Dopamine D2 receptors modulate intrinsic properties and synaptic transmission of parvalbumin interneurons in the mouse primary motor cortex

Dopamine (DA) plays a crucial role in the control of motor and higher cognitive functions such as learning, working memory and decision making. The primary motor cortex (M1), which is essential for motor control and the acquisition of motor skills, receives dopaminergic inputs in its superficial and deep layers from the midbrain. However, the precise action of DA and DA receptor subtypes on the cortical microcircuits of M1 remains poorly understood. The aim of this work was to investigate how DA, through the activation of D2 receptors (D2R), modulates the cellular and synaptic activity of M1 parvalbumin-expressing interneurons (PVINs) which are crucial to regulate the spike output of pyramidal neurons (PNs). By combining immunofluorescence, ex vivo electrophysiology, pharmacology and optogenetics approaches, we show that D2R activation increases neuronal excitability of PVINs and GABAergic synaptic transmission between PVINs and PNs in layer V of M1. Our data reveal a mechanism through which cortical DA modulates M1 microcircuitry and might participate in the acquisition of motor skills. Significance Statement Primary motor cortex (M1), which is a region essential for motor control and the acquisition of motor skills, receives dopaminergic inputs from the midbrain. However, precise action of dopamine and its receptor subtypes on specific cell types in M1 remained poorly understood. Here, we demonstrate in M1 that dopamine D2 receptors (D2R) are present in parvalbumin interneurons (PVINs) and their activation increases the excitability of the PVINs, which are crucial to regulate the spike output of pyramidal neurons (PNs). Moreover the activation of the D2R facilitates the GABAergic synaptic transmission of those PVINs on layer V PNs. These results highlight how cortical dopamine modulates the functioning of M1 microcircuit which activity is disturbed in hypo- and hyperdopaminergic states.

Sunnyvale, CA, USA) at 34°C. Data were acquired at 20 kHz and low-pass filtered at 219 4 kHz. Whole-cell patch clamp recordings with CsCl-or K-Glu-filled electrodes were 220 corrected for a junction potential of 4 mV and 13 mV, respectively. In voltage clamp 221 experiments, series resistance was continuously monitored by a step of -5 mV. Data 222 were discarded when the series resistance increased by >20%. sIPSCs and mIPSCs 223 were recorded at a holding potential of -64 mV. 224 To evaluate their intrinsic excitability, neurons were injected with increasing 225 depolarizing current pulses (50 pA steps, ranging from 0 to +550 pA, 1000 ms 226 duration). Action potential firing frequency was calculated for each current pulse. To 227 measure the input resistance, a hyperpolarizing -100 pA pulse current of 1 s was 228 applied and the voltage response was measured at steady state. Input-output curves 229

Electrophysiological characterization of M1 D2R-expressing cells 295
To determine the intrinsic properties of layer V M1 D2R neurons, whole-cell patch-296 clamp recordings were performed using ex vivo slices from Drd2-Cre:Ai9T ( Figure 2). 297 We patched neurons in acute brain slices and among the 19 neurons we recorded, 298 found 3 types of D2R-positive neurons differing in their electrophysiological 299 properties and the shape of their soma ( Figure 2). 55% were fast spiking 300 interneurons (FS), 30% were regular spiking non-pyramidal (RSNP) and 15% were 301 PNs. The FS neurons had a mean resting potential of -83.86 ± 2.02 mV (n=11) and 302 were able to fire fast action potentials at a high constant rate. Their action potentials 303 had a short duration and a large afterhyperpolarization (AHP) ( Figure 2B, inset) 304 which are general characteristics of FS neurons. The discharge frequency increased 305 as a function of the stimulation intensity and the maximal frequency, measured for 306 high intensities of depolarizing currents ranging from 100 to 230 Hz ( Figure 2D). 307 Their rheobase differed from one neuron to another and were on average 154.5 ± 308 17.13 pA. In addition, FS cells had a small input resistance (between 80 to 200 MΩ, 309 except for a neuron). We performed immunohistochemistry to detect the expression 310 of PV in 7 neurons filled with biocytin during whole-cell recording. 6 of 7 were PV-311 immunoreactive (not shown). 312 The second cell type did not maintain high frequency repetitive discharges and was 313 classified as RSNP because of the shape of the soma ( Figure 2B, central panel). 314 Action potentials evoked by current injection in RSNP cells had a longer duration and 315 a relatively smaller AHP than those recorded in FS cells. All RSNP cells displayed a 316 resting potential close to -81.23 ± 0.93 mV (n = 6). They had a low maximal 317 frequency of discharge associated with a low rheobase. At a low discharge 318 frequency, RSNP cells emitted action potentials with moderate or no accommodation. 319 Finally, a few PNs were identified by the triangular shape of their soma. They 320 exhibited a sustained action potential discharge in response to depolarizing current 321 pulses with a low maximal frequency of discharge ( Figure 2C). PNs had a mean 322 resting potential of -71.67 ± 6.45 mV, a mean input resistance of 309.5 ± 89.97 323 MΩ and a mean rheobase of 75.00 ± 43.30 pA (n = 4). 324  K-S test revealed that D2R activation increased mIPSC amplitude with no effect on 379 their frequency. The decay time of IPSC was significantly increased from 7.66 ± 0.65 380 to 8.88 ± 0.70 ms (p = 0.0273, n = 10; WSR test). The frequency was unchanged on 381 average, but it is important to note that quinpirole had a variable effect on individual 382 neurons. 383

D2R activation enhances GABAergic transmission at PVIN-PN synapses 385
Our results on GABAergic IPSCs suggested that D2R activation by quinpirole 386 induced more activity in the inhibitory network. However, the increase observed may 387 be due to any type of inhibitory IN. To determine whether quinpirole changes synaptic 388 transmission between PVINs and PNs, we used optogenetics to selectively study 389 PVIN-PN synapse properties ( Figure 5). We expressed the channelrhodopsin ChR2 390 in PVINs via local viral transfection in M1 of PV-Cre:Ai9T mice using an AAV2.5-391 EF1a-DIO-hChR2(H134R)-EYFP vector ( Figure 5A). We used 473 nm light flashes to 392 stimulate PVINs while recording from PNs. We first confirmed that 1 ms flashes of 393 light were able to reliably trigger action potentials in PVINs. As illustrated by the 394 raster plot in Figure 5B, each flash in the train evoked one or two action potentials in 395 the transfected PVIN. In a second step, we recorded the optically-evoked IPSCs from 396 PNs ( Figure 5C). PNs were identified as described previously (Figure 4) and 397 displayed a PN-typical firing pattern upon depolarizing current steps ( Figure 5C). 398

Light flashes reliably elicited evoked inhibitory post-synaptic currents (eIPSCs) in 399
PNs, which were potentiated by bath application of 2 µM quinpirole ( Figure 5D), 400 increasing their mean amplitude from 280.3 ± 68.52 pA to 321.6 ± 75.67 pA (p = 401 0.0371, n = 10; WSR test). This result strongly suggested that GABAergic synaptic 402 transmission between PVINs and PNs was enhanced by quinpirole. We further 403 characterized the short-term plasticity of the PVINs-PNs synapses using 10 flashes 404 of 1 ms at 10 Hz; Figure 5E). The inhibitory inputs to PNs showed pronounced 405 synaptic short-term depression, but bath-applied quinpirole did not change the profile 406 of synaptic transmission, which remained depressed ( Figure 5F). 407 release (Lupica, 1995). Since the excitability of PVINs was increased by quinpirole, 477

Discussion
we expected an increase in IPSC frequency, but this was not the case. Here, 478 quinpirole increased the amplitude of both sIPSCs and mIPSCs recorded in PNs with 479 no effect on their frequency. One possible explanation is that since the PVINs are not 480 spontaneously firing at their resting potential, the 5 mV depolarization generated by 481 quinpirole may not be large enough to raise the resting membrane potential to the 482 spike threshold in the absence of excitatory transmission. Another possible 483 explanation is the various origins of the GABAergic IPSCs. We studied all the 484