Measurement of anti-beta amyloid antibodies in human blood

Abstract

The human IgG repertoire contains endogenous antibodies against beta amyloid (Aβ) that may be relevant to the pathogenesis and treatment of Alzheimer's disease. There have been widely disparate estimates of the levels of these antibodies in human plasma. We identify factors that have contributed to these disparities and describe improved methods for measuring anti-Aβ antibodies in blood. These methods include isolating immunoglobulin by thiophilic chromatography and using chaotropic salts to dislodge weakly bound antibodies without significantly reducing the binding of specific anti-Aβ antibodies. Using these methods, we show that human blood contains polyvalent IgG antibodies that bind to Aβ with relatively low avidity and specificity, as well as IgG antibodies that bind to linear and conformational epitopes on amyloid monomers and aggregates with moderate to high avidity.

Introduction

The human immunoglobulin G (IgG) repertoire contains endogenous antibodies against the Aβ peptide that arise in the absence of vaccination or passive immunization. Endogenous anti-Aβ antibody activity has been detected in the blood of normal adults of various ages and patients with AD (Weksler et al., 2002, Hyman et al., 2001, Mruthinti et al., 2004, Nath et al., 2003, Sohn et al., 2009). The human plasma-derived antibody preparation known as intravenous immunoglobulin (IVIG) has been reported to contain endogenous anti-amyloid antibodies (Dodel et al., 2004) and is under study as a potential treatment for AD (Dodel et al., 2004, Relkin et al., 2009).

Published studies provide widely disparate estimates of the relative titers of anti-Aβ antibodies in AD patients versus aged normal controls. Initial studies of intact plasma specimens by standard ELISA methods ascribed lower titers of endogenous antibodies against Aβ monomers to AD patients than age-matched non-demented controls (Weksler et al., 2002). Subsequent studies reported equal or increased titers of circulating anti-Aβ monomer antibodies in AD patients (Mruthinti et al., 2004) based on ELISAs performed on plasma immunoglobulin eluted from Protein G columns at low pH (Akerström and Björck, 1986). The higher titers obtained by this method were hypothesized to reflect the presence of anti-amyloid antibodies bound to Aβ antigen that were undetected in assays of whole plasma but measurable when they were freed from antigen by acidification in the course of protein G purification. However, as first shown in murine models, exposure of polyclonal IgG to low pH results in partial denaturation of antibodies and generates large artifactual increases in apparent anti-Aβ antibody titers (Li et al., 2007). Similar phenomena have been reported with other IgG antibodies in human plasma (Bouvet et al., 2001, Djoumerska et al., 2005). Assays employing IgG isolated from human plasma by protein G chromatography and other methods that require acid elution may therefore elevate anti-amyloid activity artifactually.

The measurement of endogenous anti-Aβ antibody activity is further complicated by the existence of multiple classes of human antibodies that recognize linear as well as conformational neo-epitopes on aggregated forms of Aβ. Endogenous human antibodies include molecules that bind conformational epitopes on Aβ fibrils (O'Nuallain et al., 2006), Aβ oligomers (Relkin et al., 2008, O'Nuallain et al., 2008), chemically cross-linked oligomers (Moir et al., 2005) and Aβ monomers (Du et al., 2003, Istrin et al., 2006, Baumketner et al., 2006). Other types of amyloid binding antibodies and even catalytic antibodies against Aβ have been reported (Taguchi et al., 2008). Conventional ELISAs may under- or over-estimate total anti-amyloid activity if they fail to take into account the diversity of antibody types present as well as phenomena such as cross-reactivity and polyvalency.

Methods to increase sensitivity and signal-to-noise ratio of anti-Aβ antibody assays have been described, including a radio-immunoprecipitation assay (Brettschneider et al., 2005) and a novel peptide microarray method that was recently used to characterize endogenous human anti-Aβ antibodies in plasma as a function of age and AD (Britshgi et al., 2009). Although the latter approach yielded improved signal-to-noise ratios compared to ELISA, its use of whole plasma may be problematic owing to the effects of other plasma proteins and macromolecules on anti-Aβ antibody measurements.

These and other phenomena such as blank ELISA plate binding (Klaver et al., 2010) that may confound measurements of endogenous anti-Aβ antibody levels have important implications for studies of treatment of AD by IVIG. IVIG was first reported to contain anti-Aβ monomer antibodies (Dodel et al., 2004) at relatively low levels. The reported paucity of anti-Aβ antibodies compared to those generated by active vaccination has brought into question the potential for IVIG to exert anti-amyloid effects in AD patients. Nevertheless, human clinical trials of IVIG in AD patients resulted in positive clinical outcomes as well as altered levels of Aβ in plasma and cerebrospinal fluid (Dodel et al., 2004, Relkin et al., 2009). Although endogenous anti-amyloid antibody activity provided the original rationale for testing IVIG in AD, it is not known with certainty whether this is the major or exclusive mechanisms of action of IVIG in this disease. The question of whether true endogenous anti-amyloid antibodies are present in human plasma at levels sufficient to exert biologically meaningful effects has yet to be answered. Improved methods of assaying anti-amyloid activity in human plasma are needed to better understand the possible role of endogenous antibodies in the pathogenesis and treatment of AD and other amyloid-associated disorders.

In this report, we identify several factors that complicate measurement of endogenous anti-amyloid activity and describe improved methods for preparing and assaying endogenous anti-amyloid antibodies in human plasma and intravenous immunoglobulin preparations. We use these techniques to show that human plasma contains endogenous polyvalent antibodies that bind to Aβ with relatively low avidity and specificity, as well as anti-amyloid antibodies that have moderate to high avidity for Aβ and its aggregates.