To selectively activate striatal cholinergic interneurons (CINs), we utilized knock-in mice that express Cre-recombinase downstream of the native choline-acetyltransferase promoter and an internal ribosomal entry site (IRES) 
. ChAT-IRES-Cre mice express Cre-recombinase in cholinergic neurons of the striatum, basal and septal nuclei, and neocortex (Rossi et al., 2011). Intracranial injection of adeno-associated virus encoding a Cre-dependent (double floxed inverted) Channelrhodopsin2-mCherry fusion protein (DFI-ChR2-mCherry) 
into the striatum of these mice enabled the CIN-specific expression of the light-activated excitatory ChR2, which was clearly observed 7–14 days after injection (, see methods
ChR2-mediated activation of striatal cholinergic interneurons.
We characterized the ability of ChR2 to drive CIN activity in an acute brain slice using cell-attached recordings obtained from ChR2-positive CINs that were visually identified under epifluorescence (). Recordings were made at room temperature to reduce spontaneous firing of CINs 
. Brief pulses (2–5 ms) of whole-field blue light delivered through the microscope objective reliably evoked action potentials (APs) in CINs with a mean latency measured from the start of the light pulse of 4.2±0.5 ms (n
8), demonstrating that virus mediated Cre-conditional expression of ChR2 provides a viable tool for driving activity in this population of genetically-defined local interneurons.
Previous studies of CIN function in the striatum have focused on the actions of nicotinic and muscarinic ACh receptors, which modulate glutamate, GABA, and dopamine release at striatal synapses and directly depolarize the principal striatal cells, medium spiny neurons [MSNs, 20,21,22]. To determine whether activation of CINs could produce a direct postsynaptic response, we made current-clamp recordings from MSNs in brain slices containing CINs expressing ChR2 (). Pulses of blue light applied through the microscope objective to the area surrounding the recorded MSN evoked a transient depolarization from a resting potential of −80 mV (). On average (n
10), the evoked response had an amplitude of 1.9±0.5 mV and occurred with a latency of 7.0±0.5 ms from the light pulse onset.
Light-evoked CIN action potentials evoke glutamatergic responses in MSNs.
The light-evoked response was resistant to blockade of GABAA
receptors (50 µM picrotoxin) as well as nicotinic and muscarinic ACh receptors (1 µM mecamylamine and 10 µM scopolamine, respectively), indicating that it was not mediated by acetylcholine receptor-dependent direct depolarization or by indirect reversed inhibition via activation of nicotinic receptor-expressing GABAergic interneurons (). Instead, the depolarization was sensitive to application of the AMPA/kainate-type glutamate receptor (AMPAR) antagonist NBQX (10 µM), which significantly reduced the response magnitude to 22.4 ± 2.5% of control (n
5, Student's T-test p<0.001, ). The remaining response was eliminated by the NMDA-type glutamate receptor (NMDAR) antagonist CPP (10 µM, data not shown). Thus, our results indicate that firing of CINs generates glutamatergic excitatory postsynaptic potentials (EPSPs) in nearby MSNs.
As CIN-evoked EPSPs exhibited a NBQX-resistant component, we further examined the relative contribution of NMDA-type glutamate receptors (NMDARs) to CIN-evoked currents. Voltage-clamp recordings were obtained from MSNs at room temperature using a cesium-based internal solution, in the prensence of picrotoxin, mecamylamine, and scopolamine. Light-evoked excitatory postsynaptic currents (EPSCs) were measured at hyperpolarized and depolarized membrane potentials. At a holding potential of −70 mV, CIN stimulation evoked an average (n
7) inward peak current of 16.7±4.4 pA (), estimating the contribution of AMPARs. It is possible that these measurements are contaminated with a contribution of current flow through NMDARs, although this is likely due to the Mg block of NMDARs and the lack of a prolonged phase of the EPSC at this potential. In the same cells, we quantified the NMDAR contribution as the current evoked at a holding potential of +40 mV, measured 150 ms after the light pulse (35.2±15.5 pA). The average ratio (+40:−70) of these measures was 1.9 ± 0.4. ().
To directly compare optically-evoked glutamatergic inputs from CINs with similarly evoked inputs from cortical afferents, we expressed a Cre recombinase-independent version of the ChR2-mCherry fusion protein in motor cortex neurons projecting to the striatum 
. At 7–14 days following injection, axonal fibers descending from the overlying cortex could clearly be seen branching throughout the dorsal striatum (). As above, we made voltage-clamp recordings in MSNs and delivered light pulses through the microscope objective (), evoking EPSCs with an average (n
5) latency to 4.1±0.3 ms. From a holding potential of -70 mV, the average peak inward current amplitude was 50.5±17.3 pA, while from a holding potential of +40 mV, the average current at 150 ms was 28.2±6.4 pA (). Calculating the ratio of currents at the two holding potentials (+40:−70) yielded an average value of 0.86 ± 0.3, significantly smaller than for currents evoked by stimulation of CINs (p<0.05, Student's T-test, n
5, ). These experiments, performed at room temperature, indicate that ChR2-mediated activation of striatal CINs from a quiescent state is capable of evoking glutamatergic responses in striatal MSNs that is independent of cholinergic actions.
In order to confirm both the robustness of these findings and that they represent vesicular release of glutamate by CINs, several additional experiments were performed (). First, analysis of ChR2-mediated glutamatergic responses in MSNs was repeated at near physiological temperatures (32–34°C) at which CINs are spontaneously active (mean firing frequency in cell-attached recordings
1.6±0.3 Hz, n
6). At this temperature, brief blue light illumination also triggered EPSCs in MSNs at holding potentials of −70 mV (116.4±18.7 pA) and +40 mV (29.7±5.4 pA, ). Importantly, the smaller relative NMDAR contribution, in comparison to data in , was due to the faster kinetics of the evoked EPSC at near-physiological temperatures which results in substantial decay by 150 ms after stimulus.
CIN-mediated glutamatergic currents in MSNs require VGluT3 expression.
Second, in order to confirm the robustness of our findings in an independent mouse line, similar experiments were performed in the GENSAT GM60 mouse line in which Cre is expressed in cholinergic interneurons following integration of a bacterial artificial chromosome spanning the ChAT genetic locus and in which the ChAT coding sequence has been replaced with that of Cre 
. As above, blue light stimulation of acute slices prepared from DFI-ChR2-mCherry AAV-infected mice elicited EPSCs in MSNs at both −70 and +40 mV (). On average (n
7), these currents were 46.7±3.6 pA and 52.8±5.9 pA, for −70mV and +40mV, respectively, .
Third, although the lack of sensitivity of the observed responses in MSNs to cholinergic and GABAergic antagonists suggests direct release of glutamate from CINs, we examined the dependence of CIN-mediated glutamatergic EPSCs in MSNs on the expression of the vesicular glutamate transporter VGluT3. Importantly, in the striatum, this transporter is expressed exclusively in CINs. GM60 mice were bred with VGluT3−/−
to generate GM60;VGluT3−/−
animals. Blue-light stimulation of DFI-ChR2 AAV infected striatal slices of these mice failed to trigger EPSCs at holding potentials of −70 or +40 mV (n
6, ). Thus, glutamatergic EPSCs in MSNs triggered by optogenetic stimulation of ChR2-expressing CINs requires VGluT3, strongly suggesting that VGluT3 expression in CINs is required to package glutamate within vesicles that are subsequently released in an activity-dependent manner.