Validation and Characterization of a Novel Peptide That Binds Monomeric and Aggregated β-Amyloid and Inhibits the Formation of Neurotoxic Oligomers*

Although the formation of β-amyloid (Aβ) deposits in the brain is a hallmark of Alzheimer disease (AD), the soluble oligomers rather than the mature amyloid fibrils most likely contribute to Aβ toxicity and neurodegeneration. Thus, the discovery of agents targeting soluble Aβ oligomers is highly desirable for early diagnosis prior to the manifestation of a clinical AD phenotype and also more effective therapies. We have previously reported that a novel 15-amino acid peptide (15-mer), isolated via phage display screening, targeted Aβ and attenuated its neurotoxicity (Taddei, K., Laws, S. M., Verdile, G., Munns, S., D'Costa, K., Harvey, A. R., Martins, I. J., Hill, F., Levy, E., Shaw, J. E., and Martins, R. N. (2010) Neurobiol. Aging 31, 203–214). The aim of the current study was to generate and biochemically characterize analogues of this peptide with improved stability and therapeutic potential. We demonstrated that a stable analogue of the 15-amino acid peptide (15M S.A.) retained the activity and potency of the parent peptide and demonstrated improved proteolytic resistance in vitro (stable to t = 300 min, c.f. t = 30 min for the parent peptide). This candidate reduced the formation of soluble Aβ42 oligomers, with the concurrent generation of non-toxic, insoluble aggregates measuring up to 25–30 nm diameter as determined by atomic force microscopy. The 15M S.A. candidate directly interacted with oligomeric Aβ42, as shown by coimmunoprecipitation and surface plasmon resonance/Biacore analysis, with an affinity in the low micromolar range. Furthermore, this peptide bound fibrillar Aβ42 and also stained plaques ex vivo in brain tissue from AD model mice. Given its multifaceted ability to target monomeric and aggregated Aβ42 species, this candidate holds promise for novel preclinical AD imaging and therapeutic strategies.

model mice. Given its multifaceted ability to target monomeric and aggregated A␤42 species, this candidate holds promise for novel preclinical AD imaging and therapeutic strategies.
Alzheimer disease (AD) 4 is a progressive neurodegenerative disorder, accounting for 50 -70% of late-onset dementia cases (2). The major biochemical hallmarks of AD are extracellular amyloid plaques and intracellular neurofibrillary tangles. The predominant species present in amyloid plaques is ␤-amyloid (A␤), a 4.5-kDa peptide generated through amyloidogenic processing (sequential cleavage by ␤and ␥-secretases) of the parent amyloid precursor protein (APP). This produces two forms of A␤, comprising either 40 or 42 amino acids, where the relative amount of the longer 42-amino acid form (A␤42) is especially critical for AD progression, given its higher propensity to aggregate and form neurotoxic species (3,4).
In AD brain, the monomeric A␤42 peptide is known to aggregate and form various ordered assemblies, which precede plaque formation. These include low-n oligomers (dimers to octamers, reviewed in Ref. 5), high molecular weight oligomers such as A␤-derived diffusible ligands (6) and globulomers (7), protofibrils (8), and fibrils (9). Much evidence has indicated that soluble A␤42 oligomers, rather than mature amyloid plaques, correlate with disease severity (10,11) and contribute to synaptic degeneration and neurotoxicity (12,13). In particular, recent work has highlighted a role for dimers, trimers, and dodecamer forms of A␤ oligomers in neuronal dysfunction (see Refs. 14 and 15 and reviewed in Ref. 5). * This work was supported in part by the National Health and Medical The accumulation of A␤ is thought to occur early in the disease process. Studies in transgenic mice show that cognitive deficits are associated with small A␤ oligomeric assemblies (16,17), which precede amyloid deposition and promote Tau phosphorylation and tangle formation (18,19). Further evidence for a pivotal role for A␤ oligomers in the neurodegenerative process is provided by the development of transgenic mice expressing oligomers, which show synaptic degeneration, Tau phosphorylation, and neuronal loss in the absence of plaques (20). Although these findings suggest interventions targeting A␤ oligomers have the potential to halt disease progression, oligomers have proved to be difficult therapeutic targets. This is partly due to the complex inter-conversion between different A␤42 assemblies and the fact that A␤42 oligomers of different structure, stability, and concentration may all be neurotoxic (reviewed in Ref. 21).
We have previously reported a novel 15-amino acid peptide isolated by subtractive phage display screening, which bound the toxic human A␤42 peptide, but not the related, non-toxic rat A␤42 peptide. This peptide was shown to target A␤42 and attenuate its neurotoxicity (1). In an effort to generate a novel, specific and stable candidate peptide capable of targeting A␤42 oligomers, we have exploited the known ability of this peptide to target A␤42 and aimed to increase its utility using two approaches: (i) reduced length for improved permeability across the blood-brain barrier and peptidomimetic design; and (ii) increased proteolytic resistance. The stability of these analogues was assessed in vitro and their A␤42 binding and activity was assessed biochemically. We have further provided proofof-principle experiments to demonstrate that the lead candidate peptide shows potential as a preclinical AD imaging agent, given its ability to bind A␤42 oligomers in vitro, amyloid plaques ex vivo, and its potential for transport across the bloodbrain barrier in vivo.
In Vitro Assay of Peptide Stability-Peptides were prepared as either 1 or 4 mM solution in PBS. 20 l of the peptide solution was diluted in either 10% rat brain homogenate (in PBS ϩ 0.5% Triton X-100) or 100% serum. The solution was incubated at 37°C for different times, and the reaction was stopped by adding cOmplete protease inhibitor mixture (Roche Molecular Biochemicals, Mannheim, Germany). For 9-mer and 9M S.A., the bulk of the brain proteins (but not the peptides) were precipitated in cold methanol (1:4 (v/v) mixture/methanol) for 1 h at Ϫ20°C. The precipitated proteins were pelleted by centrifugation (10,000 ϫ g, 10 min, 4°C). The supernatant containing the peptide was concentrated five times under vacuum and separated by reversed-phase HPLC (RP-HPLC). Due to recovery issues with the longer 15-mer and 15M S.A. peptides, samples were instead lyophilized, extracted in TFA, centrifuged to remove insoluble material, and separated by RP-HPLC. The area of the peak (UV absorbance at 205 nm) corresponding to the intact peptide was measured and compared with an equivalent sample incubated in PBS.
Thioflavin T Assays-This method was adapted from Ref. 23. Briefly, A␤42 oligomers were centrifuged (18,000 ϫ g, 4°C, 1 min) to pellet insoluble material. The clarified supernatant (4.5 g/10 l of 100 M stock) was added to a black-walled, clear bottom 96-well microplate (PerkinElmer Life Sciences) in triplicate. 200 l of thioflavin T (ThT) (5 M in 50 mM glycine NaOH, pH 8.5, 0.22 M filtered) was added and the plates were read at 3-5 min post-addition in a FLUOSTAR OPTIMA instrument (excitation filter: 450 nm; emission filter: 490 nm; 30 s mix before reading; gain-adjust to highest reading). Samples were assayed in triplicate and the blank ThT fluorescence was subtracted from all readings. Candidate peptides were also assayed in the absence of A␤42 for interference in the assay.

SDS-PAGE and Western
Immunoblotting-Samples for gel analysis were prepared in two ways, i.e. denaturing and nondenaturing conditions. Non-denatured samples were separated by electrophoresis using a modified Blue Native-PAGE protocol, as described in Ref. 24. Samples were diluted in MOPS loading buffer (50 mM MOPS, 50 mM Tris, 20% glycerol, 0.05% Coomassie, pH 7.7) without reducing agent and were not heat denatured prior to PAGE separation. Where insoluble material was present in the samples, denaturing conditions were employed to solubilize this material. In the denaturing conditions, samples were diluted in LDS sample buffer plus reducing agent (NuPAGE), and heat denatured (72°C, 10 min) prior to separation of proteins by SDS-PAGE using NuPAGE Novex 4 -12% BisTris gradient gels and MES SDS running buffer (Life Technologies). Samples were transferred onto either nitrocellulose (denatured samples) or PVDF (non-denatured samples) membranes using the iBlot blotting system (Life Technologies). Samples were subjected to Western immunoblotting with WO2 antibody (25) (kindly provided by Prof. Colin Masters, University of Melbourne, Australia), to detect A␤ species. Bands were visualized using enhanced chemiluminescence (Amersham Biosciences ECL) and exposure to Hyperfilm ECL films (GE Healthcare). Films were scanned on a Bio-Rad GS-800 densitometer and densitometry was conducted with Bio-Rad Quantity One software.
Atomic Force Microscopy (AFM)-Samples for AFM were prepared and analyzed according to Ref. 22. Briefly, A␤ samples were spotted on freshly cleaved grade V1 muscovite mica and incubated for 5 min. Samples were then rinsed with 0.22-m syringe-filtered double-distilled water and blow dried with several gentle pulses of compressed air (N 2 ). Samples were then visualized under the AFM (NT-MDT) using semi-contact mode with the following parameters: a minimum contact force, amplitude between 0.5 and 2 V (depending on the cantilever) to generate magnitude ϳ20 nA, and scan rates ϳ0.5-1 Hz. All data were processed using Nova NT-MDT Software version 1.1.0.1780.
Surface Plasmon Resonance/Biacore Assays-These experiments were performed using a Biacore 3000 Instrument (GE Healthcare) with a protocol modified from Ref. 26. Either 15M S.A. peptide, or A␤42 monomer/aggregates, were immobilized on separate CM5 sensorchips (GE Healthcare). The 15M S.A. peptide was immobilized on a CM5 sensorchip via amine coupling, according to the manufacturer's instructions, to a level of 245 resonance units (RU) (1 RU ϭ 1 pg of protein/mm 2 ). Monomeric, oligomeric, and fibrillar A␤42 preparations were immobilized on another CM5 sensorchip (10 M in 10 mM sodium acetate buffer (pH 4.0), injection for 5 min at a flow rate of 30 l/min) to final levels of 9,084, 11,467, and 4,924 RU, respectively. Note that the oligomeric A␤42 and fibrillar A␤42 preparations were prepared 24 h prior to immobilization, whereas the monomeric A␤42 was prepared immediately prior to immobilization to minimize its aggregation. For both sensorchips, a reference surface was prepared in parallel (with no addition of peptide), and used for subtraction of nonspecific binding. Sensorgrams were obtained using standard conditions of 30 l/min flow rate and HBS-EP running buffer (0.01 M HEPES, pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% (v/v) surfactant P20 (GE Healthcare)) as outlined under "Results." Coimmunoprecipitation Analysis-Alexa 488-labeled oA␤42 was prepared as described in Ref. 27 and incubated with TMRlabeled 15M S.A. peptide in 300-l reactions in TBS ϩ 0.05% Tween 20 (TBST) with gentle rotation for 16 h at 4°C. Complexes were captured using 1 g of 6E10 antibody (Covance) for A␤ (2 h, 4°C) and then ␥-bind-Sepharose/Protein G-Mag-Sepharose (GE Healthcare) (2 h, 4°C). Following removal of the supernatant, the beads were washed (3 ϫ 500 l of TBST) and samples were separated by denaturing SDS-PAGE on 4 -12% BisTris NuPAGE gels (Life Technologies). Fluorescently labeled peptides were visualized by in gel fluorescence using a Typhoon FLA 9500 imaging system (GE Healthcare).
Immunohistochemistry-Brains were obtained from 8-monthold 5xFAD or TgCRND8 AD model mice (28) or age-matched non-transgenic controls and post-fixed and embedded in the freezing medium as described in Ref. 29. Serial sagittal cryosections were cut at a thickness of 10 m. Positive control staining for amyloid plaques was performed using thioflavin S (Thio S) as described in Ref. 30. After the removal of the freezing medium by serial changes of TBS (pH 7.4), the slide was incubated for 1 h at room temperature in blocking buffer (10% goat serum in TBS). The slides were then stained with 1 M TMR-labeled 15M S.A. peptide in 2% goat serum/PBS for 2 h at room temperature. Excess unbound peptide was removed prior to mounting by extensive washing with TBS. The coverslip was mounted using Prolong Gold mounting media (Life Technologies).
Brain Uptake of 3 H-Labeled 15M S.A. following Intravenous Administration-Animal experiments were approved by the Monash Institute of Pharmaceutical Sciences Animal Ethics Committee and were performed in accordance with the Australian National Health and Medical Research Council (NHMRC) guidelines for the care and use of animals for scientific purposes. Male Swiss Outbred mice (6 -8 weeks of age; 25-30 g) were used in the studies and had free access to food and water during the experimental periods. An aliquot (50 l) containing 10 Ci of 3 H-labeled 15M S.A. peptide was administered to mice by tail vein (intravenous) injection. Brain and plasma samples were collected over a 0.5-4-h period and the concentration of 15M S.A. peptide in plasma and brain homogenate was performed by liquid scintillation counting (Tri-Carb 2800 TR; PerkinElmer Life Sciences). The brain concentrations were corrected by subtracting the brain microvascular volume (0.035 ml/g) using [ 14 C]sucrose as a vascular marker (31). As described previously (32), the brain to plasma ratio of 3 H-labeled 15M S.A. peptide was determined using the following formula: (corrected number of disintegrations per minute (dpm) per g of brain tissue)/(number of dpm per ml of plasma). The intactness of 3 H-15M S.A. at post-dose time points following intravenous administration was assessed by HPLC. Briefly, brain samples were homogenized in a volume of MilliQ water (in ml) equal to twice the weight (in g) of the tissue using a glass rod. To 300 l of brain homogenate or 100 l of plasma, the same volume of acetonitrile was added prior to centrifugation at 14,100 ϫ g for 5 min. An aliquot (100 l) of the supernatant was then loaded onto a Waters Symmetry C 18 column (5 ϫ 4.6 mm). Mobile phase A consisted of 0.1% (v/v) TFA in MilliQ water and mobile phase B consisted of 60% (v/v) acetonitrile in 0.1% (v/v) TFA in MilliQ water. 3 H-15M S.A. peptide was analyzed using the following gradient profile: 0 min, 95% A; 0 -10 min, 70% A; 10 -12 min, 95% A. The eluant from the column was collected every 0.5 min and the dpm of each fraction was measured by liquid scintillation counting. The profiles (dpm versus time) of brain and plasma samples were then generated and compared with those of 3 H-15M S.A. peptide solution as control.

Results
In Vitro Stability of Analogue Peptides with Increased Therapeutic Potential-To increase the membrane permeability and potential for peptidomimetic mimicry, we used in silico modeling to reduce the size of the parent 15-mer to a 9-mer. In addition, we employed modifications to both the parent 15-mer and the 9-mer to increase their resistance to proteolytic degradation, generating candidates denoted as 15M S.A. and 9M S.A., 5 respectively. The in vitro stability of these peptides was measured following incubation in dilute rat brain homogenate or serum and HPLC quantification. For brain homogenate, we found that trace amounts of unmodified 15-mer and none of the unmodified 9-mer were present at t ϭ 30 min (Table 1). In contrast, the concentrations of both stable analogue peptides remained relatively unchanged following an extended incubation of 300 min (Table 1). Similar findings were observed in 100% serum, where compared with unmodified peptide concentrations the stable analogues remained relatively unchanged across the incubation time points (Table 1). These results indicate that modifications to the 9-mer and 15-mer peptides reduced degradation and thus improved stability. These stable analogues were investigated further to determine whether they retained the activity of the parent peptides.
Effects on A␤42 Aggregation and Neurotoxicity-Given the neurotoxicity attributed to oligomeric A␤42, we next assessed the ability of the stable analogues to influence their formation and toxicity. We incubated monomeric A␤42 peptide under conditions specifically favoring oligomerization for 24 h (22), in the presence or absence of the candidate peptides. We then assayed the A␤42 aggregation in these samples using ThT fluorescence assays and Blue Native-PAGE/Western blotting. We additionally diluted samples to 20 M A␤42 concentration, treated M17 neuroblastoma cells, and quantified neurotoxicity following 4 days of treatment. ThT analysis revealed that both the parent 15-mer and 15M S.A. peptide reduced ThT fluorescence (and thus A␤42 aggregation and oligomerization) in a dose-dependent manner and with similar potency (Fig. 1A). In comparison, the 9-mer peptide showed similar activity, but reduced potency, and the 9M S.A. peptide showed overlapping activity with a scrambled control peptide (CTL1) (Fig. 1A). Similar findings were observed when samples were analyzed by Blue Native-PAGE and Western blotting (Fig. 1B), where the presence of 15-mer and 15M S.A. resulted in a reduction in the A␤42 aggregation "smear" evident at the 1:5 and 1:10 ratios of A␤ to peptide. The LDH assay indicated that a reduction in A␤42 aggregation correlated with reduced neurotoxicity, with the parent 15-mer and 15M S.A. reducing neurotoxicity in a dose-dependent manner and with similar potency (Fig. 1C). In contrast, the 9-mer offered less potent neuroprotection and the activity of the 9M S.A. peptide was again similar to the scrambled control peptide (Fig. 1C). The neuroprotection offered by the candidate peptides was also confirmed using MTS assays, which indicated increases in cell viability in line with the reduction in toxicity evident in the LDH assays. 6 Together these findings indicated that the 15M S.A. candidate most effectively mimicked the activity and potency of the parent 15-mer peptide. Thus, this stable analogue was further characterized to provide insight into how it attenuates A␤42-induced neurotoxicity.
15M S.A. Promoted the Formation of Non-toxic, Insoluble Aggregates during A␤42 Oligomerization-The ThT and Western blotting analysis described above indicated that the 15M S.A. peptide reduced the formation of soluble A␤42 oligomers (Fig. 1, A and B). However, we noticed a corresponding increase in the formation of non-toxic, insoluble material when A␤42 was incubated in the presence of the 15M S.A. candidate. We hypothesized that the reduction in soluble A␤42 may be concomitant with an increase in insoluble A␤42 and used denaturing SDS-PAGE and Western blotting analysis to assess the relative amounts of soluble versus insoluble A␤42 species. As in Fig. 1B, we found that the presence of 15M S.A. resulted in a dose-dependent reduction in soluble A␤42 aggregates ( Fig. 2A,  left panel), whereas the control peptide resulted in similar soluble A␤42 aggregates as the A␤42 only sample ( Fig. 2A, left  panel). Increasing concentrations of 15M S.A. peptide resulted in increasing amounts of insoluble material, whereas no insoluble deposits were seen for the A␤42 only sample, or in the presence of the control peptide. Western blotting revealed that A␤42 aggregates were indeed present in the insoluble deposits, and the dose-dependent reduction in soluble A␤42 aggregates in the presence of 15M S.A. was concurrent with an increase in insoluble A␤42 aggregates ( Fig. 2A, right panel). The amounts of insoluble A␤42 generated in the presence of the peptides were further characterized using denaturing SDS-PAGE, followed by Western blotting and densitometric analysis. Soluble A␤42 was measured to quantify the effect as A␤42, which does not remain in the soluble fraction is by definition insoluble and measurements made from the soluble fraction have greater  accuracy and precision; measurements made from the insoluble fraction display elevated non-sedimenting background and variance, attributable to small amounts of residual/surface contamination after supernatant removal from the pellet (where present). As before, dose-dependent effects of the 15-mer and 15M S.A. peptides were observed, with increasing amounts shifting A␤42 of the soluble fraction and into the insoluble fraction. The maximal effect was seen in the presence of 15M S.A. peptide at the 1:10 ratio, where one-fourth of A␤42 remained in the soluble fraction (Fig. 2C). The amount of insoluble material present was also assayed, and whereas, as expected, non-sedimenting background was observable, the amount of A␤42 in this fraction increased in line with decreases seen in the soluble fraction and the presence of observable pellets of insoluble material (Fig. 2C).
The utility of denaturing SDS-PAGE for analysis of A␤ assemblies is somewhat limited, given that most protein complexes and oligomers dissociate when treated with SDS, and thus observation of monomers and low-molecular weight oligomers by SDS-PAGE does not reveal whether larger assemblies existed prior to SDS addition (33). Thus, we chose to better characterize the soluble and insoluble A␤ assemblies formed in the presence of 15M S.A. using AFM. We found that A␤42 oligomers formed in the absence of 15M S.A. were predominantly 1.5-4 nm in diameter, with a few larger aggregates 8 -12 nm in diameter (Fig. 2B, left panel). However, A␤ assemblies formed in the presence of 15M S.A. ranged from smaller (2-6 nm diameter) to extreme (25-30 nm) size (Fig. 2B, center  panel). The absence of aggregates in the presence of 15M S.A. only (Fig. 2B, right panel) suggests that aggregation of this peptide does not contribute to formation of larger A␤ assemblies. Taken together, it appears that the 15M S.A. peptide binds A␤42, alters its oligomerization, and promotes the formation of non-toxic amorphous aggregates.
15M S.A. Binds Pre-formed A␤42 Oligomers, as Shown by Surface Plasmon Resonance/Biacore and Coimmunoprecipitation Assays-The above assays indicated that 15M S.A. could influence the formation of A␤42 oligomers. We next extended these findings by demonstrating that 15M S.A. could also interact with pre-formed A␤42 oligomers (oA␤42) using two complementary measures of protein-protein interactions; surface plasmon resonance/Biacore assays and coimmunoprecipitation analysis. For these assays, we utilized two stable analogue control peptides (CTL1 S.A. and CTL2 S.A., "Experimental Procedures"), which were validated in ThT and LDH assays and found to mimic the activity of the CTL1 peptide used in our previous assays. 6 For surface plasmon resonance assays, 15M S.A. was immobilized on a Biacore sensorchip and oA␤42 concentrations over 5-40 M were injected and monitored for binding to 15M S.A. It was evident that oA␤42 bound 15M S.A. in a dose-dependent manner (Fig. 3A). To confirm the specificity of this interaction, we next performed solution competition assays where free 15M S.A. peptide or control peptides were co-injected with oA␤42. For these studies, we used an oA␤42  concentration of 20 M, which was found to give reproducible responses in replicate injections over multiple cycles in a given experiment. We found that increasing concentrations of free 15M S.A. in solution resulted in a dose-dependent reduction in oA␤42 binding the immobilized 15M S.A. on the sensorchip (Fig. 3B). However, equivalent concentrations of two different control peptides could not mimic this action and did not reduce the amount of oA␤42 binding to immobilized 15M S.A. (Fig.  3B). To confirm these findings, we performed complementary studies involving coimmunoprecipitation analysis. Here, Alexa 488-labeled oA␤42 was incubated with TMR-labeled 15M S.A. peptide to promote the formation of a protein complex, which was captured by immunoprecipitation using 6E10 antibody to bind oA␤ species. Denaturing SDS-PAGE was used to separate the protein complexes and oA␤42 and 15M S.A. were visualized via their respective fluorescent labels. The 15M S.A. peptide coimmunoprecipitated with oA␤42 in a dose-dependent manner (Fig. 3C). Furthermore, the amount of labeled, coimmunoprecipitated 15M S.A. peptide could be reduced by competition in solution with increasing quantities of unlabeled 15M S.A. peptide (Fig. 3C). These findings confirmed the results of the surface plasmon resonance assays and collectively indicate that the 15M S.A. peptide directly interacts with oA␤42.
We extended the surface plasmon resonance analysis to obtain an estimate of the affinity of the interaction between 15M S.A. and oA␤42, by fitting the experimental data to models within the BiaEvaluation v4.1 software. It is crucial to highlight that this value is only an overall estimate of binding affinity across the entire spectrum of oligomers present in the oA␤42 preparation, which is a heterogeneous mixture of aggregates. The best fit occurred for a 1:1 binding model with drifting baseline correction ( 2 ϭ 0.214) and yielded a K D value of low micromolar value (11 M), indicating a moderate affinity between the 15M S.A. peptide and oA␤42, and similar to other reports of A␤-peptide interactions (34).
15M S.A. Binds Monomeric, Oligomeric, and Fibrillar A␤42 in Biacore Assays-Thus far, we had demonstrated that 15M S.A. could influence oligomer formation and directly bind pre-formed oA␤42. We further investigated the ability of this peptide to bind less aggregated A␤ preparations (monomeric (m) A␤42) and more aggregated preparations (fibrillar (f) A␤42). To allow a comparison of 15M S.A. binding to different A␤ species, a sensorchip was generated with immobilized monomeric, oligomeric, and fibrillar A␤42 preparations as described under "Experimental Procedures." A series of 15M S.A. concentrations were injected across the surfaces, monitored for binding to the respective A␤42 species, and the data were corrected for the relative amount of immobilized material on the individual flow cells. We found that 15M S.A. interacted with all of the A␤42 species in a concentration-dependent manner, but the highest magnitude of binding was seen for the A␤42 fibrils (Fig.  4). This most likely reflects the fact that the A␤42 fibrils have multiple binding sites for 15M S.A., given that they are composed of repeating units of monomeric A␤42. Overall, this data shows that the 15M S.A. peptide is able to bind unaggregated, monomeric A␤42 (perhaps accounting for its earlier effects ( Fig. 1) in reducing the formation of oA␤42) as well as oligomeric and fibrillar A␤42 assemblies. The binding of 15M S.A. to more mature amyloid species was investigated further below.
15M S.A. Can Detect Amyloid Plaques in ex Vivo Immunohistochemical Staining of AD Brain Slices-Given the ability of 15M S.A. to interact with A␤42 fibrils, we extended this finding to investigate whether TMR-labeled 15M S.A. could be used to stain amyloid plaques ex vivo using brain tissue sections from AD model mice, in comparison with thioflavin S, which binds mature amyloid deposits and readily detects these plaques. In ex vivo staining of brain tissue from 8-month-old 5xFAD AD model mice, thioflavin S staining revealed extensive plaques within the brain (Fig. 5A). In comparison, serial sections treated with the TMR-labeled 15M S.A. peptide also resulted in staining of some amyloid deposits, but to a lesser extent than thio- A␤42 was incubated under conditions favoring formation of monomers (mA␤42), oligomers (oA␤42), or fibrils (fA␤42) as described under "Experimental Procedures." These were immobilized on individual flow cells of a CM5 sensorchip, where an unmodified surface was reserved for blank-subtraction. A series of concentrations of 15M S.A. peptide were injected across these surfaces and monitored for binding. In each case, the binding response was corrected for the relative quantity of immobilized material (RU bound/RU immobilized). 15M S.A. bound all of the immobilized A␤42 assemblies, with the highest binding response observed for A␤42 fibrils. flavin S (Fig. 5B). Notably, the TMR-labeled 15M S.A. peptide did not stain control (non-AD) brain tissue from age-matched control mice in related experiments (Fig. 5C). Additionally, a TMR-labeled control peptide did not result in any comparable staining of amyloid deposits in the 5xFAD brain tissue (Fig. 5D). These findings were replicated in brain tissue from another transgenic AD mouse model, the TgCRND8 (Fig. 6). The TMRlabeled 15M S.A. peptide (Fig. 6B), but not the control (CTL2 S.A., Fig. 6D) peptide, stained amyloid deposits in TgCRND8 brain tissue but not in tissue from non-transgenic littermates (Fig. 6A). These data in brain tissue from two independent AD mouse models indicate that, in addition to binding A␤42 monomers, oligomers, and fibrils in vitro, 15M S.A. was also capable of binding mature amyloid deposits in brain tissue ex vivo. Tritiated 15M S.A. Peptide Is Detected in the Brains of Mice following Intravenous Administration-The demonstration of binding of the 15M S.A. peptide to aggregated A␤42 in vitro and ex vivo suggested it may also have the potential to bind A␤42 in vivo. However, the blood-brain barrier (BBB) represents a major hurdle for the delivery of molecules into the brain from the periphery. To make a preliminary assessment of the ability of the 15M S.A peptide to cross the BBB in vivo, 10 Ci of tritiated peptide was administered by intravenous injection to mice as described under "Experimental Procedures," and its concentration in brain and plasma was determined in samples collected over a 0.5-4 h period. Scintillation counting revealed that the plasma concentrations of tritiated peptide decreased with time, as expected following intravenous administration. However, the tritiated peptide was detected in brain homogenate at the initial t ϭ 30 min time point and remained relatively constant until t ϭ 4 h, suggesting that the peptide remains stable in the brain (Fig. 7A). We further performed control experiments to ensure that the measured radioactivity in brain and plasma samples corresponded to intact 3 H-15M S.A. peptide, rather than free label or degradation products (Fig. 7, B and C). There was no apparent shift in the retention time of 3 H-labeled 15M S.A. in brain and plasma samples collected at designated time points, suggesting that the majority of detected radioactivity was associated with intact peptide. This provided initial evidence that the 15M S.A. peptide can cross the blood-brain barrier in vivo following intravenous administration and potentially access brain-based A␤42 targets.

Discussion
Here, we extend our earlier findings (1) and report the generation of a 15-mer stable analogue peptide that targets A␤42. We provide evidence that the peptide is multifaceted in its actions, in that it can bind monomeric, oligomeric, and fibrillar A␤42 and also reduce the formation of neurotoxic A␤42 oligomers.
The ability of the 15M S.A. peptide to interfere with formation of oligomers is of particular interest, given their likely role in mediating the neurotoxicity of A␤42. We noted that in the presence of 15M S.A., the formation of soluble A␤42 oligomers was decreased, with a concurrent increase in the formation of non-toxic, amorphous aggregates. Thus, it appears that the 15M S.A. peptide can shift the equilibrium between different A␤42 assemblies and favor the formation of non-toxic, amorphous aggregates rather than soluble oligomers. A recent report demonstrated that the aggregation of A␤42 is promoted by a positive feedback loop that originates from the interactions between the monomeric and fibrillar forms of this peptide (35). More specifically, once a small but critical concentration of amyloid fibrils has accumulated, the toxic oligomeric species are predominantly formed from monomeric peptide molecules through a fibril-catalyzed secondary nucleation reaction (35). inset ϭ ϫ10 magnification) is shown. DAPI was used as a nuclear counterstain. A, TMR-labeled 15M S.A. peptide did not stain cortical brain tissue from Non-Tg control mice. B, TMR-labeled 15M S.A. peptide bound to amyloid deposits in TgCRND8 cortical brain tissue. C and D, TMR-labeled control peptide (CTL2 S.A.) did not result in any detectable staining of amyloid deposits in Non-Tg or TgCRND8 brain tissue (n ϭ 1).
Given our data showing that 15M S.A. binds monomers and fibrils, it is feasible that it could interfere with such a nucleation reaction. Further validation is required to determine whether 15M S.A. can inhibit the formation of soluble A␤42 oligomers via interactions with monomers, fibrils, or a combination of these A␤42 moieties.
The ability of 15M S.A. to bind preformed A␤42 oligomers is also highly significant, particularly with respect to developing agents that target early events in the disease process. As mentioned earlier, the formation of soluble aggregates is thought to occur prior to plaque and tangle formation and promote the neurodegenerative process. Current evidence suggests that A␤ deposition may precede mild cognitive impairment (the first clinical manifestation of AD) by 10 -20 years (36 -38) and so the accumulation of A␤ oligomers could occur more than 20 years prior to the development of a clinical AD phenotype. It would be advantageous to have the ability to image some of the earliest events in the development of AD, when therapeutic approaches to neutralize and/or clear neurotoxic oligomers might prevent the resulting neuronal injury and cognitive decline. Given the multifaceted action of 15M S.A. in inhibiting oligomer formation and binding preformed oligomers, it may prove useful in both diagnostic and therapeutic approaches targeting preclinical AD.
Peptide-based drugs offer several advantages compared with other popular alternatives, including antibodies. Their small size means they can be produced at a lower cost, with higher activity per mass and often less immunogenicity (39). A further hurdle for antibody-based approaches is penetration of the BBB, which allows only ϳ0.1% of peripheral antibody to gain access to the central nervous system (40) and thus smaller peptides may be more amenable to BBB transport. Certainly, our preliminary data presented here suggests that the 15M S.A. peptide is capable of crossing the BBB in vivo to some degree (Fig. 7). Furthermore, a recent study designed short 20-mer peptides to bind the C terminus of APP-C99 and following intraperitoneal injection of mice showed efficacy at reducing cerebral A␤ levels (41), providing initial evidence for the potential of similar sized peptides to cross the BBB. In our ongoing studies, we are further investigating the extent of BBB transport of 15M S.A. and its ability to bind A␤42 aggregates in vivo, to better define its diagnostic and therapeutic capabilities.
Several peptide-based agents have been designed to target A␤ and inhibit its aggregation. Many of these peptides can reduce A␤ aggregation and neurotoxicity in vitro and modified versions with improved stability have further been shown to be effective in vivo (e.g. Ref. 26 and 42-49). Other peptides that have recently been reported to target oligomeric A␤ in particular include a peptide based on the N-terminal fragment of the cellular prion protein (N1) that blocked the formation of amyloid fibrils in vitro and bound A␤42 oligomers and inhibited their neurotoxicity in vivo (50) and the ABP-p-4 -5 synthetic peptide corresponding to a region of the human pericentriolar material 1 protein that preferentially bound A␤42 oligomers and protected against their neurotoxicity in vitro (51). Few of FIGURE 7. Tritiated 15M S.A. peptide is detected in the brains of mice following intravenous administration. Male Swiss Outbred mice were administered 10 Ci of 3 H-labeled 15M S.A. peptide by tail vein injection. Brain and plasma samples were collected over a 0.5-4-h period and the determination of compound in plasma and brain homogenate was performed by liquid scintillation counting (A, n ϭ 3). Data are mean Ϯ S.E. B and C, assessment of the intactness of 3 H-15M S.A. peptide in samples from brain (B) and plasma (C), respectively. HPLC/scintillation counting indicated a similar retention time for radioactivity detected in brain/plasma samples, and undegraded 3 H-labeled 15M S.A. peptide (Control), suggesting that the majority of detected radioactivity in samples was due to intact peptide.

Novel A␤-binding Peptide Targets Neurotoxic Aggregates
these agents have progressed to the clinic, often as a result of undesirable effects (insolubility, toxicity) at the relatively high doses required to observe their protective effects. However, it is feasible that this may be circumvented by chemical modifications such as conjugation with polyethylene glycol, which can improve solubility, increase circulatory half-life, reduce immunogenicity, and improve delivery (52)(53)(54). Targeted delivery methods may also improve the efficacy of peptide agents to cross the BBB, such as "Trojan horse" approaches (55) and liposomal nanoparticles (56). Another approach to enhance BBB transport is to periodically and reversibly modulate tight junctions. This has recently been demonstrated through the use of siRNA directed against claudin-5 (57).
Although dose limitations have tended to preclude the use of peptides as therapeutic and preventative agents thus far, their high specificity in binding to the A␤42 target may render them useful as diagnostic AD imaging probes, where only tracer quantities are required. The ability of fluorescently tagged 15M S.A. to bind fibrillar A␤42 in vitro, and amyloid plaques ex vivo, highlights its potential as an AD imaging agent. However, detection of mature amyloid plaques can already be achieved by positron emission tomography imaging with current amyloid imaging agents such as the "gold-standard" Pittsburg compound B (58) (a radiolabeled analogue of ThT) and newer 18 F analogues (e.g. Refs. 59 and 60). It is important to highlight here that the existing amyloid imaging tracers are unable to detect oligomeric A␤42. Although oligomer-specific antibodies have been identified (61,62), antibody-based positron emission tomography imaging has not proved useful due to poor brain penetration of the probes (63). Here, we have used several biochemical assays to indicate that the 15M S.A. peptide can indeed bind oligomeric A␤42. Thus, it seems feasible that when conjugated to an appropriate radioisotopic tracer (e.g. 18 F), 15M S.A. has the potential to bridge an important gap in the existing diagnostic techniques and non-invasively highlight the presence of oligomeric A␤42 in vivo.
In our ongoing studies, we are attempting to clarify the binding sites on the 15M S.A. and A␤42 peptides that mediate their interaction. This may clarify how 15M S.A. binds monomeric and aggregated A␤42 and reduces the formation of toxic A␤42 oligomers in vitro, as reported here. It may further indicate other mechanisms by which 15M S.A. might be neuroprotective with respect to A␤42-induced toxicity in vivo. For example, specific domains of A␤ have been reported to facilitate aggregation on and association with lipid bilayers, with implications for bilayer stability and cell viability (64). Thus, if 15M S.A. can make contact with such A␤42 domains and interrupt lipid binding, this could potentially result in neuroprotection. Finally, clarification of the binding interface between 15M S.A. and A␤42 may assist in the design of novel peptidomimetics that mimic the action of 15M S.A. and better satisfy the physicochemical requirements associated with in vivo therapeutics.
Overall, we have validated and characterized the binding profile of a novel A␤ binding peptide that has the potential to bridge an important gap in the existing diagnostic techniques and permit timely intervention to reduce or prevent the neuro-degeneration and cognitive decline associated with the accumulation of oligomeric A␤42.
Author Contributions-R. M. conceived the study and together with G. V., K. T. and R. K. B. co-ordinated the study. R. K. B. and G. V. wrote the manuscript. R. K. B., L. K. W., M. M., V. G., and S. P. performed and analyzed the experiments in Figs. 1-5. L. J. and J. A. N. conceived, performed, and analyzed experiments in Fig. 7. P. E. F. conceived, performed, and analyzed experiments in Table 1. P. E. F. and E. K. performed experiments in Fig. 6. All authors reviewed the results and approved the final version of the manuscript.