Synaptic Zinc Enhances Inhibition Mediated by Somatostatin, but not Parvalbumin, Cells in Mouse Auditory Cortex

Abstract

Cortical inhibition is essential for brain activity and behavior. Yet, the mechanisms that modulate cortical inhibition and their impact on sensory processing remain less understood. Synaptically released zinc, a neuromodulator released by cortical glutamatergic synaptic vesicles, has emerged as a powerful modulator of sensory processing and behavior. Despite the puzzling finding that the vesicular zinc transporter (ZnT3) mRNA is expressed in cortical inhibitory interneurons, the actions of synaptic zinc in cortical inhibitory neurotransmission remain unknown. Using in vitro electrophysiology and optogenetics in mouse brain slices containing the layer 2/3 (L2/3) of auditory cortex, we discovered that synaptic zinc increases the quantal size of inhibitory GABAergic neurotransmission mediated by somatostatin (SOM)- but not parvalbumin (PV)-expressing neurons. Using two-photon imaging in awake mice, we showed that synaptic zinc is required for the effects of SOM- but not PV-mediated inhibition on frequency tuning of principal neurons. Thus, cell-specific zinc modulation of cortical inhibition regulates frequency tuning.

Introduction

Cortical inhibition gates the flow of sensory information and flexibly shapes cortical network dynamics (Isaacson and Scanziani 2011; Hattori et al. 2017). Many different interneuron classes have been identified in mammalian cortical circuits (Jiang et al. 2015; Kubota et al. 2016; Tremblay et al. 2016). The two most prevalent classes are somatostatin (SOM) and parvalbumin (PV)-expressing neurons, which differ in their morphological, electrophysiological, biophysical, and connectivity properties (Rudy et al. 2011; Harris and Shepherd 2015; Tremblay et al. 2016; Hattori et al. 2017). SOMs synapse onto dendrites to inhibit principal neurons (PNs) and PVs (Fino and Yuste 2011; Cottam et al. 2013; Pfeffer et al. 2013; Xu et al. 2013; Karnani et al. 2016), whereas PVs synapse onto somata and proximal dendrites to inhibit PNs and PVs (Kubota et al. 2016; Yavorska and Wehr 2016). PVs and SOMs are under robust neuromodulatory control (Kawaguchi and Shindou 1998; Munoz and Rudy 2014; Tremblay et al. 2016; Hattori et al. 2017), but the precise cell-specific effects of neuromodulators onto these cell types and their resultant impact on sensory processing are not fully understood.

One newly discovered neuromodulator of auditory cortical circuits is synaptic zinc. Synaptic zinc is coreleased with glutamate to modulate excitatory neurotransmission via inhibition of NMDA receptors and inhibition or enhancement of AMPA receptor excitatory postsynaptic currents, depending on the presynaptic stimulation frequency (Pan et al. 2011; Vergnano et al. 2014; Anderson et al. 2015; Kalappa et al. 2015; Kalappa and Tzounopoulos 2017). In the auditory cortex (AC), synaptic zinc signaling shapes sound processing via cell-specific modulation of response gain and frequency tuning of PNs, SOMs and PVs (Anderson et al. 2017; Kumar et al. 2019). This modulation is associated with improved sound frequency discrimination (Kumar et al. 2019). In the somatosensory cortex, synaptic zinc signaling contributes to whisker-mediated fine texture discrimination (Patrick and Dyck 2018). Together, these findings support a general role of synaptic zinc in enhancing acuity for sensory input discrimination across different sensory modalities; yet, the synaptic mechanisms via which zinc signaling shapes cortical sensory processing remain unknown.

The vesicular zinc transporter (ZnT3) that loads zinc into the presynaptic vesicles is mostly expressed in glutamatergic terminals (McAllister and Dyck 2017); thus, synaptic zinc modulation of cortical neurotransmission has been studied mostly in excitatory neurotransmission (McAllister and Dyck 2017). However, recent data showed that the ZnT3 mRNA is highly and specifically expressed in a subclass of SOMs, the Martinotti cells (Paul et al. 2017). This finding suggests that synaptic zinc might be coreleased with GABA from SOMs to provide cell-specific modulation of cortical inhibitory neurotransmission and sensory processing. As such, investigation of the mechanisms via which synaptically released zinc modulates cortical inhibition and cortical sensory processing holds the potential for unmasking novel cortical synaptic mechanisms and linking cell-specific synaptic modulation with distinct changes in sensory processing.

To explore the synaptic mechanisms via which synaptic zinc modulates cortical inhibition in auditory cortex and their effects on cortical sound processing, we combined in vitro electrophysiological recordings from synaptically coupled cortical neurons and optogenetic approaches with in vivo two-photon imaging of specific neuronal types in the auditory cortex of awake mice. Our results reveal that synaptically released zinc is a powerful modulator of SOM- but not PV-mediated inhibition. This cell-specific synaptic modulation shapes the distinct contribution of different interneurons to sound processing.

Materials and Methods

Animals

All mice handling was approved by the Institutional Animal Care and Use Committee at the University of Pittsburgh. Postnatal (P) 29-35 male and female SOM-GFP (GIN) mice (Jackson, Bar Harbor, ME) were used for experiments shown in Figures 1 and 6 and Supplementary Figure S2. P29-35 male and female PV-GFP mice (Jackson, Bar Harbor, ME) were used for experiments shown in Supplementary Figure S3. P29-35 male and female SOM-Cre mice (Jackson, Bar Harbor, ME) were used for experiments shown in Figures 2 and 7 and Supplementary Figure S1a–e,k,l. P42-64 male and female SOM-Cre mice were used for experiments shown in Figures 8 and Supplementary Figures S4 and S5. P29-35 male and female SOM-Cre/PV-GFP mice were used for experiments shown in Figure 4 and Supplementary Figure S1f–j. P29-35 male and female PV-Cre mice (Jackson, Bar Harbor, ME) were used for experiments shown in Figure 5. P42-64 male and female PV-Cre mice were used for experiments shown in Figure 9 and Supplementary Figure S6. P29-35 SOM-Cre Het/ZnT3 Het or SOM-Cre Het/ZnT3 KO mice were used for experiments shown in Figure 3. These mice were generated by crossing ZnT3 KO mice with the F1 generation of the ZnT3 KO x SOM-Cre cross (Jackson, Bar Harbor, ME).

Figure 1.

Paired recordings from synaptically-coupled AC L2/3 SOMs and PNs: zinc signaling enhances SOM-mediated synaptic inhibition to PNs. (a) Left: schematic illustration of stereotaxic injections of retrograde microspheres of different colors to label corticocallosal (CCal, red) PNs and corticocollicular neurons (CCol, green). CCols were used to identify AC. Right: schematic illustrating slice electrophysiology experiment involving recordings between connected AC L2/3 SOMs and adjacent PNs (red). (b) Images in ×4 magnification showing, (1) the extent of AC in bright-field (left), (2) merged labeled CCols, CCal and SOMs (3) labeled CCols, and SOMs and (4) labeled CCals. (c) Images of L2/3 in ×40 magnification (originating from b1) showing, (1) L2,3 of AC in bright-field (left), (2) merged labeled CCals and SOMs (3) labeled SOMs and (4) labeled CCals. (d) Schematic illustration of an L2/3 SOM → PN connection. (e) Representative action potentials (Pre-AP) in a presynaptic SOM in current-clamp (C-clamp) mode and the elicited GABA_AR uIPSCs (post-IPSCs) in the connected PN in voltage clamp (V-clamp) mode in control (black), after 100 μM ZX1 (red), after washing out of ZX1 (blue), and after 20 μM SR 95531 (magenta, path application). (f) Time course of the average amplitude of GABA_AR uIPSCs before and after ZX1, and after ZX1 wash out and subsequent application of SR 95531. Note that after washing out ZX1, the responses tend to recover to control levels. Application of SR 95531 eliminates the responses. (g) Average effect of ZX1 (red) on L2/3 PN GABA_AR uIPSCs amplitudes, normalized to control (control vs. ZX1: 37.8 ± 2.8%, t = 6.149, df = 9, P = 0.000169, paired t-tests, n = 10 cells from nine mice). Asterisks (^***) denote a statistically significant difference at P < 0.001.

Figure 6.

Presynaptic mechanisms do not contribute to the zinc enhancement of SOM-mediated inhibition. (a) Schematic illustrating slice electrophysiology experiment involving recordings between synaptically-coupled pairs of AC L2/3 SOMs and adjacent PNs (red) (same as in Fig. 1). (b) Example paradigm of a unitary L2/3 SOM → PN connection. (c) An example of AP train in presynaptic L2/3 SOM eliciting a series of uIPSCs in a synaptically-connected adjacent PN. (d) Average peak amplitudes of the uIPSCs from trains, normalized to the peak amplitude of the first uIPSC. No statistically significant difference was observed between control (black) and ZX1 (red) (see detailed values for main figures). (e) Summary graph of paired-pulse ratio (for control: 0.66 ± 0.04 vs. ZX1:0.74 ± 0.06, t = −0.893, df = 7, P = 0.402, paired t-tests, n = 8 cells from six mice). (f) Summary graph of 1/CV² (for control first pulse: 24.1 ± 4.2 vs. ZX1 first pulse: 19.4 ± 3.7, t = 1.481, df = 9, P = 0.173, paired t-tests, n = 10 cells from nine mice).

Figure 2.

Optogenetic approach: zinc signaling enhances the SOM-mediated inhibition to PNs in AC L2/3. (a) Left: schematic illustration of stereotaxic injections, in SOM-Cre mice, of retrograde microspheres of different colors to label CCal PNs, and CCols (red) neurons. CCols were used to identify AC and viral vector (AAV) for expression of ChR2 in AC L2/3 SOMs. Right: schematic illustrating slice electrophysiology experiment involving photostimulation of ChR2 expressing AC L2/3 SOMs while recording from adjacent L2/3 PNs. (b) Representative traces of L2/3 PN GABA_AR Lev-IPSCs evoked by a 0.15-ms duration pulse photostimulation of adjacent SOMs in control (black), after 100 μM ZX1 (red), after washing out of ZX1 (blue) and after 20 μM SR 95531 (magenta, path application). Note that after washing out ZX1, the responses tend to recover to control levels. Application of SR 95531 eliminates the responses. (c) Time course of the average amplitude of SOM-mediated Lev-IPSCs before and after ZX1, and after ZX1 wash out and subsequent application of SR 95531. (d) Average effect of ZX1 (red) on L2/3 PN GABA_AR Lev-IPSCs amplitudes, normalized to control (control vs. ZX1:55.9 ± 2.6%, z = −3.621, df = 16, P = 0.000293, Wilcoxon sign-rank test, n = 17 cells from 12 mice). Asterisks (^***) denote a statistically significant difference at P < 0.001.

Figure 7.

Synaptic zinc enhances the quantal size of SOM-mediated inhibition. (a) Left: schematic illustration of stereotaxic injections, in SOM-Cre mice, of retrograde microspheres of different colors to label CCal PNs, and CCols neurons (red). CCols were used to identify AC and viral vector (AAV) for expression of ChR2 in AC L2/3 SOMs. Right: schematic illustrating slice electrophysiology experiment involving photostimulation of ChR2 expressing AC L2/3 SOMs while recording from adjacent PNs. (b) Example traces of SOM LEv-Sr²⁺-qIPSCs recorded from L2/3 PN in control (Ca²⁺), Sr²⁺, after 100 μΜ ZX1 and after washing out ZX1. A 0.15-ms duration LED light pulse was delivered at the arrowhead to activate SOM inhibitory terminals. LEv-Sr²⁺-qIPSCs were measured during a 400-ms window 100 ms after the LED stimulation (black solid line). (c) Average of 50 consecutive LEv-Sr²⁺-qIPSCs in Sr²⁺ (black), after 100 μΜ ZX1 (red) and after washing out of ZX1 (grey). (d) Average effect of ZX1 (red) on LEv-Sr²⁺-qIPSC amplitude, normalized to control (control vs. ZX1: 61.8 ± 2.5%, t = 9.806, df = 7, P = 0.000024, paired t-tests, n = 8 cells from five mice). (e) Distribution histogram of LEv-Sr²⁺-qIPSC amplitudes in control Sr²⁺ (black), after ZX1 (red) and after washing out ZX1 (blue) (n = 5 cells from three mice). Asterisks (^***) denote a statistically significant difference at P < 0.001.

Figure 8.

Synaptic zinc is required for the SOM-mediated sharpening of A1 L2/3 frequency tuning. (a) Schematic illustration of viral vectors (AAV) injections in SOM-Cre mice. (b) Left: schematic of experimental setup illustrating 2-photon imaging and photo-activation of SOMs with 660-nm LED in SOM-Cre mice expressing GCaMP6f in L2/3 PNs and ChrimsonR-td-Tomato in SOMs. Right: image of a population of A1 L2/3 labeled PNs and SOMs (red). (c) Representative example of the receptive fields of an A1 L2/3 PN in control LED-off (black) and LED-on (red) conditions. (d) Average effect of photo-stimulation of SOMs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in control (for LED-off: 145.26 ± 10.52 vs. LED-on: 71.36 ± 8.83, z = 12.09, df = 283, P = 1.1 e − 33, Wilcoxon sign-rank test, corrected for multiple comparisons, n = 284 neurons from eight mice). (e) Normalized effect of photo-stimulation of SOMs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in control (58.11 ± 2.9%, t = 19.43, df = 283, P = 1.2e − 53, one sample t-test, compared to 0, n = 284 neurons from eight mice). (f) Average delta bandwidth (ΔBW) representing change in the frequency tuning of A1 L2/3 PNs with photo-stimulation of SOMs in control (−0.22 ± 0.06, t = −3.66, df = 283, P = 3.2 e − 4, one sample t-test, compared to 0, n = 284 neurons from eight mice). (g) Representative example of the receptive fields of an A1 L2/3 PN in ZX1 LED-off (black) and LED-on (red) conditions. (h) Average effect of photo-stimulation of SOMs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in ZX1 (for LED-off: 69.57 ± 6.55 vs. LED-on: 52.24 ± 6.07, z = 2.88, df = 283, P = 0.003, Wilcoxon sign-rank test, corrected for multiple comparisons, n = 284 neurons from eight mice). Overall control compared to ZX1: one-way repeated measure ANOVA, F = 21.33, P = 2.4 e − 13. (i) Normalized effect of photo-stimulation of SOMs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in ZX1 (28.53 ± 7.5%, t = 3.77, df = 283, P = 2.24 e − 4, one sample t-test, compared to 0, n = 284 neurons from eight mice). (j) Average delta bandwidth (ΔBW) representing change in the frequency tuning of A1 L2/3 PNs with photo-stimulation of SOMs in ZX1 (0.07 ± 0.08, t = 0.99, df = 283, P = 31, one sample t-test, compared to 0, n = 284 neurons from eight mice). Asterisk (^*) and asterisks (^***) denote a statistically significant difference at P < 0.05 and P < 0.001, respectively.

Figure 4.

Synaptic zinc enhances the SOM-mediated inhibition to PVs in AC L2/3. (a) Left: schematic illustration of stereotaxic injections, in SOM-Cre/PV-GFP mice, of red retrograde microspheres to label CCols (red) PNs to identify AC and viral vector (AAV) for expression of ChR2 in AC L2/3 SOMs. Right: schematic illustrating slice electrophysiology experiment involving photostimulation of ChR2 expressing AC L2/3 SOMs while recording from adjacent L2/3 PVs. (b) Representative traces of L2/3 PV GABA_AR Lev-IPSCs evoked by a 0.15-ms duration pulse photostimulation of adjacent SOMs in control (black), after 100 μM ZX1 (red), after washing out of ZX1 (blue) and after 20 μM SR 95531 (magenta, path application). (c) Time course of the average amplitude of GABA_AR Lev-IPSCs before and after ZX1. Note that after ZX1 wash out the responses recover to control levels. Application of SR 95531 eliminates the responses. (d) Average effect of ZX1 (red) on L2/3 PV GABA_AR Lev-IPSCs amplitudes, normalized to control (control vs. ZX1:62.9 ± 4.3%, t = 2.625, df = 4, P = 0.043, paired t-tests, n = 5 cells from three mice). Asterisk (^*) denotes a statistically significant difference at P < 0.05.

Figure 5.

Synaptic zinc does not affect the PV-mediated inhibition to PNs in AC L2/3. (a) Left: schematic illustration of stereotaxic injections, in PV-Cre mice, of retrograde microspheres of different colors to label CCal PNs, and CCols neurons (red). CCols were used to identify AC and viral vector (AAV) for expression of ChR2 in AC L2/3 PVs. Right: schematic illustrating slice electrophysiology experiment involving photostimulation of ChR2 expressing AC L2/3 PVs while recording from adjacent L2/3 PNs. (b) Representative traces of L2/3 PN GABA_AR Lev-IPSCs evoked by a 0.15-ms duration pulse photostimulation of adjacent PVs in control (black) and after 100 μM ZX1 (red). (c) Time course of the average amplitude of GABA_AR Lev-IPSCs before and after ZX1. (d) Average effect of ZX1 (red) on L2/3 PN GABA_AR Lev-IPSCs amplitudes, normalized to control (control vs. ZX1:98.8 ± 1.3%, t = 0.706, df = 4, P = 0.519, paired t-tests, n = 5 cells from three mice).

Figure 9.

Synaptic zinc is not required for the PV-mediated sharpening of A1 L2/3 frequency tuning. (a) Schematic illustration of viral vectors (AAV) injections in PV-Cre mice. (b) Left: schematic of experimental setup illustrating 2-photon imaging and photo-activation of PVs with 660-nm LED in PV-Cre mice expressing GCaMP6f in L2/3 PNs and ChrimsonR-td-Tomato in PVs. Right: image of a population of A1 L2/3 labeled PNs and PVs (red). (c) Representative example of the receptive fields of an A1 L2/3 PN in control LED-off (black) and LED-on (red) conditions. (d) Average effect of photo-stimulation of PVs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in control (for LED-off: 121.184 ± 12.21 vs. LED-on: 22.51 ± 5.39, z = 12.11, df = 233, P = 8.4 e − 34, Wilcoxon sign-rank test, corrected for multiple comparisons, n = 234 neurons from five mice). (e) Normalized effect of photo-stimulation of PVs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in control (81.14 ± 2.5%, t = 31.26, df = 233, P = 2.5 e − 85, one sample t-test, compared to 0, n = 234 neurons from five mice). (f) Average delta bandwidth (ΔBW) representing change in the frequency tuning of A1 L2/3 PNs with photo-stimulation of PVs in control (−0.28 ± 0.08, t = −3.2, df = 233, P = 0.001, one sample t-test, compared to 0, n = 234 neurons from five mice). (g) Representative example of the receptive fields of an A1 L2/3 PN in ZX1 LED-off (black) and LED-on (red) conditions. (h) Average effect of photo-stimulation of PVs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in ZX1 (for LED-off: 43.36 ± 5.39 vs. LED-on: 28.32 ± 3.36, z = 2.78, df = 233, P = 0.005, Wilcoxon sign-rank test, corrected for multiple comparisons, n = 234 neurons from five mice). Overall control compared to ZX1: one-way repeated measure ANOVA, F = 40.61, P < 0.0001. (i) Normalized effect of photo-stimulation of PVs on the sound-evoked responses of A1 L2/3 PNs to BF sounds in ZX1 (34.1 ± 5.94%, t = 5.73, df = 233, P = 85.5 e − 8, one sample t-test, compared to 0, n = 234 neurons from five mice). (j) Average delta bandwidth (ΔBW) representing change in the frequency tuning of A1 L2/3 PNs with photo-stimulation of PVs in ZX1 (−0.30 ± 0.08, t = −3.47, df = 233, P = 6.25 e − 4, one sample t-test, compared to 0, n = 234 neurons from five mice). Asterisks (^**) and asterisk (^***) denote a statistically significant difference at P < 0.01 and P < 0.001, respectively.

Figure 3.

ZnT3-dependent (synaptically-released) zinc enhances the SOM-mediated inhibition to PNs in AC L2/3. (a) Left: schematic illustration of stereotaxic injections, in SOM-Cre/ZnT3 Het or KO mice, of retrograde microspheres of different colors to label CCal PNs, and CCols neurons (red). CCols we used to identify AC, and viral vector (AAV) for expression of ChR2 in AC L2/3 SOMs. Right: schematic illustrating slice electrophysiology experiment involving photostimulation of ChR2 expressing AC L2/3 SOMs while recording from adjacent L2/3 PNs. (b) Top: representative traces of L2/3 PN GABA_AR Lev-IPSCs evoked by a 0.15-ms duration pulse photostimulation of adjacent SOMs in control (black), after 100 μM ZX1 (red), after washing out of ZX1 (blue) and after 20 μM SR 95531 (magenta, path application) in SOM-Cre/ZnT3 Het mice. Bottom: same as the top panel but without washing out of ZX1 in SOM-Cre/ZnT3 KO mice. (c) Time course of the average amplitude of GABA_AR Lev-IPSCs before and after ZX1 in SOM-Cre/ZnT3 Het (red) and SOM-Cre/ZnT3 KO mice (light blue). Note that after ZX1 wash out the responses tend to recover to control levels. Application of SR 95531 eliminates the responses. (d) Average effect of ZX1 on L2/3 PN GABA_AR Lev-IPSCs amplitudes, normalized to control (ZnT3 KO ZX1 effect = 97.3 ± 5.5%, t = 1.132, df = 6, P = 0.309, paired t-tests, n = 7 cells from five mice; ZnT3 Het ZX1 effect = 57.8 ± 2.1%, t = 7.094, df = 8, P = 0.000103, paired t-tests, n = 9 cells from six mice). (The difference is statistically significant at P = 0.01, t = 6.8925, df = 14, Unpaired t-test). Asterisk (^*) denotes a statistically significant difference at P < 0.05.

Stereotaxic Injections for Electrophysiology

P21-23 male or female, SOM-GFP (GIN), PV-GFP, SOM-Cre, PV-Cre, SOM-Cre/PV-GFP, SOM-Cre Het/ZnT3 Het or SOM-Cre Het/ZnT3 KO mice were anesthetized with inhaled isoflurane (induction: 3% in oxygen, maintenance: 1.5% in oxygen) and secured in a stereotaxic frame (Kopf, Tujunga, CA). Core body temperature was maintained at ~ 37 °C with a heating pad and eyes were protected with ophthalmic ointment. Lidocaine (1%) was injected under the scalp and an incision was made into the skin at the midline to expose the skull. Projection PNs in the AC, in SOM-Cre, SOM-GFP, PV-Cre, SOM-Cre Het/ZnT3 Het or SOM-Cre Het/ZnT3 KO mice, were retrogradely labeled by injecting different colored fluorescent latex microspheres (Lumafluor) in the contralateral AC (in a small craniotomy drilled 4 mm posterior to bregma and 4 mm lateral, injection depth 0.4–0.8 mm) and the ipsilateral inferior colliculus (IC) (1 mm posterior to lambda and 1 mm lateral, injection depth 0.75 mm). A volume of ~0.12 μL of microspheres was pressure injected (25 psi, 10–15 ms duration) from capillary pipettes (Drummond Scientific) with a Picospritzer (Parker–Hannifin). In PV-GFP and SOM-Cre/PV-GFP mice, it was not necessary to inject in contralateral AC. After injection, the pipette was held in the brain for 2 min before slowly withdrawing.

To stimulate SOMs or PVs, SOM-Cre or PV-Cre P21-23 mice were intracranially injected at the right AC with recombinant adeno-associated virus (AAV) encoding Cre-dependent ChR2- td-Tomato fusion protein (AAV9.CAGGS.Flex.ChR2-tdTomato.WPRE. SV40; titer 2.724e¹³ genome copies/mL;Penn Vector Core) at the same time with the injection of microspheres. A small craniotomy (~0.4 mm diameter) was made over the temporal cortex (~4 mm lateral to lambda). A glass micropipette, containing the viral vector, was inserted into the cortex 0.4–0.8 mm past the surface of the skull with a micromanipulator (Kopf). The glass micropipette was backfilled with mineral oil and connected to a 5-μL glass syringe (Hamilton, Reno, NV). We used a syringe pump (World Precision Instruments, Sarasota, FL) to inject 400 nL of the virus solution (diluted 1:1 in PBS) over the course of 5 min. The pipette was left in place for 2 min after the end of the injection. The pipette was then removed, and the scalp of the mouse was closed with cyanoacrylate adhesive. Mice were injected with nonsteroidal anti-inflammatory drug carprofen 5 mg/kg (Henry Schein Animal Health) for 24 h prior to and 48 h after surgery. Mice were monitored for signs of postoperative stress and pain.

Stereotaxic Injections for In vivo Imaging

Male or female SOM-Cre and PV-Cre mice between P21-36 were used. We followed the same surgical procedure as described above. We used AAV9.CaMKII.GCaMP6f.WPRE.SV40 (Penn Vector Core; (Chen et al. 2013)) (diluted 1:4 in PBS) and AAV9-hSyn-FLEX-ChrimsonR-tdTomato (UNC Vector Core; (Klapoetke et al. 2014)) (diluted 1:1 in PBS) to express GCaMP6f in PNs and ChrimsonR in SOMs or PVs, respectively. We injected 200–400 nL of each virus solution separately over the course of 6–10 min.

Slice Electrophysiology

Slice electrophysiology experiments were performed in 29–35-day-old mice, at least 8 days after viral vector and colored microspheres injections. Mice were first anesthetized with isoflurane and then immediately decapitated. Brains were rapidly removed and coronal slices (300 μm) containing the right AC were prepared in a cutting solution at 1 °C using a Vibratome (VT1200 S; Leica). The cutting solution, pH 7.4, ∼300 mOsm, contained the following (in mM): 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 0.5 CaCl₂, 7 MgCl₂, 7 glucose, 205 sucrose, 1.3 ascorbic acid, and 3 sodium pyruvate (bubbled with 95% O₂/5% CO₂). The slices were immediately transferred and incubated at 34 °C in a holding chamber for 40 min before recording. The holding chamber contained artificial cerebrospinal fluid (ACSF), pH 7.4, ∼300 mOsm containing the following (in mM): 125 NaCl, 2.5 KCl, 26.25 NaHCO₃, 2 CaCl₂, 1 MgCl₂, 10 glucose, 1.3 ascorbic acid, and 3 sodium pyruvate, pH 7.4, ∼300 mOsm (bubbled with 95% O₂/5% CO₂). Contaminating zinc was removed from the ACSF by stirring the ACSF with Chelex 100 resin (Biorad) for 1 h (Anderson et al. 2015). After the filtering of ACSF from Chelex resin, using Nalgene rapid flow filters lined with polyethersulfone (0.2 μm pore size), high purity MgCl₂·6H₂O and CaCl₂·2H₂O salts (99.995% purity, Sigma Aldrich) were added. All plastic and glassware were washed with 5% high-purity nitric acid (Sigma-Aldrich). Postincubation, the slices were stored at room temperature until the time of recording. Whole-cell recordings in voltage- and current-clamp modes were performed on slices bathed in carbogenated ACSF, which was identical to the incubating solution. To reduce polysynaptic responses of PV-mediated GABA_AR IPSCs to PNs, we used High Divalent ACSF (HIDI-ACSF) containing the following (in mM): 115 NaCl, 2.5 KCl, 26.25 NaHCO₃, 4 CaCl₂, 4 MgCl₂, 10 glucose, 1.3 ascorbic acid, and 3 sodium pyruvate, pH 7.4, ∼300 mOsm (Sivaramakrishnan et al. 2013). The increase of Ca²⁺ and Mg²⁺ was followed by a reduction of the NaCl concentration to balance the osmolarity.

Borosilicate pipettes (World Precision Instruments) were pulled into patch electrodes with 2.5–5 MΩ resistance (Sutter Instruments) and filled with a potassium-based intracellular solution, which was used for assessing intrinsic properties in current clamp. This solution contained (in mM): 128K-gluconate, 10 HEPES, 4 MgCl₂, 4 Na₂ATP, 0.3 Tris-GTP, 10 Tris phosphocreatine, 1 EGTA, and 3 sodium ascorbate (pH = 7.25, 295 mOsm). For recording GABA_AR IPSCs, a cesium-based internal solution containing (in mM): 126 CsCH₃O₃S, 4 MgCl₂ 10 HEPES, 4 Na₂ATP, 0.3 TrisGTP, 10 Tris-phosphocreatine, 1 CsEGTA, 1 QX-314, and 3 sodium ascorbate (pH = 7.25, 295 mOsm) was used.

For electrophysiological recordings, we used a MultiClamp-700B amplifier equipped with Digidata-1440A A/D converter and Clampex (Molecular Devices). Data were sampled at 10 kHz and filtered at 4 kHz. In experiments used to study intrinsic properties of neurons, we added the following drugs: 20 μΜ DNQX (AMPA receptor antagonist), 50 μΜ APV (NMDA receptor antagonist), and 20 μΜ SR 95531 Hydrobromide (Gabazine—a GABAA receptor antagonist). Pipette capacitance was compensated and series resistance for recordings was lower than 15 MΩ. Series resistance (R_series) was determined by giving a −5-mV voltage step for 50 ms in voltage-clamp mode (command potential set either at −70 mV or at 0 mV) and was monitored throughout the experiments. R_series was calculated by dividing the −5-mV voltage step by the peak current value generated immediately after the step in the command potential. Recordings were excluded from further analysis if the series resistance changed by more than 15% throughout the experiment. Input resistance (R_input) was calculated by giving a −5-mV step in voltage-clamp mode (command potential set either at −70 mV or at 0 mV), which resulted in transient current responses. The difference between baseline and steady-state hyperpolarized current (ΔI) was used to calculate R_input using the following formula: R_input = −5 mV/ΔI − R_series. Representative traces of hyperpolarizing pulses in Supplementary Figures S2 and S3 were acquired by averaging 10 consecutive responses.

The average resting membrane potential (V_rest) was calculated by holding the neuron in voltage-follower mode (current clamp, at I = 0) immediately after breaking in and averaging the membrane potential over the next 20 s. V_rest was again calculated by averaging the membrane potential over the next 20 s after ZX1 bath application. In current clamp, depolarizing current pulses (0–450 pA in 50 pA increments of 1-s duration) were used to examine each neuron’s basic suprathreshold electrophysiological properties (baseline V_m was maintained at −70 mV, by injecting the required current, if necessary). Action potential (AP) width was calculated as the full width at the half-maximum amplitude of the first resulting AP at rheobase. The AP threshold was measured in phase plane as the membrane potential at which the depolarization slope exhibited the first abrupt change (Δslope > 10 V/s). Current-firing frequency function was quantified by calculating the initial (150–250 pA) slope of the frequency-current (f–I) relationship. We chose to calculate the slope between those three particular current points because they are above the nonlinear region of the curves near the rheobase, and they are below the region of large current values where the f–I function start to saturate.

Light-evoked Inhibitory Postsynaptic Currents (Lev-IPSCs)

In the postsynaptic PNs, Lev-IPSCs were evoked by optogenetic stimulation of presynaptic axons. More specifically, a collimated blue LED light source (470 nm, Thorlabs) was directed through a diaphragm and a 40× microscope objective and restricted to a small spot adjacent to the recording neuron. The minimal light intensity to elicit a reliable response was determined on a cell-by-cell basis, with 0.15-ms pulse duration. Trials included either a single pulse or a train of five pulses at 20 Hz with an intertrial interval of 30 s. The intertrial interval was changed to 10 s for the Light Evoked strontium quantal IPSCs (Lev-Sr²⁺-qIPSCs; see methods below), the paired recordings and the CV analysis of Lev-IPSCs. GABA_AR IPSCs were recorded in voltage clamp mode at 0 mV (peak values were averaged over a 1-ms time window) in the presence of DNQX (10 μM, AMPAR and kainate receptor antagonist) and CGP 55845 (10 μM, GABA_BR antagonist). To investigate the effect of ZX1, GABA_AR IPSCs were recorded under control conditions for at least 10 min and for 20 min following 100 μM ZX1 application. At the end of each recording, we applied SR 95531 hydrobromide (Gabazine) to block GABA_AR-mediated IPSCs. Representative traces for Figures 2–5 and Supplementary Figure S1 were acquired by averaging 10 consecutive Lev-IPSCs.

Light-evoked Strontium Quantal Inhibitory Postsynaptic Currents (Lev-Sr²⁺-qIPSCs)

Slices were superfused with standard extracellular solution containing 4 mM SrCI₂ and 4 mM MgCI₂ with 0 mM CaCl₂. More specifically, we used ACSF containing the following (in mM): 115 NaCl, 2.5 KCl, 26.25 NaHCO₃, 4 SrCl₂, 4 MgCl₂, 10 glucose, 1.3 ascorbic acid, and 3 sodium pyruvate, pH 7.4, ∼300 mOsm (bubbled with 95% O₂/5% CO₂). Slices were allowed to incubate in this solution for a minimum of 30 min prior to recording (Oliet et al. 1996; Petrus et al. 2014).

ChR2 was activated using collimated blue LED light source (470 nm, Thorlabs) illuminated through a 40× objective lens. Cells were held at 0 mV. Event analysis was performed using mini analysis software (see below). A 400-ms window beginning 100 ms after the stimulus onset was used for quantifying the amplitude of LED-evoked desynchronized quantal events. This 400-ms time window is consistent with previous studies (Oliet et al. 1996; Petrus et al. 2014) and ensures elimination of synchronous synaptic responses. Acquired Lev-Sr²⁺-qIPSCs were analyzed with a Mini Analysis program (Clampfit), with a detection threshold set at three times the SD of the baseline noise level (noise levels were measured during a 100-ms timing period before illumination). We excluded all Lev-Sr²⁺-qIPSCs with a rise time >3 ms, and those showing a negative correlation between amplitude and rise time (Petrus et al. 2014). All the observed consecutive Lev-Sr²⁺-qIPSCs, 5 min before and the last 5 min of ZX1 application, were analyzed from each cell, and the data are expressed as mean ± standard error of the mean.

Paired Recordings

For the paired recordings, we used the transgenic GIN mouse line SOM-GFP (Oliva et al. 2000), which gives us the advantage to record from L2/3 SOM Martinotti neurons. Whole-cell recordings were established from a green-labeled SOM neuron and a red-labeled corticocallosal PN in AC L2/3. Connectivity was assessed by evoking a train (five pulses, 20 Hz) of action potentials in SOM, while monitoring the unitary inhibitory postsynaptic current (uIPSC) in the PN (Figure 6c); at least 30 trials of the presynaptic train were delivered (10 s intertrial interval); multiple trials were averaged to detect the presence of a connection. We recorded from 80 SOMs presenting regular firing pattern (Supplementary Fig. S2g,h). The connection rate between SOMs and PNs was ~ 7% (10 connections found out of 143 tested). uIPSC amplitudes were obtained by averaging a 1-ms window around the peak response. uIPSC trains were normalized to the amplitude of the first uIPSC before analyzing them for short-term plasticity (STP). In current-clamp mode recordings, neurons were held at −70 mV. Representative traces (either uIPSCs or APs) in Figures 1 and 6 were acquired by averaging 30 consecutive responses.

Optogenetic Stimulation of L2/3

To ensure that the photostimulation of SOMs only targeted L2/3 of AC, we first expressed the red fluorescent protein, tdTomato, in SOMs, by injecting Cre-dependent tdtomato viral vector into the AC of SOM-Cre mice (AAV9.Flex.tdTomato; titer ≥1e¹³ genome copies/mL, Addgene). To photobleach the area from light applied by the 40× objective, we stimulated the tissue for 45 min at the same LED intensity we used for our experiments. Note that during our experiments the duration of stimulation was much smaller (0.15 ms per stimulation, approximately 50 stimuli per experiment). These experiments show that the photobleached area from light applied by the 40× objective was mainly isolated to layer 2/3 (Supplementary Fig. S1k,l).

In vivo Imaging Preparation

About 21–28 days after viral vectors injections, mice were prepared for in vivo calcium imaging. Mice were anesthetized with inhaled isoflurane (induction: 3% in oxygen, maintenance: 1.5% in oxygen) and positioned into a custom-made head holder. Core body temperature was maintained at ~37 °C with a heating pad and eyes were protected with ophthalmic ointment. Lidocaine (1%) was injected under the scalp, and an incision (~1.5 cm long) was made into the skin over the right temporal cortex. The head of the mouse was rotated ~ 45° in the coronal plane to align the pial surface of the right temporal cortex with the imaging plane of the upright microscope optics. The skull of the mouse was secured to the head holder using dental acrylic (Lang, Wheeling, IL) and cyanoacrylate adhesive. A tube (the barrel of a 25-mL syringe or an SM1 tube from Thorlabs, Newton, NJ) was placed around the mouse body to reduce movement, and the mouse received an injection of the sedative chlorprothixene (0.36 mg/kg intramuscular) to reduce mouse movement during in vivo imaging (Chen et al. 2013; Issa et al. 2014; Kato et al. 2015). A dental acrylic reservoir was created to hold warm (37 °C) ACSF over the exposed skull. In preparing the ACSF, we removed contaminating zinc by incubating with Chelex 100 resin (Biorad, Hercules, CA) for 1 h. Subsequently, we removed the Chelex by vacuum filtration and added high-purity calcium and magnesium salts (99.995% purity; Sigma-Aldrich, St. Louis, MO). The solution contained in millimolar: 130 NaCl, 3 KCl, 2.4 CaCl₂, 1.3 MgCl₂, 20 NaHCO₃, 3 HEPES, and 10 D-glucose, pH = 7.25–7.35, ∼300 mOsm. For better optical access of the auditory cortex, we injected lidocaine–epinephrine (2% lidocaine, 1:100 000 weight/volume epinephrine) into the temporal muscle and retracted a small portion of the muscle from the skull. Mice were then positioned under the microscope objective in a sound- and a light-attenuation chamber containing the microscope and a calibrated speaker (ES1, Tucker-Davis Davis Technologies, Alachua, FL). Acoustic stimuli were calibrated with a free-field compensated ¼ inch microphone (4954-B, Bruel & Kjaer, Narum, Denmark) placed at the location of the mouse’s ear within the chamber.

Transcranial Imaging for A1 Localization

We performed transcranial imaging to locate A1 in each mouse (Supplementary Fig. S4ab). We then removed the isoflurane from the oxygen flowing to the mouse and began imaging sound-evoked responses at least 10 min later (Issa et al. 2014; Anderson et al. 2015). Sounds were delivered from a free-field speaker 10 cm from the left ear of the mouse (ES1 speaker, ED1 driver, Tucker-Davis Technologies), controlled by a digital to analog converter with an output rate of 250 kHz (USB-6229, National Instruments, Austin, TX). We used ephus (Suter et al. 2010) to generate sound waveforms and synchronize the sound delivery and image acquisition hardware. We presented 50 or 60 dB SPL, 6 kHz tones to the mouse while illuminating the skull with a blue LED (nominal wavelength of 490 nm, M490L2, Thorlabs). We imaged the change in green GCaMP6f emission with epifluorescence optics (eGFP filter set, U-N41017, Olympus, Center Valley, PA) and a 4× objective (Olympus) using a cooled CCD camera (Rolera, Q-Imaging, Surrey, BC, Canada). Images were acquired at a resolution of 174 × 130 pixels (using 4× spatial binning, each pixel covered an area of = 171.1 μm² of the image) at a frame rate of 20 Hz to locate A1 in each mouse (see Analysis section below). To localize A1, we used 50 or 60 dB SPL, 6 kHz tones and we normalized the sound-evoked change in fluorescence after sound presentation (ΔF) to the baseline fluorescence (F), where F is the average fluorescence of 1 s preceding the sound onset (for each pixel in the movie). We applied a two-dimensional, low-pass Butterworth filter to each frame of the ΔF/F movie, and then created an image consisting of a temporal average of 10 consecutive frames (0.5 s) beginning at the end of the sound stimulus. This image indicated two sound-responsive regions corresponding to the low-frequency tonotopic areas of A1 and the AAF (Supplementary Fig. S4b).

Two-Photon Imaging

For two-photon imaging in awake mice, after A1 localization (Supplementary Fig. S4), we reanesthetized the mouse with isoflurane and created a craniotomy (~1 mm²) over A1 to improve optical access (Supplementary Fig. S4c). Using a micromanipulator (Siskiyou, Grants Pass, OR), we inserted a glass micropipette backfilled with mineral oil and connected to a 5-μL glass syringe into the cortex at the edge of this craniotomy (Figure 8b, left). The pipette contained ACSF including 100 μM of ZX1 (an extracellular, high-affinity, fast, zinc-specific chelator; (Pan et al. 2011; Anderson et al. 2015) and 50 μM Alexa-594. To photoactivate PVs or SOMs, we mounted a fiber-coupled red LED (660 nm) coupled to black-coated cannula (0.39 NA; Thorlabs) on a micromanipulator and positioned it ~ 1 mm above A1 (Figures 8b and 9b). Next, we removed the isoflurane and began two-photon imaging after 60 min of recovery from isoflurane. Mode-locked infrared laser light (940 nm, intensity at the back focal plane of the objective, MaiTai HP, Newport, Santa Clara, CA) was delivered through a galvanometer-based scanning two-photon microscope (Scientifica, Uckfield, UK) controlled with scanimage 3.8 (Pologruto et al. 2003), using a 40×, 0.8 NA objective (Olympus) with motorized stage and focus controls. We imaged green and red fluorescence simultaneously with two photomultiplier tubes behind red and green emission filters (FF01-593/40, FF03-525/50, Semrock) using a dichroic splitter (Di02-R561, Semrock) at a frame rate of 5 Hz over an area of 145 × 145 μm and at a resolution of 256 × 256 pixels. We imaged PNs in L2/3 at a depth of ~ 200 μm from pia. After identifying A1 L2/3 PNs responding to sounds, we presented sound trains consisting of 500-ms long-duration pure tones in the range of 4–40 kHz frequencies in 0.20-octave increments at 70-dB SPL. To activate ChrimsonR-expressing SOMs or PVs, at each alternate trial, red LED pulse-trains were paired with sound-trains triggering red LED 250 ms before the sound onset and lasting for 1 s (Figure 8c). Each trial was performed at least 5 times, and sound-evoked responses were imaged in both LED-off and LED-on conditions. After obtaining movies of responses to different sound stimuli in both LED-off and LED-on conditions, we began to infuse ZX1-containing solution into the cortex at a rate of 30 nL/min (Anderson et al. 2017). Once ZX1 diffused in A1, we reduced the pump speed to 9 nL/min and remeasured the responses of the same neurons to the same sounds in both LED-off and LED-on conditions. Mice were euthanized at the end of the recording session.

Two-Photon Analysis

To quantify the neuronal responses to sounds, we identified the same PNs that were present in the field of view before and after ZX1 infusion and targeted only these cells for analysis. Before extracting ΔF/F, we used the NoRMCorre software to correct motion artifacts from individual tiff movies (Pnevmatikakis and Giovannucci 2017; Wilson et al. 2018). Next, using FluoroSNNAP software (Patel et al. 2015), we selected ROIs within the soma of each L2/3 PN from the temporal average of each tiff movie (300 frames, 60-s long). The pixels in each ROI from each frame were averaged and converted into ΔF/F as described previously (Kumar et al. 2019). We then averaged the fluorescent response for 5–7 presentations of the same sound frequency for each PN. Sound-evoked responses were measured for 1 s after the sound onset and were defined as responses if the sound-evoked changes in ΔF/F were larger than the mean + 3 standard deviations (SDs) of the baseline fluorescence measured prior to the sound onset. Responses were quantified as the integral of the increased fluorescence during this 1-s period. Best frequency (BF) was defined as the sound frequency resulting in the largest response (Kato et al. 2017). Receptive field bandwidth (BW) was defined as log2 of the ratio of the highest sound frequency and the lowest sound frequency that elicited a response from the PN (Winkowski and Kanold 2013). To analyze the effect of SOMs or PVs on PNs’ frequency tuning, we calculated the delta bandwidth (ΔBW) of individual PNs, (ΔBW = BW_LED-on − BW_LED-off).

Image Processing

Image captioning was done with Qcapture software. All the images were imported and processed into ImageJ (https://imagej.nih.gov/ij/).

Experiments with KO mice

Experiments with ZnT3 Het and ZnT3 KO were blinded.

Statistics

For data that passed the Shapiro–Wilk normality test, Paired t-tests were used for all statistical analyses to compare the effect of drug application on responses. Otherwise, we used the Wilcoxon sign-rank test for non-normally distributed data. For the statistical comparisons between two independent groups that passed the Shapiro–Wilk normality test we used unpaired t-tests. Otherwise we used the Mann–Whitney rank-sum test for non-normally distributed data. For comparisons between multiple groups, we assessed overall differences with a one-way ANOVA or a one-way repeated measures ANOVA for paired data followed by pairwise comparisons using the Holm-Bonferroni correction method. Significance levels are denoted as ^*P < 0.05, ^**P < 0.01, ^***P < 0.001. The details of statistical tests are described in the figure legends. Group data are presented as mean ± SEM.

Results

Endogenous Zinc Enhances SOM-mediated Inhibition to L2/3 PNs

To investigate the effect of endogenous zinc on SOM-mediated inhibition, we recorded from synaptically coupled pairs of SOMs (presynaptic) and PNs (postsynaptic). This approach allowed us to determine whether zinc affects SOM-mediated GABA_AR uIPSCs in auditory cortex (AC) L2/3 PNs, in response to a single presynaptic AP. We used the transgenic GIN mouse line SOM-GFP (Oliva et al. 2000), which permits identification of labeled L2/3 SOM Martinotti neurons (Fig. 1a). The vast majority of GIN (GFP-positive) cells in this mouse line are L2/3 Martinotti cells (Oliva et al. 2000; Ma et al. 2006; Rudy et al. 2011), and express high levels of ZnT3 mRNA (Paul et al. 2017). To identify L2/3 PNs in brain slices from SOM-GFP mice, we labeled corticocallosal (CCal) PNs (red), by in vivo injection of red fluorescent retrograde microspheres into the contralateral cortex. Although a small proportion of these projection neurons are PVs (Rock et al. 2018), we recorded from morphologically defined CCal PNs according to their larger cell bodies and distinct shape, compared to the small and round-shaped PVs (see Discussion). While AC is stereotaxically localizable in the intact brain, AC localization is more challenging in brain slices because its areal borders are not sharply demarcated by cytoarchitectural features. To localize the AC in brain slices, we labeled corticocollicular (CCol) L5B AC PNs (green), by injecting green fluorescent retrograde microspheres to the inferior colliculus (Fig. 1a) (see Methods). Consistent with our previous reports (Joshi et al. 2015; Joshi et al. 2016), the localization of CCol PNs in the auditory cortex, along with anatomical landmarks, such as the rhinal fissure and the underlying hippocampal formation allowed us to locate AC (Fig. 1b). After AC localization, we recorded simultaneously from synaptically connected L2/3 red CCal PNs and L2/3 green SOMs (Fig. 1c,d). To determine the effect of endogenous zinc on the synapses between L2/3 SOMs and L2/3 PNs, we bath applied 100 μM ZX1, an extracellular, fast, high-affinity zinc-specific chelator (Pan et al. 2011; Anderson et al. 2015; Kalappa et al. 2015; Kalappa and Tzounopoulos 2017). ZX1 inhibited the uIPSCs in L2/3 PNs evoked by a single presynaptic SOM spike (Fig. 1e–g). Upon ZX1 removal from the bath (wash out), we observed a partial recovery of uIPSCs. Unitary IPSCs were eliminated after the application of a GABA_AR blocker SR 95531 (20 μM), indicating that the recorded currents are mediated by GABA_A receptors. Together, these results support that endogenous and extracellular, likely synaptically released, zinc elicited by a single AP enhances SOM-mediated GABA_AR inhibition in L2/3 PNs.

To further confirm the enhancing effect of zinc on SOM-mediated inhibition, we evaluated the effect of ZX1 on light-evoked SOM-mediated IPSCs in L2/3 PNs. To achieve this, we selectively expressed ChR2 in SOMs by injecting Cre-dependent ChR2 viral vector into the AC of SOM-Cre mice (Fig. 2a, left). Although all SOM subtypes are labeled in the SOM-Cre mouse line, L2/3-labeled SOMs are mainly Martinotti neurons (Munoz et al. 2017). We identified AC L2/3 PNs in brain slices as described in Figure 1. This approach allowed us to elicit light-evoked SOM-mediated GABA_AR IPSCs (Lev-IPSCs) with pulses of blue light applied to AC L2/3 (Fig. 2a, right). Consistent with the paired recordings shown in Figure 1, bath application of 100 μM ZX1 inhibited Lev-IPSCs in L2/3 PNs (Fig. 2b–d). Wash out of ZX1 led to the recovery of Lev-IPSCs, which were subsequently fully blocked by application of SR 95531 (Fig. 2b–d). Together, these results show that endogenous zinc enhances the GABA_AR IPSCs in the synapses between SOMs and CCal PNs in AC L2/3. This is an important finding, suggesting that endogenous extracellular zinc is a novel modulator of SOM-mediated GABAergic inhibition.

Zinc Enhancement of SOM-mediated Inhibition is ZnT3 Dependent

To track the origin of the extracellular endogenous zinc that enhances SOM-mediated inhibition, we performed similar optogenetic experiments to those described in Figure 2, but now ZnT3-expressing and ZnT3-lacking mice (ZnT3 KO). These mice were generated by crossing ZnT3 KO mice with the F1 generation mice of the ZnT3 KO x SOM-Cre cross (see Methods). This crossing resulted in littermate SOM-Cre mice heterozygous for ZnT3 (ZnT3 Het), which express ZnT3 and synaptic zinc (Cole et al. 1999), and ZnT3 KO mice, which lack ZnT3 and synaptic zinc (Cole et al. 1999). We performed optogenetic experiments in littermate mice in a blind fashion. We found that ZX1 reduced SOM-mediated Lev-IPSCs in PNs in ZnT3 Het mice but did not affect Lev-IPSCs in ZnT3 KO mice (Fig. 3a–d). Moreover, the magnitudes of the ZX1 inhibition of SOM-mediated Lev-IPSCs in SOM-Cre Het/ZnT3 Het and SOM-Cre/ZnT3 WT mice were not different (compare red bars between Figs 2d and 3d; P = 0.435, unpaired t-test). Taken together, these results indicate that synaptic release of ZnT3-dependent vesicular zinc mediates the enhancement of SOM-mediated inhibition in L2/3 PNs.

Synaptic Zinc Enhances SOM- But Not PV-mediated Inhibition

To investigate whether zinc enhancement of SOM-mediated inhibition depends on the postsynaptic target of SOMs, we next tested whether zinc enhances SOM-mediated inhibition to PVs. To record SOM-mediated GABA_AR IPSCs to PVs, we selectively expressed ChR2 in SOMs by injecting Cre-dependent ChR2 viral vector into the AC of SOM-Cre/PV-GFP mice (Fig. 4a). Bath application of 100 μM ZX1 inhibited the SOM-mediated Lev-IPSCs in L2/3 PVs (Fig. 4b–d). The effect of ZX1 in the amplitude of SOM-mediated Lev-IPSCs in PVs was not significantly different from the effect of ZX1 in the amplitude of SOM-mediated Lev-IPSCs in PNs (compare red bars between Figs 2d and 4d; P = 0.210, unpaired t-test), suggesting that the enhancing effect of synaptic zinc on SOM-mediated inhibition does not depend on the postsynaptic target of SOMs—synaptic zinc enhances SOM-mediated inhibition in AC.

To explore whether the effect of synaptic zinc is dependent on the presynaptic (input) neuron, we next studied whether zinc enhances the inhibition between PVs and L2/3 PNs. To achieve this, we selectively expressed ChR2 in PVs by injecting Cre-dependent ChR2 viral vector into the AC of PV-Cre mice (Fig. 5a,b) (see Methods). Application of 100 μM ZX1 did not affect the PV-mediated Lev-IPSCs in PNs (Fig. 5b–d), suggesting that the enhancing effect of zinc on cortical inhibition is cell input-specific: synaptic zinc enhances SOM- but not PV-mediated inhibition.

Synaptic Zinc Enhancement of SOM-mediated Inhibition is Mediated by Changes in Quantal Size

Next, we sought to study the mechanism by which zinc enhances SOM-mediated inhibition. To determine whether the effects of synaptic zinc were mediated by presynaptic mechanisms, we used paired-pulse ratio (PPR) and coefficient of variation (CV) analyses. These assays are sensitive to changes in the presynaptic probability of neurotransmitter release (Pr) (Faber and Korn 1991; Tsien and Malinow 1991). PPR, the ratio of the amplitude of the second IPSC to the first IPSC evoked by two stimuli applied in succession of 20 Hz (interval of 50 ms), depends on Pr. Synapses with high Pr show paired-pulse depression, whereas low Pr synapses show paired-pulse facilitation. The CV (SD of the IPSC amplitudes normalized to the mean amplitude) varies inversely with the quantal content, where quantal content = n ×Pr, n the number of release sites. The inverse square of the CV (1/CV²) is directly proportional to the quantal content. In paired recordings, we found that application of 100 μM ZX1 did not affect either PPR or 1/CV² of SOM-mediated uIPSCs (Fig. 6b–f) or Lev-IPSCs in ChR2 experiments (Supplementary Fig. S1d,e), suggesting a lack of contribution of presynaptic mechanisms in the zinc-mediated enhancement of SOM-mediated inhibition. Consistent with this conclusion, ZX1 did not affect the STP of either uIPSCs (Fig. 6c,d) or Lev-IPSCs (Supplementary Fig. S1b,c). Similarly, chelation of zinc did not affect PPR, 1/CV², or STP in the synapses between SOM and PVs (Supplementary Fig. S1g–j), suggesting that synaptic zinc does not exert its effects via presynaptic mechanisms.

To test whether ZX1 affects the spiking output of either SOMs or PVs, we measured the effect of ZX1 on the intrinsic properties of these interneurons. Input resistance (R_input), membrane resting potential (V_rest), AP width (AP_width), AP threshold (AP_threshold) and current-firing frequency function did not change after ZX1 application (Supplementary Figs S2 and S3), suggesting that extracellular zinc does not affect the intrinsic excitability of either SOMs or PVs.

To investigate whether zinc exerts its effect via postsynaptic mechanisms, we replaced calcium (Ca²⁺) with strontium (Sr²⁺) in the bath. Sr²⁺ desynchronizes the evoked release of neurotransmitter. The asynchronous release of quanta is enhanced and thus allows the analysis of quantal events from the stimulated synapses (see Methods) (Oliet et al. 1996; Petrus et al. 2014). Changes in the amplitude of quantal events correspond to postsynaptic changes. Application of 100 μM ZX1 reduced the amplitude of the SOM-mediated Lev-Sr²⁺-qIPSCs. Wash out of ZX1 led to the recovery of the amplitude of the LEv-Sr²⁺-qIPSCs (Fig. 7b–e). Together, these results suggest that zinc enhancement of SOM-mediated GABA_AR IPSCs is due to a postsynaptic increase in quantal size.

Differential Contributions of Synaptic Zinc to the Effects of SOM and PV Activation on Sound Processing

So far, our in vitro results show that synaptic zinc enhances SOM- but not PV-mediated inhibition, suggesting differential effects of zinc on SOM- and PV-mediated inhibition to sound processing. To address how zinc contributes to PV- and SOM-mediated inhibition during sound processing, we used two-photon microscopy to image the sound-evoked responses of A1 L2/3 PNs before and after SOM- or PV-optogenetic activation in the presence and absence of zinc signaling.

To interrogate the contribution of zinc signaling in SOM-mediated inhibition, we used a viral approach to drive selective expression of the calcium indicator GCaMP6f into PNs and a red-shifted channelrhodopsin variant, ChrimsonR, into SOMs (Fig. 8a,b). To achieve this, we injected two viral vectors into the AC of SOM-Cre mice, the calcium/calmodulin protein kinase 2 promoter (CaMKII)-dependent viral vector to express GCaMP6f into PNs (AAV-CaMKII-GCaMP6f), and the Cre-dependent ChrimsonR viral vector (AAV-hSyn-FLEX-ChrimsonR-tdTomato) to express ChrimsonR into SOMs (Chen et al. 2013; Klapoetke et al. 2014; Anderson et al. 2017; Kumar et al. 2019) (see Methods). Three to four weeks postviral injection, after locating A1 in head-fixed, awake mice (Supplementary Fig. S4), we anesthetized the mouse and performed a small craniotomy over A1 to insert a thin glass micropipette filled with ZX1 (Anderson et al. 2017; Kumar et al. 2019). To activate SOMs, we mounted a fiber-coupled red LED (660 nm) coupled to black-coated cannula (0.39 NA; Thorlabs) on a micromanipulator and positioned it ~1 mm above A1 (Fig. 8b). After the mouse recovered from isoflurane induced anesthesia (at least 1 h later), we performed two-photon imaging to visualize the sound-evoked responses of A1 L2/3 PNs. We presented trains of 500 ms pure tones at interstimuli interval of 3 s in pseudo-random order that spanned 4–40 kHz in 0.20-octave increments at 70 dB SPL (total train duration was 48 s). To activate ChrimsonR-expressing SOMs, during alternate trials, red LED pulse-trains were paired with sound-trains triggering red LED 250 ms before the sound onset and lasting for 1 s (Fig. 8c). Each trial was performed at least five times and sound-evoked responses were imaged in both LED-off and LED-on conditions. Consistent with previous studies (Wilson et al. 2012; Cottam et al. 2013; Seybold et al. 2015; Phillips and Hasenstaub 2016), we found that optogenetic activation of SOMs reduced the sound-evoked responses of A1 L2/3 PNs to BF sounds (Fig. 8c–e). To further explore the effects of SOM activation on different sound frequency regions within the neuronal receptive field, we created a heat map showing the average sound-evoked response amplitudes of A1 L2/3 PNs for different sound frequencies compared with BF before and after the activation of SOM interneurons (Supplementary Fig. S5f,g). We observed a general reduction in response amplitudes to BF sound frequencies, indicated by the shift in dark red color to orange, and to sound frequencies below and above BF, indicated by the shift in dark blue color to light blue (Supplementary Fig. S5f,g). To characterize whether optogenetic activation of SOMs modulates Α1 L2/3 PN tuning, we measured the change in the range of sound frequencies that elicited a significant response (3 SDs above baseline, see Methods), termed delta bandwidth (ΔBW). Optogenetic SOM activation reduced the bandwidth of PNs (Fig. 8f), suggesting that SOMs sharpen the tuning of PNs (see Discussion for the two major models explaining changes in the tuning of auditory cortical neurons). The effects of SOM activation are not attributed to nonspecific issues related to the LED light, because LED light did not affect the sound-evoked responses of A1 L2/3 PNs in mice that were not injected with the ChrimsonR containing virus (Supplementary Fig. S5a–e).

To identify the role of endogenous, extracellular zinc on SOM-mediated inhibition of A1 L2/3 PN sound-evoked responses, we infused ZX1 as described previously (Anderson et al. 2017; Kumar et al. 2019). In previous studies, we showed that effects of ZX1 are not due to nonspecific effects related to the injection of fluid into the auditory cortex, because vehicle injections containing ACSF and Alexa Fluor 594 do not affect gain and/or frequency tuning (Anderson et al. 2017; Kumar et al. 2019). Based on in vitro results (Figs 1–7), we hypothesized that endogenous zinc would enhance SOM-mediated inhibition of A1 L2/3 PN sound-evoked responses to BF sounds. To test this, we reimaged the sound-evoked responses of the same individual neurons in the presence of ZX1 in both LED-off and LED-on conditions (Fig. 8g). We found that chelation of endogenous zinc with ZX1 reduced the SOM-mediated inhibition of A1 L2/3 PN responses to BF sounds (Fig. 8g–i and Supplementary Fig. S5h–j), suggesting that endogenous zinc contributes to SOM-mediated inhibition of A1 L2/3 PNs. Importantly, SOM activation in the presence of ZX1 did not reduce the BW of PNs (Fig. 8j and Supplementary Fig. S5h–k), suggesting that synaptic zinc is necessary for the SOM-mediated sharpening of frequency tuning in A1.

Next, to identify the role of synaptic zinc in PV-mediated inhibition to A1 L2/3 PNs, we performed similar experiments in PV-Cre mice (Fig. 9a,b). Consistent with previous reports (Lee et al. 2012; Seybold et al. 2015; Phillips and Hasenstaub 2016), we found that optogenetic activation of PVs reduced sound-evoked responses in L2/3 PNs (Fig. 9c). More specifically, PV activation reduced the sound-evoked responses of Α1 L2/3 PNs to BF sounds (Fig. 9c–e and Supplementary Fig. S6a,b) and reduced tuning bandwidth (Fig. 9f). PV-mediated inhibition was greater than SOM-mediated inhibition to PNs for BF sounds (compare Figs 8e and 9e; P = 6.1 e−10, Mann–Whitney rank-sum test) and had similar sharpening effects on the frequency tuning (compare Figs 8f and 9f; P = 0.47, Mann–Whitney rank-sum test). Infusion of ZX1 reduced the PV-mediated inhibition on sound-evoked responses of A1 L2/3 PNs to BF sounds (Fig. 9g–i and Supplementary Fig. S6c–e) but did not affect frequency tuning (Fig. 9j and Supplementary Fig. S6f). Consistent with the cell-specific effects of synaptic zinc in enhancing SOM- but not PV-mediated GABAergic neurotransmission, synaptic zinc is required for the SOM but not PV effects on A1 L2/3 PNs frequency tuning.

Discussion

Zinc Modulation of GABAergic Inhibition

Our findings show for the first time that synaptically released zinc enhances SOM mediated GABA_AR IPSCs in auditory cortex synapses. Previous electrophysiological recordings in cell culture studies report that zinc inhibits (Akaike et al. 1987; Westbrook and Mayer 1987; Yakushiji et al. 1987; Mayer and Vyklicky 1989; Smart and Constanti 1990; Celentano et al. 1991; Legendre and Westbrook 1991; Kilic et al. 1993), enhances (Smart and Constanti 1983, 1990; Smart et al. 1994; Qian et al. 1997), or has no effect on GABA_AR responses (Smart and Constanti 1982, 1990). Studies in brain slices from piriform cortex or hippocampus report that zinc does not affect GABA_AR IPSPs (Smart and Constanti 1983; Hori et al. 1987; Xie and Smart 1991). In these studies, the bulk (but not localized) application of exogenous zinc might not reflect physiological conditions. In hippocampal slices, zinc inhibited GABA_AR IPSCs evoked by stimulation of mossy fibers but not IPSCs evoked by stimulation of striatum radiatum interneurons (Ruiz et al. 2004). However, this study relied on CaEDTA, which is an extracellular zinc chelator with significantly slower kinetics compared to ZX1 (Anderson et al. 2015). These slower kinetics render CaEDTA a potentially inappropriate chelator for investigating the effects of fast, transient elevations of synaptic zinc on synaptic targets. This could explain the lack of effect of synaptic zinc in these studies. Moreover, in this study, TPEN was also employed, which is an intracellular zinc chelator, and thus inappropriate for investigating the effects of extracellular synaptic zinc. In the dentate gyrus, zinc reduced GABA release by inhibiting the presynaptic T-type calcium channels, and zinc chelation with TPEN increased the excitability of dentate interneurons (Grauert et al. 2014). These effects are different from our findings, which show lack of contribution of presynaptic mechanisms and no change in the intrinsic excitability of presynaptic interneurons after zinc’ chelation. Overall, the observed differences between our findings and previous studies on the effects of zinc on GABA_AR responses are likely due either to different effects of synaptic zinc in neurons from different brain areas and/or the usage of different zinc chelators. In our study, the advantages of paired recordings from synaptically coupled neurons and the use of ZnT3 KO mice and ZX1 provide strong evidence that synaptically released zinc enhances SOM GABAergic inhibition in auditory cortex synapses. Given the presence of synaptic zinc in all sensory cortices (McAllister and Dyck 2017), it is likely that this is a general finding for neocortical synapses, but future studies in other cortical areas are needed to validate this prediction.

The cell-specific effects of zinc in SOM-mediated inhibition are consistent with the seemingly higher levels of ZnT3 mRNA in SOMs compared to PVs (Paul et al. 2017). However, differential subunit GABA_ARs composition might also contribute to these cell-specific differences in zinc modulation. In fact, in cell culture experiments, α1-containing GABA_ARs are zinc insensitive but α5-containing GABA_ARs are zinc sensitive (Draguhn et al. 1990; Smart et al. 1991; Smart et al. 1994; Burgard et al. 1996). Given that PVs target synapses onto PN somata and proximal dendrites containing mostly α1 subunits, whereas SOMs target PN dendritic synapses containing mostly α5-containing GABA_ARs (Sperk et al. 1997; Sur et al. 1999; Wainwright et al. 2000; Ali and Thomson 2008; Rudolph and Mohler 2014; Speigel et al. 2017), the differential subunit composition of the GABA_ARs targeted by these interneurons might also contribute to the observed cell-specific effects of synaptic zinc. This notion is consistent with previous cortical studies showing that exogenous application of 100 μΜ of zinc did not affect IPSPs in PNs elicited by activation of multipolar fast-spiking interneurons, presumably PVs (Ali and Thomson 2008). Further experiments are needed for determining the precise molecular mechanism underlying SOΜ- but not PV-mediated inhibition by synaptic zinc.

Caveat on the Identification of CCal PNs in Our Studies

One caveat of our approach in identifying CCal PNs is that a small proportion of callosal axons are also immunoreactive for glutamic acid decarboxylase, and thus our approach might label CCal PNs and PVs (Rock et al. 2018). To address this issue, we recorded from morphologically defined CCal PNs according to their large cell bodies and distinct shape, compared to the smaller and round-shaped PVs. Moreover, all ten paired recordings between presynaptic SOMs and postsynaptic CCals showed ZX1 inhibition of uIPSCs (Fig. 1); and all twenty-six optogenetic experiments addressing the same connections (SOM to CCals, Figs 2 and 3) also showed ZX1 inhibition, strongly suggesting that SOM inputs to either CCal PNs or PVs are enhanced by zinc. Consistent with our conclusions, ZX1 did not affect Lev-IPSCs in any of the five optogenetic experiments addressing PV to CCal connections (Fig. 5). Together, even if we included a small number of CCal PVs in our analysis, our conclusions on the cell-specific effects of zinc on SOM- but not PV-mediated inhibition of PNs are valid.

Developmentally related Considerations

We used mice of two different age groups: a group of P29-35 mice for in vitro experiments; and a group of P42-64 mice for in vivo experiments. In mice (and rats), the development of the GABA_AR and the final GABA_AR subunit composition are completed by P21 and maintained through adulthood (Laurie et al. 1992; Fritschy et al. 1994; Fritschy and Mohler 1995; Tremere 2011; Le Magueresse and Monyer 2013; Fritschy and Panzanelli 2014). Moreover, developmental studies in rodents indicate that the maturation of synaptic, cellular and network properties are also completed by P29 (Metherate and Aramakis 1999; Oswald and Reyes 2008; Froemke and Jones 2011), concurrently with changes in A1 frequency tuning (Insanally et al. 2009; Dorrn et al. 2010). Finally, cortical inhibition is already cotuned with cortical excitation by P21 (Dorrn et al. 2010; Froemke and Jones 2011). Thus, the age groups we employed in our studies have adult composition of GABA_AR subunits and do not differ in their cellular, synaptic, and network properties.

The Role of Zinc in Sensory Cortical Processing

Our results complement and expand previous studies showing the inhibitory effect of SOMs and PVs on sound-evoked responses of A1 L2/3 PNs (Lee et al. 2012; Wilson et al. 2012; Cottam et al. 2013; Seybold et al. 2015; Phillips and Hasenstaub 2016). We found that the PV-mediated inhibition is greater than SOM-mediated inhibition to sounds with frequencies near BF. This also agrees with data supporting that PVs provide stronger inhibition to somata and proximal dendrites of PNs, compared to the SOMs that target the dendritic tufts of PNs (Pfeffer et al. 2013; Li et al. 2014). Moreover, consistent with previous reports, activation of either SOMs or PVs had similar sharpening effects on frequency tuning (Seybold et al. 2015).

In agreement with our previous studies, synaptic zinc increases the gain of the responses to sounds with frequencies near BF (Fig. 8d,h black bars; Fig. 9d,h black bars) and sharpens L2/3 frequency tuning (Supplementary Figs S5f,h and S6a,c) (Anderson et al. 2017; Kumar et al. 2019). Although it is known that these actions are associated with the cell-specific mechanisms of synaptic zinc on the gain modulation of PNs and interneurons (Anderson et al. 2017; Kumar et al. 2019), the precise synaptic mechanisms via which synaptic zinc mediates these effects are less understood. These mechanisms have been assumed to involve the well-known functions of synaptic zinc on glutamatergic synapses (Vergnano et al. 2014; Anderson et al. 2015; Kalappa et al. 2015), but our results, by establishing the previously unknown role of synaptic zinc on cortical GABAergic inhibitory neurotransmission, connect synaptic zinc signaling, GABAergic signaling and cortical sound processing.

Although the bulk application of ZX1 cannot differentiate cell (direct effects) from circuit effects, the increased SOM-mediated inhibition of sound-evoked responses of PNs in the presence versus the absence (ZX1 chelation) of zinc (Fig. 8e vs. i and Supplementary Fig. S5j) is consistent with the enhancing effect of zinc on direct SOM-mediated inhibition to PNs (Figs 1 and 2).

Currently, there are two major models to explain the tuning of auditory cortical neurons. In the “iceberg” model, where excitation and inhibition are balanced or cotuned, frequency tuning is achieved because only the strongest excitatory input drives the cell to spike threshold (Wehr and Zador 2003; Zhang et al. 2003; Moore and Wehr 2013). Consistent with the “iceberg” model, synaptic zinc by enhancing SOM-mediated inhibition, it likely drops excitation, and in turn narrows the response area of PNs (Fig. 8j and Supplementary Fig. S5k). According to the “lateral inhibition” model, cortical inhibition inhibits wider regions of cortical space through network mechanisms (Priebe and Ferster 2008; Kato et al. 2017). Previous studies report that SOMs might sharpen cortical tuning via lateral inhibition (Adesnik et al. 2012; Yavorska and Wehr 2016; Kato et al. 2017). Therefore, one additional potential interpretation of our results is that synaptic zinc signaling by boosting SOM GABAergic neurotransmission modulates A1 frequency tuning via enhancement of lateral inhibition (Kato et al. 2017). These models are not mutually exclusive, but they might be engaged during different temporal or spatial patterns of cortical activation (Levy and Reyes 2011; McGinley et al. 2015). Nevertheless, our study adds zinc signaling as a key cortical synaptic mechanism that shapes inhibition and cortical frequency tuning.

Chelation of synaptic zinc increases the sound-evoked responses of PVs in a non-NMDAR-dependent manner (Anderson et al. 2017; Kumar et al. 2019), suggesting a lack of synaptic zinc effect on the excitatory input to PVs. Because zinc does not affect the intrinsic excitability of PVs (Supplementary Fig. S3), we suggest that the inhibitory effect of synaptic zinc chelation on the direct SOM-mediated inhibition to PVs (Fig. 4) might explain the increased sound-evoked PV responses during zinc chelation.

PVs contribute to the balance of excitation/inhibition by providing direct feedforward inhibition in all cortical layers (Hasenstaub et al. 2005; Rudy et al. 2011; Tremblay et al. 2016). Sound-evoked activation of PNs is followed by recruitment of PV-mediated inhibition (Xue et al. 2014), resulting in reduced PN sound-evoked responses, which is consistent with our findings (Fig. 9d). Although our results show that synaptic zinc does not affect the direct PV-mediated IPSCs on PNs (Fig. 5), chelation of zinc caused a reduction in PV-mediated inhibition on PN responses (Supplementary Fig. S6e). This reduction might be explained, albeit not tested here, by the contribution of recurrent inhibition of PVs. Namely, during zinc chelation and PV photoactivation, the reduced SOM-mediated inhibition to PVs (Fig. 4), along with the robust photoactivation of PVs (Fig. 9), might favor recurrent PV-mediated inhibition, which, in turn, leads to decreased PV-mediated inhibition onto PNs.

Overall, our in vitro studies reveal a cell-specific effect of synaptic zinc on enhancing SOM but not PV GABAergic inhibition in AC L2/3. This cell-specific enhancement is consistent with our in vivo experiments showing a significant contribution of zinc signaling on SOM- but not PV-mediated sharpening of cortical A1 L2/3 frequency tuning. As such, our results advance our knowledge of cortical inhibition and create a new framework for approaching and interpreting cortical models and data.

Funding

National Institute of Health (R01-DC007905 to T.T.). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Notes

We thank Dr Ross Williamson for critical reading of the manuscript and helpful suggestions. Conflict of Interest: None declared.

Author Contributions

S.K., M.K., T.T. designed experiments. S.K. performed and analyzed in vitro experiments and M.K. performed and analyzed in vivo experiments. S.K., M.K. and T.T. wrote the manuscript. The authors declare that there are no competing interests.