Inherent Anti-amyloidogenic Activity of Human Immunoglobulin γ Heavy Chains

Abstract

We have previously shown that a subpopulation of naturally occurring human IgGs were cross-reactive against conformational epitopes on pathologic aggregates of Aβ, a peptide that forms amyloid fibrils in the brains of patients with Alzheimer disease, inhibited amyloid fibril growth, and dissociated amyloid in vivo. Here, we describe similar anti-amyloidogenic activity that is a general property of free human Ig γ heavy chains. A γ₁ heavy chain, F1, had nanomolar binding to an amyloid fibril-related conformational epitope on synthetic oligomers and fibrils as well as on amyloid-laden tissue sections. F1 did not bind to native Aβ monomers, further indicating the conformational nature of its binding site. The inherent anti-amyloidogenic activity of Ig γ heavy chains was demonstrated by nanomolar amyloid fibril and oligomer binding by polyclonal and monoclonal human heavy chains that were isolated from inert or weakly reactive antibodies. Most importantly, the F1 heavy chain prevented in vitro fibril growth and reduced in vivo soluble Aβ oligomer-induced impairment of rodent hippocampal long term potentiation, a cellular mechanism of learning and memory. These findings demonstrate that free human Ig γ heavy chains comprise a novel class of molecules for developing potential therapeutics for Alzheimer disease and other amyloid disorders. Moreover, establishing the molecular basis for heavy chain-amyloidogenic conformer interactions should advance understanding on the types of interactions that these pathologic assemblies have with biological molecules.

Introduction

Alzheimer disease (AD)³ is approaching global epidemic proportions and is the most common of over 25 incurable protein misfolding diseases that are termed the amyloidoses (1, 2). A hallmark of the disease is deposition of amyloid fibril-containing neuritic plaques that are formed by abnormal processing of β-amyloid protein (Aβ), a proteolyzed transmembrane 39–43 fragment of amyloid precursor protein (3). Diverse experimental studies support a central pathogenic role for aggregated Aβ, which in the brain initiates a cascade of events that ultimately lead to neuronal dysfunction and death (4–6). The most neurotoxic Aβ species are low molecular weight oligomers (5), which include noncovalently linked species (7–9) and dityrosine cross-linked β-amyloid protein species (CAPS) (10).

Active vaccination with Aβ or passive administration of anti-Aβ antibodies has shown promise in transgenic AD animal models and in some AD patients by inducing neuritic plaque clearance, neutralizing neurotoxic soluble Aβ oligomers, and/or improving cognitive functioning (6, 8, 11–13). Immunotherapy has generally targeted linear sequence epitopes on Aβ (13). However, antibodies that do not distinguish between pathologic Aβ conformers and the monomeric peptide may have adverse effects because the native peptide has been implicated in normal lipid and cholesterol homeostasis (14). Of potentially greater use are antibodies that cross-react with conformational epitopes on pathogenic aggregated Aβ species that do not bind to the native precursor protein (11, 15–18). Among these molecules are a subpopulation of Aβ-reactive polyclonal IgGs in intravenous immune globulin, derived from pools of plasma from presumably healthy donors, that we have shown to cross-react with amyloid fibrils and oligomers (17, 18). These antibodies inhibited amyloid fibril growth in vitro and demonstrated amyloidolytic activity by expediting the clearance of patient-derived amyloid in mice (18). Presumably, the promising effect of intravenous immune globulin on some AD patients is due at least in part to the Aβ-reactive IgG component (19–22). However, intravenous immune globulin is limited in supply, and there is not enough to treat the entire AD patient population.

In an attempt to generate more renewable human anti-Aβ antibodies, we recently used splenic B-cells from a normal individual to create hybridomas secreting pan-amyloid fibril and oligomer cross-reactive human antibodies. This fusion resulted in the cloning of a novel free human Ig heavy chain (HC), F1, which cross-reacted with amyloid fibrils and Aβ oligomers without binding to the native precursor proteins, inhibited Aβ fibril growth, and reduced soluble Aβ-induced impairment of a cellular mechanism of rodent memory and learning, hippocampal long term potentiation (LTP). Moreover, we have demonstrated that anti-amyloidogenic activity is a general property of free human Ig γ HCs. These findings advance understanding of the types of molecular interactions that amyloidogenic conformers may be involved with in vivo and should facilitate the development of novel and much needed therapeutic reagents for patients with amyloid-associated diseases.

EXPERIMENTAL PROCEDURES

Peptides, Proteins, and Antibodies

Human Aβ40, Aβ42, and human islet amyloid polypeptide (IAPP) were purchased from Quality Controlled Biochemicals (Hopkinton, MA). The peptide preparations were >90% pure, as determined by standard mass spectrometric analysis. Before use, the lyophilized Aβ40 or IAPP peptide was disaggregated by sequential exposure to trifluoroacetic acid and hexafluoroisopropanol (Pierce) followed by the addition of 2 mm NaOH and 2 × PBS (1 × final) (Pierce) and ultracentrifuged to give a final peptide concentration of ∼0.2 mg/ml (23). Soluble Aβ42 was disaggregated using trifluoroacetic acid/hexafluoroisopropanol pretreatment and solublized to a concentration of ∼0.04 mg/ml using a modified version of an alkaline pretreatment protocol (17, 24). Peptide concentrations were determined at A_{215 nm} by reverse phase high pressure liquid chromatography using an Aβ40 standard curve or by the MicroBCA assay (Pierce).

The recombinant λ6 LC variable domain, Jto, was produced in an Escherichia coli expression system and purified using Amberlite XAD-7 (Sigma-Aldrich) (25). The soluble LC was sterile-filtered using a 0.22-μm polyvinylidene fluoride 25-mm Millex-GV syringe-driven filter unit (Millipore, Bedford, MA) and shown to be >90% pure, consisting of monomers and dimers using Sephadex G25 (GE Healthcare) gel filtration and SDS-PAGE. The protein concentration was determined by the MicroBCA assay.

Polyclonal IgGs from normal sera samples were isolated by zone electrophoresis on polyvinyl chloride-polyvinyl acetate blocks, and after passage through an agarose gel filtration column, purity was established by SDS-PAGE (26). Polyclonal human HCs were generated from purified IgGs by mild reduction and alkylation and purified under acidic conditions using size exclusion chromatography (27). Monoclonal human HCs were generated from mAbs, 13A and 30B (28, 29), by mild reduction and alkylation and purified using affinity chromatography with a column consisting of rabbit anti-human free λ and κ LC IgGs (Dako, Carpinteria, CA) conjugated to a N-hydroxysuccinimide-activated Sepharose 4 fast flow agarose matrix (GE Healthcare). The percentage of yield of each soluble HC preparation was determined by ultracentrifugation (50,000 × g for 2 h at 4 °C) to be ∼50–90% for the monoclonal HCs F1, 30B, and 13A and ∼15–25% for the polyclonal HCs. The protein concentrations were determined by the MicroBCA assay. SDS-PAGE confirmed that the polyclonal and monoclonal HCs were >85% pure.

An N-terminal Aβ reactive murine mAb, MAB1560, was purchased from Millipore. Bovine elastin fibrils, DNA, aprotinin, and recombinant human insulin were obtained from Sigma-Aldrich.

Preparation of Peptide and Protein Aggregates

Soluble CAPS were prepared from alkaline-pretreated Aβ40 (∼0.2 mg/ml) by incubating the peptide with 1.1 μm HRP and 250 μm H₂O₂ in PBS at 37 °C for 3 h (30) and purified using copper (CuSO₄) precipitation, followed by centrifugation, guanidine HCl, and PBS washes, and resolubilization of the purified CAPS pellet with 5 mm EDTA before dialysis into PBS to a final concentration of ∼0.2 mg/ml (17). CAPS were quantified using SDS-PAGE (4–12% BisTris precast gels; Invitrogen) and the MicroBCA assay. Electrospray ionization mass spectrometry (Applied Biosystems, Foster City, CA) and dityrosine fluorescence (31) confirmed that the aggregates consisted of low molecular mass (<38 kDa) cross-linked SDS stable species.

Synthetic Aβ40 and IAPP fibrils were prepared from the trifluoroacetic acid/hexafluoroisopropanol disaggregated peptides. Fibril growth in PBS containing 0.02% sodium azide (PBSA) was monitored by thioflavin T (ThT) fluorescence (32). The fibrils were harvested by centrifugation at 20,200 × g for 30 min at room temperature. Amyloid fibrils were extracted from amyloid-laden tissues derived from patients with AL or AA amyloidosis by the method of Pras et al. (33), lyophilized, and then reconstituted in PBSA. Neuritic plaque cores from AD brain were isolated as described (17). The chemical nature of patient-derived fibrils was determined by protein sequencing and tandem mass spectrometry (34). All of the amyloid and non-amyloid bovine elastin fibrils (Sigma-Aldrich) were sonicated (2 × 30 s bursts) with a probe sonicator disruptor (Teledyne Tekmar, Mason, OH), aliquoted, and stored at −20 °C.

Antibody Binding and Competition Assays

HC and antibody reactivity with amyloid fibrils, CAPS, and elastin fibrils was determined at 37 °C using a europium (Eu³⁺)-based fluoroimmunoassay (EuLISA) (35) or ELISA. Antibodies or HCs were serially diluted in assay buffer (1% BSA in PBSA) and tested (100 μl/well) in activated, high binding microtiter plate wells (COSTAR, Corning, NY) that were coated with 400 ng of target protein and blocked with 1% BSA (Sigma-Aldrich) in PBSA. For competition studies involving Aβ monomers as an inhibitor of HC or antibody binding to the plate-immobilized homologous Aβ conformer or fibrils, 1 μm HC F1 or 5 nm MAB1560 was preincubated with the competitor for 40 min. The reaction sample was serially diluted in appropriate microtiter wells. A biotinylated goat anti-human IgG (γ-chain-specific; Sigma- Aldrich) or biotinylated goat anti-mouse IgG (whole molecule; Sigma-Aldrich) reagent served as a secondary antibody. For EuLISA, a Eu³⁺-streptavidin conjugate was added, followed by the releasing enhancement solution. Eu³⁺ time-resolved fluorescence was measured using a Victor² 1420 multilabel counter (PerkinElmer Life Sciences). The amount (fm) of lanthanide released was calculated from a standard curve using known concentrations of Eu³⁺. For ELISA, HRP-streptavidin (Jackson Immunoresearch Laboratories Inc., West Grove, PA) was added followed by TMB substrate (SureBlue Reserve^TM, KPL, Gaithersburg, MD), and absorbance at 452 nm was determined using a SpectraMax M2 instrument (Molecular Devices Inc., Sunnyvale, CA). All of the measurements in this and other assays were done in triplicate (error bars in the figures represent S.D.). The values for EC₅₀ and IC₅₀ were determined from the sigmoidally fit antibody binding curves (SigmaPlot 2000, version 6; Systat Software, Chicago, IL).

Generation of mAb F2 and HC F1

Discarded human spleen tissue was obtained from a young woman with no history of neurological or immunological disease, following surgical resection of a benign pancreatic cyst, in accord with a protocol approved by the Institutional Review Board of the Massachusetts General Hospital (Boston, MA). The sample was crushed, filtered through mesh, and enriched for mononuclear cells (MCs) using gradient density centrifugation with FicollPaque Plus (GE Healthcare). Splenic MCs were stored frozen in 90% heat-inactivated fetal calf serum (Invitrogen) and 10% Me₂SO (Sigma-Aldrich) under liquid nitrogen. Prior to cell fusion, CD27-positive splenic MCs were isolated with anti-CD27 magnetic beads (Miltenyi Biotec, Auburn, CA) according to the manufacturer's instructions, and splenic MCs were cultured for 8 days on a monolayer of tCD40L cells (courtesy of Gordon Freeman, Dana Farber/Partners Cancer Care, Boston, MA) (29).

Cultured splenic MCs (2.5 × 10⁶) were fused to the B5-6T heteromyeloma cell line, which ectopically expresses the hTERT and mIL-6 genes, at a 1:1 ratio using the stirring method with 50% polyethylene glycol (Sigma-Aldrich), plated in 20 wells of a 48-well plate (Corning), and selected in HAT medium (Sigma-Aldrich) (29). Hybridoma supernatants that were positive in EuLISA against plate-immobilized Jto fibrils were selected for subcloning by limiting dilution. After four rounds of subcloning, stable hybridomas were adapted to serum-free medium (IS MAB-CD; Irvine Scientific, Santa Ana, CA), plated at a density of 5 × 10⁵ cells/ml in 100 ml culture medium, and incubated for 5 days in a 500-ml roller bottle. Filtered supernatants were purified over protein G-Sepharose (GE Healthcare). Purity was assessed by SDS-PAGE (Invitrogen) and shown to be >90% pure. Protein concentrations were determined using the NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE).

Reverse Transcription-PCR Amplification of Immunoglobulin Variable Domain Sequences

Antibody variable domain cDNA sequences were amplified with consensus primer sets specific for human HC γ, κ, and λ LCs (36). RNA was isolated from hybridomas using RNA Stat 60 (Tel-Test, Friendswood, TX). Reverse transcriptase reactions were performed with Superscript II (Invitrogen). PCRs were performed with the Expand High Fidelity PCR system (Roche Applied Science) (25 cycles of 94 °C for 15 s, 55 °C for 30 s, 72 °C for 60 s, plus a 5-s increase every cycle for cycles 16–25). The amplified sequences were isolated by agarose gel electrophoresis followed by purification with the QiaQuick gel extraction kit (Qiagen) and then sequenced at the Kimmel Cancer Center Nucleic Acid Facility (Thomas Jefferson University, Philadelphia, PA). DNA sequences were analyzed using the V-Quest program (37).

Western Blots

Standard Western blotting was performed with a 4–12% gradient SDS-PAGE (Invitrogen), transfer to Hybond (GE Healthcare), and detection with HRP-conjugated antibodies specific for human Ig γ HCs (Southern Biotechnology, Birmingham, AL), λ (Sigma-Aldrich), or κ LCs (Sigma-Aldrich).

Immunohistochemistry

4–6-μm-thick sections were cut from formalin-fixed, paraffin-embedded blocks of amyloid-laden tissues from patients with amyloid light chain (AL), serum amyloid A (AA), or Apin amyloidosis and placed on poly-l-lysine coated microscope slides for Congo red and antibody staining (18). For immunohistochemistry, the tissues were exposed for 1 h to a solution of 1.0–3.0 μg/ml HC F1 in HEPES-buffered saline and then to a biotinylated goat anti-human Ig antibody (MultiLink system; Biogenex, San Ramon, CA). The slides were developed using a Vector ABC kit and 3,3′-diaminobenzidene (Vector Labs, Burlingame, CA) and examined with a Leica DM500 microscope with an epifluorescent attachment. Congo red fluorescence was visualized by use of a 540-nm bandpass filter, and images were acquired with a cooled CCD camera (SPOT; Diagnostic Instruments, Sterling Heights, MI).

Fibril Formation and Extension

De novo Aβ fibril formation was monitored in real time using ThT, which gives a fluorescent quantitative signal when bound to amyloid fibrils (38). Wells of an ultra low binding 96-well plate (Corning) were filled with 90 μm soluble Aβ and 30 μm ThT and serially diluted HC F1 (10 nm-1.0 μm) or 0.7 μm 13A HC or whole IgG in PBSA. Control wells included ThT with inhibitor or Aβ alone. The plates were incubated at 37 °C, and fluorescence intensity was measured for ∼1 day with a FL600 microplate reader (Biotek, Winooski, VT). For measurements of fibril extension, HCs F1 and 13A or IgG 13A were serially diluted (10 nm-1.0 μm) in PBSA in high binding microtiter plate wells (COSTAR; Corning) that were coated with 400 ng of sonicated Aβ fibrils. Soluble biotinylated Aβ monomers (50 nm) were added, and after 3 h, the plates were washed, and fibril-recruited biotinyl-Aβ was detected using an Eu³⁺-streptavidin conjugate and quantified by time-resolved fluorometry using a Victor² 1420 multilabel counter.

Electron Micrographs

Aggregates (0.1–0.4 mg/ml) were adsorbed onto carbon and Formvar-coated copper grids and then negatively stained with 0.5% uranyl acetate. Stained samples were examined and micrographed using a Hitachi H-800 transmission electron microscope.

In Vivo Electrophysiology

The experiments were carried out on urethane-anesthetized adult male Wistar rats, licensed by the Department of Health and Children of Ireland. Single pathway recordings of field postsynaptic potentials (EPSPs) from the stratum radiatum in the CA1 area of the hippocampus were determined in response to stimulation of the ipsilateral Schaffer collateral-commissural pathway (39). Test EPSPs were evoked at a frequency of 0.033 Hz and at a stimulation intensity adjusted to give an EPSP amplitude of 50% of maximum. The high frequency stimulation protocol for inducing LTP consisted of 10 trains of 20 stimuli, an interstimulus interval of 5 ms (200 Hz), and an intertrain interval of 2 s. The intensity was increased to give an EPSP of 75% of the maximum amplitude during high frequency stimulation. To inject samples, a cannula was implanted in the lateral cerebral ventricle (coordinates, 1 mm lateral to the midline and 4 mm below the surface of the dura) just before electrode implantation. Soluble synthetic Aβ42 was prepared as described previously (39). Briefly, a 5-μl aliquot of 40 pmol of peptide with or without ∼3–6 μg of HC F1 or 13A IgG in ice-cold Milli-Q water (Millipore) was intracerebroventricularly injected 10 min before high frequency stimulation over a 2-min period. Control vehicle injections comprised Milli-Q water.

LTP was expressed as the mean ± S.E. percentage base-line field EPSP amplitude recorded over at least a 30-min base-line period. Similar results were obtained when the EPSP slope was measured. The statistical comparisons used paired and unpaired Student's t tests.

RESULTS

A Cloned Human HC F1 Cross-reacts with Amyloid Fibrils and Oligomers

In an effort to generate novel hybridomas secreting pan-amyloid fibril and oligomer cross-reactive human IgGs, splenocytes from a normal 16-year-old female were fused with B5-6T heteromyeloma cells (29, 36). Four hybridoma supernatants were shown to contain antibodies reactive with microtiter plate-immobilized human Ig LC λ6 Jto fibrils (25) in our standard EuLISA (35). Following limiting dilution subcloning, two independent hybridomas, F1 and F2, were isolated that bound to LC fibrils with molar antibody concentrations that obtained half-maximum binding (EC₅₀) values of ∼20 nm (Fig. 1 and Table 1). However, only F1 was a prototypic pan-amyloid fibril binder that had similar reactivity against Aβ40 and LC fibrils (Fig. 1 and Table 1). F2 bound ∼7-fold more weakly to Aβ than to LC fibrils with an EC₅₀ value of 145 nm.

FIGURE 1.

Fibril and Aβ conformer binding by HC F1 and mAb F2. A and B, binding curves for F1 and F2, respectively, against plate-immobilized amyloid fibrils: LC (○) and Aβ40 (●) fibrils, non-amyloid elastin fibrils (□), and CAPS (▵). C and D, F1 and a N-terminal Aβ-reactive mAb, MAB1560, respectively, binding to plate-immobilized Aβ40 monomer in the presence (○) or absence (●) of a 100-fold molar excess of homologous peptide conformer. Binding studies were carried out in PBSA containing 1% BSA at 37 °C.

TABLE 1.

Monoclonal human IgG and HC binding to amyloid fibrils and CAPS

The EC₅₀ values and maximum binding signal amplitudes were determined from sigmoidally fitted binding curves such as those shown in Figs. 1 and 4. ND, not determined.

Ig	HC	Aβ fibrils		CAPS		LC fibrils		IAPP fibrils
		EC50	Maximum Eu3+	EC50	Maximum Eu3+	EC50	Maximum Eu3+	EC50	Maximum Eu3+
		nm	fmol	nm	fmol	nm	fmol	nm	fmol
HC F1	γ1	23 ± 0.2	249 ± 16	28 ± 0.1	250 ± 3.0	24 ± 0.3	195 ± 14	40 ± 6.2	110 ± 11
IgG F2	γ1	145 ± 1.0	84 ± 6.0	330 ± 3.0	58 ± 7.0	21 ± 0.3	228 ± 16	>1000	>10
IgG 13A	γ1	>1000	>30	>1000	>10	>1000	>20	>1000	>15
HC 13A	γ1	14 ± 0.01	261 ± 2.0	12 ± 0.3	165 ± 4.2	65 ± 0.8	306 ± 20	ND	ND
IgG 30B	γ2	>1000	>40	ND	ND	>1000	>50	ND	ND
HC 30B	γ2	109 ± 0.3	430 ± 7	52 ± 1.4	179 ± 2.1	23 ± 0.2	23 ± 0.2	ND	ND

Sequence analysis of F1 using consensus primer amplification of Ig cDNA showed that the protein contained a somatically mutated γ₁ HC with any intact open reading frame (Table 2). A partial κ4LC sequence was amplified but was identical to the aberrant murine κ transcript that is expressed by the heteromyeloma fusion cell, B5-6T (40). In contrast, the mAb F2 expressed intact human HC and λ LC open reading frames (Table 2). Validation that F1 was a free HC was demonstrated by protein migration in SDS-PAGE that was consistent with dimeric (∼100 kDa) and monomeric HC and by electrospray ionization mass spectral analysis of excised SDS-PAGE protein bands that confirmed F1 lacked a human LC. Moreover, immunoblotting of the F1 hybridoma whole cell lysate with polyclonal antisera specific for human LCs confirmed that F1 did not contain a κ or λ LC (Fig. 2). Although free HCs are generally less soluble than intact antibodies, it was established by ultracentrifugation and size exclusion gel chromatography that the native soluble F1 homodimer and monomer were the fibril-reactive species.

TABLE 2.

Sequence analysis of HC F1 and mAb F2

HC and LC cDNA sequences were analyzed using the V-Quest program (37).

Ig	Subclass	V-Gene	J-Gene	D-Gene	CDR3 sequence
F1 HC	γ1	IGHV1–18*01	IGHJ4*02	IGHD3–10*01	AREKTMVRGAISGYSDY
F2 HC	γ1	IGHV3–23*01	IGHJ4*02	IGHD3–22*01	AKEASDDSTYNPFDY
F2 LC	λ	IGLV4–69*02	IGLJ1*01	N/A	CQTWGTGIHVF

FIGURE 2.

Western blot analysis of Ig chain expression by the F1 hybridoma. The blots show that only human HC was detected for the F1 hybridoma cell lysate. Human HC and LC were detected in cell lysates for control hybridomas expressing IgGκ or IgGλ, respectively. The cell fusion partner for the F1 hybridoma, B5-6T, was negative for HC and LC. The cell lysates were run on a 4–12% BisTris gel under reducing conditions, and Western blots were carried out using HRP-conjugated secondary antibodies specific for human HC γ, LC κ, or λ.

Further investigation of HC F1 pan-amyloid fibril reactivity showed that it bound similarly to all plate-immobilized amyloid fibrils tested, including those formed from Aβ40, Aβ42, IAPP, and LC, with an EC₅₀ value of ∼30 nm (Fig. 1 and Table 1). Specificity of F1 for amyloid fibrils was evident from its inability to bind to non-amyloid elastin fibrils (Fig. 1), soluble precursor Aβ, and LC proteins and its ability to retain binding to Aβ fibrils in the presence of murine serum (data not shown). F1 also bound to CAPS as strongly as to amyloid fibrils, with an EC₅₀ value of ∼20 nm, and had weak (μm) interactions with plate-immobilized Aβ monomers (Fig. 1 and Table 1). The HC binding to plate-immobilized Aβ monomers was not prevented by a ∼100-fold molar excess of Aβ monomers, but the same competitor was a potent inhibitor of an N-terminal Aβ-reactive mAb, MAB1560 (Fig. 1, C and D). This suggested that HC F1 targeted a plate-induced amyloid-like epitope on the monomeric peptide (17).

Having established that HC F1 bound to a conformational epitope present on synthetic amyloid fibrils and oligomers, we investigated the ability of the protein to bind to patient-derived amyloid fibrils and to stain amyloid-laden patient tissue sections. F1 had nanomolar binding to plate-immobilized amyloid extracted from patients with primary (AL) or secondary (AA) amyloidosis, or AD, which was comparable with its binding to synthetic amyloid fibrils (Fig. 3A). We next assessed binding of F1 to AL and AA amyloid-laden patient renal tissue sections. Amyloid deposition in the tissues was demonstrated by apple green Congo red birefringence, which is known to detect some, but not all, amyloid (42) (Fig. 3B). F1 staining was observed throughout the renal parenchyma, whereas no staining was observed in renal tissues from normal subjects. The HC also stained congophilic APin amyloid deposits in a calcifying epithelial odontogenic tumor (41) but did not react with a similar tumor that was devoid of amyloid (Fig. 3B).

FIGURE 3.

Ex vivo and in situ amyloid fibril binding by HC F1. A, amyloid fibril binding curves for F1 against plate-immobilized fibrils extracted from patients with ALλ6 (○) or AA (▵) amyloidosis or AD (♦). B, representative Congo red and F1 staining of amyloid-laden glomeruli kidney tissue sections from patients with ALλ6 or AA amyloidosis or with an Apin amyloid-containing CEOT tumor (a) that was surrounded by non-amyloid-containing amelobastoma tumor tissue (b). Panels on the right show negative F1 staining of control tissues that did not contain amyloid. The original magnification for tissue sections was ×200.

Amyloid Fibril and Oligomer Cross-reactivity Is a General Property of HCs

To establish whether amyloid fibril and oligomer cross-reactivity is a general and novel property of free human Ig γ HCs, we isolated HCs from two human mAbs, 13A and 30B, which are specific for botulinum neurotoxins (28, 29) and do not react with amyloid fibrils. Fig. 4 and Table 1 show that the purified free 13A and 30B HCs had novel binding to all amyloid assemblies tested (Aβ fibrils, LC fibrils, and CAPS) with EC₅₀ values of ∼10–100 nm. Further validation that amyloid fibril binding is a general property of free HCs was demonstrated by up to 60-fold stronger Aβ fibril binding by polyclonal HCs, isolated from IgGs from normal individuals, compared with the reactivity of the intact IgGs (Fig. 4C). Ultracentrifugation studies confirmed that HC binding to amyloid fibrils and oligomers was attributed to the soluble proteins. The inherent anti-amyloidogenic activity of free HCs was not an artifact of carboxymethylation, which was used to separate these Ig chains from the whole antibodies, because similar (nanomolar) binding to Aβ fibrils was obtained when these molecules were freshly prepared by reduction alone (Fig. 5). Moreover, the specificity of HCs for amyloidogenic conformers was demonstrated by their inability to bind to non-amyloid elastin fibrils and other molecules, including DNA, BSA, aprotinin, insulin, and gelatin (Fig. 5C).

FIGURE 4.

Aβ fibril-binding by monoclonal and polyclonal IgGs and γ HCs isolated from healthy individuals. A and B, Aβ40 fibril-binding curves for mAbs 13A and 30B, respectively, for whole molecule (♦) and HC (○), as well as binding by HC F1 (●). C and D, Aβ40 fibril binding by polyclonal HCs derived from five different subjects (●, ○, □, ■, and ▵) and polyclonal IgGs (♦, ◇, and +).

FIGURE 5.

Specificity of HCs 13A and 30B for amyloidogenic conformers. A, Aβ fibril binding for reduced (▵) and carboxymethylated (▴) HC 13A and for the intact IgG (●). B, Aβ fibril binding for reduced (▵) and carboxymethylated (▴) HC 30B and for the intact IgG (●). C, Aβ fibril and non-amyloidogenic molecule binding by 50 nm carboxymethylated 13A and 30B HCs (closed and open bars, respectively). IgG and HC binding was determined in PBSA containing 1% BSA by ELISA using a HRP-streptavidin detection system.

Monoclonal HCs Inhibit Amyloid Fibril Growth and Prevent Aβ-induced Impairment of Rodent Synaptic Plasticity

Given our observation that free HCs have novel binding to pathologic Aβ assemblies, we investigated whether these molecules could prevent the two growth phases of Aβ fibrils: de novo fibril formation and elongation (23, 43). Fig. 6 shows that subequimolar concentrations of HCs, F1, and 13A inhibited de novo Aβ fibril formation. However, the inhibitory potency of HC is not unique because similar inhibition was observed for the intact 13A antibody, BSA, and several other human mAbs that have no appreciable binding to Aβ (Fig. 6).⁴ These molecules may have primarily functioned by binding to microtiter plate wells that inhibited surface adsorption of Aβ and subsequent seeding of the reaction (44). Electron micrographs of the reaction products indicated that HC F1 but not IgG 13A enhanced the formation of less ordered fibrillar-like Aβ aggregates (Fig. 6, C–E). The observed inhibitory effects of the test proteins were not due to their quenching of ThT fluorescence that was used to monitor fibril growth or because they prevented fibril binding by the dye, because these molecules did not alter the ThT fluorescence of pregrown Aβ fibrils.

FIGURE 6.

HC F1 inhibition of de novo Aβ fibril growth. A, dose-dependent effect of 5 μm (○), 2.5 μm (●), 1.25 μm (▴), 0.31 μm (▵), and 0 μm (□) F1 on Aβ40 fibril formation. B, Aβ40 fibril formation carried out in the absence (□) or presence of 0.7 μm HC 13A (●), IgG 13A (○), or BSA (▴). C–E, electron micrographs of aggregate products from reactions carried out in the absence of inhibitor (C) or in the presence of 1 μm F1 (D) or IgG 13A (E). Each reaction was carried out 90 μm Aβ40 with or without inhibitor in PBSA containing 30 μm ThT at 37 °C.

In a more stringent test of anti-amyloidogenic activity, F1 was the only protein that inhibited Aβ fibril elongation, albeit without inducing the formation of alternative Abeta assembly species, with a molar protein concentration that obtained half-maximum inhibition (IC₅₀) of ∼1.0 μm (Fig. 7A). This indicated that the anti-amyloidogenic activity of a HC was partly dependent on its primary sequence.

FIGURE 7.

HC F1 inhibition of Aβ fibril elongation. A, dose-dependent inhibitory effect of F1 (▴), mAb 13A (●), 13A HC (○), and BSA (▵) on Aβ40 fibril elongation. B–D, electron micrographs of aggregate products from reactions carried out in the absence of inhibitor (B) or in the presence of 1 μm F1 (C) or IgG 13A (D). Fibril extension reactions were carried out with 30 nm biotinyl-Aβ40 with or without inhibitor in PBSA for 3 h at 37 °C.

Given the unique anti-amyloidogenic activity of HC F1, we used this molecule to investigate the in vivo therapeutic potential of free HCs using a rodent model of Aβ oligomer impairment of hippocampal LTP, which is a measurable correlate of long term memory and learning (45–47). Fig. 8 showed that soluble Aβ42 (40 pmol, intracerebroventricularly) prevented the induction of LTP of excitatory synaptic transmission in the CA1 area of anesthetized rats (103 ± 8% base line, n = 5; p < 0.05 compared with 140 ± 8% in vehicle-injected rats, n = 5). In contrast, high frequency conditioning stimulated animals that were cotreated with F1 (3 μg, intracerebroventricularly), and Aβ42 induced significant LTP (124 ± 6%, n = 5; p < 0.05 compared with base line and p > 0.05 compared with vehicle). This was a specific effect because a control molecule with an inert HC, i.e. the intact 13A antibody (3.75 μg, intracerebroventricularly), did not reverse Aβ impairment of LTP (107 ± 3%, n = 3; p > 0.05 compared with base line or Aβ42 alone) (Fig. 8B). F1 in the absence of Aβ42 did not affect LTP (n = 2, data not shown).

FIGURE 8.

HC F1 prevents Aβ42-mediated inhibition of LTP in the rat hippocampus in vivo. LTP is an activity-dependent long lasting increase in the strength of synaptic connections (EPSPs) and models the cellular mechanisms underlying learning and memory formation. A, in vehicle-injected controls (●), the application of high frequency conditioning stimulation (arrow) at 30 min triggered a robust LTP. Intracerebroventricular injection (asterisk) of synthetic Aβ42 (40 pmol, ▴) 10 min before the conditioning stimulation completely inhibited LTP. B, the inhibition of LTP by Aβ42 was prevented by intracerebroventricular co-injection of 60 pmol (3 μg) of HC F1 (▵). Co-injection of a similar amount (3.75 μg) of the control mAb, 13A (○) did not prevent the Aβ-mediated inhibition of LTP.

DISCUSSION

The pan-amyloid fibril and oligomer binding of free Ig γ HCs is analogous to the reactivity of a subpopulation of naturally occurring human antibodies that we have recently identified (17) and is similar to the conformational specificity of antibodies generated by active vaccination (11, 15, 48). However, the anti-amyloidogenic activity of HCs is unique because it is inherent in all of the molecules tested. Furthermore, in contrast with antibody binding, free HC interactions with amyloidogenic assemblies are likely to involve neo-exposed residues, which are usually buried and involved in HC-LC interactions (49) because free HC binding to Aβ conformers was prevented when the molecule was bound to a LC in an antibody molecule. Such surfaces are presumably ideal for binding to similar surfaces on the growing ends of amyloid fibrils or oligomers because the HC F1 prevented Aβ fibril elongation. By analogy to HC-LC interactions, HC binding to amyloidogenic conformers and the diverse anti-amyloidogenic potencies we determined may crucially depend on particular amino acid residues at or removed from the reactive surface (50). The exposure of a protein “hot spot” on dissociation of its natural ligand, as occurs on dissociation of antibody HCs and LCs, may be a general mechanism by which endogenous accessory molecules incorporate onto amyloidogenic assemblies (51). Nevertheless, the binding of HCs to Aβ fibrils is unique relative to other endogenous molecules because murine serum was not able to prevent nanomolar F1 binding to Aβ fibrils (data not shown). Moreover, the free LCs of 13A and 30B also had novel nanomolar binding to Aβ fibrils (Fig. 9), indicating that anti-amyloidogenic activity is an inherent property of unpaired β-sheet surfaces of free Ig chains.

FIGURE 9.

LC binding to Aβ fibrils. Binding curves for LCs (■) and intact IgGs (●) for 13A (A) and 30B (B) against plate-immobilized Aβ fibrils. Antibody binding was determined in PBSA containing 1% BSA by ELISA using a -streptavidin detection system.

The anti-amyloidogenic activity of free HCs may also have been due to a sequence-dependent subtle conformational rearrangement that occurs when these molecules are not bound to LCs or because the relatively small HCs can intercalate into grooves on amyloidogenic assemblies that are not accessible to “large” antibody molecules. For example, camelid HC antibodies, which do not contain LCs, can penetrate enzyme active sites that are not accessible to larger native antibody molecules (52). Future studies that should advance understanding of the contribution of the HCs subgroup and the molecular basis for protein interactions with amyloidogenic assemblies include the generation of phage display peptide epitope mimics (53), HC fragmentation, and/or site-directed mutagenesis (54).

Active and/or passive immunization against Aβ has decreased peptide concentration and neuritic plaque burden in the brain of some AD patients, decreased cognitive decline, and neutralized toxic Aβ species in animal models of the disease (8, 12, 55–58). The prototypic γ₁ HC, F1, has also shown therapeutic potential for AD by binding to a conformational epitope on amyloid fibrils and oligomers, preventing in vitro Aβ fibril growth and partly neutralizing the in vivo neurotoxic effect of Aβ oligomers on rodent hippocampal long term potentiation. These results and our findings with polyclonal HCs that demonstrated the general anti-amyloidgenic activity of free γ HCs suggest that these molecules could provide a benefit to AD patients. This may involve HCs binding to neurotoxic Aβ within the brain (5, 59) and/or reacting with the vascular peptide that increases Aβ efflux from the brain via a peripheral sinklike mechanism (60). A potential advantage of HCs for therapeutics is the relative ease by which they can cross the blood-brain barrier compared with full-length antibodies, and they are less prone to adverse inflammatory side effects (61). Moreover, HCs are generally robust, well suited to genetic manipulation, and easily expressed, and their solubility can be greatly enhanced with point mutations in the VL-VH interface (40). Given that F1 and not 13A, which are γ1 and γ2 HCs, respectively, inhibited Aβ fibril elongation, it is important to establish whether the anti-amyloidogenic potential of an HC is influenced by its subtype.

Free HCs are not the only antibody subunits that have shown anti-amyloidogenic activity. Single chain variable fragments have been generated against Aβ that preferentially bind to the multimeric peptide, inhibit fibril formation, reduce Aβ cytotoxicity in cultured cells, and degrade brain amyloid plaques in vivo (62–64). Murine F(ab′)₂ fragments can also clear in vivo AD plaques (65, 66). Moreover, single-chain camelid antibodies, which consist of only a HC-HC dimer, that were specifically generated against Aβ inhibited Aβ fibril growth and neutralized neurotoxic Aβ assemblies in vitro (67, 68). However, the anti-amyloidogenic activity of free HCs is distinct from other antibody subunits because it is an inherent property of all HCs tested. It is evident from our experimental findings that further characterization of the therapeutic potential for free HCs for AD and other amyloid-related diseases is warranted. Determination of the molecular basis for their reactivity with amyloidogenic assemblies could facilitate the development of novel therapeutics and diagnostics and advance understanding on the types of interactions that pathogenic amyloidogenic assemblies have with biological molecules.