1. No category

SUPPLEMENTARY MATERIAL
Closed-loop optogenetic control of thalamus as a new tool
to interrupt seizures after cortical injury
Jeanne T. Paz, Thomas J. Davidson, Eric S. Frechette, Bruno Delord, Isabel Parada,
Kathy Peng, Karl Deisseroth & John R. Huguenard
a Cortical
stroke
Camk2a:eNpHR
injection in vivo
in thalamus
0
1 week
Death of CT
cells
and axons
Chronic EEG/optrode
implants for behaving
recordings/optical stim.
Thalamic
hyperexcitability
RT TC
Death of TC
cells
and axons
Peri-stroke
Cortex
CT
4 weeks
3 weeks
Post-stroke cell death
<1 week post-stroke
b
Awake behaving recordings
and optical stim.
Stroke
Time
Epilepsy
Post-stroke epilepsy/hyperexcitability
in the surviving TC circuits
>1 week post-stroke
Peri-stroke
CT
10 months
CT
Stroke
Selective optical inhibition of TC neurons
interrupts seizures
Peri-stroke
CT
CT
Stroke
CT
Thalamus
RT
Thalamus
VB
TC
TC
VPL
VPM
TC
TC
TC
VPL
VPM
VPL
594nm
TC VPM
Camk2a:eNpHR:eYFP
c
EEG recording system
1
2
Real-time digital signal processor
(calculates line-length)
3
Laser
(594 nm)
Seizure
interruption
Seizure
onset
EEG
4
EEG
Thr.
Line-length
EEG
Thr.
Line-length
Processor detected seizure
and triggered light
3’
EEG
Thr.
Line-length
Processor detected seizure
but did not trigger light
Supplemental Figure 1. Experimental design.
a, Timeline showing sequence of events. Green and yellow boxes indicate experiments involving optogenetics. Light grey
box indicates time of in vitro recordings (2 days –6 months post-stroke). b, Diagrams of the thalamocortical loop comprised
of cerebral cortex, thalamocortical relay nuclei and the reticular thalamic nucleus (RT). Blue and black projections
correspond to GABAergic inhibitory and glutamatergic excitatory pathways, respectively. b, left: Cortical infarct results in
death (dashed lines) of cortical neurons and corticothalamic (CT) axons and, by the end of the first week, in death of TC
cells in VPL, and does not affect intra-RT inhibition19. VB: somatosensory ventrobasal complex. b, middle: The surviving
thalamocortical loop becomes hyperexcitable (red regions and thicker projections) and generates epilepsy. b, right:
Camk2a:eNpHR viral expression in TC neurons enables inhibition of these cells with yellow light and thus reduced
excitatory output to the cortex and interruption of seizures in awake freely behaving animals. c, Real-time detection and
interruption of seizures: 1) a cortical EEG channel recorded in the awake behaving rat was routed from the recording system
to a programmable real-time digital signal processor. 2) The processor calculated the line-length in a sliding window of 2
seconds (see Methods for details). Upon upward crossing of the line-length threshold (dashed line), the system randomly
triggered laser activation (3) or not (3’). Laser activation (3) resulted in light delivery in thalamus (4) that typically
interrupted the seizure.
b
Cumulative probability
HCN2 / Biocytin
Injured
Control
HCN4 / Biocytin
Injured
Control
20 mm
Cumulative probability
a
HCN2/Biocytin
1.0
Control
Injured
0.5
0.0
0
50
100 150
Volume (mm3)
HCN4/Biocytin
1.0
Control
Injured
0.5
0.0
0
50
100 150
Volume (mm3)
Supplemental Figure 2. Cortical stroke leads to a switch from predominant HCN2 to
predominant HCN4 channels in TC neurons.
a-b, HCN2 (a, Left) and HCN4 (b, Left) channel immunolabeling from representative control and
injured TC neurons filled with biocytin during electrophysiological recordings from slices 7-14 days
post-stroke. a, Right: Cumulative probability distributions of the volume of HCN2 particles from
control TC neurons (n = 3 cells, 1200 values, 400 values per cell, from 2 rats) and injured TC neurons
(n = 3 cells, 1200 values, 400 values per cell, from 2 rats) are significantly different (p<10-7 One way
ANOVA Student-Newman-Keuls test). b, Right: Cumulative probability distributions of the volume
of HCN4 particles from control TC neurons (n = 4 cells, 380 values, 95 values per cell, from 2 rats)
and injured TC neurons (n = 4 cells, 1160 values, 290 values per cell, from 2 rats) are significantly
different (p = 8.10-60, One Way ANOVA Dunn’s test). These differences in HCN subunit expression
could explain at least in part the changes in biophysical properties of Ih as suggested by [32].
a
25 mV
b
500 ms
-72.4mV
25 mV
500 ms
-65.7mV
Supplemental Figure 3. Current-clamp membrane potential traces in a Hodgkin-Huxley model
of an individual TC cell.
a, Control conditions. Note the absence of action potential upon depolarizing pulses and the moderate
sag, as found in whole-cell recordings. b, Injured conditions (reduced membrane area, depolarized
half-activation voltage and faster activation time constant of the h conductance): Excitability is
increased: action potential discharge occurs upon depolarization and the hyperpolarization-evoked
sag is enlarged, consistent with the experimental observation. (a, b) Injected currents from -1.65 to
0.55 mA.cm-2. Note that the increase in hyperpolarization induced depolarizing sag (here and in
figure 1a,b) results from a combined change in input resistance from cell shrinkage and from altered
Ih biophysical properties and not from altered Ih expression.
Oscillation duration changes
−55
<−500
>2000
0
−75
Injured (area) − Control
20 40 60 80 100
gGABA (mS.cm−2)
Oscillation duration changes
−55
−85
−65
−85
0
0
−1000
20 40 60 80 100
gGABA (mS.cm−2)
400
d
<−2000
(ms)
-1.0
-0.5
0
Iinj (µA.cm−2)
0.5
Oscillation duration
2000
>2000
−75
T/gL − Control
800
(ms)
1000
T/gh − Control
1200
area
Ih+area
0
-1.5
<−2000
−65
−75
Control
1600 Ih
duration (ms)
Vm (mV)
Vm (mV)
Injured (Ih) − Control
−65
−85
0
Vm (mV)
0
−75
Oscillation duration
2000
−65
−85
c
b
>500
Control
1600 Injured
duration (ms)
Vm (mV)
a
T/gh
1200 T/gL
T/gL+gh
800
400
0
-1.5
-1.0
-0.5
0
Iinj (µA.cm−2)
0.5
Supplemental Figure 5. Model result: mapping thalamic network response after injury shows that
membrane area and leak conductance play predominant roles in determining the oscillation duration.
a, Changes in oscillation duration for transient oscillations following modifications in Ih activation (top) and in
area (bottom), as a function of gGABA and the membrane potential (Vm). In both cases, the duration is globally
increased in the physiological range of membrane potentials ([-75; -65]) mV, dashed lines). b, Oscillation
duration profiles as a function of the input current in control and different injured conditions (gGABA=50 mS.cm2
). Changes in Ih activation properties (depolarized half-activation voltage and faster activation time constant)
strongly decrease the threshold for transient oscillation initiation; by contrast, the decrease in membrane area
induces a smaller threshold shift but powerfully increases oscillation duration. Note also that the threshold is
shifted in a supra-linear manner in the presence of both Ih and area changes. c, Changes in oscillation duration
in single therapeutic conditions, compared to the control condition. A decrease in the h conductance (gh) is
unable to restore the duration of oscillations to control values (top). By contrast, control durations are fully
restored by an increase in the leak conductance (gL) (bottom). d, Oscillation duration profiles as a function of
the input current in the control, injured and the different therapeutic conditions. The profiles illustrate how (i) a
therapeutic decrease of the gh restores the threshold but leaves enhanced oscillations duration, (ii) by contrast, a
therapeutic increase in gL does not restore the threshold but strongly lowers oscillation duration, and (iii) the
combined therapeutic modification of the gh and gL restores both threshold and duration of oscillations.
“Threshold” (q) is input current Iinj threshold for initiation of oscillations.
a
b
EEG 1
3 1
4
2
2
3
0.14
mV
4
06/18/2010 10:44:52
*Seizure End
EMG
Interictal
Ictal
20 s
Power spectrum
(x10 )
5
Ictal
4
Interictal
3
0.2
2
mV
1
0
0 5 10 15 20 25 30
Frequency (Hz)
2
-4
EEG1
1
2
2
3
3
4
4
EMG
d
c mV
0.5 s
2 mm
Bregma -2.5 mm
Supplemental Figure 6. Simultaneous EEG and EMG recordings 6 weeks following a cortical stroke.
a–b, Epileptiform activities in the EEG are associated with a behavioral arrest and a cessation of EMG activity.
The box indicates a seizure (a). The inset indicates the location of EEG electrodes contra- (3,4) and ipsilateral
(1,2) to the stroke (arrow) determined post-mortem from the same rat (scale, 2 mm). b, Expanded traces from
ictal and interictal recordings depicted in a. c, Power spectrum of ictal and interictal EEG activities from peristroke EEG recording #1. Dots indicate the typical dominant peak frequencies (~4-5 Hz and ~8 Hz; see also
Fig. 4c). Note that the peak frequency (4-5 Hz) is lower than typical absence seizures in rats. d, A Nissl-labeled
coronal section taken through the lesion from a rat from sacrificed 6 months after stroke and from which
chronic EEG and EMG recordings were obtained. The stroke appears as a scarred area of cortex (dashed line:
necrotic core). The stroke core was usually dislodged during tissue sectioning. Note that the lesion extends to
the subcortical white matter without damaging the hippocampus.
b
2
c
Medial
0
22
T4
T3
T2
T1
T4
T3
T2
T1
RT
VPM
VPL
0
Frequency (Hz)
10
2
10 Thalamus T4
10
0
11
1 mm
1
1 mm
0
10
2
10 Contra cx
10
0.18
1
RMS power (mV)
10
Thalamus
Dorsal
16
10 Ipsi cx
0.12
T1
ns
T2
T3
Cortex
T4
ns
ns
0.06
0.00
ns
RMS power (mV)
a
Ipsi. Contra.
p=0.05
0.12
ns
0.06
0.00
Pre-light
Light (low power)
d
1
Ipsi cx
0
0
12
10
2
10 Thalamus T3
Contra cx
10
1
0
0 T4
20
10
2
10 Thalamus T2
10
1
T3
0
10
2
10 Thalamus T1
0
22
T2
10
1
0
10
-5
Light
0
Time (s)
0
5
0.8
mV
T1
2s
Supplemental Figure 7. Low power (3-5 mW) 594 nm light is not sufficient to interrupt epileptic seizures
in freely behaving animals: compare to Fig. 5 c-f.
a, Averaged wavelet spectrograms from 7 seizures from one rat of cortical (ipsi- and contra-lateral to stroke)
and thalamic recordings from channels T1-4 ipsilateral to cortical stroke. The depicted cortical and thalamic
spectrograms are aligned in time and were obtained from simultaneously recorded seizures. 0s corresponds to
onset of 3-5 mW 594 nm light delivery to thalamus. Note that the low power light has a small, though not
significant effect, on T4 and T3 electrodes (located within<0.5 mm from optical fiber; see b) but does not
modulate the deep thalamic channels (T2 and T1; ~ 1 mm from the optical fiber; see b). b, Left: Tip of CMO
implant for awake behaving optical stimulation and recordings in the thalamus. Red arrowheads indicate
thalamic recording sites (T1-4); black arrow indicates tip of optical fiber. Right: Schematic diagram of the
somatosensory thalamus showing location of the CMO. c, Power quantification of cortical EEGs (ipsi and
contralateral to stroke) and thalamic LFPs ipsilateral to stroke before and during 594 nm 3-5 mW light delivery
in the right somatosensory thalamus, ipsilateral to the cortical stroke. Power was averaged 2s before and 2s
during light delivery. Bars, mean ± s.e.m.. ns, p>0.5; paired t-test or signed rank test as appropriate. d,
Representative example traces of simultaneous cortical EEG and thalamic LFP before and during 594 nm light
delivery (yellow box) in the thalamus. Arrow indicates the onset of the seizure which is not interrupted by 3-5
mW light delivery in thalamus. Note that In deep thalamic channels (T1-T2) the ictal activity is more robust
(i.e. characterized by larger LFP spikes (d) and stronger signal power (a)) than in more superficial thalamic
electrodes T3-T4. Note also that ictal activities start earlier in T1-T2 compared with T3-T4. These findings are
in agreement with the observation that the most hyperexcitable area is between VPL and VPM (also see Fig.1).
Results in a-d and Fig. 5c,d,e left, f were obtained from the same rat. RT, VPL and VPM correspond to
reticular thalamic, ventroposterolateral and ventroposteromedial thalamic nuclei, respectively. a,d are from the
same trial as Fig. 5c,d. These results suggest that low power light does not efficiently disrupt seizures because it
does not affect the particularly “active” thalamic channels (T1-T2; located far (~1 mm) from optical fiber)
which show the highest signal power in agreement with the presence of a more robust hyperexcitability in this
deep thalamic region close to VPL. In contrast, the higher light power (8-10 mW; see Fig. 5) interrupts seizures
presumably because it modulated all thalamic channels (T1-T4).
a
Ictal 1
2
10 Ipsi cx
10
1
10
0
Ictal 2
Ipsi cx
Interictal
Ipsi cx
bi
0.5
mV
EEG
0.005
Line-length 1 s
2
10 Contra cx
Contra cx
Contra cx
bii
101
Frequency (Hz)
10
Before light
0
2
10 Contra cx
Contra cx
Contra cx
During light
101
Light
100
102 Thalamus T4
10
Thalamus T4
Thalamus T4
c
EEGc
anterior
1
100
102 Thalamus T2
CMO
Thalamus T2
EEGi
anterior
Stroke
EEGc
posterior
EEGi
posterior
Thalamus T2
2 mm
101
100
0.5
mV
-5
Light
5 -5
0
Time (s)
Light
0
5 -5
Time (s)
Light
0
5
Time (s)
Supplemental Figure 8. Thalamic illumination disrupts seizures in a freely behaving rat.
a, Averaged wavelet spectrograms from the cortical EEGs ipsi- and contra-lateral to the stroke and from
thalamic LFPs ipsilateral to stroke during ictal and interictal periods. 594 nm light pulses were delivered to
thalamus at time 0. The depicted spectrograms are aligned in time vertically and were obtained from
simultaneously recorded cortical and thalamic channels. Shown are examples from stimulations (ictal 1: n=5;
ictal 2: n=1; interictal: n=11) from a 2.5 month old rat; 1.5 months post-stroke and post-viral delivery in
thalamus. Light disrupted seizure activities when presented either “late”, >5s after seizure onset (Ictal 1
spectrograms) or “early”, <1s after seizure onset (Ictal 2 spectrograms). Light had no effect on interictal EEG
activity. bi, Top: Ipsilateral cortical EEG recording. Bottom: the corresponding line-length. Upon crossing of
the line-length threshold (dashed line) the seizure onset (red box) is detected in real-time triggering a 594 nm
laser delivering light to thalamus which interrupts the seizure activity (see also Supplemental Fig. 1c). bii: 200
ms–long EEG recordings from bi are enlarged. c, Brain from the same rat sacrificed and fixed for histology 1
year post-stroke, from which chronic optrode recordings/optical stimulations were regularly obtained during a
period of 1 year. Location of CMO (see Supplemental Fig. 7b) and EEG electrodes is indicated (EEGi and
EEGc: ipsi- and contralateral EEGs, respectively). Note that cerebral cortex was not damaged by chronically
implanted device for ~1 year. a-c panels and Fig. 5e right are from the same rat.
e
RT
VPL
VPM
150
100
50
b
d
Recording
electrode
0
0
5
10
Light power (mW)
20 mm
-70 mV
Optical
fiber
ic
RT
f
120
50 pA
0
GFAP NpHR/eYFP Biocytin
240 n=9 cells
Peak I NpHR (pA)
ic
c
Peak I NpHR (pA)
a
20
mV
VPM
GFAP
-70 mV
NpHR/eYFP
Biocytin
0.5 s
Supplemental Figure 9. Functional properties of eNpHR in TC neurons in vitro.
a, Representative confocal image of a horizontal thalamic slice 3.5 months post-stroke and ~3 months after
eNpHR:Camk2a construct injection in vivo in VPL and VPM thalamic nuclei. The image was taken following
fixation after electrophysiological recordings of TC cells (arrows) from the same slice and after GFAP (blue),
eNpHR/EYFP (green) and biocytin (red) labeling. b, Low-power videomicroscopic image of the slice showing
locations of patch-clamp electrode and optical fiber through which the 594nm light was delivered to activate
eNpHR. c, eNpHR photocurrent (INpHR) activation curve from a representative TC neuron was best fitted with a
monoexponential function (grey line). Inset: the corresponding averaged outward INpHR traces induced by 1slong 594 nm light (yellow bar). Each trace corresponds to an average of 5 individual traces. d, Yellow light
inhibited action potential firing induced by a +120 (top) and a +160pA (bottom) current injection in a eNpHRexpressing TC neuron. c,e: Data correspond to mean ± s.e.m. (c) and (d) are from the same VPM TC neuron
indicated by the right arrow in (a). e, Quantification of the peak INpHR from 9 TC neurons from 4 rats. f, Highmagnification confocal image of a representative TC neuron filled with biocytin during whole-cell recording.
Overlap of eNpHR/eYFP (green) and biocytin (red) gives a yellow aspect to the cell. Inset: yellow light
inhibited the firing induced by a positive current injection in this TC neuron. ic, internal capsule; RT, reticular
thalamic nucleus; VPL and VPM, ventroposterolateral and ventroposteromedial relay thalamic nuclei.
Supplemental Table 1. Comparison of electrical membrane properties of injured and
control TC neurons.
AP amplitude
(mV)
AP duration
(ms)
AP threshold
(mV)
Rheobase
(pA)
# cells
# rats
Control
70.5 ± 1.5
2.6 ± 0.1
-52.1 ± 0.7
109 ± 16
19
6
Injured
67.7 ± 2.2
2.3 ± 0.1
-52.1 ± 0.9
56 ± 9
16
5
ANOVA
ns
ns
ns
p < 0.01
Action potential (AP) properties were similar in control and injured TC neurons. Rheobase, i.e. the minimal
current that needs to be injected in the cell to produce an action potential firing, was lower in injured TC
cells in agreement with an increased Rin (see Fig. 1). Maximal number of APs crowning the post-inhibitory
rebound low threshold spike (LTS) was similar in both groups (not shown), suggesting no robust increase
in T-channel expression in TC neurons. These results were quantified 7-14 days post-stroke. All values are
expressed as means ± s.e.m. ns, not significant (p>0.09).