LETTER doi:10.1038/nature11060 Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-b Justin M. Nussbaum1*, Stephan Schilling2*, Holger Cynis2, Antonia Silva1, Eric Swanson1, Tanaporn Wangsanut1, Kaycie Tayler3, Brian Wiltgen3, Asa Hatami4, Raik Rönicke5, Klaus Reymann5, Birgit Hutter-Paier6, Anca Alexandru7, Wolfgang Jagla7, Sigrid Graubner7, Charles G. Glabe4, Hans-Ulrich Demuth2,7 & George S. Bloom1,8 Fig. 3). The ratio of optical densities at 450 nm versus 490 nm (OD450 nm/OD490 nm) for Ab3(pE)–42 rose and peaked more rapidly than for Ab1–42, but peaked at a ,25% lower level. The fastest rise in the Aβ1–42 Vehicle a Aβ3(pE)–42 95% Aβ1–42 5% Aβ3(pE)–42 WT neurons Tau KO neurons WT glia 42 μm P < 0.01 b Viability (% control) Extracellular plaques of amyloid-b and intraneuronal neurofibrillary tangles made from tau are the histopathological signatures of Alzheimer’s disease. Plaques comprise amyloid-b fibrils that assemble from monomeric and oligomeric intermediates, and are prognostic indicators of Alzheimer’s disease. Despite the importance of plaques to Alzheimer’s disease, oligomers are considered to be the principal toxic forms of amyloid-b1,2. Interestingly, many adverse responses to amyloid-b, such as cytotoxicity3, microtubule loss4, impaired memory and learning5, and neuritic degeneration6, are greatly amplified by tau expression. Amino-terminally truncated, pyroglutamylated (pE) forms of amyloid-b7,8 are strongly associated with Alzheimer’s disease, are more toxic than amyloid-b, residues 1–42 (Ab1–42) and Ab1–40, and have been proposed as initiators of Alzheimer’s disease pathogenesis9,10. Here we report a mechanism by which pE-Ab may trigger Alzheimer’s disease. Ab3(pE)–42 co-oligomerizes with excess Ab1–42 to form metastable low-n oligomers (LNOs) that are structurally distinct and far more cytotoxic to cultured neurons than comparable LNOs made from Ab1–42 alone. Tau is required for cytotoxicity, and LNOs comprising 5% Ab3(pE)–42 plus 95% Ab1–42 (5% pE-Ab) seed new cytotoxic LNOs through multiple serial dilutions into Ab1–42 monomers in the absence of additional Ab3(pE)–42. LNOs isolated from human Alzheimer’s disease brain contained Ab3(pE)–42, and enhanced Ab3(pE)–42 formation in mice triggered neuron loss and gliosis at 3 months, but not in a tau-null background. We conclude that Ab3(pE)–42 confers tau-dependent neuronal death and causes template-induced misfolding of Ab1–42 into structurally distinct LNOs that propagate by a prion-like mechanism. Our results raise the possibility that Ab3(pE)–42 acts similarly at a primary step in Alzheimer’s disease pathogenesis. pE-Ab peptides contain an amino-terminal pyroglutamate, whose modification from glutamate is catalysed by glutaminyl cyclase (QC; also known as QPCT)10. The most prominent pE-Ab species in vivo are Ab3(pE)–40, Ab3(pE)–42, Ab11(pE)–40 and Ab11(pE)–42 (ref. 8; Supplementary Fig. 1), with Ab3(pE)–42 being most abundant11. pE-Ab is more cytotoxic12 and aggregates more rapidly13,14 than conventional amyloid-b, and QC activity and pE-Ab levels are increased several-fold in Alzheimer’s disease brain10. Alzheimer’s disease mouse models also indicate a role for pE-Ab in initiating pathology: oral administration of a QC inhibitor led to improved memory and learning, and reduced levels of pE-Ab and conventional amyloid-b10. These data imply that pE-Ab potentiates the neurotoxicity of conventional amyloid-b, but leave open the issue of molecular mechanisms. To address that issue, we compared oligomerization of Ab3(pE)–42, Ab1–42, and mixtures of the peptides in vitro, and analysed responses of primary cultured neurons and glial cells (Supplementary Fig. 2) to the oligomers. At 5 mM peptide, 5% pE-Ab aggregated faster than Ab3(pE)–42 or Ab1–42 alone, based on thioflavin T fluorescence shifts15 (Supplementary 0.1 μM 0.5 μM 1.0 μM 100 80 60 40 20 0 Aβ1–42 Aβ3(pE)–42 Oligomerized Oligomerized together separately 5% Aβ3(pE)–42, 95% Aβ1–42 Figure 1 | Tau-dependent cytotoxicity of oligomers formed by coincubation of Ab3(pE)–42 and Ab1–42. Primary mouse wild-type (WT) and tauknockout (KO) forebrain neurons, and secondary cultures of wild-type mouse glia were treated for 12 h with Ab1–42, Ab3(pE)–42, or 5% Ab3(pE)–42 plus 95% Ab1–42, which were oligomerized for 24 h at 5 mM before dilution into culture media. a, Cells were exposed to calcein-AM and imaged live by epifluorescence microscopy to assay viability16. Extensive death and detachment of cells were observed only for wild-type neurons treated with Ab3(pE)–42 or the 5% Ab3(pE)–42 plus 95% Ab1–42. b, Following peptide treatment, cell viability was analysed by the XTT plate reader assay17. Note the robust cytotoxicity of Ab3(pE)–42 containing solutions at concentrations as low as 0.5 mM, unless Ab3(pE)–42 and Ab1–42 were incubated separately during oligomerization (P , 0.01; yellow stars signify statistical significance of the indicated bar graphs versus vehicle controls; black stars signify statistical significance between the indicated bar graph pairs; mean 6 standard error of the mean (s.e.m.), n 5 9 replicates from 3 independent experiments). 1 Department of Biology, University of Virginia, Charlottesville, Virginia 22904, USA. 2Probiodrug AG, 06120 Halle (Saale), Germany. 3Department of Psychology, University of Virginia, Charlottesville, Virginia 22904, USA. 4Department of Biochemistry and Molecular Biology, University of California at Irvine, Irvine, California 92697, USA. 5Deutsches Zentrum fuer Neurodegenerative Erkrankungen, c/o Leibniz-Institut fuer Neurobiologie, 39118 Magdeburg, Germany. 6JSW Life Sciences GmbH, A-8074 Grambach, Austria. 7Ingenium Pharmaceuticals GmbH, 82152 Munich/Martinsried, Germany. 8 Department of Cell Biology, University of Virginia, Charlottesville, Virginia 22904, USA. *These authors contributed equally to this work. 3 1 M AY 2 0 1 2 | VO L 4 8 5 | N AT U R E | 6 5 1 ©2012 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER Evidence for hybrid oligomers came from immunoprecipitation of various forms of amyloid-b using aggregation-dependent M64, which does not recognize Ab3(pE)–42 (see Supplementary Fig. 4 for characterization of all anti-amyloid-b antibodies used, including M64). Immunoprecipitations were analysed on dot blots using 4G8, which equally recognizes Ab3(pE)–42 and Ab1–42, and anti-pE-Ab, which does not react with Ab1–42. M64 immunoprecipitated oligomers made from Ab1–42 or 5% pE-Ab, but it did not immunoprecipitate Ab3(pE)–42 oligomers, nor monomers of either peptide (Fig. 2a). Because anti-pE-Ab reacted with material immunoprecipitated out of 5% pE-Ab, M64 pulled down hybrid peptide oligomers. Ab3(pE)–42 accounted for ,16% of the amyloid-b in gel-filtered cytotoxic oligomers after 3 h of oligomerization, and steadily dropped to ,8% by 24 h (Fig. 2b). Ab3(pE)–42 thus acts as a template that initiates formation of cytotoxic oligomers. Cytotoxicity was sensitive to oligomerization time (Fig. 2c). Baseline cytotoxicity was observed at all time points for Ab1–42, and for 5% pEAb solutions in which Ab3(pE)–42 and Ab1–42 oligomerized separately. Pure Ab3(pE)–42 killed ,50% of the cells after 24 h of oligomerization, but was virtually non-toxic at 0 h and after 96 h of oligomerization. The most cytotoxic solutions were 5% pE-Ab, in which the constituent peptides co-oligomerized for 24 h. These solutions killed ,60% of the cells within 24 h, and lower but robust cytotoxicity was observed at 96 h. Even the 0 h co-oligomers of 5% pE-Ab exhibited low, significant cytotoxicity. Co-incubated mixtures of 5% Ab3(pE)–42 and 95% ig )– ol pE 42 3( 1– Aβ Aβ 4G8 101 ng 162 ng Anti-pE-Aβ 15.4 ng (5× load) (corrected to 1×) Per cent Aβ3(pE)–42 (of total) 15 10 5 24 h 96 h 80 Viability (% control) P < 0.01 20 0h 100 (5X) b P < 0.01 c 5% a IP: M64 42 om er s Aβ olig om 3 o ( er Aβ lig pE)– s om 42 1– + e 4 Aβ 2 m rs 95 % on 3( Aβ pE om )– 42 1– m ers 42 on om er s OD450 nm/OD490 nm ratio was for 5% pE-Ab, which peaked similarly to Ab3(pE)–42. Ab3(pE)–42, Ab1–42 and 5% pE-Ab thus oligomerized by different pathways. To test whether distinct biological activities were coupled to these oligomerization differences, we compared cytotoxicity of the peptides towards cultured neurons or glia using calcein-AM and fluorescence microscopy16. Twelve hours of Ab1–42 exposure had little effect on cell viability for wild-type or tau-knockout neurons, or wild-type glial cells (Fig. 1a). Contrastingly, most wild-type neurons died and detached from the substrate after exposure to Ab3(pE)–42 or 5% pE-Ab. Tauknockout neurons and wild-type glia, which express little tau, were resistant to Ab3(pE)-42 and 5% pE-Ab. Cytotoxicity dose dependence was examined by incubating wildtype neurons for 24 h in oligomers comprising 0.1, 0.5 or 1 mM peptides, and using the 2,3-bis-(2-methoxy-4-nitro-5-sulphophenyl)2H-tetrazolium-5-carboxanilide (XTT) reduction assay17 (Fig. 1b). Cells were unaffected by Ab1–42, but Ab3(pE)–42 and 5% pE-Ab had substantial cytotoxicity at 0.5 mM and even more at 1.0 mM. Cytotoxicity of 5% pE-Ab required Ab3(pE)–42 and Ab1–42 to incubate together for 24 h before being added to cells. When they were incubated separately for 24 h and mixed together at a 1:19 molar ratio immediately before being applied to cells, they were not cytotoxic. A small amount of Ab3(pE)–42 can thus markedly enhance the cytotoxicity of a large excess of Ab1–42, provided the two peptides oligomerize together. 60 40 20 3 0 3 12 24 48 Oligomerization time before fractionation (h) 0 Aβ1–42 Aβ3(pE)–42 Oligomerized Oligomerized together separately 5% Aβ3(pE)–42, 95% Aβ1–42 Figure 2 | Ab3(pE)–42 and Ab1–42 form metastable, cytotoxic, hybrid oligomers. a, Ab3(pE)–42 and Ab1–42 were incubated together at a 1:19 molar ratio (5% pE-Ab) for 24 h at 1 mM total amyloid-b, and were then immunoprecipitated (IP) with M64, a rabbit monoclonal antibody that specifically recognizes residues 3–7 (EFRH) of Ab1–40 oligomers or fibrils. Additional samples that were immunoprecipitated included otherwise identically treated oligomers made from pure Ab3(pE)–42 or Ab1–42, and monomeric versions of the two peptides. Immunoprecipitated oligomers were converted to monomers by lyophilization, solubilization with HFIP and dilution into PBS, and along with the other samples were dot blotted onto nitrocellulose and analysed using 4G8, a mouse monoclonal antibody that recognizes Ab3(pE)–42 and Ab1–42 equally well, and an antibody that specifically recognizes pE-Ab (see Supplementary Fig. 4 for characterization of all antibodies used here). Quantification of the dot blots using a LI-COR Odyssey imaging station indicated that the oligomers that were immunoprecipitated from the mixed peptide solution contained both Ab3(pE)–42 and Ab1–42, at a molar ratio of ,1:10. b, Solutions containing 5% pE-Ab3(pE)–42 and 95% Ab1–42 were incubated for the indicated times, and then were fractionated by gel filtration. At each time point, fractions that eluted at 12.5 ml, where most cytotoxicity resided (see Fig. 3b) were immunoprecipitated using anti-human amyloid-b (N), an amino-terminal-specific antibody that does not react with pE-Ab (data not shown). The immunoprecipitates were then lyophilized, resolubilized with HFIP, and quantitatively analysed on dot blots with 4G8 and anti-pE-Ab using the LI-COR Odyssey. The time-dependent decrease in the Ab3(pE)–42 content of the immunoprecipitated oligomers implies that Ab3(pE)– 42 initiated formation of hybrid peptide oligomers. c, Ab3(pE)–42 and Ab1–42 oligomerized for 0, 24 and 96 h either separately or together as 1:19 mixtures, and then were added to primary wild-type neuron cultures for 24 h at a final concentration of 1 mM total amyloid-b. Following peptide treatment, cell viability was analysed by the XTT plate reader assay17. The most cytotoxic species observed were the hybrid oligomers after 24 h of oligomerization (P , 0.01; yellow stars signify statistical significance of the indicated bar graphs versus vehicle controls; black stars signify statistical significance between the indicated bar graph pairs; mean 6 s.e.m., n 5 6 or 9 replicates from 3 independent experiments for panel b or c, respectively). 6 5 2 | N AT U R E | VO L 4 8 5 | 3 1 M AY 2 0 1 2 ©2012 Macmillan Publishers Limited. All rights reserved LETTER RESEARCH Ab1–42 can therefore form oligomers whose cytotoxicity is both greater and more enduring than oligomers formed by Ab3(pE)–42 alone. To identify the co-oligomer size(s) that were cytotoxic, amyloid-b solutions were oligomerized for various times from 0–96 h before fractionation by gel filtration. Total amyloid-b in all fractions was determined using 4G8 dot blots that, as shown in Fig. 3a (for 5% pE-Ab) and Supplementary Fig. 5 (for Ab1–42 and Ab3(pE)–42), illustrate the full fractionation range of the column but exclude most void volume fractions. Presumptive monomeric Ab1–42 dominated initially and persisted at 3 h, but was nearly undetectable after 12 h. The 3 h time point also marked the appearance of Ab1–42 oligomers, which gradually increased in size over the next 93 h. Ab3(pE)–42 and 5% pE-Ab oligomerized differently. Putative monomers were present at 0 h for both samples, when slightly larger species, LNOs that possibly corresponded to dimers or trimers (Supplementary Fig. 6), were also present. These persisted as the main species for 24 h for Ab3(pE)–42 and for nearly 72 h for 5% pE-Ab, and later time points were dominated by larger aggregates that eluted in void volume fractions. Cytotoxicity was assayed for individual fractions of 5% pE-Ab that oligomerized for 24 h (Fig. 3b). Most cytotoxicity was associated with the possible dimers/trimers that eluted at 12.5 ml, which at 425 nM peptide killed more than 60% of the cells. Low cytotoxicity was also observed at 554 nM peptide for the larger oligomers that eluted at 8.5 ml. 48 72 96 7.5 8.5 100 80 5 10.5 60 40 20 0 7. P < 0.01 c d N D D 42 5 9.5 0.000625% pE 60 0.0125% pE 0.25% pE 40 Ve (ml) 5% pE 10.5 11.5 20 12.5 0 Viability (% control) 9.5 10.5 11.5 3 2 sa ge as lp Se r ria ia sa as sa ge ge 1 42 1– as 80 60 40 20 12.5 V – 42 e = ,0 1 .44 Aβ 2.5 μM m 1– 5 % V 42 , l e = 0 Aβ 1 .22 2. 3 5 μM V (pE m 5% e = )–42 , l 12 0. 3 A β .5 6 m μM 3 V (pE)– l e = 42 , 12 0.1 .5 8 m μM l at 3( ed pE ct )–42 , io 1 Aβ nat μM ed 1 fra un 1– un Aβ 1– 42 5% 1 95 Aβ –42 % 3(p Aβ E)– 42 5% frac 42 , 1 t Aβ ion μM 0 14 Aβ P < 0.01 100 f 8.5 Se 5% Aβ3(pE)–42, 95% Aβ1–42 7.5 lp Se r ia pa Se ss ria ag l e 1 pa Se ss ria ag l e 2 pa Se ss ria ag l e 3 O lig se om pa er ra ize O tely d lig to om ge er th ize er d 14 Aβ Viability (%) control) 7.5 8.5 80 Ve (ml) N Peptide concentration (nM) 14 e N 50 7 55 4 45 2 12.5 D Ve (ml) 11.5 lp Ve (ml) 9.5 P < 0.01 b .5 11 .5 12 .5 14 24 10 12 5 3 Viability (% control) 0 9. 5 5% Aβ3(pE)–42, 95% Aβ1–42 oligomerization (h) 8. a The marked enhancement of Ab1–42 cytotoxicity by Ab3(pE)–42 suggested a prion-like templating mechanism of Ab1–42 misfolding initiated by Ab3(pE)–42. To test that hypothesis, 5% pE-Ab that oligomerized for 24 h was diluted into 19 volumes of monomeric Ab1–42. A 24 h incubation of this mixture yielded ‘serial passage 1’, which was followed by two equivalent, sequential dilutions into monomeric Ab1–42 to yield serial passages 2 and 3. A gradual loss of cytotoxicity was observed with successive passages, but even passage 3, which contained only 0.000625% Ab3(pE)–42, killed ,50% of the neurons within 24 h (Fig. 3c). Serially passaged gel-filtration samples contained abundant material that eluted at 12.5 ml in passages 1–3, despite the progressive dilution of Ab3(pE)–42 (Fig. 3d). Ab3(pE)–42 can therefore template formation of metastable, cytotoxic LNOs from excess Ab1–42, yielding potent bioactivity that can be serially passaged multiple times into monomeric Ab1–42 without further addition of Ab3(pE)–42. One possible explanation for why Ab1–42 LNOs were inert is that they lacked sufficient properly sized oligomers. Accordingly, we altered the oligomerization protocol from 5 mM peptide for 24 h at 37 uC to 10 mM peptide for 30 min at 4 uC to obtain abundant Ab1–42 oligomers that eluted at 12.5 ml (Fig. 3e). These LNOs were not cytotoxic (Fig. 3f), implying that they were structurally distinct from the putative dimers/trimers initiated by Ab3(pE)–42. This was confirmed by dot blots using M87, a conformation-sensitive anti-amyloid-b antibody, to compare the putative dimers/trimers used for the cytotoxicity assays shown in Fig. 3f. We first lyophilized aliquots of all the amyloid-b solutions, resuspended them with hexafluoroisopropanol (HFIP) to restore them to monomers, and then analysed them using 4G8. When parallel samples that were not lyophilized but were otherwise identical were analysed using M87, immunoreactivity was approximately twice as strong with LNOs made from Ab1–42 versus those made from 5% pE-Ab (Supplementary Fig. 7). Cytotoxic LNOs of 5% pE-Ab are thus structurally distinct from comparably sized LNOs of Ab1–42. Figure 3 | The cytotoxic species are low-n, prion-like oligomers. a, Gelfiltration chromatography was used to fractionate 5% pE-Ab after oligomerization at 5 mM at 37 uC for 0–96 h. The resulting fractions were then converted to monomers using HFIP and analysed on dot blots using monoclonal antibody 4G8. Note the metastable oligomers with an average elution volume (Ve) of 12.5 ml. b, Isolated gel-filtration fractions from the 24 h time point were added to wild-type neuron cultures for 24 h, after which the cells were assayed for cell viability using XTT17. Robust cytotoxicity was associated only with the Ve 5 12.5 ml fraction, although the Ve 5 8.5 ml fraction had low, but statistically significant cell killing activity (P , 0.01; mean 6 s.e.m., n 5 9 replicates from 3 independent experiments). ND, not detected. c, Cytotoxic hybrid oligomers made by co-incubating a 1:19 ratio of Ab3(pE)–42:Ab1–42 for 24 h at 5 mM were diluted into a 19-fold molar excess of freshly dissolved, monomeric Ab1–42, which was then incubated at 5 mM for another 24 h to yield serial passage 1. Two further iterations of this strategy yielded serial passages 2 and 3. The starting material and its serially passaged derivatives were added to wild-type neurons at 1 mM peptide for 24 h, after which cells were analysed using the XTT assay for cell viability17. Only a gradual loss of cytotoxicity was observed with each successive serial passage. d, Each serially passaged sample, as well as otherwise identically prepared oligomers made from pure Ab1–42, were fractionated by gel filtration and analysed on dot blots with 4G8. Note that all serially passaged samples contained metastable LNOs of Ve 5 12.5 ml, which were absent from the pure Ab1–42 samples. e, Ab1–42 (10 mM) that was oligomerized for 30 min at 4 uC, and 5% Ab3pE–42 plus 95% Ab1–42 (5 mM) that was oligomerized for 24 h at 37 uC were fractionated by gel filtration and analysed on dot blots exactly using 4G8. Note the isolation of fractions with Ve 5 12.5 ml from both preparations. f, Wildtype neurons were assayed for viability using the XTT plate reader assay17 following 24 h of exposure to the indicated amyloid-b preparations. Note the minimal cytotoxicity of unfractionated Ab1–42 and Ab1–42 with Ve 5 12.5 ml (P , 0.01, mean 6 s.e.m., n 5 9 replicates from 3 independent experiments). P , 0.01; yellow stars signify statistical significance of the indicated bar graphs versus vehicle controls; black stars and blue stars signify statistical significance between the indicated bar graph pairs; blue mean 6 s.e.m., n 5 9 replicates from 3 independent experiments for panels b, c and f. 3 1 M AY 2 0 1 2 | VO L 4 8 5 | N AT U R E | 6 5 3 ©2012 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER Ve (ml) b WT TBA2.1 TBA2.1 tau KO 8. 4 9. 2 10 .0 10 .8 11 .6 12 .4 13 .2 14 .0 14 .8 15 .6 a AD (62-98) Anti-pE-Aβ AD (17-01) Anti-pE-Aβ AD (10-02) Anti-pE-Aβ Normal (15-02) Anti-pE-Aβ M87 Aβ3(pE)–x M87 M87 GFAP + Hem. M87 Anti-pE-Aβ Normal (47-97) M87 Hem. Anti-pE-Aβ Normal (03-00) 100 μm M87 Figure 4 | Ab3(pE)–42 in vivo. a, Cytosol obtained from human Alzheimer’s disease (AD) and similarly aged normal brains (Supplementary Fig. 9) were fractionated by gel filtration, and analysed by dot blotting with anti-pE-Ab and M87. Note the appearance of pE-Ab in LNO fractions, including those that eluted at 12.4 ml, especially in the Alzheimer’s disease samples. b, Three- month-old TBA2.1 mice generating Ab3(pE)–4218 show amyloid-b deposits (arrows), massive astrogliosis (GFAP) and neuron loss (Hem., haematoxylin nuclear staining), none of which are evident in comparably aged wild-type (WT) mice or TBA2.1/tau-knockout19 (KO) hybrids. Several lines of evidence demonstrate in vivo relevance for the data described so far. First, we identified LNOs containing Ab3(pE)–42 in three out of three Alzheimer’s disease samples, based on gel filtration of human brain extracts followed by dot blots of resulting fractions with anti-pE-Ab and M87. In contrast, only one of three age-matched samples with normal neuropathological diagnoses was positive for Ab3(pE)–42 (Fig. 4a and Supplementary Fig. 8). Second, we crossed TBA2.1 mice18 into a tau-knockout background19. By 3 months, TBA2.1 mice accumulated small amounts (40–100 ng g21 brain weight) of Ab3(pE)–42, which formed primarily intraneuronal aggregates, and was associated with massive hippocampal neuron loss and gliosis18. Knocking out tau provided almost complete protection against neuron loss and glial activation (Fig. 4b). Additional in vivo data are shown in Supplementary Fig. 9. Long-term potentiation (LTP) of mouse hippocampal neurons in slice cultures was potently and equally inhibited by oligomers made from 5% Ab3(pE)–42 or 100% Ab3(pE)–42, whereas Ab1–42 oligomers had no effect on LTP. 1% Ab3(pE)–42 provoked mild, but statistically insignificant LTP impairment (Supplementary Fig. 9a). To evaluate the effects of increased Ab3(pE)–42 in animal models, we crossed mice with neuron-specific expression of human b-amyloid precursor protein (APP) harbouring Swedish and London mutations (hAPPSL)20, with mice expressing human QC21. Nine-month-old double (hAPPSL/hQC) and single (hAPPSL) transgenic mice were indistinguishable in terms of insoluble and soluble Abx–42 levels, but the double transgenics had approximately twofold more insoluble Ab3(pE)–42 and approximately ninefold more soluble Ab3(pE)–42 than single transgenics (Supplementary Fig. 9b). Further analysis of the soluble Abx–42 by the A4 assay22 revealed an approximately eightfold excess of oligomers in the double versus single transgenics (Supplementary Fig. 9c). Double transgenics performed more poorly in Morris water maze tests (Supplementary Fig. 9d) and had reduced hippocampal immunoreactivity for the synapse marker, synaptophysin (Supplementary Fig. 9e). Finally, peri-hippocampal injection of 5% pE-Ab at 5 mM into APPSwDI/NOS22/2 Alzheimer’s disease model mice23 led 3–5 months later to the presence of plaques containing both pE-Ab and conventional amyloid-b. Comparable plaques were rarely seen in sham-injected Alzheimer’s disease mice or in wild-type mice injected with 5% pE-Ab (Supplementary Fig. 9f). These collective in vivo results emphasize the physiological significance of the companion biochemical and cultured cell results. Our studies provide new insights into Alzheimer’s disease pathogenesis by demonstrating that hypertoxic amyloid-b oligomers can be triggered by small quantities of a specifically truncated and posttranslationally modified version of amyloid-b. Although some previous studies demonstrated that pE modification of amyloid-b considerably enhances its aggregation kinetics13,14,24, toxicity12,18,25 and resistance to degradation12, a mechanistic explanation for the unique properties of pE-Ab has been lacking until now. Prior studies suggest coincident appearance of Ab3(pE)–42 with development or progression of human Alzheimer’s disease26,27. Co-localization of QC and Ab3(pE)–42 was found in cored plaques of vulnerable regions in Alzheimer’s disease, and evidence was provided for axonal transport of Ab3(pE)–x from QC-rich neuronal populations of the entorhinal cortex and locus coeruleus28. As LNOs containing Ab3(pE)–42 are reasonably stable (Fig. 3a), they might initiate tau-dependent cytotoxicity intracellularly during axonal transport29 or extracellularly following release at remote hippocampal synapses30 of projection neurons28. The Ab3(pE)–42-induced formation of toxic mixed oligomers provides a rationale for these previous observations, and the taudependent cytotoxicity of 5% pE-Ab establishes a new functional connection between amyloid-b and tau in Alzheimer’s disease pathogenesis. METHODS SUMMARY Full descriptions of thioflavin T assays, cell culture, cell viability assays, procedures for oligomerization of amyloid-b peptides and their fractionation by gel-filtration chromatography, production and specificity of rabbit monoclonal anti-amyloid-b antibodies, immunoprecipitation, dot blots and western blots, generation of hAPPSL/hQC transgenic mice, LTP measurements of mouse hippocampal slice cultures, peri-hippocampal injection of 5% pE-Ab into Alzheimer’s disease model mice, cultured cell and brain immunohistochemistry, and collection of human brain extracts are provided in Supplementary Methods. Received 16 May 2011; accepted 16 March 2012. Published online 2 May; corrected 30 May 2012 (see full-text HTML version for details). 1. 2. 3. 4. Gandy, S. et al. Days to criterion as an indicator of toxicity associated with human Alzheimer amyloid-b oligomers. Ann. Neurol. 68, 220–230 (2010). Walsh, D. M. & Selkoe, D. J. Ab oligomers—a decade of discovery. J. Neurochem. 101, 1172–1184 (2007). Rapoport, M., Dawson, H. N., Binder, L. I., Vitek, M. P. & Ferreira, A. Tau is essential to b-amyloid-induced neurotoxicity. Proc. Natl Acad. Sci. USA 99, 6364–6369 (2002). King, M. E. et al. Tau-dependent microtubule disassembly initiated by pre-fibrillar b-amyloid. J. Cell Biol. 175, 541–546 (2006). 6 5 4 | N AT U R E | VO L 4 8 5 | 3 1 M AY 2 0 1 2 ©2012 Macmillan Publishers Limited. All rights reserved LETTER RESEARCH 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Roberson, E. D. et al. Reducing endogenous tau ameliorates amyloid b-induced deficits in an Alzheimer’s disease mouse model. Science 316, 750–754 (2007). Jin, M. et al. Soluble amyloid b-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration. Proc. Natl Acad. Sci. USA 108, 5819–5824 (2011). Mori, H., Takio, K., Ogawara, M. & Selkoe, D. J. Mass spectrometry of purified amyloid b protein in Alzheimer’s disease. J. Biol. Chem. 267, 17082–17086 (1992). Saido, T. C. et al. Dominant and differential deposition of distinct b-amyloid peptide species, AbN3(pE), in senile plaques. Neuron 14, 457–466 (1995). Jawhar, S., Wirths, O. & Bayer, T. A. Pyroglutamate amyloid-b (Ab): a hatchet man in Alzheimer disease. J. Biol. Chem. 286, 38825–38832 (2011). Schilling, S. et al. Glutaminyl cyclase inhibition attenuates pyroglutamate Ab and Alzheimer’s disease-like pathology. Nature Med. 14, 1106–1111 (2008). Tabaton, M. et al. Soluble amyloid b-protein is a marker of Alzheimer amyloid in brain but not in cerebrospinal fluid. Biochem. Biophys. Res. Commun. 200, 1598–1603 (1994). Russo, C. et al. Pyroglutamate-modified amyloid b-peptides—AbN3(pE)—strongly affect cultured neuron and astrocyte survival. J. Neurochem. 82, 1480–1489 (2002). Schilling, S. et al. On the seeding and oligomerization of pGlu-amyloid peptides (in vitro). Biochemistry 45, 12393–12399 (2006). Schlenzig, D. et al. Pyroglutamate formation influences solubility and amyloidogenicity of amyloid peptides. Biochemistry 48, 7072–7078 (2009). Levine, H. III. Thioflavine T interaction with synthetic Alzheimer’s disease b-amyloid peptides: detection of amyloid aggregation in solution. Protein Sci. 2, 404–410 (1993). Wang, X. M. et al. A new microcellular cytotoxicity test based on calcein AM release. Hum. Immunol. 37, 264–270 (1993). Scudiero, D. A. et al. Evaluation of a soluble tetrazolium/formazan assay for cell growth and drug sensitivity in culture using human and other tumor cell lines. Cancer Res. 48, 4827–4833 (1988). Alexandru, A. et al. Selective hippocampal neurodegeneration in transgenic mice expressing small amounts of truncated Ab is induced by pyroglutamate–Ab formation. J. Neurosci. 31, 12790–12801 (2011). Tucker, K. L., Meyer, M. & Barde, Y. A. Neurotrophins are required for nerve growth during development. Nature Neurosci. 4, 29–37 (2001). Rockenstein, E., Mallory, M., Mante, M., Sisk, A. & Masliaha, E. Early formation of mature amyloid-b protein deposits in a mutant APP transgenic model depends on levels of Ab1–42. J. Neurosci. Res. 66, 573–582 (2001). Jawhar, S. et al. Overexpression of glutaminyl cyclase, the enzyme responsible for pyroglutamate Ab formation, induces behavioral deficits, and glutaminyl cyclase knock-out rescues the behavioral phenotype in 5XFAD mice. J. Biol. Chem. 286, 4454–4460 (2011). Tanghe, A. et al. Pathological hallmarks, clinical parallels, and value for drug testing in Alzheimer’s disease of the APP[V717I] London transgenic mouse model. Int. J. Alzheimers Dis. 2010 (2010). Wilcock, D. M. et al. Progression of amyloid pathology to Alzheimer’s disease pathology in an amyloid precursor protein transgenic mouse model by removal of nitric oxide synthase 2. J. Neurosci. 28, 1537–1545 (2008). 24. He, W. & Barrow, C. J. The Ab 3-pyroglutamyl and 11-pyroglutamyl peptides found in senile plaque have greater b-sheet forming and aggregation propensities in vitro than full-length Ab. Biochemistry 38, 10871–10877 (1999). 25. Wirths, O. et al. Intraneuronal pyroglutamate-Ab 3–42 triggers neurodegeneration and lethal neurological deficits in a transgenic mouse model. Acta Neuropathol. 118, 487–496 (2009). 26. Güntert, A., Dobeli, H. & Bohrmann, B. High sensitivity analysis of amyloid-b peptide composition in amyloid deposits from human and PS2APP mouse brain. Neuroscience 143, 461–475 (2006). 27. Piccini, A. et al. b-Amyloid is different in normal aging and in Alzheimer disease. J. Biol. Chem. 280, 34186–34192 (2005). 28. Hartlage-Rübsamen, M. et al. Glutaminyl cyclase contributes to the formation of focal and diffuse pyroglutamate (pGlu)-Ab deposits in hippocampus via distinct cellular mechanisms. Acta Neuropathol. 121, 705–719 (2011). 29. Vossel, K. A. et al. Tau reduction prevents Ab-induced defects in axonal transport. Science 330, 198 (2010). 30. Wilcox, K. C., Lacor, P. N., Pitt, J. & Klein, W. L. Ab oligomer-induced synapse degeneration in Alzheimer’s disease. Cell. Mol. Neurobiol. 31, 939–948 (2011). Supplementary Information is linked to the online version of the paper at www.nature.com/nature. Acknowledgements The authors are grateful for support from the following sources: the Alzheimer’s Association (grant 4079 to G.S.B.); the Owens Family Foundation (G.S.B.); the Cure Alzheimer’s Fund (G.S.B., C.G.G.); NIH/NIGMS training grant T32 GM008136, which funded part of J.M.N.’s PhD training; NIH/NIA grant R01 AG033069 (C.G.G.); and the German Federal Department of Science and Technology grant 03IS2211F (H.-U.D.). Funding for the UCI-ADRC was provided by NIH/NIA grant P50 AG16573. We also thank H. Dawson and M. Vitek of Duke University for providing the tau-knockout mice. This work fulfilled part of the requirements for the PhD earned by J.M.N. at the University of Virginia. The technical assistance of A. Spano, H.-H. Ludwig, E. Scheel and K. Schulz is gratefully acknowledged. Author Contributions J.M.N. performed most of the biochemical and cell biological experiments; S.S. was the principal force behind the experiments involving hAPPSL/ hQC and TBA2.1/tau-knockout mice, and was aided by B.H.-P. and H.C.; A.S. and T.W. fractionated and analysed human brain extracts; E.S., K.T. and B.W. performed the peri-hippocampal injection experiments; A.H. and C.G.G. produced and characterized the M64 and M87 antibodies; R.R. and K.R. performed the electrophysiology experiments; A.A., W.J. and S.G. performed and analysed the immunohistochemical experiments on TBA2.1 and tau-knockout/TBA2.1 mice; G.S.B. and H.-U.D. initiated and directed the project; G.S.B. was the principal writer of the paper; all of the authors participated in the design and analysis of experiments, and in editing of the paper. Author Information Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of this article at www.nature.com/nature. Correspondence and requests for materials should be addressed to G.S.B. (gsb4g@virginia.edu) or H.-U.D. (Hans-Ulrich.Demuth@probiodrug.de). 3 1 M AY 2 0 1 2 | VO L 4 8 5 | N AT U R E | 6 5 5 ©2012 Macmillan Publishers Limited. All rights reserved SUPPLEMENTARY INFORMATION doi:10.1038/nature11060 FULL METHODS Thioflavin T Assays Peptides were diluted from monomeric solutions in hexafluoroisopropanol (HFIP) into PBS to a final peptide concentration of 5 µM and then were incubated at 37° C for the indicated times. At each time point, samples of 100 µl each were placed in 96-well black-walled plates (Corning) taken, 5 µl of 200 µM thioflavin-T (Sigma-Aldrich) were added to each well, which were then assayed for fluorescence (450 nm excitation, 490 nm emission) using a SPECTRAmax Gemini EM fluorescent plate reader (Molecular Devices). Cell culture Neurons were extracted from the forebrains of E17-E20 C57BL/6 or Tau KO mouse embryos1 trypsinization and mechanical disruption. Cells were plated plated on a poly-L lysine (Sigma-Aldrich) coated substrate in Neurobasal medium supplemented with Glutamax, glucose, B27 supplement, penicillin/streptomycin (Invitrogen) and Cosmic Calf Serum (Thermo Scientific). Four hours after plating, the medium was removed and replaced with otherwise identical medium lacking cosmic calf serum. Cells were grown for 7-8 days prior to Aβ treatment. Glial cells were extracted from whole brain homogenates of 2 day-old pups by trypsinization and plated in Dulbecco’s modified Eagle’s medium supplemented with 10% Cosmic Calf Serum and gentamycin. Following 7 days of growth, cultures were shaken to remove loosely attached cells, split into fresh medium by trypsinization, and grown for 3-4 days prior to Aβ treatment. Cell Viability Assays For light microscopic assays (as in Figure 1), primary neurons grown on poly-L lysine coated glass cover slips were exposed to 1 µM calcein AM (Invitrogen) for 15 minutes at 37° C after exposure to Aβ. Cells were then quickly washed in PBS and inverted onto slides for imaging. Live cells were imaged using epifluorescence illumination with a YFP filter set WWW.NATURE.COM/NATURE | 1 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION mounted on a Zeiss Axiovert 100 inverted microscope. Images were captured by a Hamamatsu 9100-­‐13 ImagEM cooled EMCCD. For quantitation of cell viability primary neurons grown in black-walled 96-well tissue culture plates (Corning) were assayed using Cell Proliferation Kit II (Roche) according to the vendor’s instructions. This method measured reduction of XTT (sodium 3 ́-[1-(phenylaminocarbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro) benzene sulfonic acid hydrate) by live, but not dead cells, and spectrophotometric detection of the reduction product using an iMark Microplate Absorbance Reader (Bio-Rad) to measure absorbance at 450nm. A reference reading at 690 nm was subtracted from all measurements. The mean from a control condition, in which all cells were killed, was also subtracted. Complete cell death was induced by addition of 10 µl of 1M HCl at the same time as peptide treatment. Immediately prior to addition of XTT reagent, 10 µl of 1M NaOH was added to normalize pH. Unpaired, two-tailed t-tests were used to judge statistical significance. Aβ oligomers Peptides were synthesized in 50 µmol scale on an automated Symphony synthesizer (Rainin) applying a Fmoc-strategy. Aβ1-42 was synthesized on Fmoc-Ala-NovaSyn ® TGA resin (Merck Biosciences). Gly-25 and Ser-26 were incorporated using isoacyl dipeptide BocSer(Fmoc-Gly)-OH (4 eq.). Amino acid coupling was achieved using HOBt (4 eq.) / DIPCDI (4.4 eq.) for 2 × 45 min. The resulting 26-O-Isoacyl-β-amyloid(x-42) was purified by RP-HPLC after deprotection. Subsequently, the depsipeptides were dissolved in 0.1 M ammonium bicarbonate (pH 7.4) for 1 hour to initiate isoacyl conversion. The reaction was monitored by analytical RP-HPLC. Analytical HPLC analysis was performed on a 4.6 × 150 mm Source 5RPC column (5 µm; GE Healthcare) with a gradient made of solvent A (0.1% NH4OH in H2O at pH 9) and solvent B (acetonitrile / solvent A 60:40). In case of all N-terminal pyroglutamated peptides the pGlu was incorporated as Boc-pGlu-OH. Lyophilized, synthetic Aβ peptides were dissolved to 1 mM in 1,1,1,3,3,3-hexafluoro-2propanol (HFIP; Sigma-Aldrich). Oligomerization was initiated by adding the dissolved peptides directly to neurobasal medium at final peptide concentrations of 1-10 µM, evaporating the HFIP WWW.NATURE.COM/NATURE | 2 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION using an air stream, and incubating the solutions at 37° C or 4° C for the indicated times. Supplemental cell culture reagents were added following oligomerization, and when applicable, after subsequent fractionation by gel filtration chromatography. Peptide-containing media were added to cultures to achieve the noted final concentrations of total Aβ. Rabbit anti-Aβ monoclonal antibodies Rabbit monoclonal antibodies M64 and M87 were made under contract to Epitomics (Burlingame, CA) using fibrillar Aβ1-42 as an antigen and immunizing New Zealand white rabbits as previously described for preparing OC polyclonal serum2. Pools of hybridomas (approximately 10,000) were screened for those expressing antibodies against Aβ1-42 fibrils, and prefibrillar oligomers or monomeric Aβ, and 120 pools giving at least a 3-fold higher absorbance than background in ELISA assays were selected for further analysis. Secondary screening consisted of probing blots of a medium density array of 130 different preparations of fibrils, prefibrillar oligomers and monomers of Aβ1-42, Aβ1-40, islet amyloid polypeptide, polyQ40, overlapping 15 residue peptide segments of Aβ and amyloid-forming random peptides, and by immunohistochemistry on human AD brain tissue. Pools giving a unique pattern of immunoreactivity on the array or on immunohistochemistry were selected for cloning and further characterization by western blotting and ELISA. Both M64 and M87 recognize aggregated, but not monomeric Aβ, and their epitopes are 3EFRH6 and 3EFRHD7, respectively (Suppplementary Fig. 4). For epitope mapping, a peptide array (PepSpotsTM) consisting of a series of overlapping 10 mers from the -4 position of the Aß sequence to residue 46 covalently bonded via the carboxyl terminus to a cellulose membrane was prepared by JPT Peptide Technologies, GmbH, Berlin, Germany and used according to the manufacturer’s recommendations. Membranes were incubated with 100 ng/ml of primary antibody and then 1 ug/ml goat anti-rabbit secondary conjugated with alkaline phosphatase and visualized with TMB substrate (Promega, Madison, WI). For aggregation kinetics, Aβ1-40 was dissolved in a 1:1 mixture of water and acetonitrile, aliquoted into 0.3 mg samples and lyophilized overnight. Prep A: 33 µL of 10% SDS was added WWW.NATURE.COM/NATURE | 3 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION to 0.3 mg lyophilized Aβ1-40 to obtain a 2 mM Aβ1-40 solution. This solution was then boiled for 10 minutes and diluted with 1.5 ml of water to obtain a 47 uM Aß solution, which was incubated at room temperature without stirring for 7 days. Prep B: 0.3 mg lyophilized Aβ1-40 was dissolved in 200 µl of HFIP and HFIP was evaporated under a stream of air to leave a dry film of peptide in the centrifuge tube. This film was then dissolved in 33 µl of 100 mM NaOH to yield a 2 mM solution. This solution was incubated at room temperature for 15 minutes and then diluted with 1.5 ml 20 mM sodium phosphate buffer to yield a final concentration of 47 uM Aβ. This prep was then incubated at room temperature without stirring for 7 days. Prep C: 200 µl HFIP was added to 0.3 mg Aβ1-40. This solution was allowed to incubate for 15 minutes at room temperature in a centrifuge tube with the cap closed. Following the incubation, 700 µl of water was added to the tube to yield a 47 uM Aβ solution in 22% HFIP. This prep was then kept in a centrifuge tube with a perforated top under a fume hood at room temperature with stirring at 180 RPM for 7 days. Prep D: 33 µL 100 mM NaOH was added to 0.3 mg lyophilized Aβ1-40 to yield a 2 mM solution. This solution was then incubated at R/T for 15 minutes. Following the incubation, the solution was diluted with 1.5 mL of 10 mM HEPES/ 100 mM NaCl to yield a 47 uM solution and incubate at room temperature for 7 days without stirring. Oligomer fractionation and immunoprecipitation 1 ml oligomer solutions of synthetic peptides were fractionated on a 30 x 0.8 cm Superdex-75 column (GE Healthacare) using 50 mM ammonium acetate or neurobasal medium (when added to cultures) as the mobile phase. Fractions of 1 ml each were collected, and native samples from each fraction were lyophilized and re-solubilized with HFIP to restore peptides to the monomeric state, as described earlier in the "Aβ oligomers" section. The monomeric peptides were then applied directly onto nitrocellulose using a BioDot (Bio-Rad) dot-blot manifold. Gel filtration fractions of human brain extracts (Figure 4a) were blotted directly onto nitrocellulose, without prior lyophilization and re-solubilization with HFIP. All dot blots were imaged on a Odyssey (LI-COR Biosciences) imaging station, and a standard curve of known monomeric Aβ concentration was used for quantitation of gel filtration fractions of synthetic peptides. Signal intensity was linear from a few ng to a few hundred µg of peptide, with the exact range depending on which primary anti-Aβ antibody was used. Unpaired, two-tailed t-tests were used to judge statistical significance for dot blot samples that were compared quantitatively. WWW.NATURE.COM/NATURE | 4 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION The mouse monoclonal antibodies, 4G8 and anti-human amyloidβ (N), were purchased from Covance and Immuno-Biological Laboratories (clone 82E1), respectively. Immunoprecipitations were performed by binding purified M64 or anti-human amyloidβ (N) IgG to protein A or protein G magnetic beads (New England Biolabs) according to the manufacturers instructions. Following antibody binding, peptides in tissue culture media were incubated with the beads and then washed with PBS before collection of the beads using a magnet. Peptides were removed from beads by addition of HFIP, which also returned them to the monomeric state. The 23.6 ml (30 x 1.0 cm) Superdex-75 column was calibrated with both globular protein standards (Sigma-Aldrich: aprotinin, MW 6500; horse heart cytochrome c, MW 12400; bovine erythrocyte carbonic anhydrase, MW 29000; bovine serum albumin, MW 66000; and blue dextran, MW 2000000) and globular dextran standards (Pharmacosmos: MWs 4400, 9890, 21400, 43500, 66700 and 123600). Equations relating elution volumes to MW were generated by plotting Kav versus MW for each MW standard, where Kav = (Ve-Vo)/Vt-Vo), and Ve = elution volume, Vo = void volume of the column and Vt = total column volume. Elution volumes were determined by measuring absorbance at 280 nm for proteins and by a colorimetric assaay for dextrans3 Tau-KO/TB2.1 Animals The tau-KO mouse line Mapttm1(EGFP)Klt/J (Stock No. 004779) was purchased from The Jackson Laboratory (JAX, Maine, USA). These mice have a C57BL/6 x 129S4/SvJae hybrid background and were generated by knock-in of the EGFP coding sequence into the first exon of the MAPT gene4. Homozygotes are viable, fertile, normal in size and do not display any gross physical or behavioral abnormalities. No MAPT gene product is detected, while cytoplasmic EGFP signal is detected in the CNS during development and at an adult age (JAX strain datasheethttp://jaxmice.jax.org/strain/004779.html, The Tau-KO line was used in crossbreeding experiments with TBA2.1 mice (genetic background C57Bl/6/DBA hybrids) yielding double homozygous mice (Tau-KO/TB2.1) and control genotype combinations. TBA2.1 mice over expresses the peptide AβQ3-42 driven by the Thy-1 promotor. The regulatory element flanks the coding sequence for a fusion protein consisting of the pre-pro-peptide of murine thyrotropin releasing hormone (TRH, Thyroliberin), fused to the N terminus of the modified human Aβ WWW.NATURE.COM/NATURE | 5 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION polypeptide, AβQ3–42. Prohormone convertase (PC) cleavage within the trans-Golgi and secretory vesicles liberates the N-truncated Aβ species5. Immunohistochemical Studies For immunohistochemistry, mice were deeply anesthetized, transcardially perfused with phosphate-buffered saline (PBS) and brains were removed. Tissue was then immersion-fixed in IHC zinc fixative (BD Pharmingen), dehydrated, embedded in low-melting-point paraffin (DCS Innovative Diagnostic Systems) and sectioned at 8 µm on a rotating microtome. After deparaffinization, sections were incubated with the primary antibodies overnight at 4°C. The following antibodies were used in this study: glia-specific antibody: GFAP (rabbit polyclonal, Z0334; DAKO Cytomation, Glostrup, Denmark, dilution 1:5,000), pE3-Aβ-specific antibody: Pyro-Glu Abeta (rabbit polyclonal, 218003; Synaptic Systems, Göttingen, Germany, dilution 1:500). For immunodetection, a biotinylated rabbit-specific IgG (Vector Laboratories, Burlingame, CA, USA) was used, followed by diaminobenzidine staining (Vectastain ABC-Kit, Vector Laboratories). 3,3’-diaminobenzidine (ImmPACT DAB Peroxidase Substrate, Vector Laboratories) was used for visualization. Injection of AD model mice with 5% pE-Aβ and immunmohistochemical analysis Mice were anesthetized with isoflurane and mounted in a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA, USA). The scalp of each animal was retracted and the skull was adjusted to place bregma and lambda in the same horizontal plane. Small burr holes were drilled at the appropriate injection sites. Injectors (23 gauge) were inserted bilaterally at the following positions relative to bregma (mm): AP -2, ML ±1.5, DV: -2.3. Solutions of co-oligomerized 5% Aβ3(pE)-42 plus 95% Aβ1-42, or 0.1M PBS was infused into the hippocampus (1 µl/side; 0.1 µl/min). Infusion was followed by a ten minute diffusion period, during which injectors remained in place. A dental cement cap (Harry J. Bosworth Company, Skokie, IL, USA) was used to secure the incised scalp and protect the infused area. Mice were transcardially perfused with 0.1 M PB followed by 4% paraformaldehyde at 8 or 18 weeks following the infusion procedure. Brains were post-fixed in paraformaldehyde for 24 hours. Analysis of hAPPSL/hQC mice WWW.NATURE.COM/NATURE | 6 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Transgenic APPSL mice6 (C57BL/6xDBA background) were intercrossed with human QC transgenic mice7 (C57Bl/6 background) for generation of double transgenic mice and littermate control animals. Mice were housed in individually ventilated cages (IVCs) under a constant light-cycle (12 hours light/dark). Normal tap water was available to the animals ad libitum. Animals were housed in individual ventilated cages (IVCs) on standardized rodent bedding. Each cage contained a maximum of five mice. The room temperature during the study was maintained at 24°C and the relative humidity was maintained between 40 to 70 %. Animals were housed under a constant day/night cycle (12 hours light/dark). Dried, pelleted standard rodent chow (Altromin®) and normal tap water were available to the animals ad libitum. The Morris water maze task was conducted in a black circular pool of 100 cm diameter. Tap water was filled and a temperature of 22 ± 1°C was maintained. During the whole test session the platform was located in the southwest quadrant of the pool. Each mouse had to perform three trials on four consecutive days. A single trial lasted for a maximum of one minute. During this time, the mouse had the chance to find the hidden, diaphanous target. If the animal did not find the platform the investigator guided to or placed the mouse on it at the end of each trial. After each trial, mice were allowed to rest on the platform for 10–15 sec. For the quantification of escape latency (the time [second] - the mouse needed to find the hidden platform and therefore to escape from the water), of pathway (the length of the trajectory [meter] to reach the target) and of the abidance in the goal quadrant a computerized tracking system was used. Monitoring stops when the mouse sits on the platform for more than 2 sec. At least one hour after the last trial on day 4, mice had to fulfill a so-called probe trial (PT). During the probe trial (PT), the platform was removed from the pool and the number of crossings over the former platform position and the abidance in this quadrant was measured. After the last behaviour test animals were sacrificed and blood, CSF and brains were collected. Therefore, mice were sedated by standard inhalation anesthesia. Following blood sampling, mice were transcardially perfused with physiological (0.9%) saline. Thereafter, brains were removed and hemisected. The hemisected brain regions were frozen immediately and stored at -80°C until shipment to the sponsor. The right hemisphere of all mice was prepared for histology. WWW.NATURE.COM/NATURE | 7 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION To perform ELISA analysis, brain tissue was homogenized in TBS (20 mM Tris, 137 mM NaCl, pH 7.6) containing protease inhibitor cocktail (Complete Mini, Roche, Switzerland), sonicated and centrifuged at 75,500 x g for 1 hour at 4°C. The supernatant was stored at –80°C and Aβ peptides were sequentially extracted with TBS/1% Triton X-100 (TBS/triton fraction), 2% SDS in distilled water (SDS fraction), and 70% formic acid (formic acid fraction). The combined SDS and FA fractions were considered as the insoluble pool of Aβ. Aβx–42 and Aβ3(pE)– 42 specific sandwich ELISAs (IBL-Hamburg, Germany) were performed according to the manufacturer’s manual. In addition, the TBS fractions were applied to oligomer concentration analysis, which is based on matrix-based isolation of aggregates, dissociation and immunodection8. Briefly, using a proprietary sample enrichment protocol, only the aggregated Aβ was isolated from TBS fractions. Each sample was then disaggregated to allow detection of monomeric Aβ using an immunoassay based on an europium fluorescent beads coupled to the 4G10 antibody (N-terminal) and magnetic beads coupled to the antibodies 1F8 and 2H12 (Cterminal) recognizing Aβ1-40 and Aβ1-42, respectively. The europium fluorescence intensity was measured using Time Resolved Fluorescence (TRF). The signal/noise cutoff value for all experiments was 2.0, equaling two times the background signal from buffer alone. The mean of three determinations from one TBS sample was calculated and applied for comparison of the experimental groups (Figure 5). The right hemispheres of all mice were fixed by immersion in freshly prepared 4% paraformaldehyde/PBS (pH 7.4) for one hour at room temperature. Thereafter brains were transferred to a 15% sucrose PBS solution for 24 hours to ensure cryoprotection. On the next day brains were frozen in isopentane and stored at -80° C until used for histological analysis. 15 cryo-sections per layer (altogether 5 layers), each 10µm thick were sagittally cut (Leica CM 3050S). Brain levels were chosen according to the morphology atlas “The Mouse Brain” from Paxinos and Franklin (2nd edition). The cut of the five levels started with a random slice corresponding to figure 105 (total appearance of the dentate gyrus), then sampling was continued uniformly and systematically, always retaining 15 slices per level in series and discarding 100 µm in between the levels. Plaque load was determined with 6E10 primary antibody (Covance, SIG-39320; 1:1000 dilution) directed against the human amyloid peptide (amino acids 1-16) as WWW.NATURE.COM/NATURE | 8 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION well as pGlu Abeta (anti-Abeta N3pE; antibody provided by Probiodrug) clone 6 and ThioflavinS (Sigma ®) staining against beta-sheet structures in a double incubation. For analysis of synaptic density, hippocampal CA3 and dentate gyrus regions were quantitated in sections stained with synaptophysin monoclonal antibody (Chemicon; 1:5000 dilution). The number of synapses was quantified in 9 images per animal and region, whereas three 1000-fold magnified images were recorded from the granular layer of the dentate gyrus (medial blade) regions. Object count was classified into constituent and cohesive synapses, total area of cohesive synapses was divided by the mean measured size of constituent synapses in the image, the outcome was added to the number of constituent synapses and built the number of total synapses. Electrophysiological analysis of hippocampal slice cultures Hippocampal slices (400 µm thick) were prepared from 4-months-old male C57/Bl6 mice (Institute breeding stock) as described previously9. Briefly, both hippocampi were isolated and transferred into a pre-chamber containing 8 ml permanently carbogen-gasified artificial cerebrospinal fluid (ACSF), to allow Aβ application. The lyophilized Aβ peptides were dissolved in HFIP to 1 mM. The solution was aliquoted, HFIP was evaporated and the peptide stored at 80°C. Then, Aβ was dissolved (100 µM) in dimethylsulfoxide (DMSO; Sigma, St. Louis, MO), sonicated and diluted to 20µM in F12/DMEM (without glutamine, Biochrom). Peptidecontaining media were added to the pre-chamber to achieve the final concentrations of total Aβ. Slices were transferred into a submerged-type recording chamber and were allowed to recover for at least 30 min before the experiment started. The chamber was constantly perfused with artificial cerebrospinal fluid (ACSF) at a rate of 2.5 ml/min at 33±1 °C. Synaptic responses were elicited by stimulation of the Schaffer collateral–commissural fibers in the stratum radiatum of the CA1 region using lacquer-coated stainless steel stimulating electrodes. Glass electrodes (filled with ACSF, 1–4 MΩ) were placed in the apical dendritic layer to record field excitatory postsynaptic potentials (fEPSPs). The initial slope of the fEPSP was used as a measure of this potential. The stimulus strength of the test pulses was adjusted to WWW.NATURE.COM/NATURE | 9 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION 30% of the EPSP maximum. During baseline recording, single stimuli were applied every minute. Once a stable baseline had been established, long-term potentiation was induced by applying 100 pulses at an interval of 10 ms and a width of the single pulses of 0.2 ms (strong tetanus) three times at 10 min intervals. Human brain extracts Frozen brain tissue was obtained from the Institute for Memory Impairment and Neurodegenerative Diseases, and was collected from individuals enrolled in the UC Irvine Alzheimer Disease Center in full compliance with institutional review board (IRB), state of California and US federal regulations. Soluble lysates were prepared from frozen frontal cortex (Boradman’s B11) as previously described10. Briefly, frozen tissues were weighted, diced and homogenized in freshly prepared ice-cold PBS, 0.02% NaN3, pH 7.4 with protease inhibitor cocktail (4:1 PBS volume/brain wet weight). The samples were ultracentrifuged at 100,000 x G for 1 hr at 4oC. The PBS soluble fraction was collected, the protein concentration normalized and aliquoted and stored at –80oC for further testing. Mouse husbandry All experiments involving mice were performed with full approval of the Institutional Animal Use and Care Committee of the University of Virginia; the German animal protection law §11; TVA 55.2-1-54-2531-135-07 allowance to Ingenium GmbH, Bavaria, Germany; the ethics committee of the German federal state of Sachsen-Anhalt to the Institute for Neurobiology, Magdeburg (performed in accordance with the European Communities Council Directive 86/609/EEC); or in conformity with the Austrian Animal Experiments Act BGBl. Nr. 501, especially Part III. § 11 and Part V. § 15 und § 16. References 18. Alexandru, A. et al. Selective hippocampal neurodegeneration in transgenic mice expressing small amounts of truncated Abeta is induced by pyroglutamate-­‐Abeta formation. J. Neurosci. 31, 12790-­‐12801 (2011). WWW.NATURE.COM/NATURE | 10 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION 19. Tucker, K. L., Meyer, M. & Barde, Y. A. Neurotrophins are required for nerve growth during development. Nature Neurosci. 4, 29-­‐37 (2001). 20. Rockenstein, E., Mallory, M., Mante, M., Sisk, A. & Masliaha, E. Early formation of mature amyloid-­‐beta protein deposits in a mutant APP transgenic model depends on levels of Abeta(1-­‐42). J. Neurosci. Res. 66, 573-­‐582 (2001). 21. Jawhar, S. et al. Overexpression of glutaminyl cyclase, the enzyme responsible for pyroglutamate A{beta} formation, induces behavioral deficits, and glutaminyl cyclase knock-­‐out rescues the behavioral phenotype in 5XFAD mice. J. Biol. Chem. 286, 4454-­‐ 4460 (2011). 22. Tanghe, A. et al. Pathological Hallmarks, Clinical Parallels, and Value for Drug Testing in Alzheimer's Disease of the APP[V717I] London Transgenic Mouse Model. Inter. J. Alzheimer's Dis. 2010, doi:10.4061/2010/417314 (2010). 31. Dawson, H. N. et al. Inhibition of neuronal maturation in primary hippocampal neurons from tau deficient mice. J Cell Sci. 114, 1179-­‐1187 (2001). 32. Kayed, R. et al. Fibril specific, conformation dependent antibodies recognize a generic epitope common to amyloid fibrils and fibrillar oligomers that is absent in prefibrillar oligomers. Mol. Neurodegener. 2, 18 (2007). 33. Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A. & Smith, F. Colorimetric Method for Determination of Sugars and Related Substances. Anal. Chem. 28, 350-­‐356 (1956). 34. Ronicke, R. et al. Early neuronal dysfunction by amyloid beta oligomers depends on activation of NR2B-­‐containing NMDA receptors. Neurobiol. Aging 32, 2219-­‐2228 (2011). WWW.NATURE.COM/NATURE | 11 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION 35. Tomic, J. L., Pensalfini, A., Head, E. & Glabe, C. G. Soluble fibrillar oligomer levels are elevated in Alzheimer's disease brain and correlate with cognitive dysfunction. Neurobiol. Dis. 35, 352-­‐358 (2009). WWW.NATURE.COM/NATURE | 12 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 1 | Comparison of pE-Aβ versus conventional Aβ peptides. Amino acid sequences of the four major pE-Aβ species, and conventional Aβ1-40 and Aβ1-42 are shown. WWW.NATURE.COM/NATURE | 13 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 2 | Cellular and protein content of primary neuron and glial cultures. a, Primary WT neuron cultures were labeled with DAPI to mark all nuclei and with mouse monoclonal anti-tau (tau-5) to identify neurons. Out of 365 DAPI-positive nuclei counted WWW.NATURE.COM/NATURE | 14 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION in a total of 10 fields of view, 303, or 82.9% were tau-positive neurons. b, Primary tau KO neuron cultures were labeled for double immunofluorescence with rabbit polyclonal anti-MAP2 and tau-5. Note the absence of detectable tau. c, Unfractionated homogenates of primary WT and tau KO brains were analyzed by western blotting with tau-5 and anti-α-tubulin. Note the absence of detectable tau in the tau KO sample. d, Primary glial cell cultures were labeled for double immunofluorescence with rabbit polyclonal anti-GFAP to identify astrocytes, and with tau-5. The few weakly tau-5-positive cells that were seen were not labeled by anti-GFAP, and their morphology suggests they were tau-positive oligodendrocytes1. Supplementary Figure 3 | Aβ3(pE)-42, Aβ1-42 and 1:19 mixtures of the two peptides oligomerize by distinct pathways. Assembly of the peptides at 5 µM beyond the stage of small oligomers was monitored by shifts in the fluorescence of bound thioflavin T2. WWW.NATURE.COM/NATURE | 15 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 4 | Characterization of antibody specificity. a, Serial dilutions of monomeric or oligomeric Aβ3(pE)-42 and Aβ1-42 were analyzed on native dot blots with the indicated antibodies. Note that anti-pE-Aβ did not recognize Aβ1-42, M87 did not recognize Aβ3(pE)-42, and that 4G8 reacted equally well with both peptides. b, Epitope mapping of M64 and M87 monolconal antibodies. A peptide array (PepSpotsTM) consisting of 40 overlapping 10mers from the -4 position of the Aβ sequence to residue 46 covalently bonded via the carboxyl terminus to a cellulose membrane was stained with M64 and M87. The first residue of the 10mer sequence is indicated above the top row of spots and below the second row of spots. The line indicates immunoreactive spots. c, Interpretation of the epitope mapping data. Peptide WWW.NATURE.COM/NATURE | 16 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION segments containing the sequence common to all immunoreactive spots is shown in red (M64) and blue (M87). Kinetics of formation of M64 and M87 immunoreactivity. Aβ40 monomer was aggregated under 4 different conditions (A-D, see Full Methods), and samples were spotted on nitrocellulose membranes at the indicated times and visualized by M64 and M87. Neither antibody recognized any of the monomeric samples at time 0 or the sample boiled in SDS (condition A) at any time examined. M64 immunoreactivity was observed after 1 day of incubation for condition D and after 4 days of incubation for condition C, but at no time for condition B. M87 immunoreactivity was observed between 1 day and 6 days of incubation for condition D, and weak immunoreactivity was seen for condition B between 3 and 6 days. Taken together, these results indicate that while the linear epitopes of M64 and M87 overlap, they recognize distinct conformations of this segment. Supplementary Figure 5 | Gel filtration of Aβ1-42 and Aβ3(pE)-42 oligomers (companion data for Fig. 3a). Gel filtration chromatography on a 23.6 ml Superdex-75 column (see Supplementary Fig. 6 for column calibration) was used to fractionate the indicated peptides after oligomerization at 5 µM total Aβ for various times from 0-96 hours at 37° C. The resulting fractions were then converted to monomers using HFIP and analyzed on dot blots using monoclonal antibody 4G8. Note the low-n oligomers with an average elution volume (Ve) of 12.5 ml formed by Aβ3(pE)-42 (b) but not by Aβ1-42 (a). WWW.NATURE.COM/NATURE | 17 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 6 | Calibration of the gel filtration column. a, A 23.6 ml (30 x 1 cm) Superdex-75 column was calibrated with globular protein and globular dextran MW standards, as described in the Full Methods section. b, The equations shown in panel a were used to calculate MW for typical column fractions. Values in red indicate material in the void volume, where MW cannot be accurately determined. Note also that the remaining values are valid for globular proteins, but apparently are ~3-fold overestimates of the MW of Aβ monomers and small oligomers. This is because there is no assurance that any of the fractionated Aβ species were also globular, so calculations of their molecular weights based on the calibration equations should be regarded as crude estimates. Indeed, the final, and thereby smallest peptides to elute from the column, with an average elution volume of 14 ml, were found only at the earliest time points and were thus most likely monomers of ~4500 molecular weight. Their calculated molecular weights based on the calibration standards were ~2-4-fold higher, however, implying that the calibration standards provide overestimates of the actual molecular weight of Aβ monomers and LNOs. WWW.NATURE.COM/NATURE | 18 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 7 | Low-n oligomers of 5% pE-Aβ and Aβ1-42 are structurally distinct. a, Dot blots are shown for three preparations each (RepA, RepB and RepC) of Ve = 12.5 ml fractions made from pure Aβ1-42 or 5% pE-Aβ. The same preparations were used in the viability assays shown in Fig. 3f. The peptides were converted to monomers with HFIP for the 4G8 blots, but not for the blots using M87, a conformation-sensitive rabbit monoclonal antibody (Supplementary Fig. 4). b, Quantitation of the dot blots in panel a using the LI-COR Odyssey. Note that M87 was ~2-fold more immunoreactive with Ve = 12.5 ml fractions made from Aβ1-42 versus 5% pE-Aβ, thus indicating conformational differences independently of oligomer size (p<0.01; yellow [ ] star signifies statistical significance of the indicated bar graph versus vehicle controls; black [ ] star signifies statistical significance between the indicated bar graph pairs; mean ± SEM, n= 6 replicates from 3 independent experiments). WWW.NATURE.COM/NATURE | 19 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 8 | Sources of human brain extracts for fractionation and Aβ oligomer analysis. The data shown in Fig. 4a were derived from these AD patients and age matched normal controls. WWW.NATURE.COM/NATURE | 20 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION WWW.NATURE.COM/NATURE | 21 RESEARCH SUPPLEMENTARY INFORMATION doi:10.1038/nature11060 Supplementary Figure 9 | Aβ3(pE)-42 in vivo. a, Influence of Aβ3(pE)-42, Aβ1-42 and 5% Aβ3(pE)-42 on LTP in hippocampal slices. Aβ3(pE)-42 and 5% Aβ3(pE)-42 reduced the LTP significantly compared to control (repeated measures ANOVA, *, P<0.05, n>21) at a final concentration of 250 nM. b-e, Characterization of hAPPSL/hQC mice compared to the parental hAPPSL strain at 9 months of age. b, Quantification of Aβ3(pE)-42 and Aβx-42 in water soluble (TBS) and insoluble (SDS and formic acid) extracts (Student’s t-test, *, P<0.05, n=12 hAPPSL, n=9 hAPPSL/hQC, mean ± SEM). c, Aβ oligomer concentration in TBS extracts from hAPPSL and hAPPSL/hQC mice. The analysis is based on template-recognition and adsorption of oligomers (Amorfix Ltd.) (Mann-Whitney test, *, P<0.05, n=6 hAPPSL, n=5 hAPPSL/hQC, mean ± SEM). d, Assessment of spatial learning and memory in a Morris water maze paradigm. hAPPSL/hQC mice showed impaired spatial memory in the probe trial of the test. Differences in the spatial learning were not obvious (inset). (One-way ANOVA followed by Newman-Keuls post-hoc analysis, *, P<0.05, n=10 APPSL, hAPPSL/hQC n=18, WT n=20, mean ± SEM). e, Quantitative image analysis of synaptophysin immunoreactivity in hippocampus. Double transgenics showed a significantly reduced immunopositive area (one-way ANOVA followed by Tukey; *, P<0.05 and ** P<0.01 hAPPSL vs. hAPPSL/hQC, ## ,P<0.01 WT vs. hAPPSL/hQC; n=7-8). f, 3 month old AD model mice3 were injected peri-hippocampally with ~1 µl each of 5% pE-Aβ oligomers. 5 months later the mice were sacrificed and brain sections were stained with mAb87 and anti-pE-Aβ. Plaques containing pE-Aß were much less abundant than plaques containing conventional Aβ (left panels), but both classes of peptides were extensively co-localized in the pE-Aβ-positive plaques (right panels). References 1. LoPresti, P., Szuchet, S., Papasozomenos, S. C., Zinkowski, R. P. & Binder, L. I. Functional implications for the microtubule-associated protein tau: localization in oligodendrocytes. Proc. Nat. Acad. Sci. USA 92, 10369-10373 (1995). 2. LeVine, H., 3rd. Thioflavine T interaction with synthetic Alzheimer's disease beta-amyloid peptides: detection of amyloid aggregation in solution. Protein Sci. 2, 404-410 (1993). WWW.NATURE.COM/NATURE | 22 doi:10.1038/nature11060 RESEARCH SUPPLEMENTARY INFORMATION 3. Wilcock, D. M. et al. Progression of amyloid pathology to Alzheimer's disease pathology in an amyloid precursor protein transgenic mouse model by removal of nitric oxide synthase 2. J. Neurosci. 28, 1537-1545 (2008). WWW.NATURE.COM/NATURE | 23