The Familial British Dementia Mutation Promotes Formation of Neurotoxic Cystine Cross-linked Amyloid Bri (ABri) Oligomers*

Background: Familial British dementia is a disorder similar to Alzheimer disease and is believed to be caused by the ABri peptide. Results: Cystine-linked oligomers of ABri are toxic to neurons and block long-term potentiation. Conclusion: The type and toxicity of ABri assemblies depends on cystine bond formation. Significance: ABri offers a tractable system to study the structure of disease-causing oligomers. Familial British dementia (FBD) is an inherited neurodegenerative disease believed to result from a mutation in the BRI2 gene. Post-translational processing of wild type BRI2 and FBD-BRI2 result in the production of a 23-residue long Bri peptide and a 34-amino acid long ABri peptide, respectively, and ABri is found deposited in the brains of individuals with FBD. Similarities in the neuropathology and clinical presentation shared by FBD and Alzheimer disease (AD) have led some to suggest that ABri and the AD-associated amyloid β-protein (Aβ) are molecular equivalents that trigger analogous pathogenic cascades. But the sequences and innate properties of ABri and Aβ are quite different, notably ABri contains two cysteine residues that can form disulfide bonds. Thus we sought to determine whether ABri was neurotoxic and if this activity was regulated by oxidation and/or aggregation. Crucially, the type of oxidative cross-linking dramatically influenced both ABri aggregation and toxicity. Cyclization of Bri and ABri resulted in production of biologically inert monomers that showed no propensity to assemble, whereas reduced ABri and reduced Bri aggregated forming thioflavin T-positive amyloid fibrils that lacked significant toxic activity. ABri was more prone to form inter-molecular disulfide bonds than Bri and the formation of covalently stabilized ABri oligomers was associated with toxicity. These results suggest that extension of the C-terminal of Bri causes a shift in the type of disulfide bonds formed and that structures built from covalently cross-linked oligomers can interact with neurons and compromise their function and viability.


Familial British dementia (FBD) is an inherited neurodegenerative disease believed to result from a mutation in the BRI2
gene. Post-translational processing of wild type BRI2 and FBD-BRI2 result in the production of a 23-residue long Bri peptide and a 34-amino acid long ABri peptide, respectively, and ABri is found deposited in the brains of individuals with FBD. Similarities in the neuropathology and clinical presentation shared by FBD and Alzheimer disease (AD) have led some to suggest that ABri and the AD-associated amyloid ␤-protein (A␤) are molecular equivalents that trigger analogous pathogenic cascades. But the sequences and innate properties of ABri and A␤ are quite different, notably ABri contains two cysteine residues that can form disulfide bonds. Thus we sought to determine whether ABri was neurotoxic and if this activity was regulated by oxidation and/or aggregation. Crucially, the type of oxidative crosslinking dramatically influenced both ABri aggregation and toxicity. Cyclization of Bri and ABri resulted in production of biologically inert monomers that showed no propensity to assemble, whereas reduced ABri and reduced Bri aggregated forming thioflavin T-positive amyloid fibrils that lacked significant toxic activity. ABri was more prone to form inter-molecular disulfide bonds than Bri and the formation of covalently stabilized ABri oligomers was associated with toxicity. These results suggest that extension of the C-terminal of Bri causes a shift in the type of disulfide bonds formed and that structures built from covalently cross-linked oligomers can interact with neurons and compromise their function and viability.
Familial British dementia (FBD) 3 is an autosomal dominantly inherited neurodegenerative disease characterized by progres-sive cognitive impairment, spastic tetraparesis, and cerebellar ataxia (1). FBD is linked to a point mutation in the stop codon of the BRI2 gene, which causes an 11-amino acid extension of the BRI2 protein (2). BRI2 is a type II transmembrane protein, which contains an evolutionary conserved ϳ100 residue long BRICHOS domain. This domain is found in 309 proteins and the term BRICHOS is derived from 3 of these proteins, BRI2, Chondromodulin-I, and surfactant protein C (3). It has been hypothesized that the BRICHOS domain acts as an intra-molecular chaperone and interacts with C-terminal ␤-sheet-rich regions (4). Although the exact physiological role of BRI2 is still unknown, several functions have been suggested including neuronal differentiation, involvement in stress response pathways, and its presence on the cell surface has led some to speculate that BRI2 may act as a receptor (5)(6)(7). Wild type BRI2 and FBD-BRI2 are cleaved close to their C termini by one or more pro-protein convertases, releasing 23-and 34-amino acid long peptides, termed Bri and ABri, respectively (8,9). Although the Bri peptide is not known to be amyloidogenic (9,10), the ABri peptide is found in perivascular and parenchymal deposits in the brain of affected individuals (2). In addition to widespread amyloidosis, neurofibrillary tangles composed of hyperphosphorylated Tau protein are an invariant feature of this disease (11). Thus, the neuropathological presentation of FDB is highly similar to that seen in Alzheimer disease (AD); indeed, even the phosphorylation sites found on neurofibrillary tangles in FBD brain are the same as those present in AD (12). These findings, together with the clinical similarities shared by FBD and AD, have promoted the notion that FBD is effectively a phenocopy of AD, with ABri the molecular equivalent of the AD-associated amyloid ␤-protein (A␤) (13)(14)(15)(16). Yet the sequences and innate biophysical properties of ABri and A␤ are quite different. For instance, in contrast to A␤, both Bri and ABri contain two cysteine residues, providing the potential for the formation of both intra-and intermolecular disulfide bonds, and ABri containing disulfide bonds has been found in FBD brain (2,17). Specifically, a ladder of cystine cross-linked, ABri oligomers were detected in FBD brain extracts (17).
Previously it was shown that oxidation of ABri promotes its aggregation and that aggregates of oxidized ABri, but not reduced ABri, were toxic to a neuroblastoma cell line (18 -20). However, the effect of ABri on neurons and the ability of Bri and ABri to form interversus intramolecular disulfide bonds and how such linkages affect their aggregation and toxicity had not been previously investigated. Here we show that oxidation of Bri predominantly results in cyclization, whereas oxidation of ABri results in formation of both cyclized ABri monomers and cross-linked oligomers. Cyclized Bri and ABri show no detectable propensity for aggregation, whereas the reduced monomers can form fibrils and other aggregates, but these assemblies are not toxic to neurons. In contrast, a mixture of cross-linked ABri multimers are potent neurotoxins, which inhibit LTP and cause the rapid demise of neurons. Our results indicate that the simple ability to form aggregates does not determine toxicity, whereas the formation of covalently stabilized assemblies does. We also discuss the possible consequences that increased intermolecular cross-linking may have on the full-length FBD-BRI2 protein, and how the toxicity of oxidized ABri and loss of BRI2 may combine to cause disease.

Experimental Procedures
Peptides, Chemicals, and Reagents-Bri (EASNCFAIRHFE-NKFAVETLICS), ABri (EASNCFAIRHFENKFAVETLICSRT-VKKNIIEEN), and A␤ 1-42 (DAEFRHDSGYEVHHQKLVFFA-EDVGSNKGAIIGLMVGGVVIA) were synthesized, purified, and characterized by Dr. James I. Elliott at Yale University (New Haven CT). All peptides had free amino and carboxyl termini. Three separate batches of ABri were used in this study, one of these was obtained from GL Biochem (Shanghai, China) and all yielded similar results. For all the peptides, mass and purity were determined by electrospray/ion trap mass spectrometry and purified by reverse-phase HPLC, respectively. All peptides had the correct mass and were Ͼ95% pure. Unless otherwise stated, chemicals and reagents were obtained from Sigma and were of the highest purity available. Tissue culture reagents were from Gibco (Carlsbad, CA).
Quantitative Amino Acid Analysis-Peptide samples were isolated in 10.9 mM HEPES, pH 7.8, and 20-l aliquots were dispensed into hydrolysis tubes. Samples were dried under vacuum in preparation for hydrolysis, hydrolyzed for 22 h in 6 N (vapor phase) HCl at 110°C, and dried prior to analysis. Amino acids were separated by reversed-phase HPLC and their elution and peak size compared with a standard mixture of amino acids.
Preparation of Oxidized Peptides-To form oxidatively cross-linked cysteines, lyophilized peptides were dissolved at 1 mg/ml (i.e. 253 M for ABri and 380 M for Bri) in 0.05% ammonium hydroxide and then diluted with 20 mM ammonium bicarbonate, pH 8.2, to produce a peptide concentration of ϳ20 M. Peptide solutions were then incubated in open tubes at room temperature for 5 days, and bubbled with molecular oxygen for 5 min at 24-h intervals. To ensure disassembly of aggregates that may have been produced during this period, peptides were lyophilized and subsequently dissolved at 1 mg/ml in disaggregation buffer (without ␤ME) and incubated overnight at room temperature. To isolate oxidized monomers SEC was used as described below. To recover the entire mixture of oxidized species, GdmHCl-treated material was buffer exchanged into either 10.9 mM HEPES, pH 7.8, or 20 mM NaP, pH 8, using a 5 ml of HP desalting column (GE Healthcare). The concentration of the oxidized mixture was determined by quantative amino acid analysis.
Size Exclusion Chromatography-Peptide solutions (1 ml) were loaded on a Superdex 75 10/300 GL column connected to an ÄKTA purifier (GE Healthcare, Uppsala, Sweden) and eluted at a flow rate of 0.8 ml/min in the appropriate buffer and UV absorbance at 220 nm was recorded. Buffers were filtered through 0.2-m nylon filters (Millipore, Bedford, MA) and bubbled with molecular nitrogen prior to use.
Reverse-phase HPLC-Peptide samples were analyzed on a Vydac Protein C4 column (Grace, Deerfield, IL) connected to a System Gold HPLC system (Beckman Coulter, Inc., Brea, CA) and eluted with a 0.1% trifluoroacetic acid/water/acetonitrile gradient (from 0 to 42% acetonitrile in 20 min) at a flow rate of 1.5 ml/min and UV absorbance at 214 nm was recorded. All buffers were of HPLC grade and were filtered through 0.2-m nylon filters (Millipore) and bubbled with molecular nitrogen prior to use.
Matrix-assisted Laser Desorption Ionization (MALDI) Mass Spectrometry-Aliquots (1 l) of samples were spotted on a MALDI target with 1 l of 10 mg/ml of sinapinic acid matrix (Fluka, Dublin, Ireland), and analyzed on a 4800ϩ MALDI TOF/TOF Analyzer (Applied Biosystems, Foster City, CA). Protein masses were acquired over two ranges: 1-10 and 10 -25 kDa, and calibration was performed using insulin (5,735 Da) and myoglobin (16,952 Da), respectively (Bruker, Billerica, MA). Spectra were summed from 2500 laser shots from an Nd:YAG laser operating at 355 nm and 200 Hz, and were acquired with an accepted signal-to-noise ratio set to Ͼ20. Spectra were processed and analyzed by the Global Protein Server Work station (Applied Biosystems, Life Technologies Corp., Carlsbad, CA).
Thioflavin T Binding Assay-Aggregation of Bri and ABri was monitored using a continuous thioflavin T (ThT) binding assay (21). Peptides were diluted to the desired concentration with the appropriate SEC elution buffer and ThT was added from a ϫ100 stock solution to a final concentration of 20 M. Aliquots (120 l) of the peptide solutions were then dispensed into the wells of an ice-cold 96-well black microtiter plate (Nunc, Roskilde, Denmark) and read immediately. Plates were then sealed with an adhesive plastic cover (VWR, Chicago, IL) and incubated at 37°C with shaking at 700 rpm in a VorTemp 56 shaker/incubator with an orbit of 3 mm (Labnet International, Windsor, UK). ThT fluorescence was measured every 20 min using a SpectraMax M2 microplate reader (Molecular Devices, Sunnyvale, CA) with excitation and emission of 435 and 485 nm, respectively. Data are presented as ThT fluorescence plotted versus time, graphs were produced by connecting data points without smoothing or fitting. The lag time was defined as the first time point showing a statistical significant difference compared with the initial reading (22), the rate of aggregation was defined as the slope of the linear phase and the 1/2t max was defined as the time required to achieve half of the plateau level of ThT fluorescence.
Electron Microscopy-Peptide samples were prepared as for the ThT binding assay, but in the absence of ThT, and analyzed by negative contrast microscopy as previously described (21). Aliquots of peptide samples (10 l) were applied to carboncoated Formvar grids (Electron Microscope Sciences, Fort Washington, PA), fixed with 0.5% (v/v) glutaraldehyde, and stained with 2% (w/v) uranyl acetate (Electron Microscope Sciences, Fort Washington, PA). Grids were air-dried and viewed under a Tecnai TM G 2 20 Twin electron transmission microscope (FEI, Hillsboro, OR).
Circular Dichroism (CD) Spectroscopy-Peptide solutions in 20 mM NaP, pH 8.0, were analyzed at 22°C in a 1-mm quartz cuvette (Starna Scientific, Hainault, UK) using a J-810 JASCO spectropolarimeter (JASCO Corp., Tokyo, Japan). Spectra were collected from three data accumulations between 190 and 260 nm with 10 nm/min continuous scanning and 1 nm bandwidth. Raw data were corrected by subtraction of the buffer alone spectrum and binomial smoothing according to the manufacturer's instructions. Data are displayed as molar ellipticity () versus wavelength.
Rat Primary Hippocampal Neuronal Cultures-Neurons were prepared and grown essentially as described previously (24,25). Hippocampi were explanted from E18 Sprague-Dawley rat pups in Hanks' balanced salt solution and dissociated using 0.25% trypsin (Invitrogen) for 16 min at 37°C. Neurons were plated at a density of 3 ϫ 10 4 cells on the inner 24 wells of poly-D-lysine-coated 48-well plates and cultivated in Neurobasal medium supplemented with 0.5 mM Glutamax-I (Invitrogen) and B27 with antioxidants (Invitrogen). Five days after plating, 20 M 5-fluoro-2Ј-deoxyuridine was added to reduce glial proliferation. Cultures were kept in a humidified incubator at 37°C and 5% CO 2 and once per week half of the medium was replaced with fresh Neurobasal medium containing 0.5 mM Glutamax-I and B27 with antioxidants (26) and treatments commenced after 21 days in vitro.
Preparation of Peptides for Toxicity Studies-A␤ 1-42 , reduced ABri, and Bri were prepared essentially as described previously (25). A␤  and reduced Bri and ABri monomers were isolated by SEC in HEPES buffer, and filtered through a 0.2-m sterile filter (Millipore) in a sterile hood. An aliquot of the filtered sample was used for concentration determination. The remainder of the sample was diluted with sterile buffer to 80 M and all further manipulations were performed on ice. ThT was added to a portion of the stock A␤ 1-42 solution and aliquots were added into a black 96-well plate (Nunc, Roskilde, Denmark) (Plate 1), and to a second sterile black 96-well plate (Plate 2). The remainder of the A␤ 1-42 stock solution that did not contain ThT was added to Plate 2 and held on ice for 60 min. Plate 1 was incubated at room temperature and fluorescence was monitored at 20-min intervals. After 60 min Plate 2 was also incubated at room temperature and fluorescence was monitored in the wells containing A␤ 1-42 plus ThT. Results from Plate 1 were used to estimate the incubation time required for the samples in Plate 2 to attain half-maximal aggregation and the replicates containing ThT in Plate 2 to confirm that the initial phase of aggregation was similar to that observed for samples in Plate 1. 1/2t max samples were then flash frozen on dry ice and stored at Ϫ80°C until used for toxicity experiments or for characterization using EM. Samples of oxidized Bri and oxidized ABri mixtures were lyophilized, dissolved at 1 mg/ml in disaggregation buffer (without ␤ME), incubated overnight at room temperature, and exchanged into sterile HEPES using 5-ml HP desalting columns (GE Healthcare). Buffer-exchanged oxidized peptides were snap frozen, kept at Ϫ80°C until used, and the concentration was determined using quantative amino acid analysis. Buffer eluted from the same SEC or desalting columns used for the peptide treatment was used as the vehicle control.
Toxicity Assay-Peptide solutions were diluted 1:1 with Neurobasal medium containing B27 without antioxidants (Invitrogen) and half of the medium in each well was replaced with peptide-containing medium. Toxicity was tested at 20 M, because this is the concentration range that ABri and A␤ had previously been studied (20). Aliquots of medium (40 l) were collected on post-treatment days 1, 3, 5, and 7 and the release of cytosolic lactate dehydrogenase (LDH) was measured using an LDH assay kit (Promega, Madison, WI). An LDH standard curve was used to determine the units of LDH released per ml of medium. Statistical comparisons were made using a two-way analysis of variance with Bonferroni's post hoc test. On day 7 neurons were fixed using PBS containing 4% (v/v) paraformaldehyde and 4% (w/v) sucrose and stained for microtubule-associated protein 2 (MAP2) with an anti-MAP2 monoclonal antibody (Sigma) using a Vectastain ABC kit (Vector Laboratories Ltd., Burlingame, CA). Briefly, fixed cells were treated with 0.3% hydrogen peroxide to inactivate endogenous peroxidase activity, incubated with blocking buffer (Vectastain ABC kit) diluted in PBS for 20 min, and then incubated for a further 30 min with anti-MAP2 antibody at a dilution of 1:2000 in PBS. Cells were rinsed 4 times with PBS and then incubated for 30 min with biotinylated anti-mouse antibody (Vectastain ABC kit) in blocking buffer. Cells were washed again and incubated with Vectastain Elite ABC reagent for 30 min. Finally cells were washed and incubated with HRP substrate (Vectastain kit) until a suitable level of staining was achieved. Thereafter stained cells were kept at 4°C in PBS until photographed.
In Vitro Electrophysiology-Six to 10-week-old male C57BL/ 6J mice (Charles River, Margate, UK) were used for electrophysiological recordings. In all cases, mice were anesthetized with isoflurane/O 2 and decapitated. The brain was rapidly removed and immersed in ice-cold sucrose-based artificial cerebrospinal fluid containing 87 mM NaCl, 2.5 mM KCl, 7 mM MgSO 4 , 0.5 mM CaCl 2 , 25 mM NaHCO 3 , 25 mM glucose, 1.25 mM NaH 2 PO 4 , and 75 mM sucrose. Parasagittal sections (350 m) containing the hippocampus were prepared on a Leica VT1000S vibratome (Leica Microsystems GmbH, Wetzlar, Germany). Slices were immediately transferred to a holding chamber (BSC-PC, Warner Instruments, Hamden, CT) containing artificial cerebrospinal fluid (119 mM NaCl, 2.5 mM KCl, 1.3 mM MgSO 4 , 2.5 mM CaCl 2 , 26.2 mM NaHCO 3 , 11 mM glucose, and 1.25 mM NaH 2 PO 4 ) and the solution continuously bubbled with a mixture of 95% O 2 and 5% CO 2 . The slices were allowed to recover for at least 90 min at room temperature prior to recording. Extracellular recordings were performed as described previously (27). Briefly, slices were submerged in a recording chamber and perfused with oxygenated artificial cerebrospinal fluid (2-3 ml/min). The perfusate was warmed to 30°C using an inline heating tube (HPT-2A, ALA Scientific Instruments, Westbury, NY) and recycled using peristaltic pumps (101U/R, Watson-Marlow, UK). A stainless steel microelectrode (FHC, Bowdoin, ME) was used to stimulate Schaffer collateral fibers and extracellular field excitatory postsynaptic potentials were recorded from stratum radiatum of CA1 using a glass microelectrode with a resistance of 2-3 M⍀. Field excitatory postsynaptic potentials were recorded using a Multiclamp 700B amplifier in tandem with a Digidata 1440A digitizer (Axon Instruments, Sunnyvale, CA). During baseline field excitatory postsynaptic potential recording, stimuli were delivered once every 30 s (0.033Hz), and the stimulus intensity was set to give a baseline field excitatory postsynaptic potential of 40 -50% of the maximal response. A stable baseline was recorded for at least 20 min prior to perfusion of peptide or vehicle preparations. In all cases peptide and vehicle were introduced to the perfusate 30 min prior to induction of LTP and remained in the bath solution for the remainder of the experiment. The effect of ABri on LTP has never been previously investigated. Thus we chose to study the effect of ABri at a concentration comparable with that at which A␤ is known to block LTP (28,29). LTP was induced by a burst stimulation (10 bursts of 4 stimuli at 100 Hz, with an interburst interval of 200 ms) given at baseline intensity. LTP is expressed as the mean Ϯ S.E. % of baseline field excitatory postsynaptic potential slope. Statistical comparisons used analysis of variance with post hoc Tukey-Kramer test. Vehicle and Bri/ABri experiments were performed on the same day such that slices from the same animal were used for both vehicle and peptide treatment; in addition treatments were alternated daily to avoid any temporal bias.

Results
Prior studies have demonstrated that oxidized ABri can aggregate (10) and damage certain cultured cells (19), but whether ABri can compromise neuronal activity and viability has not been investigated previously. Thus we thought it important to determine whether ABri is toxic to neurons and how this effect may be influenced by its aggregation state and in turn how its aggregation is affected by inter-and intra-molecular disulfide bond formation. Because the Bri peptide formed from wild type BRI2 contains the same amino acid sequence as the first 23 amino acids of ABri, including Cys-5 and Cys-22, we also examined the effect of oxidation on the ability of this peptide to aggregate and to compromise neuronal activity.
Oxidation of Bri Leads to Formation of a Cyclized Monomer, Whereas Oxidation of ABri Leads to the Formation of Both Cyclized Monomer and Cross-linked Oligomers-Cysteine-containing peptides are prone to atmospheric oxidation (30), and ABri is known to aggregate (10,31). Thus to reduce any oxidized cysteines and to disassemble any pre-existing aggregates we incubated both Bri and ABri peptides in a buffer containing ␤ME and GdmHCl prior to analysis by size exclusion chromatography (Fig. 1, A and B). As expected, when both Bri and ABri peptides were incubated with denaturing buffer containing ␤ME each produced a single peak, the Bri peak eluting at ϳ15.0 ml and the ABri peak eluting at ϳ13.8 ml (Figs. 1A and 2B). Based on the elution of linear dextran standards (Fig. 1B), the reduced Bri monomer had an estimated molecular mass of 2.5 kDa, whereas the reduced ABri monomer had an apparent molecular mass of 4.0 kDa. These estimates are consistent with the known molecular masses of Bri (2630 Da) and ABri (3956 Da). SDS-PAGE/silver staining of aliquots of SEC fractions of reduced Bri or reduced ABri confirmed the presence of a major species, which migrated on SDS-PAGE with molecular masses of ϳ3 and ϳ4 kDa, respectively (Figs. 1C and 2D). RP-HPLC analysis of Bri SEC fraction 11 (Fig. 1A) showed that the majority of this material was indeed reduced Bri (Ն82%) (e.g. Fig. 2, A and B). RP-HPLC analysis of ABri SEC fraction 10 ( Fig. 1B) indicated that the vast majority (Ն93%) of the material present was reduced ABri (e.g. Fig. 2, E and F). Next we deliberately oxidized Bri or ABri. In an attempt to disassemble non-covalent aggregates formed during the incubation necessary for oxidation, oxidized Bri and ABri were lyophilized and then treated with GdmHCl prior to SEC analysis. Centrifugation (3000 ϫ g for 5 min) of the samples prior to injection onto SEC produced a visible pellet in the ABri but not the Bri sample. This observation indicates that oxidation leads to formation of a discernible amount of GdmHCl-insoluble aggregates. When supernatants were analyzed by SEC, oxidized Bri produced one major peak that eluted at ϳ15.7 ml (Fig. 3A), ϳ0.7 ml later than reduced Bri (Fig. 1A). The later elution of oxidized Bri is consistent with formation of an intra-molecular disulfide bond resulting in a cyclized more compact structure than the linear reduced Bri. Indeed, comparison with linear standards accurately predicted the molecular mass of reduced Bri (2.5 kDa versus the known mass of 2.6 kDa), whereas it underestimated the size of oxidized Bri (1.9 kDa versus the known mass of 2.6 kDa). SDS-PAGE/silver staining of the same SEC fractions Oxidation of ABri Regulates Its Aggregation and Toxicity JULY 3, 2015 • VOLUME 290 • NUMBER 27 revealed the presence of a single species that migrated on SDS-PAGE with a molecular mass of ϳ3 kDa (Fig. 3C). RP-HPLC analysis of fraction 12 (Fig. 4A) confirmed the presence of a single species that eluted in a manner consistent with oxidized Bri monomer (Fig. 2, C and D). These results indicate that the major product of Bri oxidation appeared to be a cyclized monomer. In contrast, the supernatant of centrifuged oxidized ABri produced four broad peaks eluting at ϳ9, ϳ12, ϳ14.3, and ϳ16.1 ml (labeled a through d, Fig. 3B). These peaks likely represent different oxidized ABri species, which by comparison with linear dextrans had apparent molecular masses centered around 28.7, 8.4, 3.4, and 1.6 kDa, respectively. SDS-PAGE/ silver staining of the same SEC fractions revealed the presence of several different species (Fig. 3D). Peak "a" (Fx 5 and 6) contained an array of distinct species with molecular masses ranging from ϳ4 to Ͼ50 kDa (Fig. 3D). Peak "b" (Fx 8 and 9) contained species migrating at ϳ4, ϳ7-8 and ϳ10 kDa, whereas peaks "c" (Fx 11) and "d" (Fx 13) contained only a ϳ4 kDa species (Fig. 3D). Thus, peaks a and b appear to contain SDS-labile aggregates of various species that migrated on non-reducing SDS-PAGE gels with molecular weights consistent with monomer and covalently cross-linked low-n oligomers of ABri. Indeed, prior incubation of fractions containing these species with ␤ME resulted in their breakdown (Fig. 3E). Peaks c and d seem to contain only oxidized monomer, and RP-HPLC analysis of fraction 13 confirmed the presence of a single peak that eluted at 23.1 min (e.g. Fig. 2G). When fraction 13 was supplemented with ␤ME a single peak eluted after 25.8 min (e.g. Fig.  2H) in line with where reduced ABri had previously been found to elute. For further studies on cyclized ABri monomer we focused on peak d as this was the major oxidized monomercontaining peak. Given the limited resolution of SDS-PAGE and the possibility of non-ideal behavior of differently folded oligomers, we analyzed the unfractionated oxidized ABri sample by mass spectrometry in an attempt to identify cross-linked oligomers of different size. Peaks corresponding to monomer (3952.3), dimer (7906.1), trimer (11851.6), tetramer (15791.4), and pentamer (19744.0) were detected (data not shown).  Analysis of "reduced Bri monomer" (fraction 11 from A) produced 2 peaks; a major peak that eluted at 20.3 min and a minor peak that eluted at 19.0 min (A). Addition of 5% ␤ME to the reduced Bri monomer resulted in the disappearance of the 19.0-min peak (B). Analysis of "oxidized Bri monomer" (fraction 12 from Fig. 4A) produced a single peak eluting at 19.0 min (C) and treatment of this material with 5% ␤ME caused the disappearance of the 19.0-min peak and the appearance of a peak at 20.3 min (D). Injection of reduced ABri monomer (fraction 10 from Fig. 2B) produced two peaks: a major peak at 25.6 min and a minor peak at 23.1 min (E). Treatment of this material with 5% ␤ME caused the disappearance of the 23.1 min peak (F). Analysis of oxidized ABri monomer (fraction 13 from Fig. 4B) revealed a single peak eluting at 23.1 min (G), whereas treatment of oxidized ABri monomer with 5% ␤ME results in a single peak eluting at 25.8 min (H). The minor peak detected in the reduced monomer of both ABri and Bri likely represent monomers that re-oxidized prior to analysis by RP-HPLC. JULY 3, 2015 • VOLUME 290 • NUMBER 27

Oxidation of ABri Regulates Its Aggregation and Toxicity
Reduced ABri Monomer, but Not Oxidized ABri Monomer, Aggregates to Form Amyloid Fibrils-We assessed the propensity of SEC-isolated reduced ABri monomer to form aggregates using a continuous ThT binding assay (21,25) (Fig. 4). Aggregation was analyzed in two different buffers, 10.9 mM HEPES, pH 7.8, and 20 mM sodium phosphate, pH 8.0. These buffers were used because of their compatibility with subsequent toxicity and structural studies, respectively. When incubated in HEPES or NaP (Figs. 4A and 5B), there was a lag before the reduced ABri monomer began to aggregate, a characteristic of nucleation-dependent polymerization reactions (32,33). To analyze this process, we used three parameters: lag time (the time required to have a significant increase in ThT fluorescence), rate of aggregation (the slope of the linear phase of the sigmoidal curve), and 1/2t max (the time required to achieve half of the maximal level of ThT fluorescence). Aggregation was readily detectable at ABri concentrations Ն5 M, and both the rate of aggregation and 1/2t max (Figs. 4A and 5B) increased with increasing peptide concentration, but it was only at 40 M that the lag phase showed a significant (p Ͻ 0.001) decrease (Figs. 4A and 5B). In contrast, oxidized ABri monomer, collected from SEC peak d (Fig. 3B), did not bind ThT even when incubated at high concentrations (50 M) for prolonged periods of time (up to 1000 min) (Fig. 4C). Because the aggregation of reduced ABri monomer was essentially identical in both HEPES and NaP buffers we were reassured that use of these buffers for our toxicity and structural studies would allow correlation between the two.

. ABri forms both intra-and inter-molecular disulfide bonds, whereas Bri forms intra-molecular bonds.
Bri and ABri peptides were subjected to atmospheric oxidation, lyophilized, and then resuspended in denaturing buffer (20 mM sodium phosphate buffer, pH 8.0, containing 5 M guanidium hydrochloride and 200 M EDTA) and used for SEC on a Superdex 75 10/300 GL column eluted with 10.9 mM HEPES buffer, pH 7.8. Oxidized Bri (A) produced one predominant peak, whereas SEC of oxidized ABri (B) revealed the presence of several different sized species. Four peptide-containing peaks are indicated by the letters a-d. The peaks labeled e and f were also detected when buffer alone was injected and are obvious in B but not A because of the much smaller range of the y axis in B. Arrows in B indicate the void volume (V o ), estimated from the elution of blue dextran, and the elution of linear dextrans of known molecular weight. SDS-PAGE/silver stain of SEC fractions (Fx) show that only monomer was recovered in the oxidized Bri sample (C), whereas higher molecular weight species were present in the oxidized ABri sample (D). SDS-PAGE/silver stain of the same samples used for non-reducing electrophoresis but this time electrophoresed under reducing conditions demonstrated that the oligomers detected in Fx 5 to 9 are held together by disulfide bonds (E).
Freshly isolated reduced and oxidized ABri monomers produced similar CD spectra, with a minimum at ϳ200 nm (201 nm for reduced ABri and 202 nm for oxidized ABri), but no other significant features (Fig. 4D). Because the reduced ABri monomer showed a time-dependent aggregation, but the oxidized ABri monomer did not, we monitored the CD spectra of reduced ABri at time points corresponding to 1/2t max and t max using the same incubation conditions as for our ThT-binding assay (Fig. 4C). At 1/2t max the CD spectrum was similar to that of the monomer, with a minimum at 200 nm (Fig. 4D). At 1/2t max the CD spectrum indicated a blend of different secondary structures. It is noteworthy that at the three time points analyzed no aggregates were visible to the eye, but that centrifugation of the t max sample at 16,100 ϫ g for 10 min pelleted the majority of the ThT-positive material (not shown). Accordingly, negative contrast EM of an unspun sample revealed no detectable material in the t ϭ 0 reduced ABri monomer sample (not shown), but abundant structures were evident in both 1/2t max and t max samples (Figs. 4E and 5F). At 1/2t max a significant amount of material resembling twisted ribbons with a diameter between 7 and 11 nm (average diameter 9.1 Ϯ 0.4 nm) and of indeterminate length were detected, together with some rough and short fibrils with diameters between 13 and 19 nm (average diameter 14.6 Ϯ 1.0 nm) (Fig. 4E). At t max EM analysis revealed the presence of fibrils with diameters ranging from 13 to 17 nm (average diameter 14.8 Ϯ 0.5 nm) and between 51.4 and 108.9 nm in length, and large globular assemblies (which may contribute to the mixed structure detected by CD) with diameters between 48 and 68 nm (average diameter 56.3 Ϯ 4.1 nm) (Fig. 4F).
Reduced Bri Monomer, but Not Oxidized Bri Monomer, Aggregates Forming ␤-Sheet-rich Fibrils-Next, we investigated if Bri monomer could aggregate under the same conditions that ABri was found to readily aggregate. Freshly SEC-isolated  A and B). The aggregation propensity of SEC-isolated reduced (f) and oxidized (•) ABri monomer, each at 50 M, was assessed over a period of 1000 min (C). Samples from C at various time points were diluted to 10 M and used for circular dichroism spectroscopy. Spectra of these indicate that irrespective of the oxidation state the ABri monomer lacks a significant secondary structure (10 M reduced ABri monomer t ϭ 0, -; 10 M oxidized ABri monomer t ϭ 0, dot -⅐ -), but that upon aggregation reduced ABri attains a mixed structure, with unstructured, ␣-helical and ␤-sheet components (10 M reduced ABri t ϭ t max , ---) (D). Negative contrast electron microscopy revealed that aggregation of 50 M reduced ABri resulted in the production of large twisted ribbons, protofibrils, and globular structures (E and F). ThT binding experiments are representative of at least 4 experiments and each data point is the average Ϯ standard error of 6 replicates. Where error bars are not visible the standard error is smaller than the size of the symbol.
reduced Bri monomer (as in Fig. 1A) at concentrations Ͻ15 M produced no ThT binding even when incubated for up to 1500 min (not shown). However, at concentrations Ն30 M reduced Bri monomer did aggregate, but with lag times much longer than that observed for lower concentrations of ABri (140 versus 60 min for 60 M Bri and 40 M ABri) (Figs. 5A and 6A). Moreover, the extent of ThT binding for Bri was much lower than that observed for ABri (compare Figs. 5A and 6B with Figs. 4A and5B).AswasthecaseforABri,reducedBrimonomeraggregated similarly in both HEPES and NaP (Figs. 5A and 6B). Like oxidized ABri monomer, oxidized Bri monomer did not bind ThT even when incubated for periods up to 1000 min and at a concentration of 75 M (Fig. 5C). CD spectroscopy revealed little discernible structural elements in either reduced Bri monomer or oxidized Bri monomer, and although both had minima at ϳ200 nm (200 nm for reduced Bri monomer and 202 nm for oxidized Bri monomer) their spectra were somewhat different (Fig. 5D). Reduced Bri monomer produced a spectrum typical for unstructured peptides and proteins (34), whereas the spectrum for oxidized Bri monomer was more flattened suggesting at least some structural components (Fig. 5D). Interestingly, aggregation of reduced Bri monomer was accompanied by formation of significant amounts of ␤-sheet structure (Fig.  5D). At 1/2t max a strong minimum at 216 nm and a maximum at 197 nm were evident, and upon further incubation (t max ) both the minimum and maximum increased, suggesting further strengthening of the ␤-structure (Fig. 5D). Consistent with the emergence of the ␤-structure, EM revealed the presence of amyloid fibrils (Figs. 5E and 6F). The fibrils detected at 1/2t max and t max had diameters ranging from 4.5 to 8.4 nm (average diameter 6.1 Ϯ 1.0 nm) and were Ն400 nm in length. The periodicity of the 1/2t max samples tended to be a little shorter than that seen in t max samples, ranging from 29 to 45 nm (average periodicity 34.2 Ϯ 4.8 nm) at 1/2t max and from 38 to 62 nm (average periodicity 48.5 Ϯ 5.8 nm) for t max (Figs. 5E and 6F).  A and B). The aggregation propensity of SEC-isolated reduced (f) and oxidized (•) Bri monomers, each at 75 M, was assessed using a continuous ThT binding assay (C). Samples from C at various time points were diluted to 10 M and used for circular dichroism spectroscopy. Aggregation of reduced Bri was associated with an increase in ␤-sheet content (10 M reduced Bri monomer t ϭ 0, -; 10 M reduced Bri t ϭ 1/2t max , --; 10 M reduced Bri t ϭ t max , ---; 10 M oxidized Bri monomer t ϭ 0, ---) (D). Negative contrast electron microscopy of 75 M reduced Bri (t ϭ 1/2t max and t ϭ t max ) indicated that this was accompanied by the formation of amyloid fibrils (E and F, respectively). As described in the legend to Fig. 3, ThT binding experiments are representative of at least 4 experiments and each data point is the average Ϯ standard error of 6 replicates. Where error bars are not visible the standard error was smaller than the size of the symbol.

Solutions of ABri Containing a Mixture of Different Oxidized and Aggregated Species Bind
ThT-We have shown that oxidation of ABri leads to the formation of aggregates that contain covalently cross-linked ABri oligomers and ABri monomer and sedimentable aggregates (Fig. 3D). However, because the low abundance of these various species precluded direct characterization of individual species, we decided to study the mixture. To do this we prepared the sample in a manner identical to that used in Fig. 3B, but instead of fractionating by SEC, we used a desalting column to remove the GdmHCl-containing denaturing buffer and exchanged the crude oxidized ABri mixture into HEPES. Material produced in this manner bound ThT (Fig.  6A), although the extent of ThT binding was not as large as that seen for aggregates of reduced ABri monomer. Similarly prepared oxidized Bri, which by SEC contained mostly cyclized Bri monomer (Fig. 3A), did not bind ThT (Fig. 6A). Fresh bufferexchanged samples of oxidized ABri and oxidized Bri were then analyzed by CD spectroscopy. As with the SEC-isolated oxidized Bri monomer (Fig. 6D) the crude oxidized Bri yielded a spectrum typical of disordered peptides and proteins (34). The spectra for crude oxidized ABri was somewhat broader and more shallow than that seen for crude oxidized Bri (Fig. 6D) or SEC-isolated oxidized ABri (Fig. 4D), suggestive of a mixture of structural components. Interestingly, negative contrast EM of the crude oxidized ABri mixture sample showed the presence of small, curved filamentous structures (Fig. 6C). No discernible structures were detected in the crude oxidized Bri mixture sample (not shown). These results are consistent with our prior observation that oxidation of Bri leads almost exclusively to formation of cyclized monomer; whereas oxidation of ABri leads to formation of a mixture of cross-linked species.
Oxidized ABri Disrupts Synaptic Plasticity-Because we have shown here that under certain conditions both ABri and Bri can aggregate to form ThT-binding supramolecular structures (Figs. 4 and 6) and that both the aggregation kinetics and the type of aggregates found are influenced by oxidation, we sought to determine which, if any, of these different assemblies of ABri and Bri had disease-relevant biological activity. By analogy with our previous work using a defined mixture of A␤ species (24,25,35), we studied the effects of t1 ⁄ 2max reduced ABri (Fig. 4) and 1/2t max reduced Bri (Fig. 5), which also contained mixtures of different sized assemblies. Moreover, we tested the crude unfractionated mixture of aggregates produced by the oxidation of ABri (Fig. 6) and crude oxidized Bri, which contained mostly monomer (Fig. 6). We examined these samples in two experimental paradigms: one to test the effect of acute treatment and the other to test the effect of chronic application. Accordingly, we investigated their acute administration on long term potentiation using hippocampal slices of mouse brain. Because HEPES buffer, but not sodium phosphate buffer, was compatible for use with hippocampal slices and primary neurons, all peptides were prepared and characterized in HEPES. Perfusion of hippocampal slices with 1 M of buffer-exchanged crude-oxidized ABri mixture strongly inhibited LTP (123 Ϯ 12%, n ϭ 7; Fig. 7A) as compared with slices treated with vehicle alone (159 Ϯ 8%, n ϭ 7; Fig. 7A). Importantly, 1/2t max reduced ABri and 1/2t max reduced Bri, which contained a mixture of aggregated and monomeric species, failed to inhibit hippocampal LTP (166 Ϯ 9%, n ϭ 5; 154 Ϯ 4%, n ϭ 4; Fig. 7B). Perfusion with 1 M buffer-exchanged oxidized Bri, which contained mainly monomeric cyclized Bri also had no effect on LTP (160 Ϯ 12%, n ϭ 6; Fig. 7B). These results demonstrate that: 1) the ability of a peptide to aggregate and form ThT-positive structures does not always predict if the peptide will be synaptotoxic, and 2) inter-molecular cystine-linked ABri oligomers are synaptotoxic. Importantly, SDS-PAGE examination of oxidized ABri before and at the end of LTP experiments indicated that oxidized ABri oligomers remained intact throughout the course of the experiment (data not shown).
Oxidized ABri Is Toxic to Primary Hippocampal Neurons-Next, we investigated the effect of chronic exposure of different ABri and Bri preparations on the viability of primary hippocampal neurons. For these experiments we used two positive controls, 400 M glutamate and 20 M 1/2t max A␤  . Incubation with vehicle caused a slight but non-statistically significant (n ϭ 6, p Ͼ 0.05) time-dependent increase in the release of LDH (Fig.  8A). As expected, treatment of neurons with glutamate caused a significant increase in the level of released LDH from day 1 (ϩ 0.56 Ϯ 0.02 units of LDH versus vehicle control, n ϭ 6, p Ͻ 0.001) onwards, whereas treatment of neurons with A␤ 1-42 caused a significant increase in the level of released LDH from day 5 (ϩ0.28 Ϯ 0.02 units of LDH versus vehicle control, n ϭ 6, p Ͻ 0.001) onwards (Fig. 8A). Remarkably, buffer-exchanged crude oxidized ABri caused rapid and extensive neuronal loss with a significant increase in LDH apparent at day 3 (ϩ 0.17 Ϯ 0.01 units of LDH versus vehicle control, n ϭ 6, p Ͻ 0.001) onwards (Fig. 8A). Oxidized ABri was more detrimental to neuronal viability than A␤ 1-42 and on day 7 the level of LDH in oxidized ABri-treated cells was comparable with that of cells treated with glutamate. In contrast, oxidized Bri, 1/2t max reduced ABri, and 1/2t max reduced Bri had no adverse effect on neuronal form (not shown) or release of LDH (Fig. 8B). At the end of the experiment, neurons were fixed and stained for MAP2. Treatment with oxidized ABri (like treatment with glutamate) caused the loss of virtually all neurons, and only the remnants of dead cells were detected (Fig. 8C). Thus, as in the LTP experiments, oxidized ABri was the only Bri-related material to exert a detrimental effect. SDS examination of oxidized ABri at days 1, 3, 5, and 7 indicated that oxidized ABri oligomers persisted throughout the course of the experiments; however, some of this material also forms higher molecular weight assemblies (data not shown).

Discussion
Mutations in the stop codon of the BRI2 gene are associated with FBD and familial Danish dementia (FDD) (as shown in Fig.  9A) and lead to production of the C terminally extended mutant Bri peptides, ABri and ADan, which accumulate as amyloid in the brains of affected individuals (2,36). Because the normal 23-residue long Bri peptide is not associated with disease, it seems reasonable to assume that the 11-amino acid extensions present in ABri and ADan play an important role in pathogenesis. However, the sequence of the extensions present in ABri and ADan are different (2,36,37) indicating that the associated toxicity is unlikely to be dependent on primary sequence per se, but rather is more likely to be mediated by structural changes in BRI2 and/or common secondary or tertiary structures accessible to both ABri and ADan.
In the current study we assessed the aggregation state of ABri and Bri peptides using a combination of biophysical techniques and then went on to test for disease-relevant activity using two distinct paradigms. To study acute effects we used LTP, a well accepted correlate of learning and memory (38,39). The use of in vitro LTP as a read-out for toxic activity has two major benefits. First, to allow correlation between activity and the structures studied it is essential that the duration of the assay is sufficiently short so as to preclude or minimize changes in the structures under investigation (40,41). The LTP paradigm we employ takes only 90 min from the initial application of peptide until the end of the experiment and compared with most other commonly used assays of toxicity is relatively rapid. Second, loss of synapses is an early and invariant event in AD and FBD (15,16,42) and is anticipated to be preceded by changes in synaptic function. Indeed, in the case of AD it has been suggested that neuronal loss may occur as a consequence of persistent disruption of plasticity (43)(44)(45). Thus the LTP paradigm used is particularly well suited for assessing disease relevant activity over a relatively brief period. To study the effects of chronic exposure we measured the ability of various peptide preparations to kill mature rat hippocampal neurons.
Here we show that reduced Bri and reduced ABri can aggregate, but that acute application of their assemblies had no effect on neuronal plasticity. Similarly chronic application of reduced Bri or reduced ABri did not cause neuronal death. Whether this reflects an inability of reduced Bri and ABri to form toxic assemblies or that they form such assemblies but at insufficient concentrations to exert toxicity remains unclear. The fact that we found reduced ABri to aggregate faster and to a greater extent than Bri is in accord with observations that ABri, but not Bri, is deposited in the brains of individuals with FBD ( Fig. 9, B-D) (2). We also demonstrated that oxidative cyclization of both Bri and ABri results in monomers that show no propensity to assemble or form amyloid. So why then is ABri associated with disease? Our data indicate that ABri is more prone to form inter-molecular disulfide bonds than Bri and that formation of covalently stabilized ABri oligomers is associated with toxicity. This suggests that extension of the C-terminal of Bri causes a shift in the type of disulfide bonds formed, from intra-molecular to inter-molecular disulfide bonds, and that structures built from covalently cross-linked oligomers can interact with neurons and compromise their function and viability (Fig. 9, B-D). FIGURE 8. Oxidized ABri is toxic to primary hippocampal neurons. Cells were incubated for 7 days with crude oxidized ABri (f, 20 M), A␤ 1-42 (छ, 20 M t ϭ 1/2t max aggregated) and glutamate (•, 400 M) or HEPES buffer alone (OE). Culture medium was collected from neurons on days 1, 3, 5, and 7 and used to monitor the amount of cytosolic LDH released from dying cells (A). LDH release became significant for neurons treated with glutamate (ϩ, 0.56 Ϯ 0.02 units of LDH versus vehicle control; ***, p Ͻ 0.001), oxidized ABri (0.17 Ϯ 0.01 units of LDH versus vehicle control, ***, p Ͻ 0.001), and A␤ 1-42 (0.28 Ϯ 0.02 units of LDH versus vehicle control, ***, p Ͻ 0.001) after 1, 3, and 5 days, respectively. Cells were also incubated with oxidized Bri monomer (20 M), 1/2t max reduced ABri (20 M), and 1/2t max reduced Bri (20 M) but these treatments did not lead to significant release of LDH (B). Each data point is the average of values from at least 6 replicate wells and expressed as units of LDH released. When not obvious standard errors are smaller than the symbol. Treatments that caused significant increases (p Ͻ 0.001) are indicated with *** and treatments that caused no change in LDH are indicated with ns. On the final day of the experiment, cells were fixed and used for immunocytostaining with anti-MAP2 antibody and representative images are shown (C) with magnified fields identified with a black border. Treatments were tested in at least 4 individual experiments and similar results were obtained in each experiment.

Oxidation of ABri Regulates Its Aggregation and Toxicity
Besides direct toxicity of ABri, burgeoning evidence suggests that loss of BRI2 underlies at least some facets of FBD. Specifically the levels of FBD-BRI2 are reduced in both the brains of FBD patients and knock-in mice (46,47) prompting the suggestion that the FBD mutation causes formation of aberrantly folded FBD-BRI2 that is rapidly degraded by the unfolded protein response system of the endoplasmic reticulum (47) and that disease results due to a loss of BRI2 protein (Fig. 9). Using knock-out and knock-in models the D'Adamio group (47,48) have elegantly demonstrated that FBD and FDD mutations result in a reduction of BRI2 and like BRI2 hemizygous mice (47,49,50), knock-in FBD-BRI and FDD-BRI2 mice evidence impairment of synaptic plasticity and recognition memory (47,50). Yet the knock-in FBD-BRI2 and FDD-BRI2 mice show no signs of amyloid accumulation, aberrant Tau phosphorylation, or neuronal death (47,49). Our results may help explain the limitations of the loss-of-function hypothesis by expanding it to embrace a "two-hit mechanism" (Fig. 9, B-D). In such a scenario the C-terminal 11 amino acids of FBD-BRI2 interferes with intra-molecular disulfide bond formation between resi-dues Cys-5 and Cys-22 of the ABri domain resulting in FBD-BRI2 molecules that contain a misfolded C-terminal domain or a C-terminal domain cross-linked to neighboring FBD-BRI2 or BRI2 molecules. Thus, whereas precursor proteins with a misfolded C-terminal are targeted for degradation, those which dimerize at the ABri/ABri (or ABri/Bri) level could be more stable and give rise to toxic cross-linked oligomers. The C-terminal extension also results in the liberation of ABri, which under appropriate conditions forms toxic cystine cross-linked ABri oligomers and high molecular weight assemblies built of such oligomers. Therefore the FBD mutation has the capacity to affect both a loss of the normal function mediated by BRI2 and a toxic gain of function brought about by noxious assemblies formed from covalently linked ABri oligomers. In addition, a certain fraction of unoxidized ABri may be produced and contribute to the abundant amyloid deposits that typify FBD. Indeed, our finding that reduced ABri readily aggregates, but did not negatively alter LTP or neuronal viability offers a plausible explanation as to why transgenic over-expression of FDD-BRI2 results in abundant amyloid deposition, but no tangle  (5), is shown with an arrow (A). The 23-amino acid long wild type Bri peptide spontaneously oxidizes forming intra-molecular disulfide bonds (thick arrow); producing a compact cyclized structure that exhibits no propensity for aggregation and neither alters nerve cells viability nor LTP. Although theoretically possible, we found no evidence that Bri could form significant amounts of inter-molecular crosslinked multimers (thin arrows) (B). In contrast, ABri oxidizes forming both inter-(thick arrows) and intra-(thin arrow) molecular disulfide bonds (C). As with Bri, cyclized ABri appears not to aggregate, whereas cross-linked ABri oligomers are found in ThT-positive amyloid structures (D). Importantly, mixtures of oxidized ABri that include ABri monomer (shaded in orange), non-sedimentable cross-linked oligomers (shaded in green and blue), inhibit LTP and are potent neurotoxins. These results suggest that decreasing the formation of inter-molecular disulfide bonds in ABri would prevent or reduce ABri toxicity. A similar increase in the propensity to form inter-molecular cross-links in the parent FBD-BRI2 may adversely affect the stability and normal function of BRI2.
pathology and little FDD-relevant functional change (51,52). It is conceivable that over-expression of FDD-BRI2 leads to elevated levels of reduced ADan, and by analogy to reduced ABri, this material readily aggregates to form relatively innocuous insoluble macroassemblies.
In summary, based on our observation that ABri exhibits a higher propensity to form inter-molecular cross-links than Bri, and extensive data from others using genetically manipulated models of FBD and FDD we propose a pathogenic process that involves both the simultaneous loss of BRI2 function and gain of toxic ABri activity (Fig. 9). Thus in FBD we hypothesize that reduced levels of mature BRI2 contribute to impaired synaptic plasticity, and that cystine-linked ABri oligomers account for aberrant changes in Tau and consequent neuronal loss. Simultaneously, unoxidized ABri aggregates and forms amyloid plaques. Clearly further in vitro, cell culture, and animal modeling studies will be required to validate this over-arching hypothesis. In particular it will be important to determine whether oxidized ABri can induce disease-relevant changes in Tau. It will also be interesting to determine whether the sequence of the carboxyl-terminal extensions present in ABri and ADan is important or if any 11-amino acid extension would have similar effects (53). Nonetheless, the fundamental observation that only oxidized oligomers of ABri are toxic to neurons suggests that high resolution analyses of these structures will provide important insights about ABri and FBD, and possibly also about A␤ and AD.