Lateralized local circuit tuning in female mouse auditory cortex

Abstract

Most offspring are born helpless, requiring intense caregiving from parents especially during the first few days of neonatal life. For many species, infant cries are a primary signal used by parents to provide caregiving. Previously we and others documented how maternal left auditory cortex rapidly becomes sensitized to pup calls over hours of parental experience, enabled by oxytocin. The speed and robustness of this maternal plasticity suggests cortical pre-tuning or initial bias for pup call stimulus features. Here we examine the circuit basis of left-lateralized tuning to vocalization features with whole-cell recordings in brain slices. We found that layer 2/3 pyramidal cells of female left auditory cortex show selective suppression of inhibitory inputs with repeated stimulation at the fundamental pup call rate (inter-stimulus interval ~150 msec) in pup-naïve females and expanded with maternal experience. However, optogenetic stimulation of cortical inhibitory cells showed that inputs from somatostatin-positive and oxytocin-receptor-expressing interneurons were less suppressed at these rates. This suggested that disynaptic inhibition rather than monosynaptic depression was a major mechanism underlying pre-tuning of cortical excitatory neurons, confirmed with simulations. Thus cortical interneuron specializations can augment neuroplasticity mechanisms to ensure fast appropriate caregiving in response to infant cries.

1. Introduction

Parental animals must recognize and respond to distress calls and other cues produced by offspring, in order to adequately attend to the needs of children (Fleming et al., 1999; Kuroda et al., 2011, 2024; Rilling and Young, 2014; Kohl et al., 2017; Miyamichi, 2024). In many mammalian species including mice, pups are born helpless- they are unable to feed themselves and have poor thermoregulatory abilities. When pups are cold and/or isolated from nest and caregivers, they make ultrasonic cries in a frequency range of approximately 40–100 kHz with individual cries and vocal motifs produced over a limited range of inter-syllable intervals (Sewell, 1970; Noirot, 1972; Ehret, 2005; Castellucci et al., 2018; Coffey et al., 2019). Nulliparous (‘virgin’) animals usually avoid pups and can find these cries aversive (Schiavo et al., 2020; Autry et al., 2021; Mei et al., 2023), but with exposure and/or experience with pups, adults can become effective caretakers and respond appropriately to pup calls (Koch and Ehret, 1989; Marlin et al., 2015).

Many previous studies have described the responses of mouse auditory cortex to pup call sounds. These studies show that auditory cortex contains cells that selectively process pup call sounds or acoustic features, and that experienced parental females have more robust responses to pup calls than pup naïve nulliparous (‘virgin’) females (Ehret, 2005; Liu et al., 2006; Lin et al., 2013; Dunlap et al., 2020). We reported that cortical pup call responses are left lateralized, with left auditory cortex having more cells responding to pup call sounds than right auditory cortex, and left auditory cortex but not right auditory cortex in-activations disrupt pup retrieval (Marlin et al., 2015; Schiavo et al., 2020). This lateralization may be analogous to the left-lateralization of some aspects of conspecific vocalization processing in many species including speech processing in the human temporal lobe (Ehret, 1987; Poremba et al., 2004, Hickok and Poeppel, 2007; Wilson et al., 2015; Long et al., 2016). We identified a molecular correlate of left lateralization, showing that receptors for the neuropeptide oxytocin were expressed by inhibitory cells more in left than right female mouse auditory cortex (Marlin et al., 2015; Mitre et al., 2016). Oxytocin is a nine amino acid peptide made and released by neurons in the paraventricular hypothalamus, which projects to auditory cortex among many other regions of the central nervous system (Carter et al., 2001; Froemke and Young, 2021; Grinevich and Neumann, 2021; Carter, 2022; Son et al., 2022; Yukinaga et al., 2022; Lawson, 2024; Li et al., 2024; Miyamichi, 2024). Oxytocin can promote synaptic plasticity in target brain regions (Theodosis et al., 1986; Tomizawa et al., 2003; Fang et al., 2008; Marlin et al., 2015; Mitre et al., 2016; Nardou et al., 2019; Carcea et al., 2021; Osakada et al., 2024), and can affect emergence of neural and behavioral responses to pup calls and pup distress (Marlin et al., 2015; Schiavo et al., 2020; Watarai et al., 2020; Tsuneoka et al., 2022).

We recently discovered that even in pup-naïve virgin females, there was an initial sensitivity of many auditory cortical neurons to the fundamental inter-stimulus interval of pup call sounds. Pup calls are multi-syllabic vocalizations, with individual whistles or motifs separated by silent periods usually ranging between 100 and 400 msec (Ehret, 2005; Castellucci et al., 2018; Schiavo et al., 2020). While there is considerable variability in acoustics of individual distress calls (and this increases with postnatal age), the most common inter-stimulus-interval (ISI) for pup calls particularly at early ages is ~150–175 msec, which we refer to as the ‘prototypical’ pup call ISI (Schiavo et al., 2020). This temporal interval is the most important feature for the behavioral response of parents (more so than the ultrasonic frequency components); three or more syllables around the prototypical interval range are required for animals to recognize these sounds as distress calls (Ehret, 2005). We found that left auditory cortical neurons have a heightened response to stimuli presented at the prototypical ISI rate, even in brain slices made from naïve virgin females. This seemed due to a specific reduction of inhibitory postsynaptic currents (IPSCs) relative to excitatory postsynaptic currents (EPSCs), leading to an enhanced excitation-to-inhibition ratio which might promote spiking (Schiavo et al., 2020). This appeared to be a specialized form of short-term synaptic IPSC depression, but other synaptic and circuit mechanisms could also produce this effect. The goal of our current study was use whole-cell recordings and optogenetics in brain slices, combined with simulations of synaptic and spiking responses to determine the mechanisms and identity of inhibitory inputs most sensitive to 150 msec temporal intervals.

2. Materials and methods

2.1. Animals

All procedures were approved by the NYU Langone Institutional Animal Care and Use Committee in compliance with the US National Institutes for Health Guidelines for the Care and Use of Laboratory Animals. Animals were housed in fully-equipped facilities in the NYU Grossman School of Medicine Science Building (New York City). Pup-naïve virgin animals were housed in a separate room without mothers and pups to minimize exposure to pup cues and parenting before use. The facilities were operated by the NYU Division of Comparative Medicine. Mice were maintained on a normal 12-h light/dark cycle (dark cycle starts at 6 PM) and given food and water ad libitum. Mice used for this study were originally sourced from Jackson Laboratory (Bar Harbor, ME) and were either wildtype C56BL/6 J or Cre expressing mice directed with specific promoters: for somatostatin-positive (SST+) neurons, Sst^{tm2.1(cre)Zjh}/J; for PV+ neurons, B6.129P2-Pvalb^tm1cre(Arbr)/J; for vasoactive intestinal peptide-positive (VIP+) neurons, Vip^tm1(cre)Zjh/J; and for oxytocin-receptor-positive (OXTR+) neurons, OXTR-cDNA (HA)-Ires-Cre mice that were originally generated in the Nishimori lab in a 129 x C57BL/6 J background (Hidema et al., 2016) and backcrossed to C57BL/6 J. Mice were either inbred or crossed with B6.Cg-Gt(ROSA) 26Sor^{tm32(CAG-Cop4}*^{H134R/EYFP)Hze}/J to express channelrhodopsin-2 co-tagged with EYFP (ChR2-EYFP). Mice were aged postnatal day (P) 26 to P226 of both sexes.

2.2. Viral injections

For some animals, stereotaxic injections of adeno-associated virus encoding AAV1.EF1α.FLOX.ChR2.EYFP (Addgene) (ChR2-EYPF) were necessary. Animals were anesthetized with 1.5–2 % isoflurane and head fixed on a stereotaxic to make targeted injections of the primary auditory cortex (distance in mm from Bregma and brain surface, A/P: −2.54, M/L: 4.5, D/V: −0.3). Injections were made with a Nanoject III (Drummond Scientific) and a freshly-pulled micropipette to a ~20 μm diameter. Animals were allowed to recover and virus allowed to express for at least 2 weeks prior to sacrifice and experimentation.

2.3. Slice preparation

Acute brain slices were prepared from mice after deep anesthetization with 5 % isoflurane. Mice were then transcardially perfused with ice cold, oxygenated (95 % O2/ 5 % CO2), sucrose cutting buffer containing (in mM): 87 NaCl, 75 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 0.5 CaCl₂, 25 NaHCO₃, 1.3 ascorbic acid and 10 D-Glucose. After quick decapitation and brain extraction, coronal slices containing auditory cortex (250 μm thick) were made on a vibrating blade microtome (Leica VT1200s) while submerged in ice-cold and oxygenated sucrose cutting buffer. Slices were then transferred to an incubation chamber containing artificial cerebral spinal fluid (ACSF) consisting of (in mM): 124 NaCl, 2.5 KCl, 1.5 MgSO₄, 1.25 NaH₂PO₄, 2.5 CaCl₂ and 26 NaHCO₃. The chamber was heated (~35 °C) and oxygenated while the slices incubated for approximately 30 min before allowing to adjust to room temperature for at least an additional 30 minutes.

2.4. Whole-cell recordings

For recordings, methods were similar to those we have previously performed (Field et al., 2020; Schiavo et al., 2020). Brain slices were transferred to a holding chamber that was superfused (2–3 mL/min) with oxygenated and heated ACSF. Patch pipettes were made with borosilicate glass pulled to a resistance of 5.5–7.5 MΩ on a Flaming/Brown P-1000 micropipette puller (Sutter Instruments) and filled with (in mM): 127 K-gluconate, 8 KCl, 10 phosphocreatine, 10 HEPES, 4 Mg-ATP, 0.3 Na-GTP. Recordings were made using a Multiclamp 200B amplifier (Molecular Devices), filtered at 2 kHz, digitized at 10 kHz, and acquired with Clampex 10.7 (Molecular Devices). Whole-cell recordings were made from pyramidal neurons in layer 2/3. Input resistance, action potential shape and rectification ratio were used to assess pyramidal neuron identity. A bipolar stimulating electrode pulled from theta glass was placed in superficial layer 2, proximal to the recorded neuron. Electrical stimulus currents (pulse duration 0.5 ms, amplitudes of 3–700 μA) were generated by an A365 stimulus isolator (WPI) triggered by a TTL pulse from the acquisition software. Optical stimulation was done with a PLS series 5500 K Cool White LED (Mightex), projected through the objective, and filtered (Semrock) to ChR2 activation wavelength. Data were analyzed using custom written code on Matlab (Mathworks) or Prism (GraphPad).

In order to assess short-term plasticity of EPSCs and IPSCs, voltage-clamp recordings were acquired while a train of 5 pulses was evoked either electrically or optically at ISIs of 50, 150, or 550 msec (bin±25 msec). Subsequent 5-pulse trains were repeated with intervals of 1 minute in between stimulus presentation to prevent long-term changes from occurring to excitation and inhibition. Our study of spike-timing-dependent plasticity of electrically and optically-evoked specific inhibitory inputs showed that baseline synaptic strength was stable for stimulus repeats at this low rate (Song et al., 2022). Patch pipettes (3–8 MΩ) were filled with the following intracellular solution (in mM): 130 Cs-methanesulfonate, 1 QX-314, 4 TEA-Cl, 0.5 BAPTA, 4 MgATP, 0.3 Na-GTP, 10 phosphocreatine, 10 HEPES, pH 7.2). EPSCs were acquired at −70 mv, while IPSCs were acquired from −40–0 mV. The peak amplitude of evoked EPSCs and IPSCs were measured and normalized to the event evoked by the first stimulation on each trial (S_n / S₁).

2.5. Simulations

We used a straightforward conductance-based integrate-and-fire model neuron to simulate the effects of excitatory and inhibitory inputs on spike generation (Dorrn et al., 2010; Marlin et al., 2015). On each of five successive trials, one EPSC and one IPSC were chosen from the responses to stimulus pulses S1-S5. Excitatory and inhibitory synaptic conductances ge and gi were computed from currents (Dorrn et al., 2010), and were initially both set to 5.0 nS for S1 assuming input resistance R_i of 100 MΩ. This was divided by 10 for both excitation and inhibition, representing an input population of 10 excitatory and 10 inhibitory inputs, each with latency 10 ± 3 msec (for EPSCs; SD) and 13 ± 4 msec (for IPSCs; SD). Membrane voltage was computed as: ${\tau }_m\frac{\mathit{\text{dV}}}{\mathit{\text{dt}}}=V_{\mathit{\text{rest}}}-V+g_e(t)(E_e-V)+g_i(t)(E_i-V)$, with τ_m: 10 msec, resting membrane potential V_rest: −60 mV, excitatory reversal potential E_e: 0 mV, and inhibitory reversal potential E_i: −70 mV. A spike was evoked in the postsynaptic neuron if the membrane voltage reached threshold of −45 mV, at which point the membrane potential was set to −65 mV in the next time step over a total of 100 msec of simulated time for each of the five EPSC/IPSC events per trial. Spike rates were determined over 10 trials. For simulations of inhibitory short-term depression, EPSCs and IPSCs for stimulus pulses S2-S5 were scaled by the values for EPSC and IPSC suppression of OXTR+ inputs from Fig. 3d. For simulations of disynaptic inhibition, EPSCs were scaled as for the synaptic depression model, and we ran another set of similar simulations using the change in spiking S_n / S₁ of a putative intermediate interneuron as the scaling factor for IPSCs rather than the data from Fig. 3d.

Fig. 3.

Inputs from SST+ and OXTR+ cortical interneurons are not selectively suppressed at 105 msec ISIs. a, Optogenetic stimulation of PV+ interneurons to evoke IPSCs with extracellular stimulation to evoke EPSCs onto layer 2/3 pyramidal cells in naïve female left auditory cortex. Similar inhibitory and excitatory suppression for all ISIs (IPSCs and EPSCs not significantly different for each stimulus pulse; n = 8, p > 0.1 for each PSC pair across ISIs, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons). b, Optogenetic VIP+ IPSCs. Similar inhibitory and excitatory suppression for all ISIs (n = 11, p > 0.06 for each PSC pair across ISIs). c, Optogenetic SST+ IPSCs. Less suppression of IPSCs at 150 and 50 msec ISIs (n = 9, p < 0.02 for S2 at 50 msec and S2,S3 at 150 msec) but not 550 msec ISIs (p > 0.4 for each PSC pair). d, Optogenetic OXTR+ IPSCs. Less suppression of IPSCs at 150 msec ISI (n = 11, p < 0.05 for S2,S3) but not 50 msec or 550 msec ISIs (p > 0.0.5 for each PSC pair).

3. Results

3.1. Suppression of inhibition at prototypical pup call rates in left female auditory cortex

We made whole-cell voltage-clamp recordings from layer 2/3 pyramidal cells in brain slices of adult mouse auditory cortex (Fig. 1a), and measured the synaptic responses to repeated patterns of 4–5 stimulus pulses (Fig. 1b, left) around the prototypical pup call rate (150 msec ISI), faster stimuli (50 msec ISI) or slower stimuli (550 msec ISI) outside the usual distribution of ISIs maternal animals hear from pups (Schiavo et al., 2020). Initially we made brain slices from left auditory cortex of pup-naïve virgin females, using electrical stimulation in layer 2 to evoke EPSCs measured at −70 mV and IPSCs measured at −40–0 mV (Field et al., 2020). As expected, EPSCs tended to adapt with repeated stimulus pulse number, likely due to short-term synaptic depression (Fig. 1b, right). IPSCs also adapted and had lower amplitudes later in the pulse trains, but showed a different profile relative to excitation depending on the ISIs.

Fig. 1.

Inhibitory inputs adapt faster than excitatory inputs at the pup call prototypical tempo in left female mouse auditory cortex brain slices. a, Schematic of local cortical circuitry. Whole-cell voltage-clamp recordings were made from layer 2/3 (L2/3) pyramidal cells; excitatory and inhibitory inputs from various sources were evoked by electrical stimulation and measured at −70 mV or 0 mV. b, Experimental design. Left, electrically stimulation with pulse trains of 4–5 events (S1-S5) at 50, 150, or 550 ms inter-stimulus intervals (ISIs). Right, example evoked EPSCs. c, Example recording. Left, example trains of EPSCs and IPSCs evoked at 150 ms ISIs. Middle, average EPSC (blue) and IPSC (red) amplitudes for each of the five stimulus pulses at 150 ms ISI. Right, normalized PSC amplitudes for 150 vs 550 ms ISIs. PSC magnitude normalized to amplitude of stimulus 1. Note more suppressed inhibition relative to excitation at 150 ms ISI. d, Top, average PSC amplitudes for cells recorded from left auditory cortex of naïve virgin females showed more suppression for inhibition than excitation at 150 ms ISI (IPSCs lower than EPSCs at S2, S3, and S4; n = 8, p < 0.006 for each PSC pair, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons), but not 550 ms (IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.07) or 50 ms (IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.1). Bottom, cells recorded from left auditory cortex of experienced maternal females showed more inhibitory than excitatory suppression at ISIs of 150 ms (IPSCs lower than EPSCs at S3 and S4; n = 8, p < 0.04 for both PSC pairs) and 550 ms (IPSCs lower than EPSCs at S2 and S4, p < 0.03 for both PSC pairs), but not 50 ms (IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.2).

An example recording is shown in Fig. 1c. EPSCs and IPSCs both decreased in amplitude from stimulus pulses S1 to S5. We quantified the degree of suppression by normalizing the amplitudes evoked by stimuli S2 to S5 by the amplitude of S1. While the suppression of EPSCs was fairly similar for 150 and 550 msec ISIs, the degree of suppression for inhibition was more severe for 150 msec ISIs (Fig. 1c, right). The relative suppression of PSCs was comparable between excitation and inhibition for 550 msec ISIs, but selective suppression of IPSCs relative to EPSCs was obvious for 150 msec ISIs. This is similar to our previous observations for recordings in slices from pup-naïve females (Schiavo et al., 2020).

We made recordings from eight pyramidal cells in slices of pup-naïve females and another eight recordings in slices from experienced females that had been co-housed with a mother and pups and behaviorally confirmed for successful pup retrieval. At 150 msec ISIs, cells from pup-naïve animals and experienced retrieving animals showed relatively less inhibition than excitation particularly for the final stimuli S3 and S4 in the pulse train (Fig. 1d, left; top, cells from pup-naïve virgin females, IPSCs lower than EPSCs at S2, S3, and S4; n = 8, p < 0.006 for each PSC pair, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons; bottom, cells from experienced maternal females, IPSCs lower than EPSCs at S3 and S4; n = 8, p < 0.04 for both PSC pairs). 550 msec ISI pup calls are rarely but occasionally experienced by parental animals in the homecage (Schiavo et al., 2020). Correspondingly, relative magnitudes of EPSCs and IPSCs showed similar adaptation for naïve virgins at 550 msec ISIs (Fig. 1d, middle top; IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.07), although some IPSCs were relatively reduced in experienced females at 550 msec ISIs (Fig. 1d, middle bottom; IPSCs lower than EPSCs at S2 and S4, p < 0.03 for both PSC pairs). 50 msec ISIs are quite uncommon and at these faster rates, EPSCs and IPSCs were similarly reduced in relative amplitude (Fig. 1d, right; top, cells from virgin females, IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.1; bottom, cells from experienced females, IPSCs and EPSCs not significantly different for each stimulus pulse, p > 0.2). This confirms and extends our earlier finding of similar temporal interval tuning in vitro in naïve female left auditory cortex for prototype-call ISIs (Schiavo et al., 2020), and also shows that experience with pups affects local circuit dynamics by extending this mechanism for other experienced pup call intervals such as 550 msec ISIs in some cases.

3.2. Similar suppression of excitation and inhibition for males and right auditory cortex

We next asked how specific this temporal interval tuning was for female left auditory cortex, examining synaptic responses in brain slices from male mice or female right auditory cortex. We focused on responses to 150 msec ISIs, as this seemed to be the most sensitive interval in female left auditory cortex. In male left auditory cortex, there was no selective suppression at 150 msec ISIs (Fig. 2a; IPSCs and EPSCs not significantly different for each stimulus pulse; n = 10, p > 0.5 for each response pair, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons).

Fig. 2.

No difference in excitatory vs inhibitory suppression at 150 msec ISIs in male cortex or female right cortex. a, Similar inhibitory and excitatory suppression for ISIs of 150 ms for cells recorded from male left auditory cortex (IPSCs and EPSCs not significantly different for each stimulus pulse; n = 10, p > 0.5 for each response pair, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons). b, Male right auditory cortex (n = 6, p > 0.3 for each PSC pair). c, Naïve female right auditory cortex (n = 6, p > 0.08 for each PSC pair). d, Maternal female right auditory cortex (n = 7, p > 0.7 for each PSC pair).

This was also the case for cells recorded from right auditory cortex of males (Fig. 2b; n = 6, p > 0.3 for each PSC pair), naïve females (Fig. 2c; n = 6, p > 0.08 for each PSC pair), or experienced maternal females (Fig. 2d; n = 7, p > 0.7 for each PSC pair). This indicates that the circuit properties that might convey an initial sensitivity for pup call sounds is both left-lateralized and sexually dimorphic, perhaps because of the predominant female presence and importance for providing care and nutrition to pups at the very earliest postpartum period.

3.3. Optogenetic stimulation of cortical inhibitory inputs did not reveal selective suppression at 150 msec ISIs

We next aimed to determine what local circuit elements might provide this specialized inhibitory suppression at the prototypical pup call stimulus rate of ~150 msec ISIs. We hypothesized that a subpopulation of inhibitory interneurons might have stronger short-term depression at that rate, whereas other inhibitory cell types might not. Many studies have shown that one factor contributing to the diversity of cortical inhibitory cell types is differential short-term plasticity with some interneurons having facilitating synapses and others exhibiting more short-term depression (Takesian et al., 2010; Oswald and Reyes, 2011; Ma et al., 2012; Seay et al., 2020).

Three of the main cortical interneuron types are PV+ , VIP+ , and SST+ (Rudy et al., 2011). To specifically study synaptic transmission from these different cell types (Fig. 1a), we virally expressed Cre-dependent channelrhodopsin-2 in the left auditory cortex of pup-naïve virgin female transgenic Cre mice (PV-Cre, VIP-Cre, SST-Cre, or OXTR-Cre). We made whole-cell voltage-clamp recordings from slices of these animals similar as before, also using electrical stimulation to evoke EPSCs but optogenetically evoking IPSCs from a given interneuron type via blue light stimulation (Chiu et al., 2018; Valtcheva et al., 2023).

In contrast to the results of non-specific electrically-evoked IPSCs shown in Fig. 1, none of these inhibitory inputs showed selective suppression at 150 msec ISIs. For PV+ and VIP+ inputs, the amount of suppression was comparable between excitation and inhibition at all ISIs (Fig. 3a, PV+ inputs, IPSCs and EPSCs not significantly different for each stimulus pulse; n = 8, p > 0.1 for each PSC pair across ISIs, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons; Fig. 3b, VIP+ inputs, n = 11, p > 0.06 for each PSC pair across ISIs). However, for SST+ inputs, we instead observed less adaptation or even facilitation of evoked IPSCs, particularly for later pulses in the five-pulse train for 150 msec ISIs (Fig. 3c, SST+ inputs, n = 9; less suppression of IPSCs relative to EPSCs at 150 and 50 msec ISIs, p < 0.02 for S2 at 50 msec and S2,S3 at 150 msec; but not 550 msec ISIs, p > 0.4 for each PSC pair). Thus contrary to our expectations, none of the major inhibitory cell types seemed to provide inputs that selectively depressed during 150 msec stimulus trains.

We wondered if, instead, a different subset of interneurons might underlie sensitivity to prototypical pup call rates. A subset of cortical interneurons, largely SST+ or PV+ , express oxytocin receptors (Nakajima et al., 2014; Marlin et al., 2015; Yao et al., 2021), and oxytocin directly depolarizes interneurons in cortex and hippocampus, leading to increased spontaneous and reduced evoked inhibitory transmission (Owen et al., 2013; Marlin et al., 2015; Tirko et al., 2018). Given the importance of oxytocin in promoting onset of maternal behaviors and enhancing the sensitivity of cortical neurons to the statistics of pup call sounds (Marlin et al., 2015; Schiavo et al., 2020; Miyamichi, 2024), we asked if this subpopulation of interneuron might be responsible. We expressed channelrhopsin-2 in OXTR-Cre mice and optogenetically evoked IPSCs from OXTR+ interneurons onto layer 2/3 pyramidal cells. We observed less inhibitory suppression specifically at 150 msec ISIs (Fig. 3d, OXTR+ inputs, n = 11; less suppression of IPSCs at 150 msec ISI, p < 0.05 for S2,S3; but not 50 msec or 550 msec ISIs, p > 0.0.5 for each PSC pair).

3.4. Simulations suggest disynaptic inhibition leads to temporal interval tuning for prototypical pup call rates

These experimental results showed that optogenetic stimulation of specific interneuron subtypes (including OXTR+ cells) showed different responses to repetitive stimulation compared to electrical stimulation, which might be expected to activate the local circuit more generally. As the results of optogenetic stimulation did not seem to reveal how cortical neurons have heightened responses to 150 msec ISI trains, we designed a straightforward simulation to test different hypotheses about how excitatory and inhibitory inputs evoked at different rates could be integrated to produce enhanced spiking around the prototype rate but not at faster or slower rates (Fig. 4a).

Fig. 4.

Two models of enhanced responses to 150 ms ISI pup call stimuli from simulations of excitatory-inhibitory integration. a, At 150 msec ISIs, IPSCs adapt more than EPSCs, leading to more spikes due to an increase in excitatory-inhibitory ratio with increasing pulse number. But at slower or faster ISIs, EPSCs and IPSCs adapt at similar rates, leading to general reduction of spiking. b, Two mechanistic models for enhanced suppression at 150 msec ISIs: 1) direct short-term IPSC depression (top), or 2) disinhibition via enhanced short-term facilitation from some inhibitory cells onto other inhibitory cells (bottom). Left, schematic. Right, result of simulations (10 trials per stimulus pulse for each ISI and model). Shown is the simulated number of spikes relative to S1 with increasing stimulus number. For direct short-term depression model, spiking decreases due to lower excitation over time; inconsistent with experimental data in vivo. For disynaptic disinhibition model, spiking stays high with stimulus number, more consistent with experimental data.

We used a conductance-based integrate-and-fire model postsynaptic neuron, computing the membrane potential and number of evoked spikes from the simulated currents of each stimulus pulse S1-S5. We compared the predicted numbers of spikes from two different formulations of the model. In one case (‘hypothesis 1’, direct monosynaptic depression), the model postsynaptic neuron received excitatory and inhibitory inputs directly with no other intervening circuit elements. In the other case (‘hypothesis 2’, disynaptic inhibition), the model neuron received inhibitory inputs that were themselves scaled by the measured change in PSC amplitude from stimulus pulses S1-S5. As most auditory cortical neurons responding to pup calls tend to respond late in the stimulus period (Marlin et al., 2015), in agreement with behavioral findings that several repetitions of pup call syllables are required for dams to seek out pups (Ehret, 2005), we compared the predictions of the direct depression vs disynaptic inhibition models on the number of spikes evoked by pulses S3, S4, and S5 (Fig. 4b, left). Given the presumed involvement of oxytocin signaling in auditory cortex, we used results from Fig. 3d on electrically-evoked excitatory events and OXTR+ inhibitory events in both sets of simulations.

In both cases, excitatory inputs were modeled similarly- we simulated 10 inputs that each had the same connection strength (0.5 nS assuming a model neuron input resistance of 100 MΩ), and latencies of 10 ± 3 msec (mean±SD). Differential arrival times provided trial-by-trial variability in terms of excitatory integration and postsynaptic spiking while allowing us to control the impact of changing synaptic strengths in the model. We simulated responses to each of the pulses S1-S5 on each trial. Excitatory input weights were scaled by the measured reduction in EPSC amplitude from Fig. 3d (e.g., for 150 msec ISIs, excitatory input strength at S1 was 0.5 nS/input, S2 was 0.42 nS/input, S3 was 0.39 nS/input, S4 was 0.37 nS/input, and S5 was 0.37 nS/input).

In the first model (direct monosynaptic inhibitory depression), inhibitory inputs were modeled similar to excitatory inputs: 10 inhibitory connections each with initial connection strength of 0.5 nS for S1 and scaled according to the results of Fig. 3d (S2 was 0.48 nS/input, S3 was 0.44 nS/input, S4 was 0.39 nS/input, and S5 was 0.36 nS/input), with arrival times of each input of 13 ± 4 msec (mean±SD). We then computed the number of spikes evoked by each stimulus pulse S1-S5, for 10 trials each for ISIs of 150, 550, and 50 msec. As expected, the most spikes were evoked by S1 (1.9 ± 0.1 spikes/trial), with fewer spikes at S2-S5 for all temporal intervals (Fig. 4b, upper right; solid line, 150 msec ISI; dashed line, 550 msec ISI; dotted line, 50 msec ISI). In general, EPSCs depressed substantially at 50 msec ISIs, whereas at 550 msec ISIs IPSCs were only mildly depressed; consequentially in either case, spiking was equivalently lower across ISIs at later stimulus pulses. These simulations show that direct depression of excitatory and inhibitory inputs is unlikely to produce the higher spike responses observed specifically at 150 msec ISIs.

We then considered an alternative model with disynaptic inhibition, a major motif of cortical circuits (Fishell and Kepecs, 2020). Here the excitatory inputs were simulated as in the first model above, but inhibitory inputs were simulated differently. We assumed OXTR+ interneurons made connections onto an intermediate interneuron type (of unspecified identity); this intermediate interneuron node received 10 excitatory and 10 inhibitory (OXTR+) inputs, and made 10 connections onto the final postsynaptic cell of interest (each with weight of 0.5 nS). We then computed the number of spikes produced by the intermediate interneuron node (essentially the same as model 1), and used the reduction in spikes for stimulus pulses S2-S5 to reduce the inhibitory output connections onto the final postsynaptic cell (e.g., for one run of the simulation with 150 msec ISIs, inhibitory input strength at S1 was 0.5 nS/input, S2 was 0.34 nS/input, S3 was 0.34 nS/input, S4 was 0.32 nS/input, and S5 was 0.29 nS/input). This disynaptic disinhibition amplified this difference between the 150 msec ISI stimuli and the 50 or 550 msec ISI stimuli. Consequentially, more spikes were produced in the postsynaptic cell especially at later pulses in the trains for 150 msec ISI stimuli but not 550 or 50 msec ISI stimuli (Fig. 4b, lower right; more spikes at S2-S5 for model 2 than model 1 for 150 msec ISIs, p < 0.03, Student’s paired two-tailed t-test with Benjamini-Hochberg correction for multiple comparisons; for 550 or 50 msec ISIs, p > 0.1 across all stimulus pulses for model 2 vs model 1). This indicates that inhibitory circuits in auditory cortex can have specializations that distinguish between different temporal patterns of stimuli, which might be advantageous for detecting pup distress call sounds by new parents.

4. Discussion

Sound processing across species is not uniform in time but highly dependent on context, features, and experience (Froemke and Schreiner, 2015; Kudo et al., 2020; Chen and de Hoz, 2023). Here we examined potential synaptic and circuit mechanisms that might provide a sensitivity or pre-experiential basis for temporal interval tuning we previously documented in the female mouse left auditory cortex for pup distress calls (Schiavo et al., 2020). At the most common (i.e., prototypical) call rate of ~150 msec ISIs, auditory cortical neurons show a peak of tuning relative to natural or synthesized calls presented at faster or slower rates. We showed here that even in brain slices- in absence of longer-range inputs, subcortical processing, and other circuit elements in vivo- an analogue of this temporal interval tuning could be observed in the synaptic inputs onto layer 2/3 pyramidal neurons. A stronger suppression of inhibition than excitation was observed in slices from left auditory cortex of both pup-naïve nulliparous females and experienced maternal females, with experienced females also showing more inhibitory suppression at the slower temporal interval as well.

Statistical learning of experienced pup call distributions likely results from inhibitory plasticity in the cortex, to coordinate excitatory-inhibitory tuning for sensory features important for behavior (Froemke and Martins, 2011). Less inhibition should boost spike responses, and thus perhaps left-lateralization of this feature could help prevent over-excitability of the cortical network or an exaggerated (and perhaps negative) parental response to infant cries. Differences between male and female left auditory cortex do not suggest that male mice are neglectful or poor fathers; many strains of mice show rapid onset of paternal behaviors after mating or experience with pups (Tachikawa et al., 2013; Wu et al., 2014; Mitre et al., 2017). Instead we hypothesize that the specialized circuit mechanism for boosting responses to prototypical calls might function to shape maternal behavior at the earliest timepoints when care is most critical, but with longer periods of experience with pups, paternal males also can respond appropriately to pup calls.

The two models of specialized internal tuning we examined here involved mechanisms of short-term synaptic depression and disynaptic inhibition (also referred to as disinhibition), but are not mutually exclusive. The model of disynaptic inhibition depends on short-term plasticity, and our simulations show that this intervening interneuron population might amplify smaller biases to promote a clearer distinction between different input patterns. Previous work from our group and others have shown how forms of excitatory and inhibitory plasticity at a range of timescales can reduce noise in inputs patterns and improve sensory perception for detection or recognition of specific stimuli (Varela et al., 1997; Dobrunz and Stevens, 1999; Froemke and Schreiner, 2015; Phillips et al., 2017; Insanally et al., 2019, 2024; Nocon et al., 2023).

One caveat to our simulation results is that we assumed similar dynamics of excitatory and inhibitory inputs onto the intermediate interneuron as we measured for layer 2/3 pyramidal cells. Inhibitory cells can have different forms of short-term plasticity, and separate postsynaptic contacts from the same presynaptic inhibitory neurons can have separate profiles of short-term depression and facilitation (Reyes et al., 1998; Ma et al., 2012; Blackman et al., 2013; Seay et al., 2020; Campagnola et al., 2022). It was not our aim here to exhaustively survey all inhibitory connections in female left auditory cortex, but our results make predictions for what synaptic motifs might lead to enhanced cortical responses to certain temporal features or prototypical pup call sounds. We focused on adaptation of IPSCs due to the heterogeneity of cortical inhibitory cell types, with distinct patterns of short-term synaptic plasticity and inhibitory-inhibitory cell connections in some cases. While there may be subsets of excitatory cells and synapses, there is as-of-yet no clear markers for different subtypes that might be more or less important for vocalization processing. In the absence of an experimentally-tractable system for parsing different excitatory input and output cell types here, instead we focused on major inhibitory cell types. In addition, we and others have found that the main cell types directly modulated by oxytocin in the mouse cortex are SST+ and PV+ interneurons (Nakajima et al., 2014; Marlin et al., 2015). Thus it was natural to ask how these different input types might contribute to overall inhibitory adaptation we observed from non-specific electrical stimulation.

We quantified the relative differences between EPSCs and IPSCs, as it was not necessarily obvious how these patterns of synaptic inputs might support improved detection and recognition of pup calls- necessarily a function of the spiking output of auditory cortex rather than synaptic input. This motivated the simulations of synaptic integration and spike generation we performed here, to ask how depressing excitatory synapses can support improved sensory perception. The modeling revealed that when excitation remains at least high enough to occasionally evoke some spikes, if OXTR+ interneurons synapse directly onto pyramidal cells, spiking responses were not sustained over all five simulated pup call sounds. But if OXTR+ interneurons produced a disynaptic inhibition, then spike responses remained high at 150 msec ISIs. This also motivated our analysis of IPSCs relative to EPSCs rather than just comparing EPSCs across stimulus rates or IPSCs across stimulus rates.

It also remains to be determined if these mechanisms apply in vivo, where spontaneous activity and complexities of local and long-range circuit dynamics might provide additional means for enhancing responses to specific temporally-modulated stimuli (de Hoz and McAlpine, 2024). There is some previous evidence that mechanisms of short-term suppression are specific to cortical processing to alter responses to repetitive stimuli (Wehr and Zador, 2005; Froemke and Schreiner, 2015). Functionally, our results indicate that inhibitory adaptive processes might help first-time parental animals perceive these calls by either lowering threshold for detection and/or recognizing the significance of these specific vocalizations relative to other sounds or calls with similar statistics, e.g., male mouse vocalizations (Liu et al., 2003; Schiavo and Froemke, 2019). Parental behaviors in many species including mice per se are unlikely to be fully innate and require some degree of experience or learning, given the complexity and need for highly-context dependent performance, with many first-time or even multiparous animals failing to provide adequate pup care (Carcea et al., 2021; Schuster et al., 2023). An initial bias at the level of sensory perception and/or the affective components of infant cries might greatly accelerate mechanisms of plasticity in neural circuits of new parents (Zador, 2019), which could be critical for ensuring proper behaviors rapidly emerge to take care of helpless neonates within minutes to hours after parturition.