Characterization of a Single-Chain Variable Fragment Recognizing a Linear Epitope of Aβ: A Biotechnical Tool for Studies on Alzheimer’s Disease?

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder with devastating effects. Currently, therapeutic options are limited to symptomatic treatment. For more than a decade, research focused on immunotherapy for the causal treatment of AD. However, clinical trials with active immunization using Aβ encountered severe complications, for example meningoencephalitis. Consequently, attention focused on passive immunization using antibodies. As an alternative to large immunoglobulins (IgGs), Aβ binding single-chain variable fragments (scFvs) were used for diagnostic and therapeutic research approaches. scFvs can be expressed in E. coli and may provide improved pharmacokinetic properties like increased blood-brain barrier permeability or reduced side-effects in vivo. In this study, we constructed an scFv from an Aβ binding IgG, designated IC16, which binds the N-terminal region of Aβ (Aβ(1-8)). scFv-IC16 was expressed in E. coli, purified and characterized with respect to its interaction with different Aβ species and its influence on Aβ fibril formation. We were able to show that scFv-IC16 strongly influenced the aggregation behavior of Aβ and could be applied as an Aβ detection probe for plaque staining in the brains of transgenic AD model mice. The results indicate potential for therapy and diagnosis of AD.


Introduction
Alzheimer's disease (AD) is a devastating, progressive, neurodegenerative disorder, which affects more than 35 million people world-wide [1]. Currently, AD treatment is restricted to palliative care due to the lack of disease arresting or modifying therapies [2]. Several lines of evidence have indicated that the amyloid-bpeptide (Ab) plays a pivotal role in the pathology of AD. Ab is produced throughout life as a 38 to 43 residue peptide derived from the amyloid precursor protein (APP) after cleavage by two distinct proteases, called band c-secretase [3,4,5]. The function of Ab needs to be clarified, but recent studies suggest neuroprotective effects of monomeric Ab (for review, see ref. [6]). Senile plaques in the brain of the patient, one of the typical histopathological hallmarks of AD, consist mainly of Ab . They are thought to play a crucial role in the pathology of AD, and according to the original amyloid cascade hypothesis, Ab deposited in plaques has been thought to be responsible for neuronal dysfunction [7,8]. However, Ab can adopt a variety of neurotoxic conformers (for review, see ref. [9]), and more recent studies indicate that diffusible Ab oligomers are the major toxic species during disease development and progression [10,11,12,13]. Consequently, agents that interfere with Ab oligomerization or increase Ab clearance from the brain are expected to be valuable for application in therapy or prevention of AD. Passive immunotherapeutic approaches, i.e. direct administration of Ab antibodies peripherally, were shown to be effective in transgenic mouse models of AD [14,15,16,17]. A variety of humanized monoclonal antibodies are currently investigated in clinical trials (for review, see ref. [18]). However, two phase III clinical trials with antibodies were discontinued as they failed to improve cognitive functions in the treated patients [19].
Additionally, Ab binding ligands can be valuable for the investigation of the plaque load by in vivo imaging methods. Currently, only a few amyloid PET ligands have been applied in clinical studies (for review, see ref. [20,21]). Numerous efforts are devoted to develop new, target-specific imaging agents for the detection of amyloid plaques in vivo. To be suitable, such substances should exhibit highly specific binding to Ab aggregates, very selective labeling and efficient brain penetration.
Targeting Ab using scFvs have been shown to be a suitable alternative to IgGs [22]. ScFvs are genetically engineered constructs composed of the variable regions of the heavy-(V H ) and light chain (V L ) domains of a respective antibody, connected by a flexible linker to prevent dissociation. They usually retain the specific and monovalent antigenic binding affinity of the parent IgG, but can exhibit improved pharmacokinetic properties like increased blood-brain barrier permeability [23]. Because of their small size, they can be recombinantly expressed in living systems, offering advantages for protein expression in microbial hosts and purification with high yield [24], as well as for gene therapeutic approaches that could avoid repeated infusions of expensive antibodies like performed in passive immunotherapy. Due to the lack of the constant domain (F C ) in scFvs, F C induced cellular responses, microhemorrhages and inflammatory processes are prevented [25]. These advantages prompted the generation of several Ab-binding scFvs that interact with Ab, influence Ab aggregation and partially reduce Ab cytotoxicity in cell culture [26,27,28]. Recently, adeno-associated virus (AAV)-mediated intramuscular expression of a gene encoding for a scFv against the N-terminal part of Ab, isolated from a human scFv library [29], was demonstrated to be effective in removing Ab from the brain of AD transgenic mice without inducing hemorrhages or inflammation. Additionally, the cognitive performance of treated mice was improved in comparison with control mice [30]. Furthermore, another scFv binding to the C-terminus of Ab, constructed using an antibody that was generated by immunizing mice with Ab (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42), has been shown to reduce congiophilic angiopathy as well as plaque burden in APP transgenic mice after direct chronic intranasal treatment [31].
We have constructed a scFv derived from an Ab binding monoclonal antibody IC16, designated scFv-IC16. ScFv-IC16 was expressed in E. coli, purified and characterized with respect to its interaction with different Ab species and its influence on Ab fibrilization. Additionally, scFv-IC16 was used to stain and characterize plaques in brain slices of transgenic AD model mice by immunohistochemistry. scFv-IC16 was conclusively found to be a remarkable antibody fragment for the investigation of molecular pathological mechanisms, therapy or diagnosis of AD.

Molecular Modeling
The structure of IC16 in complex with Ab(1-8) was generated with Modeller [32] using the crystal structure of the Ab(1-8)-bound antibody PFA1 as template (pdb code 2IPU; [33]). The model was completed using Sybyl7.3 (Tripos Inc., St. Louis, MO, USA) by addition of Ab-residue D1, which was not resolved in the PFA-Ab crystal structure, followed by 100 steps of conjugate gradient energy minimization. The quality of the resulting model was checked using WhatCheck [34] and intermolecular contacts were analyzed with LigPlot [35].

Generation of Ab(1-16)-GB1-NHS-Sepharose for Purification of scFv-IC16
For expression of Ab(1-16)-GB1 (Ab-GB1), human Ab(1-16) was ligated into the Nde1 site of pET22b-GB1 [39]. Ab-GB1 was expressed at 37uC in BL21(DE3) Rosetta (Merck, Darmstadt, Germany) grown to mid-logarithmic phase in 2YT-medium. Expression was induced by the addition of IPTG to a final concentration of 1 mM, and cells were then grown for a further 4 h. After harvesting cells by centrifugation, they were lysed in 50 mM Tris-HCl (pH 8.0), 5 mM EDTA, 1% Triton X-100, 2 mM phenylmethanesulfonyl fluoride (PMSF), 20 mM MgCl 2 , 200 U DNase1 and 0,25 mg/ml lysozyme. The cleared lysate was first dialyzed to 20 mM Tris-HCl (pH 8.0) and 1 mM EDTA and the first purification step was then performed via Q-Sepharose (GE Healthcare, Little Chalfont, UK). The column was washed with dialysis buffer and bound Ab-GB1 was eluted with 20 mM Tris-HCl (pH 8.0), 1 mM EDTA and 75 mM NaCl. In a final purification step, Ab-GB1 was separated from minor impurities using IgG-Sepharose (GE Healthcare). After washing with 20 mM Tris-HCl (pH 8.0), 150 mM NaCl and 1 mM EDTA, Ab-GB1 was eluted with 50 mM glycine pH 2.5 and immediately neutralized by adding Tris-HCl (pH 8.0) to a final concentration on 100 mM. After dialysis against PBS the protein was coupled to NHS-Sepharose (GE Healthcare, Little Chalfont, UK) according to the manufacturer's recommendations.

Expression and Purification of scFv-IC16
Escherichia coli strain BL21(DE3)pRARE2 was used as an expression host. Bacteria were grown to high density (OD600$1.6) at 37uC, and cooled on ice for 1 h before induction with IPTG at a final concentration of 0.1 mM. Cells were further incubated for 24 h at 18uC, and subsequently harvested by centrifuging for 30 min at room temperature at 50006g. Cells were lysed in 20 mM Tris-HCl pH 8.0, 0.4 mM EDTA, 5 mM imidazole, 500 mM NaCl, 20 mM MgCl 2 , 10 mM CaCl 2 , Protease Inhibitor Cocktail Tablet Complete EDTA free (Roche, Grenzach-Wyhlen, Germany), 1 mg/ml lysozyme and 500 U DNase. The lysate was cleared by centrifugation at 20 0006g, and the soluble protein in the supernatant was purified via Ni-NTA chromatography (Ni-NTA Agarose, AppliChem GmbH, Darmstadt, Germany). After loading the sample onto the Ni-NTA Agarose, the column (3 ml) was washed with 10 column volumes (CV) 20 mM Tris-HCl pH 8.0, 5 mM imidazole, 500 mM NaCl and 1% TX-100, followed by a second wash with 10 CV 20 mM Tris-HCl pH 8.0, 5 mM imidazole, 500 mM NaCl. Bound scFv-IC16 was eluted by four CV elution buffer (20 mM Tris-HCl pH 8.0, 300 mM imidazole, and 300 mM NaCl). A second purification step using affinity chromatography was performed, because the purity of the eluted protein was lower than 50%. The previously generated Ab1-16-GB1 NHS sepharose was used for a subsequent purification step. After loading scFv-IC16 (in elution buffer), the column was washed with 10 CV TBS (137 mM NaCl, 2.7 mM KCl, 2.5 mM Tris-HCl, pH 7.4). The protein was eluted with 50 mM glycine, pH 2.5 and immediately neutralized with a final concentration of 100 mM Tris-HCl, pH 8.0. ScFv-IC16 was dialyzed against PBS (137 mM NaCl, 2.7 mM KCl, 1.8 mM KH 2 PO 4, 10 mM Na 2 HPO 4 , pH 7.4) and stored at 220uC until further usage.

Binding Constant Determination Using Surface Plasmon Resonance (SPR)
Binding kinetics were determined by SPR using a Biacore TM X (GE Healthcare, UK). Synthetic Ab peptides were dissolved in 10 mM NaAc, pH 4.0. The CM5 sensor chip surface (GE Healthcare, UK) was activated using N-ethyl-N9-3 (diethylaminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS) chemistry followed by 2-(2-pyridinyldithio)ethaneamine hydrochloride (PDEA), in order to introduce a reactive thiol group. Ab was coupled via the C-terminal cysteine to the chip at a flow rate of 5 ml/min, and the remaining active groups were blocked by injecting cysteine. The immobilization procedure was performed according to manufacturer's recommendation. All kinetic analyses were performed at a flow rate of 20 ml/min in PBS. Varying concentrations of scFv-IC16 (10 to 5000 nM) and IC16 (10 to 1000 nM) were injected. Association was observed for 180 s whereas the dissociation was observed for 120 to 360 s. When required, the surface was regenerated by injecting 20 ml 50 mM glycine, pH 11.0. The data evaluation was performed using Biaevaluation Software 4.1.1. ScFv-IC16 data were fitted according to the Langmuir 1:1 binding model, whereas IC16 data were fitted according to the bivalent binding model. Standard errors of equilibrium dissociation constants (K D ) were calculated using standard errors of the corresponding association and dissociation rate constants [40]. Seedless-preparation of Ab1-42 1 mg of Ab  or N-terminally/C-terminally biotinylated Ab(1-42) was dissolved in 1 ml 1,1,1,3,3,3-hexafluorosiopropanol (HFIP) and incubated overnight at room temperature. Afterwards, Ab(1-42) was mixed in the favored ratio with N-terminally/Cterminally biotinylated Ab(1-42) and aliquoted. HFIP was evaporated using a SpeedVac (Concentrator 5301, Eppendorf, Germany) at room temperature for 30 min. To evaporate remaining HFIP, the sample was incubated overnight at room temperature with an open lid.

Preparation of Ab(1-42) Monomers, Oligomers and Fibrils
The preparation of Ab(1-42) monomers and oligomers was carried out by size exclusion chromatography (SEC) as already described by Johansson et al [41] with minor modifications. 250 mg of Ab peptide was dissolved in 130 ml SEC buffer (50 mM NaPi, pH 7.4; 150 mM NaCl; 0.6% Tween-20) and briefly centrifuged (1 min, 16 1006g, room temperature) to remove insoluble fibrillar material. 100 ml of the supernatant was applied on a Superdex75 10/300 column (GE Healthcare, UK) connected to an Ä kta purifier system (GE Healthcare, UK). The sample was eluted at a flow rate of 0.6 ml per min at room temperature and detected at wavelengths of 214, 250 and 278 nm. Fractions of 200 ml were collected. For the preparation of Ab fibrils, 125 mg of Ab(1-42) peptide was dissolved in 200 ml PBS and incubated for 24 h (3006g, 37uC). Samples were centrifuged (16 1006g, 20 min, room temperature) and the supernatant containing soluble Ab species was discarded. Insoluble fibrils were resuspended in 200 ml SEC buffer. The concentration of the Ab preparations was determined using the Micro BCA Protein Assay Kit (Thermo Scientific, Waltham, MA, USA). Before performing the standard protocol recommended by the manufacturer, 80 ml 6 M urea was mixed with 80 ml monomer-and oligomer preparation, respectively. As the concentration of the fibril preparation was usually high, 80 ml 6 M Urea was mixed with 40 ml SEC buffer and 40 ml fibril fraction. The samples were incubated at 60uC for 30 min. Afterwards they were mixed with BCA reagent at a ratio of one to one and again incubated at 60uC for 20 to 30 min. Bovine serum albumin (BSA) at concentrations of 10, 20 and 40 mg per ml was used as a standard. The samples were measured at a wavelength of 570 nm with a Polarstar Optima plate reader (BMG, Ortenberg, Germany).

Enzyme-linked Immunosorbent Assay (ELISA) Analysis of scFv-IC16 Binding to Ab-monomers, -oligomers andfibrils
An ELISA was applied to characterize the binding affinity of scFv-IC16 to different Ab conformers. After concentration determination of the Ab(1-42) monomer, oligomer and fibril preparations, each conformer was diluted in immobilization buffer (0.1 M NaHCO 3 , pH 9.3) to a concentration of 5 mg/ml, and coated on Nunc immobilizer streptavidin F96 clear plates (250 ng/well) (Nunc, Thermo Scientific, Waltham, MA, USA). Alternatively, Ab monomers were prepared seedless in the same concentration by dissolving 50 mg N-terminally biotinylated Ab(1-42) in 1 ml immobilization buffer and diluting it further to 5 mg/ ml. N-or C-terminally biotinylated Ab(1-10) peptides were dissolved in immobilization buffer to a final concentration of 1.2 mg/ml (60 ng/well). 50 ml of these Ab solutions were incubated in the well for 1 h. Each well was washed with 100 ml immobilization buffer and the plate was stored at 4uC overnight in 100 ml immobilization buffer per well. Unspecific binding sites were blocked with 50 ml blocking buffer (PBS, 1% BSA) at room temperature for 1 h. After removal of the blocking buffer, a solution of either scFv-IC16 (12 and 23 nM in PBS-T (0.1% Tween-20) and 0.1% BSA respectively), IC16 (0.67 nM PBS-T and 0.1% BSA) or mAB-6E10 (0.67 nM PBS-T and 0.1% BSA) was added to each conformer and incubated for 1 h. Subsequently, each well was washed with PBS-T (three times, 100 ml) and incubated with Penta-His antibody (Qiagen, Hilden, Germany, dilution 1:500, in 2 ml TBS-T) for 1 h at room temperature. This was followed by another three washing steps with PBS-T and detection of the antibodies with HRP-conjugated anti-mouse IgG (Thermo Scientific, Waltham, MA, USA, dilution 1:10000, in 2 ml TBS-T). 3,3,5,5-tetramethylbenzidine (Sigma Aldrich, St. Louis, MO, USA) was used as the substrate for the HRP. The reaction was terminated by adding sulfuric acid to a final concentration of 1 M. The absorbance was recorded at 450 nm on Polarstar Optima plate reader (BMG, Ortenberg, Germany).

Immunoprecipitation of Cell Culture Derived Ab
Purified scFv-IC16 was dialyzed against PBS and coupled to NHS-Sepharose (GE, Buckinghamshire, UK) according to the manufacturers recommendations. Conditioned medium (CM) was prepared from confluent wild type CHO cells or 7PA2 cells cultured in DMEM (Life Technologies, UK) in presence or absence of FCS. Ab was immune-precipitated from cleared CM with scFv-IC16-NHS-Sepharose overnight at 4uC. After washing with PBS, captured proteins were separated by 10-20% tricine peptide PAGE (Biorad, Hercules, CA) and transferred onto a 0.2 mm nitrocellulose membrane at 400 mA for 2 h. The filter was boiled for 5 min in PBS [42] and blocked overnight at 4uC with 5% fat-free milk in PBS containing 0.05% Tween 20 (PBS-T). After washing in PBS-T, the membrane was probed with 1:200 diluted monoclonal 4G8 (Signet, Dedhem, MA). Bound antibody was detected with horseradish peroxidase conjugated goat antimouse Ig (at 1:25000) (Thermo Scientific, Bonn, Germany). Subsequently, the Amersham ECL Western Blotting Detection Reagent (GE, Buckinghamshire, UK) was applied for visualization.

Ex vivo Staining of Plaques in Mouse Brain Slices
Deep frozen horizontal brain cryosections (20 mm thickness) from 9 or 12 months old Tg2576 mice and the respective wildtype control were dried in ambient air for 15 min, and then fixed in icecold 4% (w/v) paraformaldehyde for 20 min. Incubation with 70% formic acid for 5 min was performed to enhance immunoreactivity. Sections were permeabilized with 1% Triton X-100 in TBS for 10 min, followed by blocking with Mouse on mouse

Results
Expression of scFv-IC16 in E. coli and Purification via Ni-NTA and Ab-affinity Chromatography To characterize scFv-IC16, the protein was produced in E. coli and purified via Ni-NTA and Ab-affinity chromatography. Expression was performed using the BL21DE3 pRARE2 strain. ScFv expression often results in formation of inclusion bodies [43,44]. Therefore, we have developed a protocol for slow expression, leading to an increase of correctly folded and soluble protein. Cells were shocked on ice for 1 h after reaching the stationary phase (OD 600 $1.6) in order to slow down the metabolism. Protein expression was induced with a final IPTG concentration of 0.1 mM. Under these conditions more than 10% of the expressed protein was soluble. Attempts to purify the remaining unsoluble protein after denaturation were successful and yielded pure protein after Ni-NTA affinity chromatography. However, refolding attempts remained unsuccessful. Purification of natively folded protein was performed using Ni-NTA affinity chromatography, followed by Ab affinity chromatography using Ab(1-16)-GB1 NHS Sepharose. IC16 was originally raised against Ab (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16). Therefore, the fusion protein contains the binding epitope of scFv-IC16. In addition, this step ensured that only active and correctly folded scFv was purified and thus used for further studies. The complete procedure resulted in 0.460.1 mg of pure protein as judged by SDS-PAGE (Fig. 1) and UV/Vis spectrometry, derived from 1 L cell culture.

Characterization of the Binding Site of scFv-IC16 at Ab(1-16) -Homology Modeling and Surface Plasmon Resonance Measurements
The amino acid sequence of the constructed scFv-IC16 was compared to antibody fragments that were already described in the literature. The IgG2a monoclonal anti-Ab protofibril antibodies PFA1 and PFA2 were derived from mice challenged with a stabilized protofibril form of Ab . The derived Fab fragments exhibit binding to Ab monomers in the nM range, but the binding to aggregated Ab forms is significantly impaired in comparison to the full IgG molecules. Structural characterization of the Fabs in complex with the Ab(1-8) peptide as well as binding studies revealed a significant influence of N-terminal Ab truncations on binding [33]. Our alignments exhibited sequence identities of 82% and 94% for the V H and V L regions of scFv-IC16 and the Fab fragment PFA1, respectively (Fig. 2, [33]). In addition, the complementary determining regions (CDR), which are hypervariable and mainly determine the binding properties of an antibody [45], showed high similarity of scFv-IC16 to PFA1 (see also Tables 1 and 2). Therefore, we assume that the binding of scFv-IC16 is similar to Ab as shown for PFA1 and Ab(1-8) by Xray crystallography. To understand the binding of scFv-IC16 to Ab and the differences to the binding of PFA1 to Ab in more detail, the structure of scFv-IC16 in complex with Ab(1-8) was modeled using the crystal structure of the Ab(1-8)-PFA1 complex [33] as a template. As expected from the high degree of sequence similarity, both antibodies form similar contacts with Ab(1-8) (Fig. 3). Differences are only detected for sequence positions 100 and 108 of the heavy chain, as well as 98 of the light chain. Of these three sites, only residue 108 forms significant interactions with Ab(1-8). In PFA1, D108(H) forms a salt-bridge with H6 of Ab, while V108(H) of IC16 forms nonpolar interactions with H6 (Fig. 3).
Our next aim was to confirm the prediction of the modeling. Consequently, the equilibrium constants (K D ) of scFv-IC16 to various N-terminal Ab fragments were determined using SPR. For comparison, we also performed the measurements with the monoclonal parent antibody IC16 ( Fig. 4 and 5, Table 3). The monoclonal antibody was expected to exhibit more structural stability caused by the constant domains. Additionally, an avidity effect may influence the binding constants of IC16 to the different Ab peptides. Measurements on both Ab(1-8) and Ab (8)(9)(10)(11)(12)(13)(14)(15) were used to confirm the localization of the linear binding epitope at the N-terminal. No interaction of IC16 and scFv-IC16 with Ab(8-15) was observed. For Ab(1-8) the injection curves showed a fast association rate of IC16 and a very slow dissociation rate. This result is reflected in the low K D of 51 nM (Table 3, bivalent fit). Comparison of the binding curves of IC16 and scFv-IC16 revealed that the association rates of scFv-IC16 and IC16 to Ab(1-8) were comparable, whereas the dissociation rate of scFv-IC16 was one order of magnitude higher than the corresponding IC16 dissociation rate. These results are reflected in the K D for scFv-IC16 (K D of 0.55 mM, Table 3, Langmuir 1:1 binding model), demonstrating a loss of affinity of the scFv in comparison to the monoclonal antibody, which may be due to the lost avidity of the scFv as compared to the monoclonal antibody.
A comparison of the affinities of both IC16 and scFv-IC16 for N-terminally truncated Ab-peptides (Table 3) showed that the highest affinities were observed for Ab (2)(3)(4)(5)(6)(7)(8). Deletion of A2 leading to Ab(3-8) resulted in significant reduction of affinity for scFv-IC16 ( Table 3). Truncation of the two N-terminal residues and modification of E3 into pyroE3 led to an extensive loss of affinity and no binding could be detected for scFv-IC16.

Influence of scFv-IC16 on Ab(1-42) Fibrilization
To characterize the influence of scFv-IC16 on Ab(1-42) fibrilization, ThioflavinT (ThT) assays were carried out. ThT is a benzothiazole dye, which exhibits a shift and increase in quantum yield while binding to b-sheet rich fibrils [46]. During the fibrilization process of Ab, an increase of ThT fluorescence can be observed until a saturation level is reached. Whereas the  Ab sample showed a strong increase in ThT fluorescence after 10 h, ThT assays with Ab and scFv-IC16 yielded a scFv-IC16 dose-dependent lower ThT signal (Fig. 6A), indicating a strong influence of scFv-IC16 on Ab fibril formation. With the ratio of 1:1 for Ab  and scFv-IC16 the fibrilization of Ab(1-42) was completely inhibited. At the ratios of 2:1 and 5:1, scFv-IC16 reduced the ThT fluorescence significantly. The sample that contained only scFv-IC16 was used as a negative control and did not show ThT fluorescence after 10 h.

ELISA Analysis of the Binding Activities of scFv-IC16 to Ab Monomers, Oligomers and Fibrils
The monoclonal antibody IC16 was raised against Ab (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16), and binding activity of scFv-IC16 to N-terminal Ab fragments was confirmed using SPR. Nonetheless, we intended to analyze the binding activities of scFv-IC16 to different Ab conformers in order to detect potential binding preferences. To perform ELISA studies, Ab monomers, oligomers and fibrils were prepared and equal molar amounts with respect to monomeric Ab were immobilized in wells of a 96 well plate. The commercially available monoclonal antibody 6E10 is known to bind to all Ab conformers with similar affinities as it recognizes human Ab(3-10) epitope, which is freely accessible in Ab monomers as well as in oligomers and fibrils [47]. It is frequently used as a standard in ELISA experiments [48,49,50] and was used here as a control to compare the immobilized amounts of Ab monomers, oligomers and fibrils. The different Ab conformers were incubated with scFv-IC16. Subsequently, bound scFv-IC16 was detected using an anti-His antibody that was then detected by an HRP conjugated secondary antibody. Fig. 6B depicts the relative quantification of the binding of scFv-IC16 and 6E10 to different Ab conformers. The 6E10 control demonstrated the presence of similar amounts of Ab monomers, oligomers and fibrils with respect to monomeric Ab in the wells of the microtiterplate. Furthermore, scFv-IC16 showed a preferential binding to Ab fibrils. The highest relative fluorescence was detected for fibril binding; the relative fluorescence for oligomer binding was weaker (p = 0.0014, N-bio oligomers to Nbio fibrils). Signals for monomer binding were not detected if Ab monomers were immobilized onto the streptavidin functionalized surface of the well via an N-terminal biotin tag, indicating interference of the N-terminal biotin tag with the binding of scFv-IC16 to 100% N-terminally biotinylated monomers. Ab monomers or Ab(1-10) fragments that were immobilized via a C-terminal biotin tag were bound by scFv-  IC16 very similar to Ab oligomers in the ELISA experiment, which can be shown by similar absorption values (p = 0.038, Cbio monomers to N-bio oligomers; p = 0.19, C-bio Ab1-10 to N-bio oligomers).
To confirm that scFv-IC16 also binds Ab that was naturally secreted by 7PA2 cells, scFv-IC16 was coupled to NHS-Sepharose and incubated with culture medium derived from wildtype CHO cells or 7PA2 cells. 7PA2 cells secrete high amounts of monomeric and lower oligomeric Ab species. Previously it was demonstrated that serum stabilizes the monomeric form of Ab, whereas oligomeric Ab accumulates in the absence of serum [51]. As demonstrated in Fig. 6C, scFv-IC16 efficiently precipitated monomeric Ab from 7PA2-CM containing FCS as well as lower oligomeric Ab species from CM of 7PA2 cells cultured without FCS. Thus, scFv-IC16 displays the same binding specificity as the monoclonal antibody IC16 [36].

Ex vivo Staining of Plaques in Mouse Brain Slices
Brain cryosections derived from nine or 12 months old transgenic (tg 2576) mice and the respective non-transgenic control were stained using scFv-IC16. Additionally, anti-Ab plaque stainings (using the 6E10 antibody) were performed on adjacent sections as well as costainings with scv-IC16 and 1-11-3 anti Ab(1-42) antibody. DAPI nuclei counterstainings were performed on the same slides, respectively. Photomicrographs of the stained slices and overlay images are shown in Figure 7. scFv-IC16 could readily be used to detect plaques specifically in AD transgenic mice with low background signals, demonstrating potential applicability as a molecular probe. The staining pattern of scFv-IC-16 was very similar to the staining pattern of 6E10, which also binds to the N-terminal part of Ab and the periphery of the plaques. The staining pattern of 1-11-3 is slightly different as the core of the plaques is stained with higher intensity. Ab(pE3-8) n.d.

Discussion
Here we have characterized a novel scFv with respect to its binding specificities to different Ab conformers and its influence on Ab fibrilization. This scFv-IC16 was constructed from an IgG2a, designated IC16, which was raised in mouse against the antigen Ab(1-16) and binds to Ab(2-8) [36].
scFv-IC16 shares 87% sequence identity with of PFA1 and PFA2 (see Figure 2), two Ab binding Fab fragments described by Gardberg et al. [33]. Especially, the three CDRs of scFv-IC16 are highly similar (see Tables 2 and 3), CDRs 1 of the heavy chain are 100% identical and CDRs 1 and CDRs 2 of the light chain are 91% and 100% identical, respectively. The striking WWDDD motif within CDR2 of PFA1 and PFA2 that recognizes the Ab binding motif EFRH (Ab(3-6)) [33,52] is almost completely present in scFv-IC16, too: FWDDD. Because of the high sequence homology between scFv-IC16 and PFA1, we assumed similar interactions between scFv-IC16 and Ab as described for PFA1 and Ab. Therefore, a homology modeling for the scFv-IC16-Ab complex based on the X-ray structure of the Fab PFA1 and Ab (1)(2)(3)(4)(5)(6)(7)(8) was performed (Fig. 3). The model yielded similar binding modes for scFv-IC16 to Ab(1-8) and PFA1 to Ab (1)(2)(3)(4)(5)(6)(7)(8). However, amino acid D1 of Ab was not resolved in the X-ray structure of the Ab(1-8)/PFA1 complex. Our homology modeling placed it in close spatial proximity with the amino acid D1 of the light chain. This close spatial proximity may generate a repulsive force and thus explains the increased affinity (K D ,112 nM) of scFv-IC16 to Ab(2-8) as compared to Ab(1-8) (Tab. 3). The model furthermore offers an explanation why the K D rises to more than 5 mM when the second amino acid A2 of Ab is truncated. A2 forms a hydrogen bond with the backbone amide group of V99 of the heavy chain and is therefore important for a tight binding. Deletion of residues one and two and modification of Ab E3 into pyroE3, as found in several truncated and more aggregation prone Ab species in the diseased brain [53,54], showed a decrease in affinity and no binding was observed for scFV-IC16.
In comparison, PFA 1 shows the strongest binding to Ab(1-40) monomers (K D ,39 nM). The binding to Ab(2-7) is in the same order of magnitude (K D ,60 nM). Similar to scFv-IC16, the binding of PFA1 to Ab(pE3-8) is decreased, which is reflected in the K D of ,3 mM [33,52]. pE3 lacks the E3 sidechain carboxyl group that forms two hydrogen bonds with S32(H) in the complex crystal structure.
Due to the high sequence similarity between PFA1 and scFv-IC16, similar interactions between scFv-IC16 and E3 of Ab are also present in our modeled complex structure. Therefore, a loss of the interactions described above may lead to a loss in affinity, so that an interaction between scFv-IC16 and Ab(pE3-8) could not be detected.
Altogether, the binding of scFv-IC16 to the Ab fragments is weaker in comparison to PFA1. This may be due to small differences in the binding pocket like the D108V replacement (Fig. 3), but also to the fact that PFA1 is a Fab fragment, consisting of the variable and one constant domain of the heavy chain and the variable and the constant domain of the light chain. The two constant domains may provide more stability for the binding pocket of PFA1 compared with the binding pocket of scFv-IC16, because its two variable domains are connected only via a peptide linker lacking the stability of the two constant domains.
In a series of ELISA experiments, the specificity of scFv-IC16 to different conformers of Ab, monomers, oligomers and fibrils, was characterized. The non-conformation specific monoclonal antibody 6E10 was used to control the relative amount of bound Ab conformer in the wells. 6E10 was previously characterized to bind an Ab N-terminal epitope at residues 3 to 10. Similar ELISA readouts for 6E10 and Ab monomers, oligomers and fibrils proved that all three conformers were loaded in similar amounts into the wells. The binding of scFv-IC16 to its epitope at residues 2-8 of Ab monomers that were immobilized using an N-terminal biotin tag was sterically hindered. As soon as the N-terminus was freely accessible, e.g. in C-terminally biotinylated Ab monomers, scFv- Analysis of an ELISA experiment to quantify the binding specificity of scFv-IC16 to different Ab1-42 conformers and Ab1-10 peptides (B). 250 ng Ab1-42 (100% N-biotinylated monomers, 100% C-terminally biotinylated monomers, 10% N-biotinylated oligomers and fibrils) and 60 ng Ab peptides 1-10 (100% C-terminally biotinylated Ab1-10, 100% N-terminally biotinylated Ab1-10) were immobilized on streptavidin coated 96 well plates. 6E10 was used to monitor the immobilized amount of Ab conformers. Similar absorption values of 6E10 to the different Ab conformers except for 100% N-terminally biotinylated monomers, indicate similar molar amounts of immobilized Ab regarding monomeric Ab for monomers, oligomers and fibrils. Background absorption was subtracted from all samples. A highly significant difference in the relative fluorescence value between Nbiotinylated monomers and C-biotinylated monomers, as well as between C-bio Ab1-10 and N-bio Ab1-10 was calculated by Student's t-test (***: p,0.001). N-bio = N-terminally biotinylated; C-bio = C-terminally biotinylated. Western Blot of Ab monomers and low-n oligomers immunoprecipitated by scFv-IC16 from conditioned medium (CM) of CHO or 7PA2 cells (C). The latter were grown either in medium with or without FCS. Monomeric Ab is stabilized by the presence of FCS, whereas low-n oligomeric Ab is eriched in CM-FCS. No Ab was precititated from the supernatants of CHO cells. Monomeric Ab (CM 7Pa2+ FCS) and low-n oligomeric Ab (CM 7PA22 FCS) were equally bound by scFv-IC16. Detection antibody: 4G8. doi:10.1371/journal.pone.0059820.g006 IC16 bound Ab monomers with the same affinity as oligomers, as judged by the ELISA results. A slightly higher ELISA signal was detected for the binding to Ab fibrils. Given the fact that scFv-IC16 binds to residues 2-8 of the N-terminus of Ab, specificity for a special Ab conformer was not expected. In aggregated Ab conformers, the N-terminus of Ab (Ab(1-9)) was described to be disordered and accessible, whereas residues 10-22 and 30-40 adopt a b-strand conformation and are involved in fibrilization [55]. As both, scFv-IC16 and 6E10, bind to the N-terminus of Ab, a similar binding pattern after usage for ex-vivo plaque staining in AD transgenic mice was expected. A significantly higher concentration of scFv-IC16 was needed to perform proper plaque staining in comparison to the monoclonal antibody 6E10. The need of higher amounts of protein, however, can be balanced by the fact that the scFv can be expressed in E. coli and produced in reasonable amounts at low costs. An interesting option could also be to enhance the affinity of scFv-IC16 to Ab by rational design or directed evolution methods. In conclusion, scFv-IC16 has several interesting features. It can be produced recombinantly, a fact that also provides possibilities for sequence optimization. scFv-IC16 prevents the formation of ThT positive Ab(1-42) fibrils. It recognizes all conformers of Ab as it binds to a linear N-terminal epitope. Furthermore, it is suitable as a molecular probe, which was demonstrated by ex vivo plaque staining in brain slices of AD transgenic mice. Therefore, scFv-IC16 is interesting for therapeutic, imaging and mechanistic studies. Figure 7. In vitro staining of brain sliced from transgenic 2576 AD mice using scFv-IC16, 6E10-Ab-antibody and DAPI. (A). Pretreated deep frozen horizontal brain cryosections (20 mm thickness) from 9 months old Tg2576 mice were fixed in 4% paraformaldehyde, treated with 70%formic acid and either incubated with scFv-IC16 in a concentration of 400 nM or 6E10 in a concentration of 33 nM. As scFv-IC16 is fused with a His tag, the respective sections were incubated with Penta-His Mouse antibody. Detection was performed using goat anti mouse-Alexa488 using a fluorescent microscope (Leica LMD6000. Excitation Range: UV; blue; green. Excitation Filter: Bp 420/30; Bp 495/15; Bp 570/20) with camera (Leica DFC 310 FX). Images were processed with LAS software (Leica Application Suite, V.4.0.0) and ImageJ (1.45 s). Scale bar: 50 mm. B: In vitro staining of brain sliced from transgenic 2576 AD mice using scFv-IC16 and a respective wildtype control mouse. Scale bar:100 mm. C: Co-staining of brain sliced from transgenic 2576 AD mice using scFv-IC16 and anti Ab(1-42) 1-11-3. As secondary antibodies, a mix of goat anti mouse-Alexa488 and goat anti rabbit-Alexa-568 was used. Scale bar: 50 mm. doi:10.1371/journal.pone.0059820.g007