amplitude modulated photostimulation for probing neuronal...

1
Fig. 1 Incoming external (e.g. thalamic) EPSCs sum to a ~Gaussian conductance at input PY neurons (1,2) . We bypass the first step by introducing a randomly varying, non-specific cation conductance is induced into ChR2-expressing PY cells using amplitude modulated blue light (3). Then the network response can be measured with an MEA. PY PY PY PY PY PY References/acknowledgements Conclusions Introduction Experimental methods Amplitude modulated photostimulation for probing neuronal network dynamics J.P. Newman 1, *, T. Tchumatchenko 2, *, M.-F. Fong 1,4 , S.M. Potter 1 1 Laboratory for Neuroengineering, Dept. of Biomedical Engineering, Georgia Institute of Technology, Atlanta , GA, 30332 2 Center for Theoretical Neuroscience, Columbia University, *equal contributions 4 Department of Physiology, Emory University School of Medicine, Atlanta, GA, 30322 Light → photocurrent in uncoupled neurons Light → population activity in the temporally irregular, low-rate regime References: [1] Dayan, P. and Abbott, L.F. Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems. Cambridge: MIT Press; 2001; [2] Tchumatchenko T, Malyshev A., Wolf F, Volgushev M (2011) Ultrafast Population Encoding by Cortical Neurons. J. Neurosci, 31:12171 12179. [3] S.M Potter and T.DeMarse (2001) A new approach to neural cell culture for long- term studies. J. Neurosci. Methods 110. [4] Nikolic K, et al. (2009) Photocycles of channelrhodopsin-2. Photochemistry. and Photobiology 85: 400411. [5] J. Mattis, et. al. (2011) Principles for applying optogenetic tools derived from direct comparative analysis of microbial opsins. Nat. Methods 9:2. Acknowledgements: This work was supported by NSF COPN grant 1238097 and NIH grant 1R01NS079757 from NINDS, NSF GRFP Fellowship 08-593 to J.P.N., NSF GRFP Fellowship 09-603 to M.F., NSF IGERT Fellowship DGE-0333411 to J.P.N and M.F. , and Computational Sciences Fellowship from the Volkswagen Foundation to T.T. The viral vectors used in this study were provided by Karl Deisseroth via the UNC atChapel Hill Gene Therapy Center. We thank J.T. Shoemaker for performing tissue harvests. Finally, we gratefully acknowledge all those who have contributed to NeuroRighter's hardware forums and supplied bug reports to the NeuroRighter code repository. Sensory stimuli arrive in the cortex as fluctuating currents to populations of cortical neurons. How cortical networks encode incoming synaptic currents ultimately determines an animal’s cognitive and behavioral response. A number of theoretical studies have addressed this transduction process using model cortical networks [1]. However, while dynamical rate responses have been studied experimentally in single neurons [2], few experimental systems offer the access/manipulability required to examine network responses. Optogenetic techniques in dissociated cortical networks (DCNs) that are grown on microelectrode arrays (MEAs) offer the opportunity to test key aspects of network response dynamics [3]. In this study, we use amplitude modulated photostimulation to impose specific, irregular spiking patterns on DCNs. By using continuous time inputs to control aspects of the network response, we create viable biological analog to simulated random cortical networks. Neurons expressing ChR2(H134R) support non-specific cation currents in response to blue light [4]. We use ChR2 to introduce amplitude modulated, temporally random currents into CaMKllα- expressing (pyramidal) cells embedded within a randomly connected cortical network grown on an MEA. This provides: Simultaneous manipulation of many neurons Non-invasive, long-term recording Easy manipulation of connectivity using drugs Subthreshold control of neurons using light Temporally irregular, low-rate network firing is achieved using an Ornstein-Uhlenbeck (OU) stimulation process. data model Fitted values: quantum efficiency · photon flux Rate of channel closure Rate of channel recovery Fraction of channels in the open state (O), in the desensitized state (D) and closed state (C=1-O-D) is governed by the following 3-state model: New findings: A continuum of ChR2 (H134R) channels exhibit a resonance located at ~10Hz. A continuum of ChR2 (H134R) channels reliably transmits input frequencies up to ~100 Hz Linear response function in the 3-state-model: Linear non-linear (LN) cascade model Linear filter Static non-linearity Predict the population spiking activity. New finding: Continuous low-intensity stimulation selectively targeting PY neurons entrains reliable network activity patterns Light intensity [mW/mm 2 ] Open probability Of ChR2 Photocurrent [pA] Time [s] These results indicate that continuous optogenetic stimulation with LEDs is an effective tool for providing correlated input currents to thousands of pyramidal cells embedded in cortical networks while simultaneously recording spiking activity in hundreds of individual neurons. Because continuous photostimulation allows for complex current waveforms to be injected into large neural populations, this technique opens a new level of control for testing theories of cortical response properties and small signal representation under different network parameter regimes. Trial # (1-50) 1 2 3 Fig. 2 Amplitude modulated ChR2-mediated photocurrents in uncoupled PY cells. (top) recorded light intensity, (middle) result of 3-state model, (bottom) somatic photocurrent. Fig. 3 Population spiking response to OU input signal. The top trace shows the amplitude modulated 457 nm input signal, which was repeated for 50 trials. Units are sorted by decreasing firing rate and the top twenty units from a single culture (Xeko) are shown. Different neurons show variation in mean firing rate and the temporal precision of their response to input signals. Fig. 4 Firing statistics of three cultures presented with repeated, 30 second OU stimuli (same statistics, but different 30 second realization for each culture). (left) Neuronal firing rate distribution. (right) Neuronal ISI distribution. Fig. 5 Population firing rate (grey) averaged across 50 repeats of a 30 second OU realization and the resulting LN model prediction ( black). Firing rate averages and estimates were made with a 2.5 msec bin size. Fig. 7 The open-source NeuroRighter electrophysiology platform uses inexpensive NI A/D, D/A , and DIO boards for recording and signal generation. It provides online spike sorting via a fully automated Gaussian mixture modeling procedure. It was used for all MEA recordings and to drive the stimulation system. Fig. 8 Optical feedback LED driver circuit. (A) Photodiode measures light scattered from the Köhler apparatus. A high bandwidth transimpedance amplifier converts the resulting photocurrent to a voltage (VPD) which is compared to a DAC supplied reference voltage (VREF). These two voltages enter the non- inverting and inverting pins of an op-amp whose output drives the gate of a power-FET which in turn pulls current through the LED. (B) A 1 millisecond command pulse and resulting VPD signal (proportional to light power). Fig. 6 Recorded linear response function and the model fit. The recorded linear response was estimated using repeated logarithmic chirp inputs from 0.1 to 1000 Hz. The fitted parameters are in agreement with previously reported values [4]. A B Fig. 9 Recording and stimulation system. (A) NeuroRighter software; (B) ‘Plug-in' library; (C) 4-channel LED current source; (D) 4:1 LED-fiber coupler; (E) Köhler illumination train; (F) 60- channel MEA amplifier (MEA1060, MCS) and Peltier temperature control system; (G) MEA showing culturing chamber; (H) ChR2-mCherry- expressing PY cell. Culturing methods E18 cortical tissue Laminin-coated ITO MEAs (MCS) Cultures infected with 10 9 cfu ml -1 AAV2- CaMKllα::ChR2(H134)- mCherry 1-5 DIV [5]. Cultures stored in gas- permeable FEP chambers [3]. Grown in incubator regulated to 35° C, 5% C0 2 , 65% Humidity. All experiments conducted within incubtator.

Upload: others

Post on 13-Jun-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Amplitude modulated photostimulation for probing neuronal ...jpnewman/doc/cns2012_JPN_print.pdfcognitive and behavioral response. A number of theoretical studies have addressed this

Fig. 1 Incoming external (e.g. thalamic) EPSCs sum to a

~Gaussian conductance at input PY neurons (1,2) . We bypass

the first step by introducing a randomly varying, non-specific

cation conductance is induced into ChR2-expressing PY cells

using amplitude modulated blue light (3). Then the network

response can be measured with an MEA.

PY

PY

PY

PY

PY

PY

References/acknowledgements

Conclusions

Introduction Experimental methods

Amplitude modulated photostimulation for probing neuronal network dynamicsJ.P. Newman1,*, T. Tchumatchenko2,*, M.-F. Fong1,4, S.M. Potter1

1 Laboratory for Neuroengineering, Dept. of Biomedical Engineering, Georgia Institute of Technology, Atlanta , GA, 303322 Center for Theoretical Neuroscience, Columbia University, *equal contributions

4Department of Physiology, Emory University School of Medicine, Atlanta, GA, 30322

Light → photocurrent in uncoupled neurons

Light → population activity in the temporally irregular, low-rate regime

References: [1] Dayan, P. and Abbott, L.F. Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems.Cambridge: MIT Press; 2001; [2] Tchumatchenko T, Malyshev A., Wolf F, Volgushev M (2011) Ultrafast Population Encoding byCortical Neurons. J. Neurosci, 31:12171 12179. [3] S.M Potter and T.DeMarse (2001) A new approach to neural cell culture for long-term studies. J. Neurosci. Methods 110. [4] Nikolic K, et al. (2009) Photocycles of channelrhodopsin-2. Photochemistry. andPhotobiology 85: 400411. [5] J. Mattis, et. al. (2011) Principles for applying optogenetic tools derived from direct comparativeanalysis of microbial opsins. Nat. Methods 9:2.Acknowledgements: This work was supported by NSF COPN grant 1238097 and NIH grant 1R01NS079757 from NINDS, NSF GRFPFellowship 08-593 to J.P.N., NSF GRFP Fellowship 09-603 to M.F., NSF IGERT Fellowship DGE-0333411 to J.P.N and M.F. , andComputational Sciences Fellowship from the Volkswagen Foundation to T.T. The viral vectors used in this study were provided byKarl Deisseroth via the UNC at Chapel Hill Gene Therapy Center. We thank J.T. Shoemaker for performing tissue harvests. Finally, wegratefully acknowledge all those who have contributed to NeuroRighter's hardware forums and supplied bug reports to theNeuroRighter code repository.

Sensory stimuli arrive in the cortex as fluctuating currents topopulations of cortical neurons. How cortical networks encodeincoming synaptic currents ultimately determines an animal’scognitive and behavioral response. A number of theoretical studieshave addressed this transduction process using model corticalnetworks [1]. However, while dynamical rate responses have beenstudied experimentally in single neurons [2], few experimentalsystems offer the access/manipulability required to examinenetwork responses. Optogenetic techniques in dissociated corticalnetworks (DCNs) that are grown on microelectrode arrays (MEAs)offer the opportunity to test key aspects of network responsedynamics [3]. In this study, we use amplitude modulatedphotostimulation to impose specific, irregular spiking patterns onDCNs. By using continuous time inputs to control aspects of thenetwork response, we create viable biological analog to simulatedrandom cortical networks.

Neurons expressing ChR2(H134R) support non-specific cationcurrents in response to blue light [4]. We use ChR2 to introduceamplitude modulated, temporally random currents into CaMKllα-expressing (pyramidal) cells embedded within a randomlyconnected cortical network grown on an MEA. This provides:• Simultaneous manipulation of many neurons• Non-invasive, long-term recording• Easy manipulation of connectivity using drugs

Subthreshold control of neurons using light

Temporally irregular, low-rate network firing is achieved using an Ornstein-Uhlenbeck (OU) stimulation process.

data

model

Fitted values:

quantum efficiency · photon fluxRate of channel closureRate of channel recovery

• Fraction of channels in the open state (O), in the desensitized state (D) and closed state (C=1-O-D) is governed by the following 3-state model: New findings:

• A continuum of ChR2 (H134R) channels exhibit a resonance located at ~10Hz.

• A continuum of ChR2 (H134R) channels reliably transmits input frequencies up to ~100 Hz

Linear response functionin the 3-state-model:

Linear non-linear (LN) cascade model• Linear filter• Static non-linearity • Predict the population spiking activity.

New finding: • Continuous low-intensity stimulation selectively

targeting PY neurons entrains reliable network activity patterns

Ligh

t in

ten

sity

[mW

/mm

2]

Op

enp

rob

abili

tyO

f C

hR

2P

ho

tocu

rren

t[p

A]

Time [s]

These results indicate that continuous optogenetic stimulation withLEDs is an effective tool for providing correlated input currents tothousands of pyramidal cells embedded in cortical networks whilesimultaneously recording spiking activity in hundreds of individualneurons. Because continuous photostimulation allows for complexcurrent waveforms to be injected into large neural populations, thistechnique opens a new level of control for testing theories of corticalresponse properties and small signal representation under differentnetwork parameter regimes.

Tria

l # (

1-5

0)

1 2 3

Fig. 2 Amplitude modulated ChR2-mediated photocurrents in uncoupled PY cells. (top) recorded light

intensity, (middle) result of 3-state model, (bottom) somatic photocurrent.

Fig. 3 Population

spiking response to

OU input signal. The

top trace shows the

amplitude modulated

457 nm input signal,

which was repeated

for 50 trials. Units are

sorted by decreasing

firing rate and the top

twenty units from a

single culture (Xeko)

are shown. Different

neurons show

variation in mean

firing rate and the

temporal precision of

their response to input

signals.

Fig. 4 Firing statistics of three cultures presented with repeated, 30 second OU

stimuli (same statistics, but different 30 second realization for each culture).

(left) Neuronal firing rate distribution. (right) Neuronal ISI distribution.

Fig. 5 Population firing rate (grey) averaged across 50 repeats of a 30 second

OU realization and the resulting LN model prediction (black). Firing rate

averages and estimates were made with a 2.5 msec bin size.

Fig. 7 The open-source NeuroRighter electrophysiology

platform uses inexpensive NI A/D, D/A , and DIO boards for

recording and signal generation. It provides online spike

sorting via a fully automated Gaussian mixture modeling

procedure. It was used for all MEA recordings and to drive the

stimulation system.

Fig. 8 Optical feedback LED driver circuit. (A)

Photodiode measures light scattered from the

Köhler apparatus. A high bandwidth

transimpedance amplifier converts the resulting

photocurrent to a voltage (VPD) which is

compared to a DAC supplied reference voltage

(VREF). These two voltages enter the non-

inverting and inverting pins of an op-amp

whose output drives the gate of a power-FET

which in turn pulls current through the LED.

(B) A 1 millisecond command pulse and

resulting VPD signal (proportional to light

power).

Fig. 6 Recorded linear response function and the model fit. The

recorded linear response was estimated using repeated logarithmic

chirp inputs from 0.1 to 1000 Hz. The fitted parameters are in

agreement with previously reported values [4].

A

B

Fig. 9 Recording and

stimulation system. (A)

NeuroRighter software; (B)

‘Plug-in' library; (C) 4-channel

LED current source; (D) 4:1

LED-fiber coupler; (E) Köhler

illumination train; (F) 60-

channel MEA amplifier

(MEA1060, MCS) and Peltier

temperature control system; (G)

MEA showing culturing

chamber; (H) ChR2-mCherry-

expressing PY cell.

Culturing methods• E18 cortical tissue

• Laminin-coated ITO MEAs (MCS)

• Cultures infected with 109 cfu ml-1 AAV2-CaMKllα::ChR2(H134)-mCherry 1-5 DIV [5].

• Cultures stored in gas-permeable FEP chambers [3].

• Grown in incubator regulated to 35° C, 5% C02, 65% Humidity.

• All experiments conducted within incubtator.