Altered synaptic plasticity of the longitudinal dentate gyrus network in noise-induced anxiety.

Anxiety is characteristic comorbidity of noise-induced hearing loss (NIHL), which causes physiological changes within the dentate gyrus (DG), a subfield of the hippocampus that modulates anxiety. However, which DG circuit underlies hearing loss-induced anxiety remains unknown. We utilize an NIHL mouse model to investigate short- and long-term synaptic plasticity in DG networks. The recently discovered longitudinal DG-DG network is a collateral of DG neurons synaptically connected with neighboring DG neurons and displays robust synaptic efficacy and plasticity. Furthermore, animals with NIHL demonstrate increased anxiety-like behaviors similar to a response to chronic restraint stress. These behaviors are concurrent with enhanced synaptic responsiveness and suppressed short- and long-term synaptic plasticity in the longitudinal DG-DG network but not in the transverse DG-CA3 connection. These findings suggest that DG-related anxiety is typified by synaptic alteration in the longitudinal DG-DG network.

Introduction

Hearing loss is a common affliction in modern society and occurs in over 5% of the world’s population (430 million). It is often caused by exposure to loud noises during daily activities such as work and recreation. The prevalence of hearing loss is expected to increase to 700 million people in the next 30 years, in turn increasing the related economic burden and depreciating the quality of life (Nelson et al., 2005). Noise-induced hearing loss (NIHL) can develop into hearing disorders, such as tinnitus, hyperacusis, and sensory overload. It can predispose individuals to psychiatric disorders, such as cognitive decline, dementia, depression, anxiety, and psychoses (Bhatt et al., 2017; Deal et al., 2017; Blazer and Tucci, 2019; Yang and Bao, 2013).

The dentate gyrus (DG) is a hippocampal subfield that modulates anxiety levels in response to environmental stresses. Neurogenesis and the neuronal activity of DG granule cells (DGGCs), the principal neurons in the DG, are closely correlated with stress-induced anxiety (Anacker et al., 2018; Hill et al., 2015; Kheirbek et al., 2013; Koch et al., 2020; Weeden et al., 2015). In particular, NIHL is associated with increased anxiety and impaired social interaction (Guitton, 2009; Lauer et al., 2018; Zheng et al., 2011). DG has an immense influence on these types of behaviors (Kheirbek et al., 2013; Leung et al., 2018; Zou et al., 2016). Indeed, hearing loss induces oxidative stress, tau phosphorylation, aberrant neurotransmitter receptor expression, and impaired neurogenesis and decreases the dendritic complexity of DGGCs (Nadhimi and Llano, 2021). However, it is unknown how the DG transmits incoming signals to neural networks to affect cognition subsequent to hearing loss.

The DG is an initial processing region in the cortico-hippocampal network that receives information from the second layer of the entorhinal cortex (EC) (ECII-DG-CA3 or CA2-CA1-subiculum) (Blaabjerg and Zimmer, 2007). The DG has three distinct layers: molecular, granule cell, and polymorphic (including the adult-born granule cell layer and the dentate hilus) (Amaral et al., 2007). The molecular layer is cell-sparse and mostly comprises dendrites of DGGCs. It is divided into three parts: inner, middle, and outer. The somata of granule cells are densely packed within the granule cell layer. Neurons in the molecular and granule cell layers receive inputs from the EC and dentate hilus and then send signals to the dentate hilus and CA3 and CA2 regions, respectively (Kohara et al., 2014; Blaabjerg and Zimmer, 2007; Swanson et al., 1978). In our previous study, we found that primary axons from the DG project transversely (in-lamellar) to the hilus and/or CA3 and bifurcate longitudinally (trans-lamellar) towards the molecular layer, targeting adjacent DGGCs (Choi et al., 2021). This creates diverse DG axonal projection patterns that contain granule and mossy cells (Lømo, 2009; Claiborne et al., 1986; Scharfman and Pierce, 2012; Jonas and Lisman, 2014; Choi et al., 2021), as further evidenced by the longitudinal spread of electrical signals along the dorsoventral axis of the hippocampus (Pare and Llinas, 1994; Lømo, 2009). Like CA1-CA1 longitudinal axons, DG-DG longitudinal axons show more varicosities and smaller axon widths than DG-CA3 transverse axons (Choi et al., 2021; Yang et al., 2014a).

This study aimed to investigate the effects of NIHL on the DG network along two directions: the transverse DG-CA3 and the longitudinal DG-DG projections. We found that mice that experienced NIHL displayed abnormal anxiety-like behaviors, which was associated with altered synaptic properties in the DG-DG network but not in the DG-CA3 network. Furthermore, in a chronic stress-induced anxiety animal model, we observed alterations of the short- and long-term synaptic properties of the DG-DG network, similar to the effects of NIHL. Our findings suggest that alterations of the DG-DG network, but not of the DG-CA3 network, can predict anxiety-like behaviors.

Results

NIHL enhances anxious behaviors

Unilateral hearing loss caused by exposure to loud noise was chosen to minimize the disruption of hearing-dependent normal behaviors. First, the auditory brainstem response (ABR) was recorded to check hearing ability before and after noise exposure (Figure 1A). The NIHL C57BL/6J mice showed elevated hearing thresholds in the noise-exposed ears in response to tones at each testing frequency (4 kHz, 8 kHz, 16 kHz, and 32 kHz) at 7 days after noise exposure (repeated-measures two-way ANOVA, pre vs. day 7, F(1,50) = 196.979, p < 0.0001). Unlike the NIHLs, the sham control mice (under anesthesia with no noise trauma) showed no threshold shifts (repeated-measures two-way ANOVA, pre vs. day 7, F(1,36) = 1.32, p = 0.285, see Table S1 for detail) (Figure 1B). These results indicate that our NIHL animal model experienced a severe degree of hearing loss.

Figure 1

NIHL enhances anxious behaviors

(A) Timeline of the experimental procedure.

(B) Hearing threshold (dB sound pressure level [SPL]) at different frequency tones, as measured by the auditory brainstem response (ABR) recorded during pre- and post-noise (D7) exposure. Noise exposure elevates the hearing threshold.

(C) The total number of entries and the latency to the first entry into each arm in the Y-maze forced alteration task. NIHL significantly increases the latency to the first entry to the novel arm.

(D) The total distance traveled in an open field arena and the exploration time with the novel or familiar object during object recognition. Both the control and NIHL have novel object preference over familiar one.

(E) The total area moved and the time (s) ratio of the center zone to the outer zone in the open field test. The ratio is reduced in NIHL. Individual data points have been eliminated for clearer visibility.

(F) The latency to the first food bite in the novelty-suppressed feeding test. In NIHL, the latency to food biting is increased. The numbers in parentheses indicate the number of animals tested.

Data are presented as means ± SD. NIHL: noise-induced hearing loss; ∗p < 0.05; ∗∗p < 0.01; n.s.: not significant. Also, see Table S1.

Next, we examined whether hearing loss impairs spatial memory, as has been previously reported (Manohar et al., 2020; Liu et al., 2018). In the Y-maze forced alteration task, NIHL mice retained the novel arm preference over the familiar arm similar to control mice (control entries: familiar, 5.17 ± 1.29 times, n = 18 vs. novel, 6.83 ± 1.95 times, n = 18, paired t test, t(17) = −3.18, p = 0.005; NIHL entries: familiar, 4.88 ± 1.96 times, n = 16 vs. novel, 7.56 ± 2.28 times, n = 16, paired t test, t(15) = −3.63, p = 0.002). This result indicates no short-term spatial memory impairment in our NIHL model (Figure 1C). The lack of close involvement of NIHL with short-term memory was supported by the results of the object recognition test, which is commonly used to assess the behavioral aspects of recognition memory (Figure 1D). Although the total distance traveled (control, 13.77 ± 3.71 m, n = 7 vs. NIHL, 13.86 ± 4.14 m, n = 7, one-way ANOVA, F(1,12) = 0.002, p = 0.968) was not significantly different between the control and NIHL groups, both groups showed significantly higher exploration time on a novel object than that on a familiar object (control: familiar, 7.69 ± 2.12 s, n = 7 vs. novel, 29.21 ± 10.12 s, n = 7, paired t test, t(6) = 5.433, p = 0.002, NIHL: familiar, 8.13 ± 2.90 s, n = 7 vs. novel, 37.36 ± 13.43 s, n = 7, paired t test, t(6) = 5.724, p = 0.001) with no significant difference on novel object preference index between them (control, 77.71 ± 8.67%, n = 7 vs. NIHL, 81.07 ± 7.74%, n = 7, one-way ANOVA, F(1,12) = 0.585, p = 0.459) which implies no prominent impairments of recognition memory by NIHL.

However, as in the Figure 1C, the NIHLs significantly increased the latency to the first entry into the novel arm compared to the controls, indicating possible changes in the emotional and motivational states of the animals (novel latency: control, 7.22 ± 3.99 s, n = 16 vs. NIHL, 11.19 ± 6.02 s, n = 15, one-way ANOVA, F(1,29) = 4.755, p = 0.037; familiar latency: control, 20.53 ± 8.90 s, n = 15 vs. NIHL, 19.13 ± 6.98 s, n = 13, one-way ANOVA, F(1,26) = 0.208, p = 0.652). To investigate whether the increased latency to the first entry into the novel arm was because of a change in anxiety levels, we performed two well-validated tests for anxiety: the open field test and the novelty-suppressed feeding test. In the NIHL group, the time (sec) ratio of the center zone to the outer zone in the open field test was reduced (control, 0.05 ± 0.03, n = 18 vs. NIHL, 0.04 ± 0.01, n = 13, one-way ANOVA, F(1,29) = 4.364, p = 0.046), whereas the total area traveled was unchanged (control, 30.44 ± 6.65 m, n = 18 vs. NIHL, 29.40 ± 6.14 m, n = 15, one-way ANOVA, F(1,31) = 0.216, p = 0.645) (Figure 1E). The latency to food biting in the novelty-suppressed feeding test was increased (control, 172.58 ± 89.78 s, n = 12 vs. NIHL, 267.45 ± 53.08 s, n = 11, one-way ANOVA, F(1,21) = 9.284, p = 0.006) (Figure 1F), These findings suggest that NIHL results in anxiety-like behaviors that have a negligible effect on short-term memory.

NIHL minimally alters the DG-CA3 network

We next investigated the effects of NIHL on the DG-CA3 network, which is an important hippocampal circuit implicated in the modulation of anxiety levels (Ha et al., 2011; Seo et al., 2015). Transverse hippocampal slices were prepared, and recordings were acquired from the stratum lucidum (SL) within the CA3 region and with stimulation of the superior blade of the DGGC layer (Figure 2A). NIHL did not significantly change the synaptic input/output (I/O) function (repeated-measures two-way ANOVA, control vs. NIHL, F(1,10) = 0.06, p = 0.8104, see Table S2 for detail) (Figure 2B), and neither did short-term synaptic plasticity evoked by paired-pulse stimulation compared to the control (repeated-measures two-way ANOVA, control vs. NIHL, F(1,10) = 1.187, p = 0.301, see Table S3 for detail) (Figure 2C). Long-term potentiation (LTP) was reliably induced by theta burst stimulation (TBS) and did not significantly change the potentiation between the control and NIHL groups (control, 162.39 ± 28.82%, before vs. after TBS, paired t test, t = −5.303, p = 0.003, n = 6; NIHL, 156.96 ± 53.82%, before vs. after TBS, paired t test, t = −2.592, p = 0.049 n = 6; control vs. NIHL, both after TBS, one-way ANOVA, F(1,10) = 0.048, p = 0.832) (Figure 2D). Field excitatory postsynaptic potentials (fEPSPs) were suppressed by bath application of the group II-selective metabotropic glutamate receptor agonist, DCG-IV (20 μM) (control: 60 min, 161.19 ± 26.88%, n = 6 vs. 84 min, 83.46 ± 47.07%, n = 6, paired t test, t(5) = 5.82, p = 0.002, NIHL: 60 min, 156.53 ± 49.17%, n = 6 vs. 84 min, 134.46 ± 51.05, n = 6, paired t test, t(5) = 3.10, p = 0.027), indicating that the evoked potentials largely arise from DG-originated mossy fibers (Kamiya et al., 1996). These results show that the LTP magnitude in the transverse DG-CA3 is not significantly altered by NIHL.

Figure 2

NIHL minimally alters the DG-CA3 network

(A) Hippocampal transverse slice preparation and configurations of the recording (Rec) and stimulating (Stim) electrodes in the slice used to examine the synaptic properties of the DG-CA3 network.

(B) The input-output function of CA3, represented as the fEPSP amplitude in response to increasing electrical stimulation of the DG granule cell layer, showed no significant differences between the control and NIHL groups.

(C) Short-term synaptic plasticity in response to a paired-pulse stimulation at various inter-pulse intervals. The P2/P1 (pulse2/pulse1) ratio is not significantly different between the control and NIHL groups.

(D) Long-term synaptic potentiation (LTP) at the DG-CA3 synapses, is induced by theta-burst stimulation (TBS). The insets display sample fEPSP traces of pre (black)- and post (green)-TBS induction (the scale bar denotes 5 ms × 0.2 mV). DCG-IV (mGluR2/3 agonist) is treated at the end of LTP experiment. The LTP amplitude is significantly increased compared to the normalized baseline (100%) in both groups and is not significantly different between the control and NIHL groups. The numbers in parentheses indicate the number of brain slices tested.

Data are presented as means ± SD. NIHL, noise-induced hearing loss; DG, dentate gyrus; fEPSP, field excitatory postsynaptic potential; ∗p < 0.05; ∗∗p < 0.01; n.s, not significant. Also, see Tables S2 and S3.

Longitudinal slicing angle preserves the intrinsic properties of DG granule neurons

The observation that NIHL did not alters long-term synaptic plasticity via transverse mossy fibers led us to consider another orientation of DG signal propagation, i.e., the longitudinal DG network along the dorsoventral axis perpendicular to the transverse projection. Based on the previous studies that suggested the existence of a longitudinal connection within the DG network (Lømo, 2009; Claiborne et al., 1986; Scharfman and Pierce, 2012; Jonas and Lisman, 2014; Choi et al., 2021), we examined the influence of NIHL on this network. We prepared longitudinal hippocampal slices by cutting the hippocampus along the dorsoventral axis to preserve longitudinal axons (Figure 3A) (Tetteh et al., 2019). Whole cell-patched DGGCs were identified based on the round shapes of their cell bodies, their relatively hyperpolarized RMPs, the dendritic spread toward the molecular layer, and their regular firing patterns in response to depolarizing current injection. To examine whether different slicing angles change the intrinsic properties of DGGCs, we measured their electrical properties in conventional transverse slices and longitudinal slices of the hippocampus. Whole-cell patch recordings of DGGCs showed that both the transverse and longitudinal slices had similarly linear relationships between the number of spikes and the depolarizing current injection (slope: transverse, 0.01314, n = 52 vs. longitudinal, 0.01623, n = 100, F(1,214) = 0.2370, p = 0.6269, ANCOVA) (Figure 3B) Notably, the longitudinal slices did not show any changes in RMP (longitudinal, −81.21 ± 8.71 mV, n = 24 vs. transverse, −77.29 ± 6.91 mV, n = 17, F(1,39) = 2.266, p = 0.14), input resistance (longitudinal, 344.13 ± 135.51 MΩ, n = 17 vs. transverse, 285.40 ± 206.19 MΩ, n = 16, F(1,31) = 0.888, p = 0.353), time constant (longitudinal, 0.02 ± 0.01 ms, n = 17 vs. transverse, 0.02 ± 0.01 ms, n = 15, F(1,30) < 0.001, p = 0.993), and threshold potential (longitudinal, −38.73 ± 5.96 mV, n = 10 vs. transverse, −43.36 ± 8.57 mV, n = 15, F(1,23) = 2.031, p = 0.168, one-way ANOVA, see Table S4 for detail) of DGGCs when compared with the transverse slices. Hence, any differences in the synaptic responses between the two orientations of hippocampal slices were not attributable to changes in the intrinsic membrane properties of presynaptic granule neurons caused by different slicing angles.

Figure 3

Longitudinal slicing angle enables signals to propagate longitudinally in the DG network

(A) Hippocampal longitudinal slice preparation and configuration of the stimulating (Stim) and recording (Rec) electrode and for whole-cell patch recording in the slice.

(B) DGGCs exhibit regular firing patterns in response to depolarizing current injection. The relationship between the number of spikes and the depolarizing current injection is similarly linear in both transverse and longitudinal slices. The floating numbers are positive progression slopes.

(C) The configurations of the recording and stimulating electrodes in hippocampal longitudinal slices when examining signal flow from the DG-DG network in the presence of picrotoxin (50 μM). The whole cell-patched DGGC generates sub-threshold responses and action potentials in response to increasing intensities of electrical stimulation in the molecular cell layer. The 5-Hz stimulation successfully induces spikes, and the 100-Hz stimulation induces refractory periods in the middle of the spike trains.

(D) The somatic and dendritic morphology of a DGGC, and a reconstructed image of a recorded DGGC filled with Alexa Fluor 594 showing axonal projections to the granule cell layer (a longitudinal axon) and to the hilus (a transverse axon).

(E) Glutamate uncaging is facilitated by local dot-photostimulation (cyan circle) of DGGCs that are hundreds of microns away from a patched cell (red) within the DG layer. Depolarization is induced in a postsynaptic patched neuron when it is linked to a presynaptic photo-activated neuron. Glutamate-induced EPSPs are far reduced in the presence of Cd2+, which blocks synaptic transmission. The numbers in parentheses indicate the number of cells tested.

Data are presented as means ± SD. DG, dentate gyrus; DGGC, dentate gyrus granule cell; EPSP, excitatory postsynaptic potential; ∗∗ p < 0.01.

Signals propagate longitudinally in the DG network

Given that longitudinal slices preserve the properties of DGGCs, we next tested whether electrical signals could be transferred along the longitudinal DG network. We recorded the sub-threshold responses and action potentials of whole-cell-patched DGGCs by stimulating the inner molecular layer (IML) of the superior blade of the DG (Figure 3C). 5 and 100 Hz stimulations successfully induced spikes, whereas a 100 Hz stimulation induced refractory periods in between the spike trains. By visualizing a single DGGC with two-photon imaging, we found that DGGCs not only have an axonal projection to CA3 but also have an axon pointing to the neighboring DG (306.80 ± 80.08 μm from the patched cell, n = 15) (Figure 3D). These results imply the existence of synaptic connections within the DG-DG network.

We further investigated whether the longitudinal axons of DG granule neurons have synaptic connections to neighboring DG granule neurons. Whole-cell patch-clamp recordings in longitudinal slices were acquired from the postsynaptic granule neurons, and caged glutamate was photolyzed in the presynaptic granule neurons to induce EPSPs in the postsynaptic neuron (176.80 ± 78.97 μm from the patched cell, n = 23) (Figure 3E). The EPSPs declined in the presence of 100 μM Cd2+, which blocks synaptic transmission (before Cd2+, 1.25 ± 0.6 mV vs. with Cd2+, 0.21 ± 0.15 mV, paired t test, t = 8.24, p < 0.0001, n = 23 cells). These data demonstrate that the longitudinal projections of DG granule neurons retain functional synapses, supporting emerging evidence that DG networks, including granule and mossy cells, are synaptically wired to neighboring DG neurons in a longitudinal manner.

Comparison of the synaptic properties of the DG-DG network in transverse and longitudinal slices of the hippocampus

Both longitudinal and transverse hippocampal slices were prepared with the stimulating and recording electrodes positioned near the borders of the molecular and granule cell layers (Figure 4A). The longitudinal slices showed higher synaptic responsiveness to a single stimulus that increased in intensity than the transverse slices (repeated-measures two-way ANOVA, longitudinal vs. transverse, F(1,15) = 10.92, p = 0.0048, see Table S2 for detail). Short-term synaptic plasticity, measured by the paired pulse response, showed no significant difference at various inter-pulse intervals (IPIs) between longitudinal and transverse slices, and paired-pulse depression was observed at the IPIs over 250 ms in both preparations (repeated-measures two-way ANOVA, longitudinal vs. transverse, F (1,15) = 0.128, p = 0.725, see Table S3 for detail) (Figure 4B)

Figure 4

Comparison of the synaptic properties of the DG-DG networks in transverse and longitudinal slices of the hippocampus

(A) A schematic of the transverse and longitudinal slices of the hippocampus. Both longitudinal and transverse slices are prepared with stimulating (Stim) and recording (Rec) electrodes positioned near the borders of the molecular and granule cell layers. The input-output function of DG, represented as the fEPSP amplitude in response to increasing electrical stimulation of the DG granule cell layer, showed higher synaptic responsiveness in longitudinal slices than in transverse slices.

(B) Short-term synaptic plasticity in response to a paired-pulse stimulation at various inter-pulse intervals. The P2/P1 (pulse2/pulse1) ratio is not significantly different between the transverse and longitudinal slices.

(C) Long-term synaptic potentiation (LTP) in both preparations, is tested by high-frequency stimulation (HFS). The insets display sample fEPSP traces of pre (black)- and post (red)-HFS induction. The LTP amplitude in the longitudinal slices is significantly increased compared to the normalized baseline (100%) and is significantly different between the transverse and longitudinal slices. The numbers in parentheses indicate the number of brain slices tested.

Data are presented as means ± SD. DG, dentate gyrus; fEPSP, field excitatory postsynaptic potential; ∗∗p < 0.01. Also, see Tables S2 and S3.

We also found that high-frequency stimulation (HFS) of the IML induced LTP at the DG-DG synapses more reliably in the longitudinal slices than in the transverse slices (Figure 4C). Pronounced LTP was observed in the longitudinal slices (148.03 ± 18.66%, before vs. after HFS, paired t test, t = −6.306, p = 0.001, n = 6), and no LTP was observed in the transverse slices (107.89 ± 16.82%, before vs. after HFS, paired t test, t = −1.049, p = 0.353, n = 8; longitudinal vs. transverse, after HFS, one-way ANOVA, F(1,9) = 13.771, p = 0.005). These results suggest that the short- and long-term strengths of the longitudinal DG-DG connections are plastic.

NIHL suppresses the synaptic plasticity of the DG-DG network

We next explored whether NIHL affects the synaptic properties of the DG-DG network. We used longitudinal hippocampal slices with the recording and stimulating electrodes placed in the molecular cell layer (Figure 5A). Synaptic responsiveness to electrical stimuli was significantly higher in the NIHL slices than in the control slices, reflecting the NIHL-induced hyperexcitability of the DG-DG network (repeated-measures two-way ANOVA, control vs. NIHL, F(1,10) = 11.36, p = 0.0071, see Table S2 for detail) (Figure 5B). We then tested LTP in the longitudinal slices obtained from NIHL mice.

Figure 5

NIHL suppresses synaptic plasticity in the longitudinal DG-DG network

(A) Hippocampal longitudinal slice preparation and configurations of the stimulating (Stim) and recording (Rec) electrodes in the slice used to examine the synaptic properties of the DG-DG network.

(B) The input-output function of DG, represented as the fEPSP amplitude in response to increasing electrical stimulation of the inner molecular layer of the DG, shows hyperexcitable responses in the NIHL group.

(C) Long-term synaptic potentiation (LTP) at the DG-DG synapses, is induced by high-frequency stimulation (HFS). The insets display sample fEPSP traces of pre (black)- and post (green)-HFS induction (The scale bar denotes 5 ms × 0.2 mV). The LTP amplitude in the control group is significantly increased compared to the normalized baseline (100%) but not in the NIHL group and is significantly different between the control and NIHL groups. NIHL eliminates LTPs in the longitudinal DG network.

(D) Short-term synaptic plasticity in response to a paired-pulse stimulation at various inter-pulse intervals. The NIHL condition shows paired-pulse depression and is significantly more depressed than the control group in the P2/P1 (pulse2/pulse1) ratio.

(E) Immunohistochemistry to detect neuronal viability in the hippocampal DG, stained with an antibody against NeuN (green) and a stain of nuclear DNA (DAPI, blue). The relative NeuN integrated density (NeuN intensity normalized to the DAPI intensity) indicates no significant neuronal cell loss in the DG after NIHL. The numbers in parentheses indicate the number of brain slices tested.

Data are presented as means ± SD. NIHL: noise-induced hearing loss; DG, dentate gyrus; fEPSP, field excitatory postsynaptic potential; ∗p < 0.05; ∗∗p < 0.01; n.s, not significant. Also, see Tables S2 and S3.

In NIHL mice, HFS failed to induce LTP in the longitudinal slices (control, 180.67 ± 45.96%, before vs. after HFS, paired t test, t = −3.925, p = 0.017, n = 5; NIHL, 100.04 ± 15.63%, before vs. after HFS, paired t test, t = −0.007, p = 0.995, n = 6; control vs. NIHL, after HFS, one-way ANOVA, F(1,9) = 16.499, p = 0.033) (Figure 5C). Given that NIHL altered short-term synaptic plasticity by the paired-pulse ratios with varying intervals (repeated-measures two-way ANOVA, control vs. NIHL, F(1,9) = 9.959, p = 0.012, see Table S3 for detail) (Figure 5D), it is possible that the NIHL-induced disruption of LTPs involves presynaptic mechanisms. We also found no significant differences in the postsynaptic intrinsic membrane properties of DGGCs when compared with the controls (RMP: control, −81.57 ± 5.50 mV, n = 7 vs. NIHL, −84.17 ± 4.79 mV, n = 6, F(1,11) = 0.807, p = 0.388; time constant: control, 11.95 ± 3.86 ms, n = 7 vs. NIHL, 9.83 ± 2.67 ms, n = 6, F(1,11) = 1.273, p = 0.283; threshold current: control, 162.90 ± 31.47 pA, n = 7 vs. NIHL, 180 ± 21.91 pA, n = 6, F(1,11) = 1.252, p = 0.287; threshold potential: control, −37.43 ± 3.88 mV, n = 7 vs. NIHL, −37.27 ± 3.22 mV, n = 6, F(1,11) = 0.006, p = 0.939; input resistance: control, 175.3 ± 39.82 MΩ, n = 7 vs. NIHL, 156.7 ± 20.27 MΩ, n = 6, F(1,11) = 1.06, p = 0.325, one-way ANOVA, see Table S4 for detail).

Moreover, no neuronal cell loss was observed in the DGGC layers (control, 0.98 ± 0.03, n = 7 vs. NIHL, 0.96 ± 0.02, n = 17, one-way ANOVA, F(1,22) = 3.397, p = 0.079) (Figure 5E). These results demonstrate that NIHL suppresses synaptic plasticity in the DG-DG connection and that this suppression might strongly involve presynaptic alterations.

Stress-induced anxiety alters the synaptic plasticity of the longitudinal DG-DG network

The chronic restraint stress (CRS) mouse model is widely used as a model of chronic psychoemotional stress for investigating anxiety-like behaviors (Huang et al., 2015). CRS does not change the short- and long-term plasticity of the CA3 in rodents upon mossy-fiber stimulation (Pavlides et al., 2002). Therefore, we examined whether CRS alters DG-DG synaptic plasticity in the context of anxiety. As previously described (Pavlides et al., 2002), CRS was induced by placing an animal in a 50-mL conical tube for 6 h per day for 14 consecutive days (Figure 6A). The latency to food biting significantly increased after CRS (control, 88.67 ± 33.06 s, n = 18 vs. CRS, 140.46 ± 41.15 s, n = 13, one-way ANOVA, F(1,29) = 15.098, p = 0.001), suggesting that CRS increases anxiety levels, as reported previously (Figure 6B) (Chiba et al., 2012). As seen in the NIHL model, the CRS mice showed hyperexcitable neuronal responses to electrical stimuli (repeated-measures two-way ANOVA, control vs. CRS, F(1,13) = 16.43, p = 0.0014, see Table S2 for detail) (Figure 6C). Paired-pulse depression was prominent at 100-ms and 250-ms stimulus intervals compared to the control (repeated-measures two-way ANOVA, control vs. CRS, F(1,12) = 7.94, p = 0.016, see Table S3 for detail) (Figure 6D). CRS also impaired the LTP induced by HFS (control, 173.23 ± 49.44%, before vs. after HFS, paired t test, t = −3.857, p = 0.012, n = 6; CRS, 132.92 ± 41.67%, before vs. after HFS, paired t test, t = −2.277, p = 0.057, n = 8; control vs. CRS, one-way ANOVA, F(1,12) = 2.825, p = 0.119) (Figure 6E). These results indicate that CRS increases anxiety levels in a manner similar to NIHL and alters the short- and long-term synaptic plasticity of the longitudinal DG-DG network.

Figure 6

Stress-induced anxiety alters synaptic plasticity in the longitudinal DG network

(A) Schematic of a chronic stress-induced anxiety model and timeline of the experimental procedure.

(B) The latency to the first food bite in the novelty-suppressed feeding test. Chronic restraint stress (CRS) mice show increased latency to the first food bite and display anxiety-like behaviors. The numbers in parentheses indicate the number of animals tested.

(C) The input-output function of DG, represented as the fEPSP amplitude in response to increasing electrical stimulation of the inner molecular layer of the DG, shows hyperexcitable responses in the CRS group. The insert shows a hippocampal longitudinal slice and the configurations of the stimulating (Stim) and recording (Rec) electrodes when examining the synaptic properties of the DG-DG network.

(D) Short-term synaptic plasticity in response to a paired-pulse stimulation at various inter-pulse intervals. The CRS condition shows paired-pulse depression and is significantly more depressed than the control group in the P2/P1 (pulse2/pulse1) ratio.

(E) Long-term synaptic potentiation (LTP) at the DG-DG synapses, is induced by high-frequency stimulation (HFS). The insets display sample fEPSP traces of pre (black)- and post (green)-HFS induction (The scale bar denotes 5 ms × 0.2 mV). The LTP amplitude in the control group is significantly increased compared to the normalized baseline (100%) but not in the CRS group and is not significantly different between the control and CRS groups. CRS impairs LTPs in the longitudinal network. The numbers in parentheses indicate the number of brain slices tested.

Data are presented as means ± SD. DG, dentate gyrus; fEPSP, field excitatory postsynaptic potential; ∗p < 0.05; ∗∗p < 0.01; n.s: not significant. Also, see Tables S2 and S3.

Discussion

Anxiety disorders, such as post-traumatic stress disorder (PTSD) and specific phobia (SP), are typically characterized by an excessive fear response and/or worry (Edition, 2013). Functional neuroimaging studies report that activation of the hippocampus increases in response to threatening stimuli in PTSD and SP (Patel et al., 2012; Lueken et al., 2014; Garrett et al., 2012). In particular, manipulation of DGGC activity is associated with anxiety-like behaviors, suggesting that DGGCs contribute to regulating anxiety levels (Kheirbek et al., 2013). However, few studies have examined the role of the DG network in regulating anxiety at the synaptic and circuit levels.

Here, we demonstrate that an NIHL mouse model enhances anxious behavior and alters synaptic responsiveness in the longitudinal DG-DG network. To investigate a possible correlation between anxiety and the longitudinal DG-DG network, we compared a CRS-induced anxiety model with our NIHL-induced anxiety model. This was necessitated by our current inability to completely separate the longitudinal DG-DG network from the transverse DG-CA3 network. Both animal models with abnormal anxiety behaviors showed synaptic alterations in the DG-DG network, suggesting that these alterations are associated with behavioral manifestations of anxiety but are not solely specific to NIHL. However, the dissimilarity in the trajectory of LTP progression might be resulted from the different pathophysiology of each animal model, presumably implicating that the synaptic alteration of the DG-DG network is more susceptible to NIHL than CRS.

Notably, NIHL-induced anxiety is not correlated with the transverse DG-CA3 network, which is known to underlie memory encoding. This may partially explain why our NIHL model does not show prominent memory impairment, as can be inferred from the results of the Y-maze and object-recognition tests. In addition, the CRS-induced anxiety model is not related to the mossy fibre-CA3 network, which is interchangeable with the transverse DG-CA3 network in rodents (Pavlides et al., 2002). Therefore, the DG-DG network could possibly predict anxiety-like behaviors.

The mechanisms by which such altered synaptic plasticity in the longitudinal network relates to anxiety at the circuit level need to be established. We speculate that, following hearing loss or chronic stress, the DG-DG network contributes to the rapid propagation of electrical signals, as evidenced by the hyperexcitable I/O function that is accompanied by paired-pulse depression. Subsequently, synaptic hyperexcitability of the DG-DG network could enhance the electrical spread to several cortices sequentially and/or simultaneously throughout the brain regions implicated in anxiety generation and/or development. Finally, this process could increase arousal and induce anxiety. However, it remains unclear whether anxiety renders the longitudinal DG network dysfunctional or whether a dysfunctional longitudinal DG network predisposes individuals to anxiety.

The hippocampus is susceptible to changes in response to noise exposure, irrespective of whether the noise is benign or traumatic. Long-term environmental noise exposure during the postnatal period (65 dB SPL) diminishes LTP in the perforant path (PP)-DG synapses and impairs hippocampus-dependent memory function (Zhang et al., 2021). Traumatic noise exposure (110–120 dB) also diminishes LTP in the CA3-CA1 and PP-DG networks, although the behavioral effects of such changes have not yet been studied (Cunha et al., 2015; De Deus et al., 2017). Our findings show that after traumatic noise, DG-CA3 synapses are not prominently altered in terms of short- and long-term plasticity. In contrast, the longitudinal DG-DG network is highly vulnerable to traumatic noise. This is closely associated with anxiety-like behavior but not with short-term memory impairment. Given that previous reports have shown memory impairment following either 3 months of post-acute noise exposure (Liu et al., 2018) or chronic repetitive noise exposure (Zhang et al., 2021), the 2-week observation period following acute noise exposure in our animal model may not be sufficient to reveal the development of a memory deficit. This difference between our study and previous studies presumably arises from disparities in the animal species (rats vs. mice), noise levels (circumstantial vs. traumatic), age (young vs. mature), and laterality (bilateral vs. unilateral).

Mossy cells are a prominent cell type in the hilus and actively interact with granule cells and interneurons in the DG. Some mossy cells have axonal projections that target the IML of the DG and run longitudinally along the dorsoventral hippocampus (Scharfman, 2016). Given the locations of the stimulating and recording electrodes, i.e., the middle molecular layer in the DG, their contributions to DG-DG activation appear to be minimal. Nonetheless, it is possible that the synaptic responses recorded in the middle molecular layer might reflect mossy axon-driven activities. There is also a possible synaptic effect of EC cells on the DG. Although the synaptic connection of the EC-DG circuitry is truncated during slice preparation, the severed axons must still possess the ability to be activated. Therefore, the synaptic responses we observed in the longitudinal DG-DG network may have been influenced by the axons of mossy and EC cells, albeit to a lesser degree. We can rule out the likelihood of inhibitory synaptic inputs to the DG-DG network, as all the experiments were performed in the presence of picrotoxin, a GABAA receptor antagonist. However, further investigations are required to separate the roles of the longitudinal axons of granule cells from those of the mossy cells in the DG. Despite these limitations, our study presents novel evidence of an association between a dysfunctional longitudinal DG network and enhanced anxiety behaviors. An abnormal longitudinal DG network is thus a potential hallmark of anxiety.

Limitations of the study

As already noted in the discussion, the increased synaptic excitability may originate from presynaptic alteration of DG granule neurons as the membrane properties of postsynaptic neurons remained the same. However, it is still uncertain whether the synaptic response is solely from DG granule neurons and what types of presynaptic mechanisms are involved. It awaits future studies.

STAR★Methods

Key resources table

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Sungchil yang (sungchil.yang@cityu.edu.hk).

Materials availability

This study did not generate new unique reagents.

Experimental model and subject details

Animal models

Male C57BL/6J mice aged 8–14 weeks were housed under a 12-h light/12-h dark cycle in a temperature-controlled environment (23 °C) and allowed access to food and water ad libitum. For the noise-induced hearing loss (NIHL) model, all experiments were performed on mice anesthetized by an intraperitoneal (IP) injection of 100 mg ketamine/kg and 10 mg xylazine/kg based on their body weights. One-third of the initial dose was additionally delivered by IP injection as necessary. A unilateral hearing lesion was induced by traumatic noise exposure for 2 h on day 0 (Pre) in the left ears of the animals. The noise (continuous pure tone of 8 kHz at 115 dB sound pressure level [SPL]) was played via a customized MATLAB program through a closed-field magnetic speaker (MF1 multi-field magnetic speaker, Tucker-Davis Technologies) amplified by a Samson Servo 120a amplifier (Samson Technologies, Hicksville, NY). The level of traumatizing tone was measured with an ICP microphone system (378B02, PCB Piezotronics, NY, USA) and calibrated with a customized LabVIEW 2019 software. The opposing (right) ears were plugged with cotton swabs to preserve function during hearing lesioning. Sham control mice underwent anesthesia for 2 h without sound exposure. Because noise stimuli were delivered to the left ears, the hippocampus contralateral to the noise-exposed ears was chosen for electrophysiological recordings. The noise-exposed and sham control mice were housed after anesthesia in individual cages for approximately 10 days before conducting the behavioral and electrophysiological measurements. Chronic restraint stress (CRS) was induced by placing the animals in 50-mL conical tubes with multiple ventilation holes at their home cages for 6 h a day for 14 consecutive days. Mice in the non-CRS group were simultaneously deprived of food and water but not subjected to restraint. Following hearing lesions or restraint stress, the animals were continuously monitored for pain/stress. If any signs of pain/stress were observed, analgesics (buprenorphine 0.05 mg/kg) were administered as necessary. All animal handling procedures were approved by the Institutional Animal Care and Use Committee of City University of Hong Kong (A-0117) and Incheon National University (INU-ANIM-2017-08). This research complies with all the guidelines of our institution’s animal welfare committee and this journal’s ethics policies.

Method details

Auditory brainstem response (ABR) recording

ABR recordings were used to assess hearing sensitivity before and 7 days after the noise exposure (pre and day 7, respectively). Sound generation and ABR recordings were performed using Tucker-Davis Technologies System 3 hardware (Medusa Pre-Amps, RZ6 Multi I/O Processor, MF1 multi-field magnetic speaker, FL, USA) and software (BioSigRZ, FL, USA). Acoustic stimuli were delivered through a closed-field magnetic speaker (MF1 multi-field magnetic speaker, Tucker-Davis Technologies) and were constituted of tone-pips (5-ms full-cycle sine waves) of various frequencies (4 kHz, 8 kHz, 16 kHz, and 32 kHz), each of which was played at intensities of 80–10 dB in decrements of 10 dB. Sound-evoked bioelectrical potentials were recorded using three-needle electrodes (S83018-R9, Rochester) placed subcutaneously in the test ear (active), the vertex of the head (ground), and the contralateral ear (reference). Hearing thresholds were evaluated by visually inspecting the ABR waveforms and defined as the lowest sound intensity at which a distinct first wave was observed. For convenience of calculation, the threshold was set up at 90 dB when no sound-evoked waveform was observed at 80 dB. The animals’ body temperature was maintained at 37 °C during the entire procedure using a temperature controller (RWD Life Science), and dryness of the animals’ eyes was prevented using tear ointment.

Hippocampal brain slice preparation for electrophysiology

The brain preparation and electrophysiology procedures have been described in detail in the previous publication (Sun et al., 2018). Briefly, male C57BL/6J mice aged 8–14 weeks (∼4 weeks for comparing transverse and longitudinal slices) were deeply anesthetized with isoflurane. The brains were quickly removed and placed into chilled (4 °C) and oxygenated (5% CO2 and 95% O2) slicing medium containing 75 mM sucrose, 87 mM NaCl, 2.5 mM KCl, 1.3 mM NaH2PO4, 25 mM NaHCO3, 25 mM glucose, 7 mM MgCl2 and 0.5 mM CaCl2. Transverse slices were cut perpendicular to the dorsoventral axis of the hippocampus, and longitudinal slices were cut parallel to the dorsoventral plane. 300-μm-thick slices were for whole-cell patch recording and 400-μm-thick slices were for extracellular field recording. The brain slices were then transferred to a holding chamber containing oxygenated artificial cerebrospinal fluid (ACSF; 125 mM NaCl, 2.5 mM KCl, 1.3 mM NaH2PO4, 25 mM NaHCO3, 25 mM glucose, 1 mM MgCl2, and 2 mM CaCl2). After ∼30 min of recovery at 32 °C, followed by at least 0.5 h at room temperature (whole-cell patch recording) or at least 1 h at room temperature (extracellular field recording), the slices in the recording chambers (Warner, MA) were submerged in ACSF (32 °C) delivered at a flow rate of 1.5 mL/min and bubbled with a mixture of 95% O2 and 5% CO2.

Whole-cell patch recording

The whole-cell patch recording procedures have been described in detail in our previous publications (Yang et al., 2016). Briefly, the transverse or longitudinal hippocampal slices were transferred from the holding beakers to the recording chambers and were submerged in ACSF (32 °C) with a mixture of 95% O2 and 5% CO2. Whole-cell patch recordings were obtained using an Axon Instruments Multi-Clamp 700B Amplifier (Molecular Devices); the recording pipettes had tip resistances of 5–7 MΩ and were filled with a solution containing (in mM) 135 K-gluconate, 5 KCl, 1 MgCl2, 0.02 CaCl2, 0.2 EGTA, 10 HEPES, 4 Na2-ATP and 0.3 Na-GTP. Alexa Fluor 594 (20–50 μM, Tocris) was included in the internal solution for morphological visualization as required. The pH and osmolarity of the intracellular solution were adjusted to 7.3 and 290 mOsm, respectively. The initial serial resistance typically ranged from 9 MΩ to 17 MΩ and was maintained at a stable value during the recording sessions. Serial resistances were also continuously monitored with a brief voltage pulse and were not compensated. Recordings were acquired when the cells showed serial resistance with <25% fluctuation during the recording sessions. pCLAMP v10.2 software (Molecular Devices) was used for data acquisition and analysis. To measure the neurons’ intrinsic properties, the resting membrane potential (RMP) was measured using the cells’ resting state membrane potential 1 min after whole-cell configuration was achieved. The cells’ input resistances were calculated from their steady-state membrane potentials, i.e., from the voltage changes induced by small hyperpolarizing current pulses (−20 pA). The membrane time constant represents the time required for the voltage to reach 63% of the peak value and was determined by fitting exponential curves to the hyperpolarized voltage responses. The threshold potential (mV) was visually determined at the voltage level, which showed a sudden increase in the slope of the action potential. The threshold current (pA) was determined at the lowest current level injected to induce a single action potential. To evoke synaptic potentials of a patched DGGC on the longitudinal slice, a stimulating electrode of the concentric bipolar microelectrode (CBAPC75, 25 mm inner pole diameter, FHC, USA) was placed in the inner molecular layer (IML) hundreds of microns away from a patched cell. 5 and 100 Hz stimulations were delivered to ensure synaptic connections between DGGCs.

Two-photon imaging and glutamate uncaging

The procedure for two-photon imaging has been described in detail in our previous publications (Yang et al., 2014b). Briefly, a Ti-sapphire laser was tuned to 810 nm (Coherent Chameleon) for morphological visualization with Alexa 594. Epi-fluorescent and trans-fluorescent signals were captured through a 60×, 1.0 N.A. objective and a 1.4 N.A. oil-immersion condenser (Olympus). The fluorescence was split into red and green channels using dichroic mirrors and band-pass filters. Red fluorescence (Alexa 594) signals were captured using R9110 photomultiplier tubes. For glutamate uncaging, another Ti-sapphire laser was tuned to 720 nm (Coherent Chameleon) to release the 4-methoxy-7-nitroindolinyl (MNI) compound from glutamate. MNI-caged-l-glutamate (Tocris) was prepared each day freshly at the final concentration (500 μM) in the physiological solution. The 3 × 3 grid of point stimulation (5 ms laser duration, 10 μm spot spacing, and various interpoint delay, > 0.12 ms) at an average power of 42 mW (ranged from 7.5 to 78.08 mW) was performed using Prairie View 5.4 software (Bruker Corporation); each spot size ranged from a diffraction-limited spot 0.2 to 1 μm (with laser intensity and objective dependency). Cadmium chloride (CdCl2; 0.1 mM) was externally applied to the brain slices to block chemical synaptic transmission at the presynaptic terminals.

Extracellular field recording

The transverse or longitudinal hippocampal slices were transferred from the holding beakers to the recording chambers and were submerged in ACSF (32 °C) bubbled with a mixture of 95% O2 and 5% CO2. In all extracellular field recordings, picrotoxin (50 μM; Sigma) was applied to the perfusate 15–20 min before the recordings were acquired. Local field potentials (LFPs) were evoked (constant current, 100 μs duration, repeated at 30-s intervals) in the transverse slices by placing a stimulating electrode (concentric bipolar microelectrode, CBAPC75, 25 mm inner pole diameter, FHC, USA) in the superior blade of the DGGC layer. LFPs were then recorded from the stratum lucidum (SL) within the CA3 region using glass electrodes containing ACSF. Long-term potentiation (LTP) was induced by theta-burst stimulation (TBS; 10 trains of 4-pulse bursts at 100 Hz, 200-ms inter-burst interval, delivered at 0.1 Hz). At the end of the experiment, the group II-selective metabotropic glutamate receptor (mGluR2/3) agonist DCG-IV (20 μM, Sigma-Aldrich, St Louis, MO) was bath-applied to examine the involvement of mossy fibers in the synaptic responses.

For the longitudinal slices, the recording and stimulating electrodes were placed in the middle molecular layer along the longitudinal axis to record and evoke LFPs (constant current, 100 μs duration, repeated at 30-s intervals). Long-term synaptic plasticity (i.e., LTP) was induced by 4 trains of tetanus stimuli (high-frequency stimulation [HFS]; 100 Hz, 1 s, 4 times at 1-s intervals) after acquiring >20 min of stable baseline recordings.

The peak amplitudes of all LFPs were analyzed offline using Clampfit software (Axon Instruments). Before LTP was induced, the input/output (I/O) curves and paired-pulse responses were identified. The short (paired-pulse response)- and long (LTP)-term synaptic plasticity were recorded using stimulation intensities producing half-maximal response amplitudes. The Pulse 2 (P2)/Pulse 1 (P1) ratio of paired-pulse responses was calculated by the peak amplitude of the second LFP divided by that of the first LFP measured at inter-pulse intervals of 50 ms, 100 ms, 250 ms, 500 m, and 750 ms. Any changes in synaptic strength (i.e., LTP) were expressed relative to the pre-conditioning baseline normalized to 100%. Averaged LFP amplitude during the last 10 min of baseline recording was compared to that during the last 10 min of post-conditioning recording.

Immunohistochemistry

After the animals were transcardially perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA), the isolated brains were immersed overnight at 4 °C in a centrifuge tube containing 4% PFA in 120 mM phosphate buffer, maintained at pH 7.4. A Leica VT1000 S vibrating blade microtome (Leica Biosystems) was used to cut 100-μm thick coronal sections, which were treated with blocking solution for 30 min at room temperature (125 mL PBS, 2.5 mL 10% Triton X-100, 1.25 mL 20% NaN3 [w/v], 100% serum [normal goat serum; NGS or normal donkey serum; NDS], and PBS to a total volume of 250 mL) to block and permeabilize the sections. The mouse monoclonal anti-NeuN antibody was used as a neuronal marker (1:250; Abcam, ab104224). The sections were incubated for 24 h at 4 °C with the primary antibodies and then incubated with the corresponding secondary antibodies conjugated with Alexa Fluor 488 (1:500; Jackson Immunoresearch, 115-545-205) and 4′,6-diamidino-2-phenylindole (DAPI, Santa Cruz, sc-3598) for 1 h at room temperature. After rinsing with PBS, the sections were mounted with Vectashield (Vector Labs) and viewed under a fluorescence microscope (Nikon Eclipse Ni-E). The right hemisphere of the hippocampus, contralateral to the noise-exposed ear, was chosen for analysis. The integrated densities (i.e., the sum of all pixels in the ROI) of the areas stained with NeuN or DAPI were calculated using ImageJ (NIH). The data are presented as a ratio of the integrated densities of NeuN to those of DAPI. For comprehensive measurement along the long axis of the hippocampus, we analyzed the septal, medial, and temporal sections of the hippocampus alongside the dorsal and ventral DG in each section.

Behavioral tests

The behavioral tests were conducted in a dimly lit behavioral room during the dark phase of the day cycle. All mice were habituated in the behavioral room for 1 h prior to testing. To minimize unnecessary stress during testing, all of the mice were handled daily, and the testing apparatus and arenas were cleaned with 70% ethanol and dried after each mouse was tested.

Y-maze forced alteration test

The Y-maze test was conducted in a Y-shaped plastic maze (34 cm × 6 cm × 10 cm) that had three identical arms (start, familiar, and novel/closed), as described previously (Dinel et al., 2014). The test consisted of video-recorded acquisition and retrieval trials separated by 30-min inter-trial intervals. During the acquisition trial, one arm was closed with a door, and a mouse was placed in an open-arm (start) end and allowed to explore the two open arms for 5 min. After 0.5 h, the retrieval trial was performed by removing the door from the closed arm (novel) and allowing the mouse to freely explore all of the arms (start, familiar and novel) for 5 min. The start and novel arms were randomized for each mouse to minimize side bias. The frequency of entries and the latency to first entry into the novel and familiar arms were analyzed using ANY-maze software (ANY-maze, IL, USA).

Object recognition test

The object recognition test consisted of two sessions: acquisition and retrieval. During the 5-min acquisition phase, each mouse was allowed to explore two identical objects at each of the two opposite corners of an open field arena (40 cm × 40 cm × 40 cm). During the 5-min retrieval phase, the mouse was allowed to encounter 2 objects after a 90-min delay: a familiar object to which the mouse had already been exposed and a new object. The total distance traveled and the interaction time (s) with the novel object were measured and analyzed from the recorded video using video tracking software (ANY-maze). Preference index of novel object recognition was calculated by [(novel object exploration time)/(total objects exploration time)] × 100.

Open-field test

The open-field test was performed as previously described, with minor modifications (Lim et al., 2016). In brief, each mouse was placed in the middle of an enclosed white plastic box (40 cm × 40 cm × 40 cm) and allowed to explore the box for 10 min. Each trial was video recorded, and the distance traveled and time spent in the center or outer areas were measured and analyzed using ANY-maze.

Novelty-suppressed feeding test

This test was conducted as previously described, with minor modifications (Yalcin et al., 2011). Briefly, food and water bottles were removed from the home cage 24 h before testing. During the test, regular chow pellets were placed on a petri dish in the center of an open field box (40 cm × 40 cm × 40 cm) illuminated under a 450-lux light. The mouse was placed in a corner of the open field, and the latency to feed (biting) was video recorded for 5 min and recorded using a stopwatch.

Quantification and statistical analysis

All data are expressed as means ± standard deviations (SD) of the mean and analyzed using SPSS v25 statistical analysis software. Graphs were plotted using Prism software v8.02 (GraphPad Software Inc). Statistical significance was determined using analysis of variance (ANOVA) between groups and paired t tests within a group unless otherwise stated (∗p < 0.05; ∗∗p < 0.01; n.s: not significant).

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-NeuN	Abcam	Cat# ab104224; RRID: AB_10711040
Secondary antibodies conjugated with Alexa Fluor 488	Jackson Immunoresearch	Cat#: 115-545-205; RRID: AB_2338854
Chemicals, peptides, and recombinant proteins
Ketamine 10%	Alfasan International B.V	https://www.alfasan.com/en/
Xylazine 2%	Alfasan International B.V	https://www.alfasan.com/en/
Isoflurane	RWD Life Science	R510-22
Sucrose	Sigma-Aldrich	S0389
NaCl	Sigma-Aldrich	S7653
KCl	Sigma-Aldrich	P9333
NaH2PO4	Sigma-Aldrich	S5011
NaHCO3	Sigma-Aldrich	S6297
Glucose	Sigma-Aldrich	G8270
MgCl2	Sigma-Aldrich	M2670
CaCl2	Sigma-Aldrich	C8106
K-gluconate	Sigma-Aldrich	P1847
EGTA	Sigma-Aldrich	E0396
HEPES	Sigma-Aldrich	H7523
Na2-ATP	Sigma-Aldrich	A9187
Na-GTP	Sigma-Aldrich	G8877
Alexa Fluor 594	Tocris	A10438
MNI-caged-l-glutamate	Tocris	1490
CdCl2	Sigma-Aldrich	655,198
Picrotoxin	Sigma-Aldrich	P1675
DCG-IV	Tocris	0975
DAPI	Santa Cruz	sc-3598
Vectashield	Vector Labs	H-1000-10
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	000664
Software and algorithms
MATLAB	The MathWorks, Inc.	https://www.mathworks.com/
BioSigRZ	Tucker-Davis Technologies	https://www.tdt.com/
pCLAMP software	Molecular Devices	v10.2
Image J	(Schneider et al., 2012)	https://imagej.nih.gov/ij/
ANY-maze software	ANY-maze	https://www.any-maze.com/
GraphPad Prism 8.02	GraphPad Software	https://www.graphpad.com/
Other
MF1 multi-field magnetic speaker	Tucker-Davis Technologies	https://www.tdt.com/
Power amplifier	Samson Technologies	120a
ICP microphone system	PCB Piezotronics	378B02
Tucker-Davis Technologies System 3 hardware (Medusa PreAmps, RZ6 Multi I/O Processor, MF1 multi-field magnetic speaker)	Tucker-Davis Technologies	https://www.tdt.com/
Three-needle electrodes	Rochester	S83018-R9
Temperature controller	RWD Life Science	https://www.rwdstco.com/
Vibratome	Leica Biosystems	VT1200S
Recording chamber	Warner Instruments	RC-26GLP
Axon Instruments MultiClamp 700B Amplifier	Molecular Devices	https://www.moleculardevices.com/
Ti-sapphire laser	Coherent Chameleon	https://www.coherent.com/
Concentric bipolar microelectrode	FHC Inc	CBAPC75