Accepted Manuscript Human secreted tau increases amyloid-beta production Jessica Bright, Sami Hussain, Vu Dang, Sarah Wright, Bonnie Cooper, Tony Byun, Carla Ramos, Andrew Singh, Graham Parry, Nancy Stagliano, Irene GriswoldPrenner PII:
S0197-4580(14)00602-2
DOI:
10.1016/j.neurobiolaging.2014.09.007
Reference:
NBA 9048
To appear in:
Neurobiology of Aging
Received Date: 4 February 2014 Revised Date:
4 September 2014
Accepted Date: 6 September 2014
Please cite this article as: Bright, J., Hussain, S., Dang, V., Wright, S., Cooper, B., Byun, T., Ramos, C., Singh, A., Parry, G., Stagliano, N., Griswold-Prenner, I., Human secreted tau increases amyloid-beta production, Neurobiology of Aging (2014), doi: 10.1016/j.neurobiolaging.2014.09.007. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Human secreted tau increases amyloid-beta production
2
Jessica Bright1,2, Sami Hussain1,2, Vu Dang1, Sarah Wright1, Bonnie Cooper1, Tony
3
Byun1, Carla Ramos1, Andrew Singh1, Graham Parry1, Nancy Stagliano1 and Irene
4
Griswold-Prenner1*
5
1
6
2
7
*Correspondence should be addressed to N.S. (
[email protected])
8
951 Gateway Blvd, SSF CA 94080
9
Telephone: (650) 872-4000
iPierian, Inc.
M AN U
SC
Co-first authors
EP
TE D
Fax: (650) 872 4701
AC C
10
RI PT
1
1
ACCEPTED MANUSCRIPT
11 12 Abstract
14
The interaction of amyloid-beta (Aβ) β) and tau in the pathogenesis of Alzheimer’s
15
disease (AD) is a subject of intense inquiry, with the bulk of evidence indicating
16
that changes in tau are downstream of Aβ β. It has been shown however, that
17
human tau overexpression in APP transgenic mice increases Aβ β plaque
18
deposition. Here we confirm that human tau increases Aβ β levels. To determine if
19
the observed changes in Aβ β levels were due to intracellular or extracellular
20
secreted tau (eTau for extracellular tau), we affinity purified secreted tau from AD
21
patient derived cortical neuron conditioned media and analyzed it by liquid
22
chromatography-mass spectrometry (LC/MS). We found the extracellular species
23
to be composed predominantly of a series of N-terminal fragments of tau, with no
24
evidence of C-terminal tau fragments. We characterized a subset of high affinity
25
tau antibodies, each capable of engaging and neutralizing eTau. We found that
26
neutralizing eTau reduces Aβ β levels in vitro in primary human cortical neurons
27
where exogenously adding eTau increases Aβ β levels. In vivo, neutralizing human
28
tau in two human tau transgenic models also reduced Aβ β levels. We show that the
29
human tau insert sequence is sufficient to cause the observed increase in Aβ β
30
levels. Our data furthermore suggest that neuronal hyperactivity may be the
31
mechanism by which this regulation occurs. We show that neuronal hyperactivity
32
regulates both eTau secretion and Aβ β production. Electrophysiological analysis
33
shows for the first time that secreted eTau causes neuronal hyperactivity. Its
34
induction of hyperactivity may be the mechanism by which eTau regulates Aβ β
35
production. Together with previous findings, these data posit a novel connection
AC C
EP
TE D
M AN U
SC
RI PT
13
2
ACCEPTED MANUSCRIPT
36
between tau and Aβ β, suggesting a dynamic mechanism of positive feed forward
37
regulation. Aβ β drives the disease pathway through tau, with eTau further
38
increasing Aβ β levels, perpetuating a destructive cycle.
RI PT
39 40
Keywords: secreted tau, extracellular tau, eTau, Amyloid-beta (Aβ β), neuronal
41
hyperactivity, Alzheimer’s disease, feed forward mechanism, sAPPα α
SC
42 1. Introduction
44
Alzheimer’s disease (AD) is characterized pathologically by amyloid-beta (Aβ) plaques
45
and tau dominated neurofibrillary tangles (NFTs). Understandably, the field has focused
46
on elucidating the mechanisms regulating these hallmark AD proteins and investigating
47
their interplay in disease. Aβ is formed by the sequential cleavage of amyloid precursor
48
protein (APP) by β- and γ- secretase. The accumulation of Aβ peptide in extracellular
49
plaques is traditionally seen as the primary event in AD, with subsequent NFT formation
50
and neuronal and synaptic loss following as a direct result; collectively known as the
51
amyloid cascade hypothesis (Hardy and Higgins, 1992; Hardy and Selkoe, 2002).
52
Work performed with frontotemporal dementia mutant tau (P301L) transgenic mice
53
injected with aggregated Aβ (Gotz et al., 2001) as well as experiments using APP and
54
tau bigenic mice (Lewis et al., 2001) indicates an upstream role for Aβ oligomerization in
55
initiating tau pathology. In vitro and in vivo models show Aβ induced neurotoxicity and
56
Aβ induced axonal transport deficits are ameliorated by dose dependent reductions in
57
tau (Rapoport et al., 2002; Vossel et al., 2010). In hippocampal slice preparations Aβ-
58
induced impairment of long-term potentiation (LTP) is inhibited in slices generated from
59
tau-/- mice (Shipton et al., 2011). In the J20 APP transgenic mouse, reducing
60
endogenous mouse tau decreased or eliminated learning and memory deficits,
AC C
EP
TE D
M AN U
43
3
ACCEPTED MANUSCRIPT
exploratory loco motor hyperactivity, interictal spiking, spontaneous seizures, and early
62
mortality (Roberson et al., 2007; Roberson et al., 2011). In the APP23 transgenic mice,
63
reducing tau dose dependently improved memory in the T-maze and reduced seizures
64
and premature mortality (Ittner et al., 2010). The exact downstream mechanism involved
65
in tau mediating protection from Aβ toxicity has yet to be shown. Elimination of mouse
66
tau in these APP transgenic animals mitigates disease phenotypes but does not alter Aβ
67
levels or plaque burden (Roberson et al., 2007; Ittner et al., 2010).
68
In contrast to these findings, work done with double transgenic mice expressing both
69
human mutant tau and APP showed greater Aβ deposition and NFT-like formation with
70
increased neuronal loss in double transgenic animals when compared to single
71
transgenic controls (Ribe et al., 2005). Furthermore, Braak and others have shown with
72
their analysis of brain slices taken from patient autopsy samples still in the
73
presymptomatic stages of AD, Braak-stage I-III (Braak and Braak, 1995), that
74
intraneuronal tau alterations precede aggregated Aβ deposition (Braak and Del, 2004;
75
Braak et al., 2013; Schonheit et al., 2004). They present evidence that tau tangles
76
temporally develop either before or independent of Aβ plaques. These divergent findings
77
indicate that the causal link between aberrant APP processing and tau alterations
78
remains controversial.
79
Recent AD research places tau in the extracellular space under physiological conditions
80
(Yamada et al., 2011) and shows endogenous tau to be actively secreted from human
81
(Chai et al., 2012) and rat neurons (Pooler et al., 2013). In concert with these
82
observations, others have found that tau pathology spreads in AD. Early in disease, tau
83
tangles are detectable in the entorhinal cortex and as the disease progresses tau
84
pathology spreads into the hippocampus and cortex (Braak and Braak, 1995).
AC C
EP
TE D
M AN U
SC
RI PT
61
4
ACCEPTED MANUSCRIPT
The spread of tau tangle pathology occurs concomitantly with clinical disease
86
progression, moving from memory loss to dementia. In in vivo models in which human
87
tau is specifically overexpressed in the entorhinal cortex, tau is secreted and spreads
88
along synaptic circuitry, resulting in tau pathology progression from the entorhinal cortex,
89
through the hippocampus and into the cortex (Liu et al., 2012; de Calignon A. et al.,
90
2012; Dujardin et al., 2014). Furthermore, comparative analysis of AD and healthy
91
patient cerebral spinal fluid (CSF) showed a clear increase of amino-terminal (N-
92
terminal) tau fragments in AD patient CSF, with no evidence of full-length or carboxyl-
93
terminal (C-terminal) tau (Meredith, Jr. et al., 2013). This is in contrast to data
94
suggesting that tau in the interstitial fluid (ISF) of P301L mice is full length (Yamada et
95
al., 2011).
96
Pre-synaptic release of secreted tau from cortical neurons has been shown to be
97
mediated by neuronal activity. Yamada et al. used in vivo microdialysis in mouse brain to
98
show that increasing neuronal and synaptic activity rapidly increases steady state levels
99
of transgenic extracellular tau (Yamada et al., 2014). Additionally, in vitro treatment of rat
TE D
M AN U
SC
RI PT
85
neurons with the sodium channel blocker tetrodotoxin (TTX) prevents AMPA-mediated
101
tau release (Pooler et al., 2013).
102
In fact both tau and Aβ have been shown to be affected by neuronal hyperactivity. Cirrito
103
et al. and others have demonstrated in vitro and in vivo that soluble Aβ levels in brain
104
are directly influenced by neuronal, and more specifically, synaptic activity (Cirrito et al.,
105
2005; Cirrito et al., 2008; Verges et al., 2011; Kamenetz et al., 2003; DeVos et al.,
106
2013). They postulated a negative feedback role for Aβ in healthy neurons, functioning to
107
keep neuronal hyperactivity in check by selectively depressing excitatory synaptic
108
transmission. Taken together, these findings indicate a direct correlation between
AC C
EP
100
5
ACCEPTED MANUSCRIPT
reductions in tau protein with reductions in neuronal hyperactivity, both of which
110
correspond to a decrease in soluble Aβ levels.
111
Here we have identified the major forms of human tau secreted from AD patient derived
112
cortical neurons to be a pool of secreted N-terminal tau fragments which we call eTau,
113
for extracellular tau. We show that eTau, and more specifically the human tau insert
114
region, increases soluble Aβ levels while simultaneously lowering the neuroprotective
115
product of α-secretase APP cleavage, soluble amyloid precursor protein-α (sAPPα);
116
suggesting that eTau modulates Aβ production. We show this eTau/Aβ relationship in
117
both primary human in vitro neuronal cultures and in vivo tau transgenic models.
118
Additionally, we observed that eTau induces neuronal hyperactivity when applied to
119
cortical cultures, and found that reducing or neutralizing eTau prevents aberrant
120
neuronal hyperactivity. We suggest that eTau induced hyperactivity is negatively
121
impacting neurons and that eTau may elevate Aβ production in AD brain.
M AN U
SC
RI PT
109
TE D
122 2. Materials and Methods
124
2.1 Primary human cortical neuron cultures
125
Human cortical neuron cultures were prepared as described (Wright et al., 2012) and
126
cultured for three weeks in Neurobasal medium with B27 and Glutamax (Invitrogen)
127
prior to start of 20 day treatments. Human fetal cerebral cortical tissue was obtained by
128
Advanced Bioscience Resources (Alameda, CA) and complied with federal guidelines
129
for fetal research and with the Uniformed Anatomical Gift Act.
130
2.2 iPSC patient derived cortical neurons (iPSC-CN)
131
Induced pluripotent stem cells (iPSC) were generated using the Yamanaka method
132
(Takahashi et al., 2007) as previously described (Dimos et al., 2008) from skin biopsies
AC C
EP
123
6
ACCEPTED MANUSCRIPT
taken from a familial presenilin-1 (A260V) AD patient aged 50 and a sporadic AD patient
134
aged 60. Briefly, iPSC were differentiated to neurons via dual SMAD (small body size
135
mothers against decapentaplegic) inhibition (Chambers et al., 2009) followed by cortical
136
neuron differentiation as described (Burkhardt et al., 2013). iPSC derived cortical
137
neurons (iPSC-CN) were grown between 70 and 108 days with media changes every
138
three days unless otherwise noted. A list of AD patient samples used to generate
139
individual iPSC clones, as well as their relevant mutations, is provided in Table 1.
140
2.3 Antibodies
141
2.3.1 iPierian Tau Antibodies
142
Tau sizes are represented based on 2N4R tau sequence length unless otherwise noted.
143
IPN001 and IPN002 (iPierian) are mouse monoclonal antibodies generated by standard
144
immunization with in vitro aggregated 2N4R tau, followed by clonal selection and
145
antibody purification from hybridoma. Both antibodies were selected for their high
146
binding affinity to both secreted eTau as well as full length tau (recombinant Tau383,
147
0N4R form) and for their ability to neutralize secreted eTau’s electrophysiological
148
activity. They both bind to the N-terminus of tau with epitopes spanning amino acids 15-
149
24.
150
2.3.2 Antibodies
151
MC1 (P. Davies) is a conformational monoclonal antibody specific for a conformation
152
change in tau found in AD brain and PHF1 (P. Davies) detects the pSer396/404 tau
153
epitope present on both normal adult brain tau and paired helical fragment (PHF) tau
154
(Jicha et al., 1997). Commercial mouse monoclonal antibodies against tau used are HT7
155
(aa 159-163, ThermoScientific), BT2 (aa 194-198, ThermoScientific) and T46 (C-
AC C
EP
TE D
M AN U
SC
RI PT
133
7
ACCEPTED MANUSCRIPT
terminal tau near amino acids 404-441, Invitrogen). Rabbit polyclonal antibodies used
157
were against tau (aa 243-441, Dako) and against APP (C-terminal amino acids, Sigma).
158
Mouse monoclonal control antibodies used were to β-Actin (Abcam), α-Tubulin (Cell
159
Signaling), mouse IgG (R&D Systems) and mouse IgG whole molecule (Jackson).
160
2.4 Secreted Tau Characterization, Purification, and Liquid Chromatography-Mass
161
Spectrometry (LC/MS) Analysis
162
2.4.1 Secreted Tau Immunoprecipitation
163
iPSC derived cortical neurons (iPSC-CN) derived from a familial presenilin-1 (A260V)
164
AD patient aged 50 were differentiated for 70 days at which point media conditioned for
165
three days was collected for tau immunoprecipitation. Conditioned media was
166
centrifuged at 15,000 rpm (4⁰C) and supernatants pre-cleared on a mouse IgG
167
(Jackson) Sepharose 4B resin (GEHealthcare) before immunoprecipitation on either a
168
N-terminal tau antibody, IPN002, a mid-region tau antibody, HT7, or a C-terminal tau
169
antibody, T46, each previously coupled to Sepharose 4B resins. After
170
immunoprecipitation, all three bead sets were washed three times with Mammalian
171
Protein Extraction Buffer (M-PER, ThermoScientific) and immunoprecipitated protein
172
extracted from pelleted beads with 2X Laemmli reducing buffer (Sigma) and separated
173
on NuPage Bis-Tris protein gels (Invitrogen). Western blots were probed for tau with
174
IPN001 (1 µg/ml), analysis by Odyssey SA software (LiCor). To ensure the tau specificity
175
of all three immunoprecipitations, iPSC-CN growth media was incubated with the three
176
antibody conjugated resins in parallel with the conditioned media, and the resultant non-
177
specific eluate ran as Media control on the IPN001 Western blot.
178
2.4.2 Secreted Tau Affinity Isolation
AC C
EP
TE D
M AN U
SC
RI PT
156
8
ACCEPTED MANUSCRIPT
iPSC derived cortical neurons (iPSC-CN) generated from skin biopsies taken from a
180
familial presenilin-1 (A260V) AD patient aged 50 and a sporadic AD patient aged 60
181
were cultured for 108 days in total, with media changes every three days. Starting at day
182
70 of iPSC-CN culture, conditioned media was collected every three days and pooled for
183
secreted tau affinity isolation. Pooled conditioned media was centrifuged at 15,000 rpm
184
(4⁰C) and supernatants pre-cleared on a mouse IgG Sepharose 4B resin before tau
185
purification on N-terminal IPN002 anti-tau antibody Sepharose 4B coupled resin. Protein
186
was eluted with 50 mM sodium citrate, pH 2.3, 150 mM NaCl and neutralized with 1M
187
TBS, pH 8.3. The eluate was concentrated and buffer exchanged into phosphate-
188
buffered saline (PBS) and here referenced as eTau. A mock protein control was
189
generated from a second batch of pooled conditioned media, also pre-cleared over the
190
mouse IgG resin before mock affinity purification using mouse IgG resin.
191
2.4.3 Liquid Chromatography-Mass Spectrometry Analysis (LC/MS)
192
Side-by-side lanes of IPN002 affinity purified eTau were separated on
193
reducing/denaturing 10% NuPage gel (Invitrogen), and analyzed by Coomassie stain
194
(LiCor) or transferred to nitrocellulose for Western blot analysis with iBlot (Life
195
Technologies) and probed for tau with IPN001 (1 µg/ml). Four Coomassie stained/anti-
196
tau confirmed bands were excised from the gel and submitted for LC/MS analysis (MS
197
Bioworks). Protein was proteolytically digested out of the gel slices with trypsin,
198
chymotrypsin or elastase and gel digests were analyzed by nano LC/MS with a Waters
199
NanoAcquity HPLC system interfaced to a ThermoFisher Orbitrap Velos Pro. Data were
200
searched using Mascot. This analysis was verified using intact protein LC/MS analysis to
201
ensure we identified the ends of the tau fragments. Liquid chromatography was
202
performed on a C8 column (Agilent) and mass spec analysis performed on a Q-ToF
203
Premiere (Waters).
AC C
EP
TE D
M AN U
SC
RI PT
179
9
ACCEPTED MANUSCRIPT
2.5 eTau Secretion Brefeldin Treatment
205
Familial AD PSEN1 (A260V) iPSC-CN were cultured for 70-108 days then washed once
206
with growth media prior to addition of fresh growth media with and without 10 µg/mL
207
Brefeldin A (Sigma); media was conditioned for four hours at 37⁰C/5%CO2 and collected
208
for analysis by IPN001 (1 µg/ml) Western blot. Lysates from treated wells were also
209
harvested with 50 µl M-PER and β-actin levels in the lysates were measured by Western
210
blot for normalization of eTau data.
211
2.6 Primary Human Cortical Neuron Treatments
212
Primary human cortical neurons were treated with either 100 nM BACE1 inhibitor
213
(BACEi) (Axon Medchem), tetrodotoxin (Tocris), 1 µM Accell siRNA pools against
214
human MAPT, BACE1, ADAM9, ADAM10, ADAM17 or Non-Targeting Control (Thermo
215
Scientific/Dharmacon, catalog E-012488, E-003747, E-004504, E-004503, E-003453
216
respectively ), 50 nM or 500 nM purified secreted tau (eTau), or mock control (iPierian),
217
E. coli generated full-length rTau383(0N4R) (rPeptide), synthetic human eTau4 peptide
218
(aa 2-68, 0N3R designation ), scrambled human eTau4 peptide control, or the equivalent
219
mouse tau sequence (aa 2-57) (Neobiolab), or with 1, 10, or 30 µg/mL protein A purified
220
tau antibodies PHF1 and MC1 (Peter Davies), IPN002, or anti-mouse IgG. Treatments
221
were diluted in basal media (Neurobasal media with Glutamax and B27, Invitrogen) and
222
incubated on neurons for a total of 20 days; conditioned media was harvested every 5
223
days and treatments were added again, for a total of 20 days of exposure. Conditioned
224
media was stored at -80⁰C until day 20 end of assay for full time course ELISA analysis,
225
and cells were lysed in 50 µl M-PER (ThermoScientific) or 50 µl Buffer RLT (Qiagen) for
226
protein or mRNA analysis.
227
2.7 Lactate Dehydrogenase Assay (LDH)
AC C
EP
TE D
M AN U
SC
RI PT
204
10
ACCEPTED MANUSCRIPT
LDH activity in in vitro conditioned media from iPSC-CN and human primary cortical
229
cultures was determined using the Lactate Dehydrogenase Activity Assay Kit (Sigma).
230
25 µl or 50 µl of five day conditioned media was assayed with an equal volume LDH
231
reagent and incubated for 30 minutes at room temperature. Plates were read on plate
232
reader at 490 nm.
233
2.8 Human sAPPα and Aβ40 and Aβ42 ELISAs
234
Human sAPPα and Aβ40 and Aβ42 protein levels were measured in primary human
235
cortical neuron conditioned media collected every 5 days over a 20 day treatment period
236
using specific ELISA assays (Invitrogen #KHB0051), (MilliporeEMD #EZHS40) and
237
(MilliporeEMD #EZHS42) respectively.
238
2.9 Mouse Aβ40 and Aβ 42 ELISAs
239
Mouse Aβ protein levels were measured in mouse cortical brain homogenate or
240
interstitial fluid (ISF) using mouse Aβ40 and/or Aβ42 ELISAs (Immuno-Biological
241
Laboratories #JP27720 and #JP27721). The mouse Aβ42 ELISA lacked the sensitivity to
242
reliably detect Aβ42 in brain tissue.
243
2.10 Gene Expression
244
Cultured neurons were lysed in Qiagen Buffer RLT and RNA isolated using an RNeasy
245
kit (Qiagen). cDNA was reverse-transcribed using the SuperScript VILO cDNA Synthesis
246
Kit (Life Technologies). Quantitative RT–PCR was performed using TaqMan Gene
247
Expression Assays and Taqman Expression Master Mix (Life Technologies) on an
248
Applied Biosystems HT 7900. Total MAPT, BACE1, ADAM9, ADAM10, ADAM17, APP,
249
PSEN1 and PSEN2 mRNA expression levels were normalized to RPLPO (housekeeping
250
gene) mRNA expression levels and analysed using the ∆∆CT method for relative
AC C
EP
TE D
M AN U
SC
RI PT
228
11
ACCEPTED MANUSCRIPT
expression. Invitrogen Taqman Gene Expression Assays used Cat#4331182 individual
252
ID # MAPT Hs00902194_m1 (for all isoforms MAPT); BACE1 Hs01121195_m1; ADAM9
253
Hs00177638_m1; ADAM10 Hs00153853_m1; ADAM17 Hs01041915_m1; APP
254
Hs01041915_m1; PSEN1 Hs00997789_m1; PSEN2 Hs01577197_m1; RPLPO
255
(Cat#4333761T). A probe recognizing only the 2N4R/1N4R/2N3R/1N3R isoforms of
256
MAPT Hs00213484_m1 was used with MAPT Hs00902194_m1 to distinguish fetal
257
0N3R MAPT from other MAPT isoforms in both primary human cultures and iPSC-CN
258
cultures.
259
2.11 Western Blots
260
Cultured primary human neurons were lysed in M-PER Buffer. Cell lysates, primary
261
cortical neuron conditioned media and iPSC-CN conditioned media, were diluted in 2X
262
Laemmli reducing buffer, boiled, separated on NuPage gels and transferred to
263
nitrocellulose using iBlot. Western blots were blocked (LiCor) and probed with
264
antibodies; IPN001 or IPN002 (1 µg/ml), HT7 (1 µg/ml), PHF1 (1 µg/ml), Dako tau
265
(1:3,000), β-actin (1:5,000), α-Tubulin (1:10,000), APP (1:1,000 Sigma) and anti-mouse
266
680 or anti-rabbit 800 secondary antibodies (1:10,000, LiCor); analysis by Odyssey SA
267
software (LiCor).
268
2.12 Animals
269
Two transgenic human tau mouse models were used in these studies. Experiments were
270
conducted in accordance with the National Institutes of Health Guide for the Care and
271
Use of Laboratory Animals guidelines and were approved by the Institutional Animal
272
Care and Use Committee. Schematic depicting overall study plans provided in
273
Supplementary Figure 3.
AC C
EP
TE D
M AN U
SC
RI PT
251
12
ACCEPTED MANUSCRIPT
The 8-week study used homozygous JNPL3 mice (Taconic, Tg(Prnp-
275
MAPT*P301L)JNPL3Hlmc) as described by (Lewis et al., 2000). The in-life study was
276
conducted by Brains On-Line. In brief, the JNPL3 line overexpresses human 0N4R tau
277
containing the most common frontotemporal dementia (FTDP-17) mutation P301L on a
278
C57BL/6, DBA/2, SW mixed background, driven by the prion promoter. Homozygous
279
JNPL3 mice express transgenic tau at levels about two-fold over endogenous tau, and
280
demonstrate motor and behavioral deficits as well as age dependent development of
281
NFT (Lewis et al., 2000).
282
The 26-week study used tau-4R/2N-P301L (van Leuven) mice as described by (Terwel
283
et al., 2005). The in-life study was conducted by reMYND. In brief, human 2N4R tau
284
containing the FTDP-17 mutation P301L was overexpressed by the mouse thy1
285
promoter on a pure FVB genetic background. Mice display an age dependent
286
hyperphosphorylation of tau, tau aggregation and tangle formation with concomitant
287
development of motor defects.
288
2.12.1 Eight week study (JNPL3 P301L mice) - 3.5 month old female JNPL3 mice were
289
treated weekly with a single dose of 60 mg/kg irrelevant IgG1 or IPN002 antibody for 8
290
weeks by an intraperitoneal (i.p.) injection volume of 10ml/kg. At the end of study mice
291
were anesthetized using isoflurane (2%, 800 mL/min O2) and stereotactically-implanted
292
with a microdialysis probe in the ventral hippocampus, a polyacrylonitrile membrane with
293
3 mm exposed surface and 1 megadalton cutoff (BrainLink, the Netherlands).
294
Bupivacain/epinephrine was used for local analgesia and fynadine or carprophen for
295
peri-/post-operative analgesia. Push-pull microdialysis sampling was performed 24 hours
296
after implantation surgery. Animal probes were connected with FEP tubing to a
297
microperfusion pump (Harvard PHD 2000 Syringe pump) and perfused with artificial
298
CSF containing 147 mM NaCl, 3.0 mM KCl, 1.2 mM CaCl2 and 1.2 mM MgCl2, and
AC C
EP
TE D
M AN U
SC
RI PT
274
13
ACCEPTED MANUSCRIPT
0.15% BSA at a flow rate of 0.75 µL/min. Microdialysis samples were collected over 60
300
minutes; basal samples were collected after stabilization. Dialysis ISF samples were
301
centrifuged at 15,000 rpm for 15 minutes at 4⁰C. The animals were sacrificed
302
immediately following microdialysis sampling and terminal brain, plasma, and CSF were
303
collected (Brains On-Line).
304
2.12.2 Twenty-six week study (tau-4R/2N-P301L van Leuven mice)- 3 to 4 month old
305
female mice were treated weekly by i.p. injection (10ml/kg) of 20 mg/kg irrelevant IgG1
306
or IPN002 antibody for up to 26 weeks (reMYND). At study end mice were sacrificed and
307
serum, CSF, and brain collected. CSF was collected via an incision in the neck muscle
308
and drawn through a puncture with a 26 gauge needle from the cisterna magna then
309
centrifuged at 10,000 x g/10 minutes at 4⁰C. Brain regions were dissected from left
310
hemisphere (cerebellum, cortex, hippocampus and hindbrain) for each mouse and
311
homogenized in Tris-Buffered Saline (TBS) (Sigma) (1:10 w/v) then centrifuged at
312
10,000 x g/10 minutes at 4⁰C. Brain fraction homogenates were diluted to 1 mg/ml in
313
TBS and total protein concentration determined by BCA Assay (Pierce); to generate a
314
soluble fraction a portion of each homogenate was further spun 1 hour at 100,000 x g
315
(4⁰C) and soluble supernatants collected. FVB wild-type cortex was provided by
316
reMYND.
317
2.13 Tau ELISAs
318
2.13.1 Free-tau ELISA – We designed a homogenous alphascreen (Perkin Elmer) assay
319
to measure tau not bound by IPN002 in mouse ISF and CSF and in primary human
320
cortical culture conditioned media and lysates. The tau antibodies used are HT7 (aa
321
159-163) and the N-terminal tau antibody IPN001 (aa 15-24) which competes with
322
IPN002 binding and therefore will not interact with tau already bound by IPN002. IPN001
AC C
EP
TE D
M AN U
SC
RI PT
299
14
ACCEPTED MANUSCRIPT
is biotinylated, and HT7 is conjugated to AlphaLISA acceptor beads (Perkin Elmer).
324
Samples are incubated overnight with both antibodies. Streptavidin coated acceptor
325
beads (Perkin Elmer) are added for 30 minutes and then the plate is read on Envision
326
plate reader (Perkin Elmer) at excitation 630nm, emission 615 nm. Tau levels were
327
interpolated from an eight point tau protein (rTau383,0N4R) standard curve and
328
analyzed using linear curve fit in Excel.
329
2.13.2 Total tau ELISA – The total tau ELISA uses the tau antibodies HT7 (aa 159-163)
330
to capture and biotinylated BT2 (aa 194-198) for detection. HT7 diluted in PBS is coated
331
on high binding ELISA plates (Costar-Corning) at 1 µg/ml overnight at 4⁰C and plates
332
blocked with 1% Casein (VectorLabs)/PBS for 2 hours at room temperature. Samples
333
are diluted in 0.1%Casein/PBS and incubated overnight at 4⁰C. Tau levels detected with
334
one hour incubation with biotinylated tau antibody BT2, followed by addition of
335
Streptavidin-HRP (Southern Biotech) and then 3,3′,5,5′-tetramethylbenzidine substrate
336
(TMB) (ThermoScientific), assay was stopped with 1M sulfuric acid (Sigma), and the
337
absorbance read on a Beckman plate reader at 450 nm. Tau levels were interpolated
338
from an eight point tau protein (rTau383, 0N4R) standard curve diluted in
339
0.1%Caesin/PBS and analyzed using 4-Parameter Sigmoidal Curve Fit (GraphPad
340
Prism).
341
2.14 Tau Binding Assay
342
To measure the kD of various tau antibodies, high binding ELISA plates were coated
343
overnight at 4⁰C with 100 ng of recombinant Tau383(0N4R) or purified eTau (iPierian)
344
diluted in PBS; wells were then blocked with 1% casein/PBS for 2 hours at room
345
temperature. Tau antibodies (IPN001, IPN002, MC1, PHF1) were diluted to 10 µg/mL in
346
PBS/0.1% casein and 12-point titrations generated, added to the plate, and incubated 2
AC C
EP
TE D
M AN U
SC
RI PT
323
15
ACCEPTED MANUSCRIPT
hours at room temperature. After washing with 0.05%Tween-20 (Sigma)/PBS, detection
348
antibody goat anti-mouse HRP (Southern Biotech) diluted 1:2,000 in PBS/0.1% casein
349
was added and incubated at room temperature for 1 hour. Wells were washed as above,
350
and TMB substrate (ThermoScientific) added, assay was stopped with 1M sulfuric acid,
351
and absorbance read at 450 nm; kD calculated based on nonlinear regression analysis
352
(GraphPad Prism).
353
2.15 Free IPN002 Levels in CSF and ISF
354
To measure the amount of free IPN002 in mouse CSF or ISF, high binding ELISA plates
355
were coated overnight at 4⁰C with 100 ng of recombinant Tau383(0N4R) diluted in PBS;
356
wells were then blocked with 1% casein/PBS for 2 hours at room temperature. CSF or
357
ISF was diluted 50 or 200 fold in PBS/0.1% casein. Coated plates were washed twice
358
with PBS, then diluted CSF or ISF samples added to the plate in duplicate and incubated
359
2 hours at room temperature. After washing with 0.05%Tween-20 (Sigma)/PBS,
360
detection antibody goat anti-mouse HRP (Southern Biotech) diluted 1:2,000 in
361
PBS/0.1% casein was added and incubated at room temperature for 1 hour. Wells were
362
washed as above, and TMB substrate (ThermoScientific) added, assay was stopped
363
with 1M sulfuric acid, and absorbance read at 450 nm; IPN002 amounts calculated
364
based on dilution factor and linear regression analysis equation of the IPN002 standard
365
titration (GraphPad Prism).
366
2.16 Electrophysiology
367
Whole cell patch clamp recording was performed on primary human cortical cultures and
368
familial AD PSEN1 (A260V) iPSC-CN cultured on a monolayer of normal human
369
astrocytes (Lonza) using micro-pipette (2-5 MOhm) filled with solution containing (mM):
370
K-methyl-sulfate (140), NaCl (10), CaCl2 (1), Mg-ATP (3); Na-GTP (0.4), EGTA (0.2),
AC C
EP
TE D
M AN U
SC
RI PT
347
16
ACCEPTED MANUSCRIPT
HEPES (10), Phosphocreatine (10) with adjusted pH= 7.3 and mOsm= 305. Neurons
372
were perfused (1-2 ml/min) with artificial cerebral spinal fluid containing (mM): NaCl
373
(140), KCl (2.5), MgCl2 (2) CaCl2 (2), Hepes (10), D-Glucose (10), sucrose (20), adjusted
374
pH= 7.3-4 mOsm = 310. Recordings were made using pClamp-10.3 data acquisition
375
software (Molecular Devices) and MultiClamp 700B amplifier (Axon Instrument; Foster
376
City CA). Puff application of eTau, full-length rTau383 (0N4R), and buffer or mock
377
protein controls was performed using MiniSquirt micro-perfusion system (AutoMate,
378
Berkeley, CA). Bath application of eTau, IPN002 antibody, and buffer controls was
379
performed for other recordings. Off-line data analysis used Clampfit 10.2 analysis
380
software (Molecular Devices). Recordings were conducted at 34-37oC.
381
2.17 Statistical Analysis
382
Statistics were determined using GraphPad Prism software. Data are expressed as
383
mean +/- standard deviation (STD) on a n=3 (unless otherwise noted in figure legends);
384
experiments repeated a minimum of four times. Analysis of tau or Aβ levels, comparing
385
the effects of one treatment versus control, at a single time point, were done by two-
386
tailed unpaired t test with Welchs correction. Analysis of tau, Aβ and sAPPα levels
387
involving repeated measures between multiple treatment groups at different time points
388
was done using one-way ANOVA with Tukeys multiple comparison correction, unless
389
otherwise noted. When significant ( 50% (Figure 1B).
409
We then asked if tau affects α- and/or β- secretase, two proteases opposed in Aβ
410
generation that compete for APP substrate. Colombo et al. show that α- and β-secretase
411
cleavage of their common substrate APP is partially coupled in primary mouse neurons.
412
We confirmed these results in primary human neurons, showing that BACE1 inhibition
413
(BACEi) results in increased production of sAPPα (soluble APP cleaved at the α-
414
secretase site) in the conditioned media, Figure 1C left, as well as increased production
415
of APP CTF-α in the lysate, Supplementary Figure 1C. Furthermore, treatment of
416
primary human neuron cultures with siRNAs directed against BACE1 and ADAM9/10/17
AC C
EP
TE D
M AN U
SC
RI PT
393
18
ACCEPTED MANUSCRIPT
resulted in opposing effects on cellular Aβ and sAPPα levels (ELISA data
418
Supplementary Figures 1E and 1F).
419
In our primary human neuron cultures we reduced MAPT gene expression and tau
420
protein (Figure 1A) levels over 70% using Accell MAPT siRNA pools. Under these
421
conditions non-targeting siRNA had no effect on sAPPα levels; MAPT knockdown
422
however resulted in a 42% increase in sAPPα (Figure 1C right).
423
Secreted Tau (eTau) identified from AD patient derived iPSC cortical neuron
424
conditioned media is composed of N-terminal tau fragments
425
In light of these findings and recent data showing tau to be secreted from human
426
induced pluripotent stem cell (iPSC) derived cortical neurons (Chai et al., 2012), we
427
characterized tau release from our iPSC cortical neurons (iPSC-CN) derived from AD
428
patients. To do so, we generated iPSC from a skin biopsy taken from a familial
429
presenilin-1 (A260V) AD patient aged 50 then differentiated and cultured iPSC-CN for up
430
to 108 days. To characterize eTau secretion from these iPSC-CN we collected
431
conditioned media beginning at day 70 over time for 96 hours. As shown in the tau
432
Western blot in Figure 2A, tau is present in the conditioned media of iPSC-CN, and
433
quickly accumulates over time.
434
To ensure that the tau found in the conditioned media was not the result of cell lysis we
435
measured the lactate dehydrogenase (LDH) activity in the conditioned media of several
436
of our iPSC-CN lines listed in Table 1, and found all to be free of cell lysis
437
(Supplementary Figure 2A). These data and an absence of other high abundance
438
cytoplasmic proteins such as β-Actin and α-Tubulin (Supplementary Figure 2B) in the
439
conditioned media together confirm the lack of contamination by intracellular proteins in
440
iPSC-CN conditioned media. The tau release we observed from our iPSC-CN is rapid,
441
detectable within 2-5 minutes, and plateaus over 24-96 hours (Figure 2A). iPSC-CN tau
AC C
EP
TE D
M AN U
SC
RI PT
417
19
ACCEPTED MANUSCRIPT
secretion is not modulated by Brefeldin A treatment, indicating that secretion is occurring
443
independently of the canonical secretory process (Figure 2B).
444
Since iPSC-CNs derived from AD patient samples secrete tau in the absence of
445
neuronal lysis, we characterized the secreted tau species identified in the media. To do
446
so we attempted to immunoprecipitate tau from the media (conditioned 3 days) of 70 day
447
old iPSC-CNs derived from a familial presenilin-1 (A260V) AD patient with either a N-
448
terminal tau antibody (IPN002), a mid-range tau antibody (HT7), or a C-terminal tau
449
antibody (T46). Each of these antibodies efficiently isolates cytoplasmic full length tau
450
(data not shown) but only IPN002 and HT7, not T46, were capable of
451
immunoprecipitating secreted tau (Figure 2C). The C-terminal tau antibody’s failure to
452
isolate secreted tau is shown by IPN001 Western blot analysis of the three
453
immunoprecipitated products; indicating that tau is released from AD iPSC-CN
454
predominately as N-terminal fragments, ranging from 20-28 kDa (Figure 2D) with no
455
evidence of C-terminal reactive fragments. This data is also consistent with reports
456
showing that tau is secreted into AD patient CSF as N-terminal fragments (Meredith, Jr.
457
et al., 2013). We have conducted similar characterizations of eTau secreted from cortical
458
neuron from healthy control subjects and found a similar molecular weight range of eTau
459
fragments as those found for AD patients (data not shown).
460
To verify the 20-28 kDa bands as tau and to identify their sequence, we affinity isolated
461
secreted tau from our iPSC-CN conditioned media with the tau antibody IPN002 and
462
conducted tryptic, chymotrypic and elastase digestion followed by LC/MS analysis on the
463
four major IPN002-affinity isolated, anti-tau reactive bands shown in Figure 2D. We also
464
used intact protein LC/MS to ensure we identified the end of the largest tau fragment.
465
This analysis confirmed that secreted tau is a series of N-terminal tau fragments; all
466
lacking the C-terminal amino acids of tau, including the microtubule binding repeats as
AC C
EP
TE D
M AN U
SC
RI PT
442
20
ACCEPTED MANUSCRIPT
shown in the sequence alignment in Figure 2E. We named the four major fragments
468
eTau 1, 2, 3 and 4 and refer to the pool of secreted tau as eTau (for extracellular tau).
469
Our iPSC-CNs only express the 0N3R embryonic isoform of tau (Supplementary Figure
470
1B), and LC/MS analysis confirmed the n1 and n2 inserts to be missing in the
471
sequenced fragments, Figure 2E. All forms of eTau start at the second amino acid,
472
alanine, with the N-terminal methionine removed and the alanine acetylated. This is a
473
covalent modification traditionally indicative of a cytosolic protein, providing further
474
evidence that eTau is released independent of the canonical endoplasmic
475
reticulum/golgi secretory process. The four major forms of eTau we identified range in
476
size from 67 amino acids (aa 2-44,103-126, eTau4) to 171 amino acids (aa 2-44, 103-
477
230, eTau1), Figure 2E. Phosphorylation of eTau was rarely detected but when found
478
the residues phosphorylated were consistent with phosphorylation of cytoplasmic tau at
479
residues threonine181 (p181), serine202 (p202), and threonine231 (p231).
480
We further characterized eTau by its reactivity with several tau antibodies, specific
481
affinities are shown in Table 2. Both 0N4R recombinant full length tau-383, and eTau
482
bind with high affinity to the N-terminal tau antibodies MC1 (a conformational specific tau
483
antibody) (Jicha et al., 1997) and IPN002 (generated against aggregated tau, iPierian).
484
Not surprisingly, eTau does not bind to the C-terminal anti-phospho-tau antibody PHF1
485
(Jicha et al., 1997) in either the tau binding assay (Table 2) or by Western blot
486
(Supplementary Figure 2C), indicating that eTau lacks this phosphoepitope, as expected
487
based on LC/MS analysis.
488
eTau regulates amyloid-beta (40 and 42) levels in vitro in primary human neurons
489
We found that this pool of N-terminal eTau fragments is also secreted from human
490
primary cortical neurons at concentrations up to 3-4 nM, when allowed to condition for 5
491
days (data not shown). Total tau ELISA of the conditioned media of human primary
AC C
EP
TE D
M AN U
SC
RI PT
467
21
ACCEPTED MANUSCRIPT
neurons confirmed this tau secretion, and demonstrated that MAPT knockdown in these
493
primary human neurons significantly reduces secreted eTau levels in the conditioned
494
media over time when compared to non-targeting siRNA control (Figure 3A).
495
To determine if eTau plays a role in regulating Aβ40 and Aβ42, we treated primary human
496
neurons for 20 days (media and treatments replaced every 5 days) with affinity purified
497
eTau from AD patient derived iPSC-CN or mock protein control, and saw a dose and
498
time dependent (as early as 10 days at 500 nM) increase in levels of Aβ40 and Aβ42
499
(Figure 3B). The changes in Aβ levels are small but reproducible over multiple
500
experiments and are statistically significant.
501
Since the conditioned media from primary human neurons, when allowed to condition on
502
cells for five days, already contains secreted eTau at concentration up to 3-4 nM, a large
503
excess of purified secreted eTau (500 nM) appears to be required to generate these
504
additional small changes. Less eTau was needed (75% with siRNA transfection (data not shown). Similar
507
experiments conducted with up to 1 µM of full length rTau383 (0N4R) did not affect Aβ
508
levels (data not shown) suggesting an eTau specific regulation of Aβ levels.
509
eTau regulates sAPPα α levels in vitro in primary human neurons
510
Our data indicate that eTau may be affecting how Aβ is produced. To understand how
511
eTau is affecting Aβ production at the mRNA level we measured the gene expression of
512
APP, BACE1, ADAM9, ADAM10 and ADAM17 in human primary neurons but did not
513
detect any significant gene expression changes after treatment with either MAPT siRNA,
514
purified secreted eTau, or tau binding antibodies (gene expression data not shown).
515
The addition of eTau to primary human neurons for 20 days caused a small but
516
significant, dose dependent decrease in sAPPα levels (Figure 3C). The simultaneous
AC C
EP
TE D
M AN U
SC
RI PT
492
22
ACCEPTED MANUSCRIPT
increase in Aβ with the decrease in sAPPα indicates that eTau is affecting the
518
production of Aβ in neurons.
519
eTau binding antibodies reduce amyloid-beta (40 and 42), increase sAPPα levels
520
in vitro in primary human neurons
521
We tested each of several antibodies for binding affinity to tau and eTau. As expected all
522
tau antibodies tested bound with good to excellent Kd’s and as expected IPN001,
523
IPN002 bound to eTau, while PHF1 did not bind eTau, Table 2. Unexpectedly, we found
524
that MC1 bound with high affinity to not only tau but also eTau, Table 2. This was
525
unexpected since MC1 is reported to bind to a conformational epitope comprised of both
526
aa 7-9 and aa313-322, but aa313-322 are not present on eTau. This suggests that MC1
527
can also bind to eTau as well as to the conformation epitope described previously. Peter
528
Davies has also found that MC1 binds to the N-termini of tau, in the absence of the C-
529
terminal epitope (personal communication). To confirm therefore that eTau regulates Aβ,
530
we treated our human primary cortical cultures with various tau antibodies. We used the
531
eTau binding antibodies IPN002 and MC1, as well as PHF1, shown not to bind to eTau.
532
We found that when added to cortical neurons, MC1 and IPN002 both time dependently
533
reduced Aβ40 and Aβ42 levels, while the IgG control and PHF1 had no effect (Figure 4A).
534
Additionally, eTau binding antibodies (MC1 and IPN002) increased levels of sAPPα over
535
time, while PHF1 did not (Figure 4B). APP Western blots of the cell lysates of IPN002
536
treated primary human cortical cultures confirm these ELISA results, showing a slight
537
increase in APP CTF-α when compared to IgG control (Supplementary Figure 1C).
538
To demonstrate that our IPN002 antibody is acting extracellularly, specifically binding
539
and neutralizing secreted and not intracellular tau, we used our IPN002-free-tau ELISA
540
assay to analyze the conditioned media and lysate of primary human cortical neurons
541
treated with IPN002. This is an IPN002 competitive tau assay, which only detects tau not
AC C
EP
TE D
M AN U
SC
RI PT
517
23
ACCEPTED MANUSCRIPT
already bound by IPN002. Analyzing the conditioned media of primary human neurons
543
treated with either IgG control or the N-terminal tau antibody IPN002 every 5 days for 20
544
days, we found that the vast majority of eTau in the conditioned media of these cultures
545
was bound to IPN002, measured as a significant decrease in IPN002-free tau (Figure
546
4C, right). Analyzing the lysates of these cultures we found no change in free tau levels
547
with IPN002 treatment (Figure 4C, left). This indicates that IPN002 is not entering the
548
cell in sufficient quantities to bind/neutralize intracellular tau, and suggests that the
549
antibody’s observed effect on Aβ levels is through its interaction with secreted eTau.
550
These results suggest that eTau plays a role in modulating Aβ levels in vitro. In contrast
551
to the β-secretase inhibitor which rapidly (1 day) and dramatically (75-85%) lowers Aβ
552
levels (Supplementary Figure 1D), IPN002 and MC1 lower Aβ levels more gradually (10
553
days) and modestly (30-40%) (Figure 4A). The gradual effect of these eTau binding
554
antibodies is indicative of a modulatory rather than a direct, inhibitory role.
555
Human synthetic eTau 4 increases Aβ β (40 and 42) levels in vitro in primary human
556
cortical neurons
557
In vivo studies in APP transgenic mice show that reductions of endogenous murine tau
558
do not impact Aβ levels (Roberson et al., 2007; Roberson et al., 2011; Ittner et al.,
559
2010). Therefore, to further understand our findings in the context of these studies
560
involving mouse tau, we asked if human and mouse tau affect Aβ production differently.
561
To do so we generated a synthetic version of human eTau 4, a 67 amino acid peptide
562
corresponding to the shortest secreted N-terminal tau fragment identified by LC/MS from
563
iPSC-CN media. We also synthesized a peptide corresponding to the equivalent amino
564
acid sequence in mouse tau (Figure 5A). Since both our AD patient derived iPSC-CN
565
cultures and our primary human cortical cultures predominately express embryonic
566
0N3R tau (Supplementary Figure 1B) we did not include the n1 and n2 inserts in these
AC C
EP
TE D
M AN U
SC
RI PT
542
24
ACCEPTED MANUSCRIPT
synthetic sequences. Notably mouse tau is missing the human tau insert, amino acids
568
17-28, which encompasses the majority of the IPN002 epitope (aa 15-24).
569
We treated primary human neurons with either the human or mouse eTau 4 peptide or a
570
scrambled control peptide, and looked for differential effects on Aβ levels. By using
571
synthetic versions of the sequences we eliminated any potential post-translational
572
differences between the two species tau, and treated the same primary human neuronal
573
cultures for direct comparison. Under these conditions human eTau 4 treatment
574
significantly increased Aβ40 and Aβ42 levels (18% and 26% respectively compared to
575
scrambled control), while treatment with mouse eTau 4 did not increase Aβ40 and Aβ42
576
levels (Figure 5B). Again, the increases in Aβ levels with human eTau4 are small but
577
reproducible and highly significant.
578
In fact, by treating the primary human neuronal cultures with an additional peptide
579
encompassing only the human tau insert, we show that the 12 amino acids missing from
580
mouse tau are capable of increasing Aβ40 and Aβ42 to the same level as the entire
581
human eTau 4 peptide (Figure 5B). These data confirm that human eTau increases Aβ
582
production and suggests that the modulation of Aβ production is human eTau specific
583
via an activity conferred by the human tau insert.
584
The eTau antibody, IPN002, reduces Aβ β (40 and 42) levels in vivo
585
To test our hypothesis in vivo, we first used the transgenic P301L tauopathy model
586
(JNPL3, Taconic) that releases mutant human tau into interstitial fluids (ISF) (Figure 6A).
587
We found that tau is present in the ISF of these mice at a truncated size of ~25 kDa,
588
Supplemental Figure 2d, similar to eTau found in the AD patient cortical neuron
589
conditioned media. This is in contrast to the findings by (Yamada et al., 2011) showing
590
full-length tau in the ISF of P301L mice. We treated 3.5 month old JNPL3 mice weekly
591
with either IgG1 control or IPN002 for eight weeks, collected ISF at 5.5 months of age,
AC C
EP
TE D
M AN U
SC
RI PT
567
25
ACCEPTED MANUSCRIPT
and measured the levels of IPN002 antibody, IPN002-free tau, and Aβ40 and Aβ42. To
593
verify that IPN002 entered the central nervous system (CNS) we measured levels of
594
IPN002 in the plasma and ISF of each animal at study termination and found IPN002
595
present in the ISF in all cases (Figure 6B). Additionally, we found that IPN002 bound
596
almost all the human mutant tau available in the ISF, measured by a 94% reduction in
597
the free tau levels (Figure 6C). Aβ40 and Aβ42 analysis of the animal’s ISF at the end of
598
the eight week treatment showed IPN002 treatment reduced endogenous mouse Aβ 40
599
and Aβ42 levels by ~25% (Figure 6D).
600
We validated these initial findings in a second transgenic P301L tauopathy model, using
601
the tau-4R/2N-P301L (van Leuven) mice, in which secreted human tau levels increase
602
with age, Figure 6E). In this model, 3.5 month old animals have < 10 ng/ml human tau in
603
their cerebrospinal fluid (CSF), by 9.5 months their CSF tau levels increase to an
604
average of 30 ng/ml (Figure 6E). We treated these mice weekly, for up to 26 weeks, with
605
20 mg/kg IgG1 control or IPN002. To verify that IPN002 entered the CNS, we measured
606
levels of IPN002 in the plasma and CSF of each transgenic animal at 9.5 months and
607
found IPN002 present in the CSF in all dosed animals (Figure 6F). CSF analysis by
608
IPN002-free-tau ELISA assay confirmed that human mutant tau was fully engaged by
609
IPN002, measured by a 98% reduction in free tau levels, Figure 6E, IPN002 treated
610
compared to IgG1 control treated mice.
611
We measured Aβ40 levels in both the total homogenate and the soluble fraction of the
612
hindbrain (data not shown) and cortex; the results were consistent between fractions and
613
brain regions. As shown in Figure 6G, Aβ40 levels in the soluble fraction of P301L mouse
614
cortex increase from 14.4 pg/ml at 3.5 months to 22.5 pg/ml at 9.5 months (IgG1 control
615
group). This substantial increase in levels of murine Aβ40 in a tau mouse model may be
616
due either to age, to continued human tau expression, or to a combination of both. To
AC C
EP
TE D
M AN U
SC
RI PT
592
26
ACCEPTED MANUSCRIPT
distinguish between these possibilities, we measured Aβ 40 levels in the soluble fraction
618
of the cortex of wild type C57/BL6 mice at 4.3 months and 9.5 months of age (Figure
619
6G) and saw no change in Aβ with age. This suggests that the observed increase in
620
mouse Aβ over time in this P301L tau mouse model is due to human tau expression.
621
To determine if elevated levels of Aβ40 could be lowered with the human eTau binding
622
antibody IPN002 we compared the Aβ 40 levels of IgG1 and IPN002 treated mice and
623
found that IPN002 treatment significantly reduced levels of soluble mouse Aβ40 by >
624
25%, Figure 6G. Together these data suggest that human tau released from neurons
625
modulates Aβ production both in vitro and in vivo.
626
TTX treatment lowers Aβ β and eTau levels in primary human cortical neurons in
627
vitro
628
How does eTau modulate Aβ production? In AD, secreted tau is hypothesized to be
629
released pre-synaptically and then taken up post-synaptically, inducing spread, tau
630
pathology, neuronal dysfunction, and neuronal hyperactivity. Kamenetz et al. showed
631
that picrotoxin induced hyperactivity in hippocampal slices dramatically elevated Aβ 40
632
and Aβ42 while TTX treatment was able to lower Aβ levels (Kamenetz et al., 2003);
633
Cirrito et al. found a similar link between heightened neuronal and synaptic activity and
634
increased Aβ production in living mouse ISF (Cirrito et al., 2005; Cirrito et al., 2008).
635
To determine whether our observed increase in Aβ could be the result of eTau induced
636
neuronal hyperactivity, we treated our primary human cortical cultures with the sodium
637
channel blocker TTX. As has been shown previously in vitro and in vivo, TTX is capable
638
of significantly reducing soluble Aβ 40 and Aβ42 in a dose and time dependent manner
639
(Figure 7A), with no change to cell viability (data not shown). In these same cultures,
640
eTau levels analyzed in the conditioned media of the TTX treated neurons were also
AC C
EP
TE D
M AN U
SC
RI PT
617
27
ACCEPTED MANUSCRIPT
reduced, by 60%, after 10 days of TTX treatment (Figure 7B). eTau in the conditioned
642
media of neurons is significantly reduced by day 10 of TTX treatment; however, Aβ
643
levels are not affected until day 15, suggesting that suppressing neuronal activity
644
reduces eTau levels and subsequently alters Aβ production.
645
eTau induces hyperactivity in human neurons
646
Increased neuronal activity potentiates Aβ levels in vitro and in vivo. This suggests that
647
one mechanism by which eTau can modulate Aβ production is through increasing
648
neuronal activity. To test this hypothesis, we performed whole cell patch clamp
649
recordings to determine the effect eTau has on cortical neuron activity. Application of
650
eTau increased the activity of both iPSCs-derived cortical neurons and primary human
651
cortical neurons (Figure 7C). Interestingly, application of full length tau (rTau383 0N4R)
652
at significantly higher concentrations (up to 3 µM) failed to produce any changes in
653
neuronal activity similar to those caused by eTau treatment. This suggests that the eTau
654
mediated neuronal activity effects may involve a specific interaction of the eTau protein
655
with cortical neurons. This interaction appears to be specific to the N-terminal region of
656
tau as it cannot be replicated with full length tau (rTau383 0N4R). To test this
657
hypothesis, we pre-incubate eTau with the neutralizing antibody IPN002 to block eTau
658
mediated hyperactivity. At a 10:1 molar ratio of IPN002 to eTau, IPN002 inhibited eTau
659
mediated hyperactivity (Figure 7D). Together, these results suggest that eTau can
660
modulate Aβ levels by promoting cortical neuron hyperactivity.
661
4. Discussion
662
In this study we describe the identification of a novel secreted N-terminally truncated tau
663
species (eTau), present both in vitro and in vivo, and show that these secreted tau
664
fragments can regulate Aβ production. We initially purified eTau from the conditioned
665
media of AD patient derived cortical neuron cultures, and found that secreted tau is
AC C
EP
TE D
M AN U
SC
RI PT
641
28
ACCEPTED MANUSCRIPT
composed predominantly of N-terminal fragments of tau. In vitro we show that treating
667
primary cortical cultures with eTau simultaneously increases Aβ levels while decreasing
668
sAPPα levels. Globally reducing tau with MAPT siRNA or treating neurons with
669
neutralizing eTau antibodies lowers Aβ levels and increases sAPPα levels in parallel.
670
Importantly, we validated these findings in vivo in two tau transgenic (P301L) mouse
671
models. Levels of human tau are elevated in the ISF and CSF of the tau transgenic
672
animals; treating these mice with an eTau-neutralizing antibody (IPN002) reduced
673
endogenous murine Aβ in both their ISF and brain. Finally, in vitro we demonstrate that
674
eTau treatment increases neuronal hyperactivity. We propose that this eTau driven
675
neuronal hyperactivity increase, results not only in increased Aβ secretion, but also in a
676
further increase in eTau secretion, and that together eTau and Aβ create a feed forward
677
disease mechanism that perpetuates the disease (Supplementary Figure 4).
678
We examined the conditioned media of AD patient derived cortical neurons for the
679
presence of tau in the absence of neuronal lysis and found tau fragments. Analysis of
680
the tau fragments by LC/MS for primary sequence and post-translational modifications
681
showed that this secreted tau is composed predominantly of four N-terminal fragments,
682
all containing a N-terminal acetylated alanine and a low percentage (