vax +) with Aβ1–42 (panel A) and nonTg unvaccinated (Vax -) mice (panel B; normal serum), preincubated with dissociation buffer at either pH 7 (open bars) or pH 2.5 (solid bars) at room temperature for 20 minutes and centrifuged through 10,000 MW filters, brought to neutral pH and reconstituted to the original serum dilution volume. The sera were then incubated on ELISA plates coated with one of the following at 5 μg/ml; human Aβ1–42, phosphate buffered saline (PBS; a "no protein" control), bovine serum albumin (BSA), recombinant alpha-synuclein (a-SYN) or Aβ peptide amino acids 11–20. ELISA assays were completed as described in methods and the optical density (OD) at 450 nm used to estimate the amount of antibody binding to the ELISA plate. Note different Y-axis scales in panels A and B. Results are presented as mean ± SEM. ** P < 0.001 compared to pH 7.0.

Figure 2

Incubation at pH 2.5 induced artifactual ELISA signal appears to be IgG associated. Panel 2A. Normal sera were collected from nonTg unvaccinated mice, diluted 1:400 and preincubated at either pH 7 (open bars) or pH 2.5 (solid bars) as described in methods. After centrifugation, neutralization and reconstitution of sera volume, the sera were incubated on the ELISA plates coated with human Aβ1–42 for 60 minutes. After washing the plate, anti-mouse gamma chain specific (left) or anti-mouse whole IgG (right) were pipetted into the different wells of plate. Panel 2B. Normal sera from nonTg unvaccinated mice were incubated at pH 7.0 (open bars) or pH 2.5 (solid bars) and then IgG was purified with protein A/G (Pierce) as described in methods. Protein was measured with BCA protein assay Kit (Pierce), and dilutions of the purified IgG were incubated as the indicated concentrations on ELISA plates coated with Aβ1–42. ELISA assays were completed as described in methods. ** P < 0.001 compared to pH 7.0.

Figure 3

Duration of pH 2.5 exposure differentially affects ELISA signals in sera from unvaccinated nonTg mice compared to vaccinated nonTg mice. Sera were collected from unvaccinated mice (3A) and mice vaccinated with human Aβ1–42 peptide (3B), diluted 1:400. Aliquots were incubated with dissociation buffer at pH 7 (open bar, 120 min) or pH 2.5 (solid bars) for 5, 20, 30, 90 and 120 minutes. The ELISA assays were completed as described in methods and the optical density at 450 nm used to estimate the amount of antibody binding to the ELISA plate. Results are presented as mean ± SEM. * P < 0.05; ** P < 0.001 compared to pH 7.0.

Figure 4

Incubation at pH 2.5 induced artifactual ELISA signal only interacts with full length Aβ; Authentic antibodies bind both full length Aβ and N terminal fragments. Epitope mapping of normal sera from unvaccinated mice (panel A;1:400 dilution), mice vaccinated with Aβ1–42 peptide (panel B; 1:400 dilution) or mouse anti-human Aβ1–16 monoclonal antibody (6E10; 4C;50 ng/ml) was performed. Full length human Aβ and human Aβ fragments (singly and in combination as indicated on graph) were used to coat the plate and identify the domain that interacted with the IgG fractions following pH 7 preincubation (full length Aβ only) or incubation at pH 2.5 for 20 minutes at RT. After centrifugation and neutralization the samples were then incubated on ELISA plates coated as indicated. Values are mean ± SEM. ** P < 0.001 compared to pH 2.5 plate coated with Aβ1–42 peptide.

Figure 5

Incubation at pH 2.5 induced artifactual ELISA signal has a lower affinity for Aβ than authentic anti-Aβ antibodies in competition assays. Sera from unvaccinated mice (panel A) or mice vaccinated with Aβ 1–42 peptide (panel B), were diluted 1:1000 and preincubated at either pH 7 or pH 2.5. After centrifugation, neutralization and reconstitution of sera volume, the sera were then incubated with human Aβ1–42 peptide, at increasing concentrations at 37°C for 60 minutes. Because the Aβ was initially dissolved in DMSO, some samples were also incubated with 2% DMSO as control. The ELISA assay was then performed as described in methods with no further separation of Aβ and antibody fractions. ** P < 0.001 compared to pH 2.5 without Aβ preincubation.

Figure 6

Incubation of sera at pH 3.5 does not induce artifactual ELISA signals nor inactivate authentic anti-Aβ antibody binding in ELISA assay. Sera collected from unvaccinated mice (6A) and mice vaccinated with Aβ 1–42 peptide (6B) were incubated with dissociation buffer at pH 7.0 (open bars), 3.5 (grid bars) or 2.5 (solid bars) for 20 minutes, separated by centrifugation through 10,000 MW filters neutralized and reconstituted to original volume. They were then incubated in a standard ELISA assay using full length Aβ to coat the plate. * P < 0.05 compared to pH 7.0; ** P < 0.001 compared to pH 7.0.

Figure 7

Dissociation of Aβ-IgG immune complexes occurs at both pH 2.5 and pH 3.5. Mouse anti-human Aβ1–16 monoclonal antibody (6E10) at 10 μg/ml (7A) and serum from mice vaccinated with Aβ diluted 1:100 were preincubated with Aβ1–42 peptide at 1 μg/ml to form antigen-antibody complexes (solid bars) or preincubated with PBS as a control (open bars) at 37°C for 60 minutes. Aliquots from each sample were then incubated with a 10-fold larger volume (1:1000 final serum dilution in panel B) of dissociation buffer at pH 7, pH 2.5 and pH 3.5, respectively, for 20 minutes, separated by centrifugation through 10,000 MW filters, neutralized and reconstituted to original volume for standard ELISA assay with full length human Aβ coating the plate.