Methods Summary (300-word limit)
Cell culture
Primary cultures of neurons and glia were derived from E18 rat neocortex and cultured onto MEAs (Multichannel Systems) or glass bottom dishes (Mattek) as described previously {Hales2010}.  Experiments began after 8-12 days in vitro (DIV).  All cultures used in ChR2 experiments, including controls, were transfected with AAV9-hSynapsin-ChR2(H134R)-eYFP (from Dr. Karl Deisseroth via the University of Pennsylvania Vector Core) at 1 DIV.  Drug concentrations were used (in μM): TTX, 1; CNQX, 40; bicuculline, 20; cyclothiazide, 100; APV, 50.  DMSO or water was used as vehicle.

Electrophysiology
MEA recordings were performed in standard growth medium in the cell culture incubator as described previously {PotterDemarse2001, Wagenaar2006, Newman2013} using the Neurorighter acquisition system {Newman2012; Rolston2009}.  Whole-cell recordings were performed in artificial cerebrospinal fluid using an EPC8 amplifier (HEKA). mEPSCs were analyzed blind to treatment condition using MiniAnalysis (Synaptosoft).  Spiking and mEPSC data was analyzed using MATLAB (The Matworks).  Statistical significance for firing and burst rate data was determined using a Kruskal-Wallis test, followed by Wilcoxon rank sum tests.  Statistical significance for mEPSC data was determined using analysis of variance followed by t-tests.  All post-hoc tests were performed using Bonferroni adjustment for multiple comparisons.

Optical stimulation
To deliver optical stimuli, a custom N-channel enhancement mode MOSFET current source was used  to drive a blue LED (465.11 nm FWHM; LEDEngin). The LED was butt-coupled to a randomized fiber bundle (Schott) which fed light to a Köhler illumination train mounted beneath the MEA amplifier. The average network firing rate was calculated every dt=10 ms according to f[t]=r[t]+(1-α) f[t-dt], where α  is 5 sec/dt and and r[t] is the number of detected spikes/dt, f[t] is the instantaneous firing at time t. The target rate, f*, was defined as f[t] averaged over a 3 hour period prior to CNQX application. Five minutes following the application of CNQX to the culturing medium, an error signal was generated between the target and measured firing rate according to ef[t]=f*-f[t] and used to determine stimulus application according to if Σt ef[t] > 0, apply 10 ms pulse.  Each stimulus pulse resulted in uniformly distributed 10.1 mW/mm2 light the plane of the culture. The rise and fall times of each LED pulse were ~10 μs.  Full details are described previously {Newman2013}.



(all references appearing in Methods Summary are included in main text reference list)









Methods (appears in supplement)
Cell culture
Primary cultures of neurons and glia were derived from E18 rat neocortex and grown on polyethylenimine- and laminin-coated microelectrode arrays (Multichannel Systems 60MEA200/30ir-Ti-pr) or glass bottom dishes (P35G-1.5-10-C) as described in Hales et al., 2010.  Growth medium contained: 90% high-glucose DMEM, 10% horse serum, 0.5mM GlutaMAX, 1mM sodium pyruvate, 2.5ug/mL insulin (pH 7.2, 315 mOsm).  Incubator was regulated at at 35°C and 5% CO2.  Cultures were transfected with AAV9-hSynapsin-ChR2(H134R)-eYFP (from Dr. Karl Deisseroth via the University of Pennsylvania Vector Core) at 1 day in vitro.  The following drug concentrations were used: TTX, 1uM; CNQX, 40uM; bicuculline, 20uM; cyclothiazide, 100uM.  DMSO or water was used as vehicle, corresponding to the stable solvent used to dilute the drug that sister cultures were treated with.  All experiments were performed during the second week in vitro.

Microelectrode array (MEA) recordings
MEA recordings were performed in standard growth medium in the cell culture incubator.  Extracellular voltage waveforms were continuously sampled at 25kHz using the Neurorighter acquisition system (Newman et al., 2012, Rolston et al., 2009).  Voltage recordings were filtered with a 3rd order Butterworth bandpass at 200-3000Hz, and action potentials were detected at threshold of ±5 times the root mean square error.  Analysis of spike data was performed in MATLAB (The Mathworks).  The pre-drug period was defined as a 3-hour segment preceding TTX, CNQX, or vehicle application.  The treatment period was defined as the entire 24 hours during TTX, CNQX, or vehicle application.  After the treatment period, all drugs were washed 4 times with standard growth medium.

Whole-cell recordings
Miniature excitatory postsynaptic currents (mEPSCs) were recorded from pyramidal-shaped cells in a continuous perfusion of artificial cerebrospinal fluid containing (in mM): 126 NaCl, 3 KCl, 2 CaCl2, 1.5 MgSO4, 1 NaH2PO4, 25 NaHCO3, and 25 D-glucose, and saturated with 95% O2 and 5% CO2 (pH 7.4, 315 mOsm).  To isolated AMPAergic mEPSCs, solution was supplemented with containing 1µM TTX and 20µM bicuculline. Temperature was regulated at 35°C using an inline heater (Warner ???).  Internal solution contained (in mM): 100 K-gluconate, 30 KCl, 10 HEPES, 2 MgSO4, 0.5 EGTA, 3 ATP (pH 7.4, 290 mOsm).  mEPSCs were recorded using an EPC8 amplifier (HEKA).  mEPSCs were analyzed, blind to the treatment condition, using MiniAnalysis (Synaptosoft). Pipette resistances were 2-8Mohms. mEPSCs with amplitudes <5pA were excluded from analysis.

Optical stimulation

To deliver optical stimuli, a custom N-channel enhancement mode MOSFET current source (https://potterlab.gatech.edu/main/newman/wiki/index.php?title=Cyclops_Driver_R2) was used  to drive a blue LED (465.11 nm FWHM; LZ4-00B200, LEDEngin, San Jose, CA). The LED was butt-coupled to a randomized fiber bundle (Schott AG, Mainz, Germany) which fed light to a Köhler illumination train mounted beneath the MEA amplifier. A full description and characterization of the closed-loop optical stimulation system used in this study is given in (Newman et al., 2013). Briefly, the average network firing rate was calculated every dt = 10 ms according to

f[t] = r[t] + (1 - α) f[t -dt]

where  = 5 sec/dt and and r[t] = no. detected spikes/dtis the instantaneous firing at time t. The target rate, f*, was defined as f[t] averaged over a 3 hour period prior to CNQX application. Five minutes following the application of CNQX to the culturing medium, an error signal was generated between the target and measured firing rate according to

ef[t] = f* - f[t].


Finally, an on-off controller was used to determine stimulus application according to

 if Σt ef[t] > 0, apply 10 ms pulse.
(note: currently there’s no way to notate summations properly in Pages; I will correct this when we port the document to a more supportive text editor)

Each stimulus pulse resulted in uniformly distributed 10.1 mW/mm2 light the plane of the culture. The rise and fall times of each LED pulse were ~10 μs.