Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-β

LETTERdoi:10.1038/nature11060

Prion-like behaviour and tau-dependentcytotoxicity of pyroglutamylated amyloid-bJustin M. Nussbaum1*, Stephan Schilling2*, Holger Cynis2, Antonia Silva1, Eric Swanson1, Tanaporn Wangsanut1, Kaycie Tayler3,Brian Wiltgen3, Asa Hatami4, Raik Ronicke5, Klaus Reymann5, Birgit Hutter-Paier6, Anca Alexandru7, Wolfgang Jagla7,Sigrid Graubner7, Charles G. Glabe4, Hans-Ulrich Demuth2,7 & George S. Bloom1,8

Extracellular plaques of amyloid-b and intraneuronal neurofibrillarytangles made from tau are the histopathological signatures ofAlzheimer’s disease. Plaques comprise amyloid-b fibrils that assemblefrom monomeric and oligomeric intermediates, and are prognosticindicators of Alzheimer’s disease. Despite the importance of plaquesto Alzheimer’s disease, oligomers are considered to be the principaltoxic forms of amyloid-b1,2. Interestingly, many adverse responses toamyloid-b, such as cytotoxicity3, microtubule loss4, impaired memoryand learning5, and neuritic degeneration6, are greatly amplified bytau expression. Amino-terminally truncated, pyroglutamylated (pE)forms of amyloid-b7,8 are strongly associated with Alzheimer’sdisease, are more toxic than amyloid-b, residues 1–42 (Ab1–42) andAb1–40, and have been proposed as initiators of Alzheimer’s diseasepathogenesis9,10. Here we report a mechanism by which pE-Ab maytrigger Alzheimer’s disease. Ab3(pE)–42 co-oligomerizes with excessAb1–42 to form metastable low-n oligomers (LNOs) that are struc-turally distinct and far more cytotoxic to cultured neurons thancomparable LNOs made from Ab1–42 alone. Tau is required forcytotoxicity, and LNOs comprising 5% Ab3(pE)–42 plus 95% Ab1–42

(5% pE-Ab) seed new cytotoxic LNOs through multiple serial dilu-tions into Ab1–42 monomers in the absence of additional Ab3(pE)–42.LNOs isolated from human Alzheimer’s disease brain containedAb3(pE)–42, and enhanced Ab3(pE)–42 formation in mice triggeredneuron loss and gliosis at 3 months, but not in a tau-null background.We conclude that Ab3(pE)–42 confers tau-dependent neuronal deathand causes template-induced misfolding of Ab1–42 into structurallydistinct LNOs that propagate by a prion-like mechanism. Our resultsraise the possibility that Ab3(pE)–42 acts similarly at a primary step inAlzheimer’s disease pathogenesis.

pE-Ab peptides contain an amino-terminal pyroglutamate, whosemodification from glutamate is catalysed by glutaminyl cyclase (QC;also known as QPCT)10. The most prominent pE-Ab species in vivo areAb3(pE)–40, Ab3(pE)–42, Ab11(pE)–40 and Ab11(pE)–42 (ref. 8; Supplemen-tary Fig. 1), with Ab3(pE)–42 being most abundant11. pE-Ab is morecytotoxic12 and aggregates more rapidly13,14 than conventionalamyloid-b, and QC activity and pE-Ab levels are increased several-foldin Alzheimer’s disease brain10. Alzheimer’s disease mouse models alsoindicate a role for pE-Ab in initiating pathology: oral administration ofa QC inhibitor led to improved memory and learning, and reducedlevels of pE-Ab and conventional amyloid-b10. These data imply thatpE-Ab potentiates the neurotoxicity of conventional amyloid-b, butleave open the issue of molecular mechanisms. To address that issue, wecompared oligomerization of Ab3(pE)–42, Ab1–42, and mixtures of thepeptides in vitro, and analysed responses of primary cultured neuronsand glial cells (Supplementary Fig. 2) to the oligomers.

At 5mM peptide, 5% pE-Abaggregated faster than Ab3(pE)–42 or Ab1–42

alone, based on thioflavin T fluorescence shifts15 (Supplementary

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 thanfor Ab1–42, but peaked at a ,25% lower level. The fastest rise in the

*These authors contributed equally to this work.

1Department 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/oLeibniz-Institut fuer Neurobiologie, 39118 Magdeburg, Germany. 6JSW Life Sciences GmbH, A-8074 Grambach, Austria. 7Ingenium Pharmaceuticals GmbH, 82152 Munich/Martinsried, Germany.8Department of Cell Biology, University of Virginia, Charlottesville, Virginia 22904, USA.

Via

bili

ty (%

co

ntr

ol)

100

0.1 μM 0.5 μM 1.0 μM P < 0.01

Aβ1–42 Aβ3(pE)–42 Oligomerized

separately

Oligomerized

together

5% Aβ3(pE)–42, 95% Aβ1–42

b

Vehicle 95% Aβ1–42

80

60

40

20

0

42 μm

a

WT glia

WT

neurons

Tau KO

neurons

Aβ1–42 Aβ3(pE)–42 5% Aβ3(pE)–42

Figure 1 | Tau-dependent cytotoxicity of oligomers formed by co-incubation of Ab3(pE)–42 and Ab1–42. Primary mouse wild-type (WT) and tau-knockout (KO) forebrain neurons, and secondary cultures of wild-type mouseglia 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 culturemedia. a, Cells were exposed to calcein-AM and imaged live by epifluorescencemicroscopy to assay viability16. Extensive death and detachment of cells wereobserved 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 bythe XTT plate reader assay17. Note the robust cytotoxicity of Ab3(pE)–42

containing solutions at concentrations as low as 0.5mM, unless Ab3(pE)–42 andAb1–42 were incubated separately during oligomerization (P , 0.01; yellow starssignify 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 independentexperiments).

3 1 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | 6 5 1

Macmillan Publishers Limited. All rights reserved©2012

www.nature.com/doifinder/10.1038/nature11060

OD450 nm/OD490 nm ratio was for 5% pE-Ab, which peaked similarly toAb3(pE)–42. Ab3(pE)–42, Ab1–42 and 5% pE-Ab thus oligomerized bydifferent pathways.

To test whether distinct biological activities were coupled to theseoligomerization differences, we compared cytotoxicity of the peptidestowards cultured neurons or glia using calcein-AM and fluorescencemicroscopy16. Twelve hours of Ab1–42 exposure had little effect on cellviability for wild-type or tau-knockout neurons, or wild-type glial cells(Fig. 1a). Contrastingly, most wild-type neurons died and detachedfrom the substrate after exposure to Ab3(pE)–42 or 5% pE-Ab. Tau-knockout neurons and wild-type glia, which express little tau, wereresistant to Ab3(pE)-42 and 5% pE-Ab.

Cytotoxicity dose dependence was examined by incubating wild-type neurons for 24 h in oligomers comprising 0.1, 0.5 or 1mMpeptides, and using the 2,3-bis-(2-methoxy-4-nitro-5-sulphophenyl)-2H-tetrazolium-5-carboxanilide (XTT) reduction assay17 (Fig. 1b). Cellswere unaffected by Ab1–42, but Ab3(pE)–42 and 5% pE-Ab had substan-tial cytotoxicity at 0.5mM and even more at 1.0mM. Cytotoxicity of 5%pE-Ab required Ab3(pE)–42 and Ab1–42 to incubate together for 24 hbefore being added to cells. When they were incubated separately for24 h and mixed together at a 1:19 molar ratio immediately before beingapplied 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.

Evidence for hybrid oligomers came from immunoprecipitation ofvarious forms of amyloid-b using aggregation-dependent M64, whichdoes not recognize Ab3(pE)–42 (see Supplementary Fig. 4 for character-ization of all anti-amyloid-b antibodies used, including M64). Immuno-precipitations were analysed on dot blots using 4G8, which equallyrecognizes Ab3(pE)–42 and Ab1–42, and anti-pE-Ab, which does not reactwith 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 reactedwith material immunoprecipitated out of 5% pE-Ab, M64 pulleddown hybrid peptide oligomers. Ab3(pE)–42 accounted for ,16% ofthe amyloid-b in gel-filtered cytotoxic oligomers after 3 h of oligomer-ization, and steadily dropped to ,8% by 24 h (Fig. 2b). Ab3(pE)–42 thusacts as a template that initiates formation of cytotoxic oligomers.

Cytotoxicity was sensitive to oligomerization time (Fig. 2c). Baselinecytotoxicity was observed at all time points for Ab1–42, and for 5% pE-Ab 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. Themost cytotoxic solutions were 5% pE-Ab, in which the constituentpeptides co-oligomerized for 24 h. These solutions killed ,60% ofthe cells within 24 h, and lower but robust cytotoxicity was observedat 96 h. Even the 0 h co-oligomers of 5% pE-Ab exhibited low, signifi-cant cytotoxicity. Co-incubated mixtures of 5% Ab3(pE)–42 and 95%

0

15

10

5

20

48 24 12

3

Per

cent Aβ 3

(pE

)–4

2 (o

f to

tal)

Oligomerization time before

fractionation (h)

3

b

a

(5X)

Aβ 1–42 o

ligom

ers

Aβ 3(pE)–

42 o

ligom

ers

5% Aβ 3(

pE)–42 +

95%

Aβ 1–

42

olig

omer

s

Aβ 1–42 m

onom

ers

Aβ 3(pE)–

42 m

onom

ers

4G8

Anti-pE-Aβ(5× load) 15.4 ng

(corrected to 1×)

101 ng 162 ng

0 h 24 h 96 h

0

20

100

40

60

80

Via

bili

ty (%

co

ntr

ol)

Aβ1–42 Aβ3(pE)–42

5% Aβ3(pE)–42, 95% Aβ1–42

Oligomerized together

c IP: M64 P < 0.01

P < 0.01

Oligomerized separately

Figure 2 | Ab3(pE)–42 and Ab1–42 form metastable, cytotoxic, hybridoligomers. a, Ab3(pE)–42 and Ab1–42 were incubated together at a 1:19 molarratio (5% pE-Ab) for 24 h at 1mM total amyloid-b, and were thenimmunoprecipitated (IP) with M64, a rabbit monoclonal antibody thatspecifically recognizes residues 3–7 (EFRH) of Ab1–40 oligomers or fibrils.Additional samples that were immunoprecipitated included otherwiseidentically treated oligomers made from pure Ab3(pE)–42 or Ab1–42, andmonomeric versions of the two peptides. Immunoprecipitated oligomers wereconverted to monomers by lyophilization, solubilization with HFIP anddilution into PBS, and along with the other samples were dot blotted ontonitrocellulose and analysed using 4G8, a mouse monoclonal antibody thatrecognizes Ab3(pE)–42 and Ab1–42 equally well, and an antibody that specificallyrecognizes pE-Ab (see Supplementary Fig. 4 for characterization of allantibodies used here). Quantification of the dot blots using a LI-COR Odysseyimaging station indicated that the oligomers that were immunoprecipitatedfrom the mixed peptide solution contained both Ab3(pE)–42 and Ab1–42, at amolar 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 mostcytotoxicity resided (see Fig. 3b) were immunoprecipitated using anti-humanamyloid-b (N), an amino-terminal-specific antibody that does not react withpE-Ab (data not shown). The immunoprecipitates were then lyophilized, re-solubilized with HFIP, and quantitatively analysed on dot blots with 4G8 andanti-pE-Ab using the LI-COR Odyssey. The time-dependent decrease in theAb3(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 finalconcentration of 1mM total amyloid-b. Following peptide treatment, cellviability was analysed by the XTT plate reader assay17. The most cytotoxicspecies observed were the hybrid oligomers after 24 h of oligomerization(P , 0.01; yellow stars signify statistical significance of the indicated bar graphsversus vehicle controls; black stars signify statistical significance between theindicated bar graph pairs; mean 6 s.e.m., n 5 6 or 9 replicates from 3independent experiments for panel b or c, respectively).

RESEARCH LETTER

6 5 2 | N A T U R E | V O L 4 8 5 | 3 1 M A Y 2 0 1 2


Ab1–42 can therefore form oligomers whose cytotoxicity is both greaterand more enduring than oligomers formed by Ab3(pE)–42 alone.

To identify the co-oligomer size(s) that were cytotoxic, amyloid-bsolutions were oligomerized for various times from 0–96 h beforefractionation by gel filtration. Total amyloid-b in all fractions wasdetermined 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), illus-trate the full fractionation range of the column but exclude most voidvolume fractions. Presumptive monomeric Ab1–42 dominated initiallyand persisted at 3 h, but was nearly undetectable after 12 h. The 3 htime point also marked the appearance of Ab1–42 oligomers, whichgradually increased in size over the next 93 h. Ab3(pE)–42 and 5% pE-Aboligomerized differently. Putative monomers were present at 0 h forboth samples, when slightly larger species, LNOs that possibly corre-sponded to dimers or trimers (Supplementary Fig. 6), were also pre-sent. These persisted as the main species for 24 h for Ab3(pE)–42 and fornearly 72 h for 5% pE-Ab, and later time points were dominated bylarger aggregates that eluted in void volume fractions. Cytotoxicity wasassayed for individual fractions of 5% pE-Ab that oligomerized for24 h (Fig. 3b). Most cytotoxicity was associated with the possibledimers/trimers that eluted at 12.5 ml, which at 425 nM peptide killedmore than 60% of the cells. Low cytotoxicity was also observed at554 nM peptide for the larger oligomers that eluted at 8.5 ml.

The marked enhancement of Ab1–42 cytotoxicity by Ab3(pE)–42

suggested a prion-like templating mechanism of Ab1–42 misfoldinginitiated by Ab3(pE)–42. To test that hypothesis, 5% pE-Ab thatoligomerized for 24 h was diluted into 19 volumes of monomericAb1–42. A 24 h incubation of this mixture yielded ‘serial passage 1’, whichwas followed by two equivalent, sequential dilutions into monomericAb1–42 to yield serial passages 2 and 3. A gradual loss of cytotoxicity wasobserved with successive passages, but even passage 3, which containedonly 0.000625% Ab3(pE)–42, killed ,50% of the neurons within 24 h(Fig. 3c). Serially passaged gel-filtration samples contained abundantmaterial that eluted at 12.5 ml in passages 1–3, despite the progressivedilution of Ab3(pE)–42 (Fig. 3d). Ab3(pE)–42 can therefore templateformation of metastable, cytotoxic LNOs from excess Ab1–42, yieldingpotent bioactivity that can be serially passaged multiple times intomonomeric Ab1–42 without further addition of Ab3(pE)–42.

One possible explanation for why Ab1–42 LNOs were inert is thatthey lacked sufficient properly sized oligomers. Accordingly, wealtered the oligomerization protocol from 5mM peptide for 24 h at37 uC to 10mM 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 putativedimers/trimers initiated by Ab3(pE)–42. This was confirmed by dot blotsusing M87, a conformation-sensitive anti-amyloid-b antibody, tocompare the putative dimers/trimers used for the cytotoxicity assaysshown in Fig. 3f. We first lyophilized aliquots of all the amyloid-bsolutions, resuspended them with hexafluoroisopropanol (HFIP) torestore them to monomers, and then analysed them using 4G8.When parallel samples that were not lyophilized but were otherwiseidentical were analysed using M87, immunoreactivity was approxi-mately twice as strong with LNOs made from Ab1–42 versus thosemade from 5% pE-Ab (Supplementary Fig. 7). Cytotoxic LNOs of5% pE-Ab are thus structurally distinct from comparably sizedLNOs of Ab1–42.

a 5% Aβ3(pE)–42, 95% Aβ1–42 oligomerization (h)

0 3 12 24 48 72 96

7.5

9.5

10.5

11.5

12.5

14

8.5

Ve (m

l)

b

Peptide concentration (nM)

NDDD

507

554

452

ND

ND

425

100

Via

bili

ty (%

co

ntr

ol)

P < 0.01

14

7.5

9.5

10.5

11.5

12.5

8.

5

Ve (ml)

80

60

40

20

0

e

Aβ 1–

42

5% Aβ 3(

pE)–

42

95%

Aβ 1–

42

7.5

9.5

10.5

11.5

12.5

14

8.5

f

Via

bili

ty (%

co

ntr

ol)

100

Aβ 1–42, 1

μM

unfra

ctio

nate

d

5% Aβ 3(

pE)–42, 1

μM

unfra

ctio

nate

d

V e =

12.

5 m

l

40

20

60

80

0

P < 0.01

7.5

9.5

10.5

11.5

12.5

14

8.5

Aβ 1–

42

Ser

ial p

assa

ge 1

Ser

ial p

assa

ge 2

Ser

ial p

assa

ge 3

d c

Via

bili

ty (%

) co

ntr

ol)

5% Aβ3(pE)–42, 95% Aβ1–42

P < 0.01

5% pE

0.25% pE

0.0125% pE

0.000625% pE

Ser

ial

passa

ge 1

Ser

ial

passa

ge 3

Olig

omer

ized

separ

ately

Olig

omer

ized

toge

ther

Ve (m

l)

Aβ 1–42, 0

.44 μM

Aβ 1–42, 0

.22 μM

V e =

12.

5 m

l

5% Aβ 3(

pE)–42, 0

.36 μM

V e =

12.

5 m

l

5% Aβ 3(

pE)–42, 0

.18 μM

V e =

12.

5 m

l

Ve (m

l)

80

60

40

20

0

passa

ge 2

Ser

ial

Figure 3 | The cytotoxic species are low-n, prion-like oligomers. a, Gel-filtration chromatography was used to fractionate 5% pE-Ab afteroligomerization at 5mM at 37 uC for 0–96 h. The resulting fractions were thenconverted to monomers using HFIP and analysed on dot blots usingmonoclonal antibody 4G8. Note the metastable oligomers with an averageelution volume (Ve) of 12.5 ml. b, Isolated gel-filtration fractions from the 24 htime point were added to wild-type neuron cultures for 24 h, after which thecells were assayed for cell viability using XTT17. Robust cytotoxicity wasassociated only with the Ve 5 12.5 ml fraction, although the Ve 5 8.5 mlfraction 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, notdetected. c, Cytotoxic hybrid oligomers made by co-incubating a 1:19 ratio ofAb3(pE)–42:Ab1–42 for 24 h at 5mM were diluted into a 19-fold molar excess offreshly dissolved, monomeric Ab1–42, which was then incubated at 5mM foranother 24 h to yield serial passage 1. Two further iterations of this strategyyielded serial passages 2 and 3. The starting material and its serially passagedderivatives were added to wild-type neurons at 1 mM peptide for 24 h, afterwhich cells were analysed using the XTT assay for cell viability17. Only a gradualloss of cytotoxicity was observed with each successive serial passage. d, Eachserially passaged sample, as well as otherwise identically prepared oligomersmade from pure Ab1–42, were fractionated by gel filtration and analysed on dotblots with 4G8. Note that all serially passaged samples contained metastableLNOs of Ve 5 12.5 ml, which were absent from the pure Ab1–42 samples.e, Ab1–42 (10mM) that was oligomerized for 30 min at 4 uC, and 5% Ab3pE–42

plus 95% Ab1–42 (5mM) that was oligomerized for 24 h at 37 uC werefractionated by gel filtration and analysed on dot blots exactly using 4G8. Notethe isolation of fractions with Ve 5 12.5 ml from both preparations. f, Wild-type neurons were assayed for viability using the XTT plate reader assay17

following 24 h of exposure to the indicated amyloid-b preparations. Note theminimal 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 graphsversus vehicle controls; black stars and blue stars signify statistical significancebetween the indicated bar graph pairs; blue mean 6 s.e.m., n 5 9 replicatesfrom 3 independent experiments for panels b, c and f.

LETTER RESEARCH

3 1 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | 6 5 3


Several lines of evidence demonstrate in vivo relevance for the datadescribed so far. First, we identified LNOs containing Ab3(pE)–42 inthree out of three Alzheimer’s disease samples, based on gel filtrationof human brain extracts followed by dot blots of resulting fractionswith anti-pE-Ab and M87. In contrast, only one of three age-matchedsamples with normal neuropathological diagnoses was positive forAb3(pE)–42 (Fig. 4a and Supplementary Fig. 8). Second, we crossedTBA2.1 mice18 into a tau-knockout background19. By 3 months,TBA2.1 mice accumulated small amounts (40–100 ng g21 brainweight) of Ab3(pE)–42, which formed primarily intraneuronal aggre-gates, and was associated with massive hippocampal neuron loss andgliosis18. Knocking out tau provided almost complete protectionagainst neuron loss and glial activation (Fig. 4b). Additional in vivodata are shown in Supplementary Fig. 9. Long-term potentiation(LTP) of mouse hippocampal neurons in slice cultures was potentlyand 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 impair-ment (Supplementary Fig. 9a). To evaluate the effects of increasedAb3(pE)–42 in animal models, we crossed mice with neuron-specificexpression of human b-amyloid precursor protein (APP) harbouringSwedish and London mutations (hAPPSL)20, with mice expressinghuman QC21. Nine-month-old double (hAPPSL/hQC) and single(hAPPSL) transgenic mice were indistinguishable in terms of insolubleand soluble Abx–42 levels, but the double transgenics had approxi-mately twofold more insoluble Ab3(pE)–42 and approximately ninefoldmore soluble Ab3(pE)–42 than single transgenics (Supplementary Fig.9b). Further analysis of the soluble Abx–42 by the A4 assay22 revealed anapproximately eightfold excess of oligomers in the double versus singletransgenics (Supplementary Fig. 9c). Double transgenics performedmore poorly in Morris water maze tests (Supplementary Fig. 9d) andhad reduced hippocampal immunoreactivity for the synapse marker,synaptophysin (Supplementary Fig. 9e). Finally, peri-hippocampalinjection of 5% pE-Ab at 5mM into APPSwDI/NOS22/2 Alzheimer’sdisease model mice23 led 3–5 months later to the presence of plaquescontaining both pE-Ab and conventional amyloid-b. Comparableplaques were rarely seen in sham-injected Alzheimer’s disease miceor in wild-type mice injected with 5% pE-Ab (Supplementary Fig. 9f).These collective in vivo results emphasize the physiological signifi-cance of the companion biochemical and cultured cell results.

Our studies provide new insights into Alzheimer’s disease patho-genesis by demonstrating that hypertoxic amyloid-b oligomers can betriggered by small quantities of a specifically truncated and post-translationally modified version of amyloid-b. Although some pre-vious studies demonstrated that pE modification of amyloid-bconsiderably enhances its aggregation kinetics13,14,24, toxicity12,18,25

and resistance to degradation12, a mechanistic explanation for theunique properties of pE-Ab has been lacking until now. Prior studiessuggest coincident appearance of Ab3(pE)–42 with development or pro-gression of human Alzheimer’s disease26,27. Co-localization of QC andAb3(pE)–42 was found in cored plaques of vulnerable regions inAlzheimer’s disease, and evidence was provided for axonal transportof Ab3(pE)–x from QC-rich neuronal populations of the entorhinalcortex and locus coeruleus28. As LNOs containing Ab3(pE)–42 arereasonably stable (Fig. 3a), they might initiate tau-dependentcytotoxicity intracellularly during axonal transport29 or extracellularlyfollowing release at remote hippocampal synapses30 of projectionneurons28. The Ab3(pE)–42-induced formation of toxic mixed oligomersprovides a rationale for these previous observations, and the tau-dependent cytotoxicity of 5% pE-Ab establishes a new functional con-nection between amyloid-b and tau in Alzheimer’s disease pathogenesis.

METHODS SUMMARYFull descriptions of thioflavin T assays, cell culture, cell viability assays, proceduresfor oligomerization of amyloid-b peptides and their fractionation by gel-filtrationchromatography, production and specificity of rabbit monoclonal anti-amyloid-bantibodies, immunoprecipitation, dot blots and western blots, generation ofhAPPSL/hQC transgenic mice, LTP measurements of mouse hippocampal slicecultures, peri-hippocampal injection of 5% pE-Ab into Alzheimer’s disease modelmice, cultured cell and brain immunohistochemistry, and collection of humanbrain 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. Gandy, S. et al. Days to criterion as an indicator of toxicity associated with humanAlzheimer amyloid-b oligomers. Ann. Neurol. 68, 220–230 (2010).

2. Walsh, D. M. & Selkoe, D. J. Ab oligomers—a decade of discovery. J. Neurochem.101, 1172–1184 (2007).

3. Rapoport,M., Dawson, H. N., Binder, L. I., Vitek, M. P.& Ferreira, A. Tau is essential tob-amyloid-inducedneurotoxicity. Proc.Natl Acad. Sci.USA 99,6364–6369 (2002).

4. King, M. E. et al. Tau-dependent microtubule disassembly initiated by pre-fibrillarb-amyloid. J. Cell Biol. 175, 541–546 (2006).

AD (62-98)

AD (17-01)

AD (10-02)

Normal (15-02)

Normal (47-97)

Normal (03-00)

Anti-pE-Aβ

M87

Anti-pE-Aβ

M87

Anti-pE-Aβ

M87

Anti-pE-Aβ

M87

Anti-pE-Aβ

M87

Anti-pE-Aβ

M87

8.4

9.2

10.0

10.8

11.6

12.4

13.2

14.0

14.8

15.6

a b

100 μm

GFAP +

Hem.

Hem.

Aβ3(pE)–x

WT TBA2.1

TBA2.1

tau KO Ve (ml)

Figure 4 | Ab3(pE)–42 in vivo. a, Cytosol obtained from human Alzheimer’sdisease (AD) and similarly aged normal brains (Supplementary Fig. 9) werefractionated by gel filtration, and analysed by dot blotting with anti-pE-Ab andM87. Note the appearance of pE-Ab in LNO fractions, including those thateluted 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., haematoxylinnuclear staining), none of which are evident in comparably aged wild-type(WT) mice or TBA2.1/tau-knockout19 (KO) hybrids.

RESEARCH LETTER

6 5 4 | N A T U R E | V O L 4 8 5 | 3 1 M A Y 2 0 1 2


5. Roberson, E. D. et al. Reducing endogenous tau ameliorates amyloid b-induceddeficits in an Alzheimer’s disease mouse model. Science 316, 750–754 (2007).

6. Jin, M. et al. Soluble amyloid b-protein dimers isolated from Alzheimer cortexdirectly induce Tau hyperphosphorylation and neuritic degeneration. Proc. NatlAcad. Sci. USA 108, 5819–5824 (2011).

7. Mori,H., Takio,K., Ogawara,M.&Selkoe,D. J. Mass spectrometry of purifiedamyloidb protein in Alzheimer’s disease. J. Biol. Chem. 267, 17082–17086 (1992).

8. Saido, T. C.et al.Dominantanddifferential depositionofdistinctb-amyloidpeptidespecies, AbN3(pE), in senile plaques. Neuron 14, 457–466 (1995).

9. Jawhar, S., Wirths, O.& Bayer, T. A. Pyroglutamateamyloid-b (Ab): a hatchetman inAlzheimer disease. J. Biol. Chem. 286, 38825–38832 (2011).

10. Schilling, S. et al. Glutaminyl cyclase inhibition attenuates pyroglutamate Ab andAlzheimer’s disease-like pathology. Nature Med. 14, 1106–1111 (2008).

11. Tabaton, M. et al. Soluble amyloid b-protein is a marker of Alzheimer amyloid inbrain but not in cerebrospinal fluid. Biochem. Biophys. Res. Commun. 200,1598–1603 (1994).

12. Russo, C. et al. Pyroglutamate-modified amyloid b-peptides—AbN3(pE)—stronglyaffect cultured neuron and astrocyte survival. J. Neurochem. 82, 1480–1489(2002).

13. Schilling, S. et al. On the seeding and oligomerization of pGlu-amyloid peptides (invitro). Biochemistry 45, 12393–12399 (2006).

14. Schlenzig, D. et al. Pyroglutamate formation influences solubility andamyloidogenicity of amyloid peptides. Biochemistry 48, 7072–7078 (2009).

15. Levine, H. III. Thioflavine T interaction with synthetic Alzheimer’s diseaseb-amyloid peptides: detection of amyloid aggregation in solution. Protein Sci. 2,404–410 (1993).

16. Wang, X. M. et al. A new microcellular cytotoxicity test based on calcein AM release.Hum. Immunol. 37, 264–270 (1993).

17. Scudiero, D. A. et al. Evaluation of a soluble tetrazolium/formazan assay for cellgrowth and drug sensitivity in culture using human and other tumor cell lines.Cancer Res. 48, 4827–4833 (1988).

18. Alexandru, A. et al. Selective hippocampal neurodegeneration in transgenic miceexpressing small amounts of truncated Ab is induced by pyroglutamate–Abformation. J. Neurosci. 31, 12790–12801 (2011).

19. Tucker, K. L., Meyer, M. & Barde, Y. A. Neurotrophins are required for nerve growthduring development. Nature Neurosci. 4, 29–37 (2001).

20. Rockenstein, E., Mallory, M., Mante, M., Sisk, A. & Masliaha, E. Early formation ofmature amyloid-bproteindeposits in a mutant APP transgenic model depends onlevels of Ab1–42. J. Neurosci. Res. 66, 573–582 (2001).

21. Jawhar, S. et al. Overexpression of glutaminyl cyclase, the enzyme responsible forpyroglutamate Ab formation, induces behavioral deficits, and glutaminyl cyclaseknock-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 testingin Alzheimer’s disease of the APP[V717I] London transgenic mouse model. Int. J.Alzheimers Dis. 2010 (2010).

23. Wilcock, D. M. et al. Progression of amyloid pathology to Alzheimer’s diseasepathology in an amyloid precursor protein transgenic mouse model by removal ofnitric oxide synthase 2. J. Neurosci. 28, 1537–1545 (2008).

24. He, W. & Barrow, C. J. The Ab3-pyroglutamyl and 11-pyroglutamyl peptides foundin senile plaque have greater b-sheet forming and aggregation propensities in vitrothan full-length Ab. Biochemistry 38, 10871–10877 (1999).

25. Wirths, O. et al. Intraneuronal pyroglutamate-Ab 3–42 triggers neurodegenerationand lethal neurological deficits in a transgenic mouse model. Acta Neuropathol.118, 487–496 (2009).

26. Guntert, A., Dobeli, H. & Bohrmann, B. High sensitivity analysis of amyloid-bpeptide 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-Rubsamen, M. et al. Glutaminyl cyclase contributes to the formation offocal and diffuse pyroglutamate (pGlu)-Ab deposits in hippocampus via distinctcellular 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 synapsedegeneration in Alzheimer’s disease. Cell. Mol. Neurobiol. 31, 939–948 (2011).

Supplementary Information is linked to the online version of the paper atwww.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 T32GM008136,which funded part of J.M.N.’s PhD training;NIH/NIA grantR01AG033069(C.G.G.); and the German Federal Department of Science and Technology grant03IS2211F (H.-U.D.). Funding for the UCI-ADRC was provided by NIH/NIA grant P50AG16573. We also thank H. Dawson and M. Vitek of Duke University for providing thetau-knockout mice. This work fulfilled part of the requirements for the PhD earned byJ.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 biologicalexperiments; 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 theperi-hippocampal injection experiments; A.H. and C.G.G. produced and characterizedthe M64 and M87 antibodies; R.R. and K.R. performed the electrophysiologyexperiments; A.A., W.J. and S.G. performed and analysed the immunohistochemicalexperiments on TBA2.1 and tau-knockout/TBA2.1 mice; G.S.B. and H.-U.D. initiatedand directed the project; G.S.B. was the principal writer of the paper; all of the authorsparticipated in the design and analysis of experiments, and in editing of the paper.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of this article atwww.nature.com/nature. Correspondence and requests for materials should beaddressed to G.S.B. ([email protected]) or H.-U.D.([email protected]).

LETTER RESEARCH

3 1 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | 6 5 5


www.nature.com/nature

www.nature.com/reprints

www.nature.com/nature

mailto:[email protected]

mailto:[email protected]

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

SUPPLEMENTARY INFORMATIONdoi:10.1038/nature11060

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-(phenyl-

aminocarbonyl)-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 Boc-

Ser(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-2-

propanol (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


SUPPLEMENTARY INFORMATIONRESEARCHdoi:10.1038/nature11060

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



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.



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



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



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.



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 (C-

terminal) 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



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). Peptide-

containing 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



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).



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).



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).



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.



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



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.



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 10-

mers 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 10-

mer 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



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).



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.



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).



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.



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).



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).



Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-β

Documents

Transcript of Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-β