Max Planck Institute for Human Cognitive and Brain Sciences Liepzig Germany National Center for Geriatrics and Gerontology Center for Development of Advanced Medicine for Dementia Aichi Japan Dr Peng Wang ID: 713605
Download Presentation The PPT/PDF document "Cortical excitability in Alzheimer‘s ..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Cortical excitability in Alzheimer‘s disease
Max Planck Institute for Human Cognitive and Brain Sciences Liepzig, Germany
National Center for Geriatrics and Gerontology, Center for Development of Advanced Medicine for Dementia, Aichi, Japan
Dr. Peng Wang,
Dr. Akinori Nakamura
Dr. Thomas
Knösche, Dr. Burkhard Maess
Presntation for CogNeuro Journal Club, Singapore
17.06.2016Slide2
Challage of diagnosis and treatment of Alzheimer`s
disease
The early stage is often overlooked, see it as „old age“.
The neurodegenerative processes starts
10-15
years before clinical symptom.
OR
1Slide3
Alzheimer`s disease
The first description of this disease was in 1907 by Dr. Alois
Alzheimer.
Senile plagues and neurofibrillary tangle were discoverd in the patient‘s brain.
. Senile plagues
. Neurofibrillary tangle
2Slide4
Beta-amyloid leads to neural hyperexcitability (1)
Presynaptic terminal
Postsynaptic component
1. promotes presynaptic glutamate
release
1.
Glial cell
2. inhibites glutamate recycling
Amyloid-ß (Aß)
(for review: Paula-Lima et al., 2013)
2.
Enhancement of the excitatory signal
pathway
(glutamatergic)
3Slide5
Beta-amyloid leads to neural hyperexcitability (2)
Presynaptic terminal
Postsynaptic component
Glial cell
Amyloid-ß (Aß)
(for review: Paula-Lima et al., 2013)
4
4.
4. Aberrant activation of NMDA
3.
3. increase of D-Serine release
Aberrant activation of NMDA receptorSlide6
Measure cortical excitability: Short-term adaptation of cortical responses
S1
R1
S2
S2
R2
S2
R2
S2
R2
A
daptation
Recover from the adaptation
ISI
Associating with the change in neural synaptic connection / neurotransmission
(Zucker & Regehr, 2002)
5Slide7
35 50 70 100 150 200
1.2
1.0
0.8
0.6
0.4
0.2
0
R2/R1
ratio
young
old
(David-Jürgens & Dinse, 2010)
ISI (
S2-S1) [ms]
6
Recovery
functions of cortical excitability are different between aged and young ratsSlide8
Effects of age on AMPA- to NMDA-mediated neurotransmission
(
Zhiyong
Yang et al
., 2008)
7Slide9
Research motivation
. Alzheimer‘s disease
Overactivation of NMDA
. Aging
Reduced activity of NMDA
. Experiment
Recovery function of cortical excitability
. Research intersts
8
Recovery function of AD ?
What is the neural mechnisms behind the observations?Slide10
Electrical stimulation of the right median nerve (wrist) at
130% of the motor threshold.
Six different recording types:
a single pulse
and pairs of pulses with an interval of 30, 60, 90, 120, and 150ms
SEF-R experiment - by Dr. Nakamura (1)
9Slide11
ISIs
S1
S2
S1
S2
Time
400-600ms
150 repetitions of each type with a randomly jittered
SOA of 400 to
600ms
Conducted
at
National Center for Geriatrics and Gerontology, Aichi, Japan
SEF-R experiment - by Dr. Nakamura (1)
10Slide12
Neuromag Vectorview (204 planar gradiometers and 102 magnetometers, 1000 Hz sampling, 0.1-120 Hz bandwidth)
Epochs -100 to 300ms centered at the first pulse (S1)
SEF-R experiment - by Dr. Nakamura (2)
11
gradiometerSlide13
Electrical brain activity measurable by MEG/EEG: Postsynaptic potentials (PSP
) of synchronized pyramidal cells
Hansen,
Kringelbach
,
Salmelin
(eds.): MEG an introduction to methods. 2010
B
Single
pyramidal cell
: 0.2
pAm,
MEG measures sources
~10 nAm:
50,000 cells (
current x distance
)
Cylinder
of
~1mm
diameter
Palisade-
like
arrangement
(open
field
)
12Slide14
Selected the pair of gradiometers with the strongest response at 20msPCA of both gradiometer signals – 1st. PCA component
for the neural dynamic modeling
(Courtesy of Dr.
Maess
)
SEF-R experiment - by Dr. Nakamura (3)
13
nose
left
right
Selected Gradiometer pair
Time (s)
ft/cm
gradiometer B
out
(Picture from Matti Hämäläinen,2009)Slide15
Estimated neural activity in S1
fT/cm
ms
S1-N20m
S2-N20m
1st. PCA component of the chosen gradiometer pair
Both R1 and R2 responses in evoked signal
14Slide16
Estimated neural activity in S1
fT/cm
ms
S2-N20m
15
R1 response suppressedSlide17
Participents
Total = 50 ParticipentsHealthy elderly = 21 (67±5), PiB-Patients = 19 (76±4) (AD = 9 & MCI = 10), PiB+
Young = 10 (24±4)
Age
80
60
40
20
Patient H. elderly Young
16Slide18
Recovery
function (SEF-R): healthy elderly vs young
17
Healthy elderly
vs
Young
:
ISI90 (p<0,05), ISI120 (p<0,05), ISI150 (p<0,05)
s1-N20m/s2-N20 ratioSlide19
Recovery
function (SEF-R):healthy elderly vs patients
18
Healthy elderly
vs
Patients
:
ISI90 (p<0,05), ISI120 (p<0,001),ISI150 (p<0,05)
s1-N20m/s2-N20 ratioSlide20
Recovery
function (SEF-R): patients vs young
19
Patients
vs
Young
:
ISI60 (p<0,05), ISI90 (p<0,01), ISI120 (p<0,01),
ISI150 (p<0,01)
s1-N20m/s2-N20 ratioSlide21
Hypothesis
. Imaging
Different SEF-R among patients, healthy elderly and young
. Computation
Mapping imaging data into model parameters
20
glutamatergic
transmission and NMDAr activity ?
We are looking for ...
A generative model for the MEG data with biological interpretable parameters
Slide22
Neuron, membrane potential and firing rate
1. receive infomation
2. change in membrane potential
3. fire spikes
10mv
50ms
21Slide23
Neural Mass Model as generative model
22
The neural mass model summarizes the behaviors of millions of interacting neurons through two states: average membrane potential and average firing rate. (Lopes da
Silva et al., 1974)
MEG/EEG signals are generated by the massively synchronous pyramidal cells.
Cortical column is the basic functional unit of the cortex. (Mountcastle, 1997) Slide24
[mV]
[ms]
[1/s]
[mV]
u
(t)
Q(t)
Q
A
(t)
A
B
Q
A
u
B
u
B
(t)
Basic Equations of NMM: Connection form A to B
23
Receptor time constant
A.
B
.
.Neurotransmission strength
.Receptor time constant
u
B
(t)
Q
A
(t)Slide25
Modeling short-term synaptic adaptation
Depletion and recycling of the neurotransmitter
Ready, W(t)
Not ready, 1-W(t)
adaptation rate
presynaptic activity
recovery rate
24
C: synaptic connection strength
W: connection efficiency of synaptic connection strength (0<W ≤ 1)
Q: presynaptic average firing rate
n1: adaptation rate
n2: recovery rateSlide26
Modeling laminar organization in a cortical column
Cortex has layered structure.
Different neurons are organized in different layers
special local circuit among the neurons/layers
(Thomason & Lamy, 2007)
Local cortical circuit model
pyramidal cells
i
nhib. interneurons
excit. interneurons
II/III
V/VI
IV
(Wang & Knösche,2013)
input
25
certain
uncertainSlide27
Parameter Estimation: Bayesian framework
Constraints on neural
parameters
Models of
mean activity of neural populations
synaptic adaptation
Bayesian estimation
posterior
prior
likelihood
Parameters
q
are optimized so that the predicted matches the measured data
y
(Friston, 2002 & Kiebel et al., 2009)
26Slide28
0 150 350(ms)
Normalized neural activity
1
0
1
Fitting using LCCM
single
ISI30
ISI60
ISI150
ISI120
ISI90
experimental
LCCM
error
27
(Data from healthy elderly)Slide29
Fitting using LCCM(all 6 conditions)
28
GoFSlide30
One way ANOVA 59 free parameters Bonferoni correction (factor 59)
II/III
IV
V/VI
pyramidal cells
excit. interneurons
EIN->dPC:
connection strength
receptor time constant
2
9
Parameter statisticSlide31
The patient group showed stronger excitatory connection
connection strength EIN->dPC
*
*
p
< 0.01
Arbitrary unit
*
p
<0.05
The Patient group showed an hyper-excitation in glutamate pathway
30Slide32
The patient group showed larger receptor time constant
receptor time constant EIN->dPC
[ms]
*
*
p
<
0.01
*
*
p
<
0.01
The time constant represented a lumped effect from both fast AMPA and slow NMDA. In the Patient group, the weight moved to the NMDA side.31Slide33
Simulated receptor dynamics
1
0
Time (ms)
0 50 150 250
aged
patient
young
0 50 150 250
1
0
Normolized neural activity
Time (ms)
measured in rat nerocortex
(single neuron level)
fitted with exponential function
AMPA
NMDA
Normolized neural activity
(Angulo et al.,1999)
estimated synaptic
dynamics in LCCM
using group mean value
3
2Slide34
Age ranking: Receptor time constant
R
2 = 0.61 p < 0.01
age
Time constant (s)
3
3
useful biomarker
using MEG to imaging aging and disease progress
follow-up studySlide35
How to explain the „overshoot“ in
SEF-R)
34
Hypothesis: „Overshoot“ of SEF is caused by different adaptation-recovery time loop of excitatory and inhibitory pathway. Aged people has slower recovery of inhibitory than young one.
Explained by hyper-excitabilitySlide36
„Young“ can be explained by model with very fast adaptation on inhibitory pathway (or extremely, no adapatation ), but the „Patient“ and „Healthly elderly “ can not.
35Slide37
Comparison of LCCM and LCCM without INN adaptation
GoF
LCCM_no_INN_adaptation
LCCM
p_value
Patient
0,84
0,94
p<0,01
Healthy
0,83
0,93
p<0,01
Young
0,93
0,95p=0,4
Average LogModelEvidence
LCCM_no_INN_adaptation
LCCM
Young
1583
1458
Evidence for that: INN adaptation is necessory to simulate the data of Patient and Healthy Elderly. But, this adaptation effect is less impacted to explain the data of young
36Slide38
Conclusion
Distingushing aging and Alzheimer‘s disease using EEG/MEG (SEF-R)Different changing courses of NMDA-mediated neurotransmittion in patients and healthy elderly
Aging effect on recovery of inhibitory connections from adaptationImaging synaptic (receptor) dynamics using LCCM
Suggestion for the follow-up study
37Slide39
Thank you for your attentionSlide40
Model evidence
(Penny et al., 2012)
The log evidence for model m can be split into anaccuracy and a complexity term:
Log(
m
) = Accuracy(
m) - Complexity(
m)Slide41