Heteromers of amyloid precursor protein in cerebrospinal fluid

Abstract

Background: Soluble fragments of the amyloid precursor protein (APP) generated by α- and β-secretases, sAPPα and sAPPβ, have been postulated as promising new cerebrospinal fluid (CSF) biomarkers for the clinical diagnosis of Alzheimer's disease (AD). However, the capacity of these soluble proteins to assemble has not been explored and could be relevant. Our aim is to characterize possible sAPP oligomers that could contribute to the quantification of sAPPα and sAPPβ in CSF by ELISA, as well as to characterize the possible presence of soluble full-length APP (sAPPf).

Results: We employed co-immunoprecipitation, native polyacrylamide gel electrophoresis and ultracentrifugation in sucrose density gradients to characterize sAPP oligomers in CSF. We have characterized the presence of sAPPf in CSF from NDC and AD subjects and demonstrated that all forms, including sAPPα and sAPPβ, are capable of assembling into heteromers, which differ from brain APP membrane-dimers. We measured sAPPf, sAPPα and sAPPβ by ELISA in CSF samples from AD (n = 13) and non-disease subjects (NDC, n = 13) before and after immunoprecipitation with antibodies against the C-terminal APP or against sAPPβ. We demonstrated that these sAPP heteromers participate in the quantification of sAPPα and sAPPβ by ELISA. Immunoprecipitation with a C-terminal antibody to remove sAPPf reduced by ~30% the determinations of sAPPα and sAPPβ by ELISA, whereas immunoprecipitation with an APPβ antibody reduced by ~80% the determination of sAPPf and sAPPα.

Conclusions: The presence of sAPPf and sAPP heteromers should be taken into consideration when exploring the levels of sAPPα and sAPPβ as potential CSF biomarkers.

Figure 1

Soluble full-length APP (sAPPf) is present in human CSF. (A) Schematic representation of full-length APP processing by α/γ-secretase (non-amyloidogenic pathway) and β/γ-secretase (amyloidogenic pathway) and the generation of extracellular large sAPPα, sAPPβ fragments, and the shorter Aβ and p3 peptides; as well as fragments containing the C-terminal APP (CTF and ICD) (not drawn in scale). The epitopes for the anti-APP antibodies used in this study are indicated. (B) Western blotting of human CSF samples from non-demented controls (NDC) subjects, resolved with the indicated anti-APP antibodies. (C, D) CSF aliquots (Total, T) were immunoprecipitated with either the anti-C-terminal APP antibody from Sigma (C, Sigma-Ct) or the pan-specific IBL-β antibody (D, IBL-β). Precipitated proteins (bound fraction, B) were immunoblotted with the antibodies as indicated. In the absence of capture-antibody (IPc), no bands were observed.

Figure 2

Characterization of APP complexes by native-PAGE and sucrose gradient ultracentrifugation. (A) APP complexes from the brain (frontal cortex from NDC subjects) and CSF samples (NDC subjects) were analyzed by blue native-PAGE. Incubation of blots with antibodies for the different APP epitopes confirmed the presence of APP dimers in brain extracts and CSF samples (~242 kDa), but also the existence of APP complexes with higher molecular weight. A CSF sample denatured by boiling at 95°C for 5 min under fully reducing conditions (Dn) was also analyzed by blue native-PAGE to warrant the migration of the monomeric sAPP band. (B) Brain extracts and CSF samples were also fractionated on 5-20% sucrose density gradients. The fractions (collected from the top of each tube) were immunoblotted for APP with the 6E10 antibody, and additionally with IBL-β and Sigma-Ct antibodies for CSF samples. Enzymes of known sedimentation coefficient, β-galactosidase (G, 16.0S; ~540 kDa), catalase (C, 11.4S; ~232 kDa) and alkaline phosphatase (P, 6.1S; ~140-160 kDa) were used as internal markers.

Figure 3

sAPP heteromers contribute to the estimation of sAPPα and sAPPβ levels by ELISA. (A) Box plot of CSF levels of classical AD biomarkers Aβ42, T-tau and P-tau from 13 NDC controls and 13 probable AD cases. (B) CSF samples were also assayed for specific sAPPα and sAPPβ using ELISA kits from IBL before (incubated with Sepharose A beads in the absence of an antibody) and after immunoprecipitation (IP) with the Sigma-Ct antibody. Before immunoprecipitation, no differences were found between NDC and AD subjects. After immunoprecipitation, levels of sAPPα and sAPPβ reduced significantly in both groups (paired Student-t test). (C) 10 fresh aliquots (from the 13 cases available) from NDC and AD groups were assayed with an alternative ELISA kit from Novex, of which the detection antibody binds to an epitope present in sAPPα and absent in sAPPβ, thus recognizing APPα and sAPPf. Before immunoprecipitation, no differences were found between NDC and AD subjects. Immunoprecipitation significantly reduced the levels of sAPPα and sAPPf in NDC and AD subjects (paired Student-t test). The data represent the means ± SEM (n.s., not significant).

Figure 4

sAPPf immunoreactivity levels in CSF from NDC and AD subjects. Immunodetection of sAPPf in CSF samples from 13 NDC and 13 AD subjects. (A) Representative blot of CSF-sAPPf resolved with the Covance-Ct antibody. The 130 kDa band was not easily visualized in all samples, and was quantified at longer exposure than the 100 kDa band. (B) Densitometric quantification of the APP-immunoreactive bands, ~110 and 130 kDa, from the NDC and AD cases (performed in duplicate). A control CSF sample was run on different gels (processed in parallel) to normalize the immunoreactive signal between immunoblots. Differences were found only for the 130 kDa band between NDC and AD subjects. The data represent the means ± SEM (n.s., not significant).