Abstract
The adolescent social experience is essential for the maturation of the prefrontal cortex in mammalian species. However, it still needs to be determined which cortical circuits mature with such experience and how it shapes adult social behaviors in a sex-specific manner. Here, we examined social-approaching behaviors in male and female mice after postweaning social isolation (PWSI), which deprives social experience during adolescence. We found that the PWSI, particularly isolation during late adolescence, caused an abnormal increase in social approaches (hypersociability) only in female mice. We further found that the PWSI female mice showed reduced parvalbumin (PV) expression in the left orbitofrontal cortex (OFCL). When we measured neural activity in the female OFCL, a substantial number of neurons showed higher activity when mice sniffed other mice (social sniffing) than when they sniffed an object (object sniffing). Interestingly, the PWSI significantly reduced both the number of activated neurons and the activity level during social sniffing in female mice. Similarly, the CRISPR/Cas9-mediated knockdown of PV in the OFCL during late adolescence enhanced sociability and reduced the social sniffing-induced activity in adult female mice via decreased excitability of PV+ neurons and reduced synaptic inhibition in the OFCL. Moreover, optogenetic activation of excitatory neurons or optogenetic inhibition of PV+ neurons in the OFCL enhanced sociability in female mice. Our data demonstrate that the adolescent social experience is critical for the maturation of PV+ inhibitory circuits in the OFCL; this maturation shapes female social behavior via enhancing social representation in the OFCL.
SIGNIFICANCE STATEMENT Adolescent social isolation often changes adult social behaviors in mammals. Yet, we do not fully understand the sex-specific effects of social isolation and the brain areas and circuits that mediate such changes. Here, we found that adolescent social isolation causes three abnormal phenotypes in female but not male mice: hypersociability, decreased PV+ neurons in the left orbitofrontal cortex (OFCL), and decreased socially evoked activity in the OFCL. Moreover, parvalbumin (PV) deletion in the OFCL in vivo caused the same phenotypes in female mice by increasing excitation compared with inhibition within the OFCL. Our data suggest that adolescent social experience is required for PV maturation in the OFCL, which is critical for evoking OFCL activity that shapes social behaviors in female mice.
Introduction
One of the hallmarks of postnatal maturation in the mammalian cortex is the maturation of GABAergic inhibitory circuits (Luhmann and Prince, 1991). Maturation is accompanied by an increase in the expression of both the parvalbumin (PV) in a subset of GABAergic neurons and the perineuronal nets (PNNs) surrounding the PV+ neurons (Celio and Blumcke, 1994; Huang et al., 1999). PV is a calcium-binding protein and a well known genetic marker for fast-spiking GABAergic neurons in the cortex (Rudy et al., 2011). PV gene expression is upregulated in the sensory cortex in an activity-dependent manner (Sugiyama et al., 2008; Krishnan et al., 2017) after mammals are born. Like PV, PNNs also increase during cortical maturation and, in the adult cortex, surround most of the PV+ neurons (Ueno et al., 2017, 2018). Along with the increase of PNNs in cortical circuits, other molecular changes reduce net plasticity and facilitate the maturation of the cortical network (Beurdeley et al., 2012). If mice are deprived of visual input during the critical period after eye opening, via monocular deprivation or dark rearing, PV+ neurons are less developed, and PNNs are reduced in the primary visual cortex (V1; Sugiyama et al., 2008). This deprivation, in turn, diminishes both the visual response of V1 neurons that receive input from the deprived eye (Sugiyama et al., 2008) and inhibition in the cortex (Dityatev et al., 2007). Therefore, the postnatal sensory experience is critical for the expression of PV and PNNs in PV+ neurons, as these are required for the maturation of PV+ inhibitory circuits and, eventually, for shaping sensory responses in the cortex (Hensch, 2005).
The maturation of the cortex occurs first in the sensory area, where the process is part of sensory experience, and later in the motor-frontal and the prefrontal cortex (PFC; Hensch, 2005). The social experience is critical for the maturation of the PFC in both humans (Blakemore, 2008; Paus et al., 2008) and rodents (Makinodan et al., 2012; Bicks et al., 2020; Tan et al., 2021), particularly during juvenile and adolescent periods. Patients with autism or schizophrenia often show abnormal social cognition as well as dysfunction in the PFC (Bicks et al., 2015). In these patients, the expression of PV and PNNs is reduced in the dorsolateral PFC (Enwright et al., 2016). Thus, it is plausible that immature PV+ circuits in the PFC cause the psychiatric symptoms underlying abnormal social behavior in mammals. However, we still do not understand whether social experience is required for PV+ inhibitory circuits in these PFC regions to mature. In particular, it is largely unknown whether social experience affects the maturation of inhibitory circuits in the orbitofrontal cortex (OFC), one of the PFC subregions critical for social cognition (Adolphs, 2003; Amodio and Frith, 2006).
Mammalian social behavior is important for maintaining a conspecific society. Social interaction during adolescence is critical for shaping social and cognitive behaviors in both humans and rodents (Spear, 2000; Makinodan et al., 2012; Hinton et al., 2019; Almeida et al., 2021; Potrebic et al., 2022). Interestingly, the mammalian brain shows hemispheric asymmetry and lateralization in executing social behaviors (Marlin et al., 2015). Even in rodents, according to recent studies, the PFC functions asymmetrically in mediating social behaviors (Lee et al., 2015; Marlin et al., 2015). However, whether such asymmetry in the PFC becomes increasingly pronounced in a sex-specific manner with social experience during adolescence is still unclear. We investigated the effect of adolescent social isolation on the maturation of PV+ neurons in the PFC in both sexes and hemispheres. Using postweaning social isolation (PWSI) and the CRISPR/Cas9 system, we revealed that social isolation during adolescence caused the defect in PV expression in the left OFC (OFCL) of female mice, which led to hypersociability and reduced OFCL activity during social contact.
Materials and Methods
Experimental model and subject details
We approved all experimental procedures from the KAIST Institutional Animal Care and Use Committee (IACUC-18–236). We maintained all mice ad libitum under light from 8:00 A.M. to 8:00 P.M. and dark from 8:00 P.M. to 8:00 A.M. All mice were weaned at postnatal day 21 (P21); mice were reared in double cages (28 × 17 × 14 cm) with three to five mates for group housing or in single cages (20 × 10 × 14 cm) alone for social isolation. We performed experiments on WT (C57BL/6J), PV-Cre (catalog #008069, The Jackson Laboratory), and the Rosa26-floxed STOP-Cas9 knock-in mice (catalog #024857, The Jackson Laboratory).
Histology
We anesthetized mice with isoflurane and then perfused them with 1× PBS and 4% paraformaldehyde (w/v in PBS). Mouse brain samples were postfixed for 4 h at 4°C; after fixation, we incubated them in 30% sucrose solution (w/v in PBS) for 1–2 d at 4°C. Then, we embedded each brain sample in O.C.T. medium (catalog #4583, Tissue-Tek) and froze each at −80°C. We sectioned frozen brains using a cryostat (Leica) and collected coronal sections either at 30 μm (for confirming injection sites) or 15 μm (for immunostaining) thickness. To confirm fluorescently labeled injection sites, brain sections were washed three times with 1× PBS and mounted in DAPI-containing mounting medium (VECTASHIELD, catalog #H-1200, Vector Laboratories). For immunostaining, we washed the brain sections three times with 1× PBS and permeabilized them with 0.3% Triton X-100 in 1× PBS for 30 min. After the sections were incubated in a blocking solution (2% normal donkey serum in 1× PBS) for 2 h at room temperature (RT), we incubated them with primary antibodies (1:500 dilution in blocking solution; rabbit anti-PV; catalog #PV25, Swant; 1:500 dilution in blocking solution; mouse anti-NeuN; catalog #MAB377, Millipore) for 2 d at 4°C. Last, we incubated them with fluorescence-tagged secondary antibodies (Alexa Fluor 594 donkey anti-rabbit IgG for PV staining; 1:500 dilution in 1× PBS; catalog #A21207, Thermo Fisher Scientific; Alexa Fluor 647 donkey anti-mouse IgG for NeuN staining; 1:500 dilution in 1× PBS; catalog #715–605-151, Jackson ImmunoResearch). For PNN staining, we treated the brain sections with the fluorescein (FITC)-labeled Wisteria floribunda agglutinin (WFA; 1:500 dilution in 1× PBS; catalog #FL-1351, Vector Laboratories) for 2 h at RT before mounting them in clear mounting medium (catalog #S3023, DAKO). We imaged fluorescent signals of mounted brain sections under a scanning microscope (Axio Scan.Z1, Zeiss).
Cell-counting analysis
When we sectioned the mouse brain samples, we collected 15 µm brain sections in every 60 µm thickness across the frontal cortex. After immunostaining the sections, we scanned the fluorescent images of the brain sections and matched the images to the mouse brain atlas (Franklin and Paxinos, 2012) to confirm the anterior-to-posterior location of the coronal brain sections. We then selected three images of sections in a row that included OFC areas (between +2.10 mm and approximately +2.34 mm anterior from the bregma; see Fig. 2C). To count the labeled neurons in the OFC, we placed a square (0.3 × 0.3 mm2) as a region of interest (ROI) at the center of the OFC (±1.3 mm lateral from the bregma; depth, −2.5 mm). To quantify the number of PV+ and NeuN+ neurons, we used automatic cell counting software (ImageJ; Grishagin, 2015). After converting the fluorescent images to 8 bit grayscale images, we further transformed the grayscale images into binary images. For the binary transition, we used distinct thresholds at the grayscale for isolating the NeuN+ neurons (see Fig. 2; total mean intensity, 11.39 ± 0.02%) and the PV+ neurons (see Fig. 2; total mean intensity, 2.12 ± 0.16%). We then automatically counted the white-filled circles as cells if they satisfied the following conditions: (1) the total size was >60 × 60 pixels; and (2) the circularity was >0.2. We set the thresholds in each sample at the level that yielded a similar number of cells from the automatic counting compared with the manual counting using a couple of template images from each sample. There was no significant difference between the number of cells counted automatically and the number of cells counted manually in a blind manner (see Fig. 2B). To validate the knock-down efficiency of single-guided RNAs targeting the parvalbumin gene or chondroitinase ABC (chABC; see Fig. 5), we manually and blindly counted the number of PV+ or WFA+ neurons in the ROI.
Sociability test
We performed the sociability test by using the following two types of mazes: one was a modified three chamber-like maze (50 × 45 × 20 cm) where two containers were located diagonally at opposite corners; and the other was a standard three-chamber maze (60 × 40 × 20 cm), which was used in other studies (Won et al., 2012). Most of the data we plotted in the article were collected in the modified maze. We performed the sociability test on mice at either P79 to P81 or P50. Before behavior experiments, all mice were habituated for 30–60 min in the habituation chamber. All mice were maintained in a group-housed (GH) or an isolated condition until the habituation. Mice were then moved to the sociability test maze. The sociability test consisted of the following two phases: (1) habituation in the sociability maze (empty containers, 10 min); and (2) sociability test (an object container and a social container with a stranger mouse, 10 min). We used a 2 week younger and same-sex mouse as a stranger mouse not to evoke any aggressively or sexually driven social behaviors in the tested mouse (Murugan et al., 2017). We used the software EthoVision XT 11.5 (Noldus) to automatically analyze interaction times and manually analyze the sniffing time. The sniffing time was defined as the time when the nose of the subject mouse contacted the container either empty or with a novel mouse. We analyzed mouse sociability by the following two indicators: time spent in interaction zones near the container (18 × 18 cm) and sniffing time. We calculated the social-sniffing and social interaction indices as follows (Nakanishi et al., 2017; Cheong et al., 2020):
Resident-intruder test
We devised the resident-intruder (RI) test as previously described (Tan et al., 2021). The mouse was single housed for 24 h and then exposed to an intruder mouse in the home cage for 10 min. We used a slightly smaller (10–20% lighter) unfamiliar sex-matched and strain-matched novel mouse as an intruder. The offensive behaviors of the subject mouse against the intruder, including lateral threat, upright posture, clinch attack, and chase, were scored as aggressive behaviors. In addition, the nonoffensive social behaviors in the RI test, such as nose-to-nose interaction and anogenital sniffing, were scored. All scoring was performed blindly.
Measuring estrous cycle in female mice
We monitored the estrous cycle of female mice using vaginal cytology (McLean et al., 2012). Vaginal smears from subject mice were collected right after the sociability test. Smears were placed on slides and stained with 0.1% crystal violet (catalog #V5265, Sigma-Aldrich). Each estrous cycle stage was determined based on the cell type that was mainly observed in the smear sample among the three types: nucleated epithelial cells, cornified squamous epithelial cells, and leukocytes. To examine the difference in estrous cycles between GH and PWSI mice, we performed the vaginal cytology for 4 consecutive days in each mouse.
In vivo extracellular recordings for measuring sniffing-induced activity
We performed all recordings in the head-fixed and awake mice. We marked the top of OFCL and right OFC (OFCR) and implanted custom-designed head plates on the skull of mice using dental cement and small screws. After 2–3 d of recovery, we habituated mice to be head fixed for 20 min/d for 3 d. After the habituation, we performed the recording in the OFC by making a small craniotomy on the marked spots and inserting 32-channel silicon probes (catalog #A1X32-Poly2-10 mm-50–177 or #A1X32-Poly3-10 mm-50-177, NeuroNexus) vertically into each OFC using a microdrive manipulator (Siskiyou). We first recorded spiking activity in the OFC without presenting a stranger mouse for 10 min (objectsniffing session) and next continuously recorded neural activity while presenting a head-fixed stranger mouse in front of the recorded mouse for 10 min (socialsniffing session). During the recording, we monitored and recorded the sniffing behavior of the mouse by using a web camera (model LifeCam HD-3000, Microsoft). All extracellular signals were recorded at a 30 kHz sampling rate, filtered between 500 and ∼5000 Hz, and amplified by a miniature digital head stage (CerePlex μ, Blackrock Neurotech). After the recording, we confirmed the recording sites by injecting red retrobeads (Lumafluor) into the recording areas and performing histology. If the recording sites were outside of the OFC, we did not use those recording data in further analysis.
Automatic quantification of sniffing behaviors.
We developed a MATLAB code to automatically analyze sniffing events during object-sniffing and social-sniffing sessions in head-fixed mice. First, we converted all the RGB image frames of a recorded video clip to grayscale using the rgb2gray function. Next, we chose the ROI, including the nose and whiskers of the mouse being recorded, made a binary matrix with 1 inside the ROI and 0 outside the ROI of each pixel, and multiplied an original image by the binary matrix of each corresponding pixel along all the frames (ROI_1). We subtracted background brightness from ROI_1 to eliminate overall brightness fluctuations (ROI_2) and calculated deviations of brightness within the ROI by subtracting the mean brightness of each pixel (ROI ΔF). To detect sniffing events, we subtracted a moving average (F0; sliding window = 50 frames) from ROI ΔF (ROI ΔF–F0), and set a threshold by multiplying a median of absolute ROI ΔF–F0 by 4. If ROI ΔF–F0 in each frame crossed the threshold, we included the frame in a sniffing event.
Neural activity analysis.
We performed the spike-sorting using Klusters software (http://neurosuite.sourceforge.net/; Hazan et al., 2006). We grouped 32-channel data into eight groups and sorted single units by principal components analysis. We only isolated single units that show a >2 ms refractory period in their autocorrelogram. We performed all other analyses using custom-made codes in MATLAB (MathWorks). We sorted sniffing trials in object and social sessions by identifying sniffing events across the recording sessions. We only used trials that satisfied the following two criteria: (1) intersniffing interval (time between the offset of sniffing in the current trial and the onset time of the next trial) >5 s; and (2) sniffing duration (time between the onset and the offset of sniffing in one trial) >1 s. To plot the perievent time histogram of all neurons as a population, we performed z-scoring by averaging firing rates (in hertz) of individual units across trials and normalized them by 1 s baseline activity (3 to 2 s before the onset of sniffing). The normalized firing rates were convolved with a Gaussian filter (σ = 50 ms) only for visualization purposes. To identify the neurons showing significant sniffing-induced activity, we statistically compared the firing rates of individual neurons during the 2 s preperiod (3 to 1 s before the onset of sniffing) and 2 s post-period (0 to 2 s after the onset of sniffing) of each sniffing trial by bootstrapping (p < 0.01, n = 5000). We quantified the sniffing-induced activity of neurons by averaging z values from 0 to 2 s after the onset of sniffing. We classified cell types of recorded neurons based on spike waveforms by quantifying spike widths of each isolated neuron (time between peak and through of the spike, in microseconds). If the spike width was <350 or >450 μs, recorded neurons were classified as fast-spiking (FS) neurons or regular-spiking (RS) neurons, respectively. For the baseline activity analysis, we quantified the average firing rates (in hertz) from 0 to 2 s before the onset of sniffing.
Perineuronal net degradation
To remove the PNN, we reconstituted protease-free chABC (Amsbio) in PBS to a final concentration of 61 U/ml. We anesthetized mice by inhalation of 1.5% isoflurane in oxygen on the stereotaxic apparatus. We injected 0.5 μl of PBS or chABC into the left OFC [anteroposterior (AP), +2.35; mediolateral (ML), +1.3; dorsoventral (DV), −2.5] of female mice at P70 to P75. We performed behavior experiments 7 d after the injection and killed the mice to confirm PNN degradation by histology. To stain the PNN in the brain slice containing the OFCL, we treated the FITC-conjugated WFA (catalog #FL-1351, Vector Laboratories) as a secondary antibody, diluted at 1:500 in 1× PBS, for 2 h at RT before mounting the slices in a transparent mounting medium (catalog #S3023, DAKO). We manually and blindly counted the number of WFA+ neurons within 0.5 × 0.5 mm2 ROIs in three representative coronal sections covering OFC.
CRISPR/Cas9 experiments
Generating the PX552-sgPV-tdTomato.
To generate the AAV-tdTomato (tdTom) vector, we first replaced the eGFP gene in PX552 (catalog #60958, Addgene) with tdTomato. The tdTomato template was obtained from pAAV-FLEX-tdTomato (catalog #28306, Addgene). Next, we cloned two AAV-tdTomato plasmids, both of which induce the targeted mutations at the parvalbumin gene (sgPV). One virus expresses two single-guide RNAs (sgRNAs) targeting parvalbumin exon4: sgRNA1, 5′-AGACAAGTCTCTGGCATCTG-3′ and sgRNA2, 5′-AAGGATGGGGACGGCAAGAT-3′ (cloned recombinant DNA fragment: AGACAAGTCTCTGGCATCTGGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGCAACAAAGCACCAGTGGTCTAGTGGTAGAATAGTACCCTGCCACGGTACAGACCCGGGTTCGATTCCCGGCTGGTGCAAAGGATGGGGACGGCAAGATGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGC; Bioneer/in-house PCR amplification), and the other virus expresses two sgRNAs targeting parvalbumin exon5: sgRNA3, 5′-GCTAAGTGGCGCTGACTGCT-3′ and sgRNA4, 5’-GGCCGCGAGAAGGACTGAGA-3’(Cloned recombinant DNA fragment: GCTAAGTGGCGCTGACTGCTGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGCAACAAAGCACCAGTGGTCTAGTGGTAGAATAGTACCCTGCCACGGTACAGACCCGGGTTCGATTCCCGGCTGGTGCAGGCCGCGAGAAGGACTGAGAGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGC; Bioneer/in-house PCR amplification).
After cloning the AAV-tdTomato vectors, we transfected them with packaging plasmid into HEK293T (catalog #HC20005, KCTC) to produce AAV-sgPV-tdTomato (exon4 and exon5) and AAV-tdTomato, using Lipofectamine 2000 (catalog #11668–019, Thermo Fisher Scientific). After 72 h, we collected adeno-associated virus [AAV (AAV2)] particles from the cell lysates and performed iodixanol gradient ultracentrifugation. Next, we concentrated the viral particles with Amicon Ultra Centrifugal Filter (Millipore) and quantified the concentration of viral particles by quantitative PCR (in genome copies/ml: AAV-tdTomato, 3.92 × 1014; AAV-sgPV(exon4)-tdTomato, 3.07 × 1012; AAV-sgPV(exon5)-tdTomato, 7.17 × 1012).
Validation of CRISPR/Cas9 in the Neuro2a cell.
To identify the efficiency of sgPV in knocking down the PV, we transfected cloned AAV vectors with the Cas9-eGFP-expressing vector (PX458) in the Neuro2a cells. After 72 h, we collected and sorted tdTomato/eGFP double-positive cells using FACSAria II (BD Biosciences). Next, we purified genomic DNA from sorted cells and performed PCR to amplify the target sites of the PV gene, adding the index sequence for sequencing. Last, we subjected the samples to sequencing (MiniSeq System, Illumina) and analyzed insertion and deletion frequencies using CRISPR RGEN Tools (http://www.rgenome.net/).
Behavior and physiological experiments with CRISPR/Cas9 mice.
To express Cas9 selectively in PV+ neurons, we crossed the PV-Cre mice (catalog #008069, The Jackson Laboratory) with the Rosa26-floxed STOP-Cas9 knock-in mice (catalog #024857, The Jackson Laboratory). We injected tdTom or AAV-sgPV-tdTomato (sgPV) into the crossed mice (PV::Cas9) on P50 to P65. From P78 to P85, we performed the behavior test and in vivo recording. To verify whether the AAV expression knocked down the PV effectively, we collected brain samples of the injected mice and performed immunohistochemistry as described above. We used a combination of the primary antibodies (rabbit anti-PV for PV staining, goat anti-GFP for Cas9 staining) and the secondary antibodies (Alexa Fluor 405 donkey anti-goat IgG and Alexa Fluor 647 donkey anti-rabbit IgG were used).
Whole-cell patch-clamp recordings in brain slices.
For electrophysiology experiments, tdTom/sgPV-expressed brain samples were collected from anesthetized PV::Cas9 mice with isoflurane (Terrell) and sliced (coronal, 300 μm thickness) using a vibratome (model VT1200, Leica). Brain extraction and sectioning were conducted in ice-cold dissection buffer containing the following (in mm): 212 sucrose, 25 NaHCO3, 5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 3.5 MgSO4, 10 d-glucose, 1.25 l-ascorbic acid, and 2 Na-pyruvate (Pyr), bubbled with 95% O2/5% CO2. The sectioned slices were incubated in a recovery chamber at 32°C with normal ACSF (in mm: 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 10 glucose, 2.5 CaCl2, and 1.3 MgCl2, oxygenated with 95% O2/5% CO2). After 60 min of recovery at 32°C, slices were incubated for 30 min at room temperature (20–25°C). The slice was transferred to a recording chamber at 27–29°C with circulating ACSF saturated with 95% O2 and 5% CO2. Recording pipettes were made from thin-walled borosilicate capillaries (catalog #30–0065, Harvard Apparatus) with resistance of 3.0–4.0 MΩ using a two-step vertical puller (model PC-10, Narishige).
Whole-cell patch-clamp recordings were conducted using an amplifier (MultiClamp 700B, Molecular Devices) and digitizer (Digidata 1550, Molecular Devices). We recorded both AAV-expressed and nonexpressed neurons in the OFCL. We confirmed the AAV expression by imaging the red fluorescence of the patched cells under the fluorescent microscope (Olympus). During the recordings, hyperpolarizing step pulses (5 mV, 40 ms) were injected into the cell to monitor membrane resistance calculated by the peak amplitude of the capacitance current. Data from cells that changed the resistance by >20% were excluded. To measure the intrinsic excitability, recording pipettes (3.0–4.0MΩ) were filled with an internal solution containing the following (in mm): 137 K-gluconate, 5 KCl, 10 HEPES, 0.2 EGTA, 10 Na-phosphocreatine, 4 Mg-ATP, and 0.5 Na-GTP, at pH 7.2 and 280 mOsm. Picrotoxin (100 μm; Sigma-Aldrich), NBQX (10 μm, Tocris Bioscience), and d-AP5 (50 μm, Tocris Bioscience) were added to ACSF to inhibit postsynaptic responses. After gigaseal and rupture, currents were clamped, and resting membrane potential was maintained at −70 mV. To measure the I–V curve, current inputs were increased from −300 to 120 pA in increments of 30 pA/sweep with a time interval of 2 s. To measure the current–firing (I–F) curve, current inputs were increased from 0 to 660 pA in increments of 60 pA/sweep with a time interval of 10 s. For measuring spontaneous EPSCs (sEPSCs) in the OFC PV+ neurons, glass pipettes were filled with an internal solution composed of the following (in mm): 100 CsMeSO4, 10 TEA-Cl, 8 NaCl, 10 HEPES, 5 QX-314-Cl, 2 Mg-ATP, 0.3 Na-GTP, and 10 EGTA, at pH 7.25 and 295 mOsm at a holding potential of −70 mV. Picrotoxin (100 μm) was added to ACSF to block IPSCs. For measuring spontaneous IPSCs (sIPSCs) in the OFC excitatory neurons, recording pipettes were filled with an internal solution containing the following (in mm): 120 CsCl, 10 TEA-Cl, 8 NaCl, 10 HEPES, 5 QX-314-Cl, 4 Mg-ATP, 0.3 Na-GTP, and 10 EGTA, at pH 7.35 and 280 mOsm holding at −70 mV. NBQX and d-AP5 were added to block EPSCs in the circulating ACSF. Data were acquired by Clampex 10.2 (Molecular Devices) and analyzed by Clampfit 10 (Molecular Devices).
Wireless optogenetic modulation of the OFC neurons in vivo
Design and fabrication of flexible optoelectronic neural probes.
Flexible optoelectronic probes were designed using electronic design automation software (Altium Designer 18.0, Altium Limited) and produced using a conventional photolithography process. Neural probes (0.1 mm thick) were designed in a bilateral form, in which the right and the left probes (for each probe: width, 0.38 mm; length, 4 mm) were separated by 2.6 mm to access both regions of the mouse brain. Copper traces and electrodes (18 μm thick) were patterned on one side (i.e., top layer) of the polyimide substrate (25 μm thick). In contrast, the other side (i.e., the bottom layer) was covered with a copper layer (18 μm thick) to stiffen the probe for penetration into brain tissue. The probe tip part, where the microscale inorganic light-emitting diodes (μ-ILEDs; TR2227, Cree; blue, 473 nm; green, 532 nm; 270 × 220 × 50 μm3) were attached, was not covered with a copper layer to allow light emission from both sides of the probe. The fabricated probes showed an optical intensity of ∼405 mW/mm2 at the top layer and ∼45.3 mW/mm2 at the bottom layer, which sufficiently exceeds the threshold intensity for channelrhodopsin-2 (ChR2) activation (∼1 mW/mm2). A three-pin plug-n-play female pin connector (model M50-3130345, Harwin) and two μ-ILEDs were mounted on the top layer electrodes after applying a low-temperature solder paste (SMDLTLFP10T5, Chip Quik) and soldered using a precision soldering iron (NASE-2C, JBC Tools) with a temperature of 250°C. After confirming that every component was robustly attached, probes were coated with Parylene C (7 μm thick) for waterproofing. Last, the probes were vertically bent and attached to a 3D-printed probe holder, followed by fixing them in place using epoxy (5 Minute Epoxy, Permatex).
Design of programmable wireless control module with detachable configuration.
A Bluetooth Low Energy (BLE)-based wireless control module was designed to be easily attached and detached with a bilateral optoelectronic probe and a battery through a plug-n-play connection (Qazi et al., 2019; Lee et al., 2020). This module controls the optical output of the μ-ILEDs through smartphone manipulation. The electronic circuit of the wireless module consists of a rechargeable lithium polymer (LiPo) battery (GM300910-PCB, PowerStream Technology; 12 mAh; 9 × 10 × 3 mm3), a BLE System-on-Chip (SoC; EYSHSNZWZ, Taiyo Yuden; 8.55 × 3.25 × 0.3 mm3), and various other electronic components (a linear voltage regulator, four decoupling capacitors, two resistors, and two indicator LEDs). The current supplied from the LiPo battery is regulated and stabilized by a voltage regulator (NCP4624DMU30TCG, onsemi) and decoupling capacitors and flows to the BLE SoC to operate the module. Two indicator LEDs [model VAOL-S4RP4, VCC; red (624 nm wavelength)] were placed at the output pins of BLE SoC, each connected in series with a resistor (6.04 kΩ; model RK73H1HTTC6041F, KOA Speer Electronics), to visually show the current operation status of the module during the animal experiment—for example, the indicator LED blinks every 5 s during optical stimulation. The wireless control module can be assembled with optoelectronic probes through three-pin plug-n-play connectors (male: model M50-3630342, Harwin; female: model M50-3130345, Harwin) and assembled with LiPo battery through two-pin connectors (male: model M50-3630242, Harwin; female: model M50-3130245, Harwin). Every component except the LiPo battery was soldered on the wireless module flexible printed circuit board of the wireless module (9.3 × 7.6 × 0.1 mm3), which was designed and manufactured through the same method used for the μ-ILED probes. The module was wirelessly programmed over the air (Nordic Device Firmware Update bootloader) to provide a pulse current (20 Hz with 25 ms pulse width; i.e., 50% duty cycle) or continuous current (i.e., 100% duty cycle) to the connected bilateral μ-ILED probes. The BLE-based smartphone app was developed using commercial app development software (Android Studio) and used to control the μ-ILED probes for wireless optogenetics.
Virus injection and probe implantation surgery.
For optogenetic activation of excitatory neurons in the left OFC, we injected the AAV2-CAMKIIα-eYFP (enhanced yellow fluorescent protein; catalog #VB1941, Vector Biolabs) or AAV2-CAMKIIα-ChR2-eYFP (Addgene) virus in the OFCL (AP, +2.35; ML, ±1.3; DV, −2.5) of female WT mice between P50 and P60. For optogenetic inactivation of PV+ neurons in the OFCL, we injected the AAV9-EF1α-DIO-eYFP (UNC Vector Core) or AAV9-EF1α-DIO-eArch3.0-eYFP (UNC Vector Core) virus into the OFCL of female PV-Cre transgenic mice. For the injection, we anesthetized mice using 1.5% isoflurane in oxygen inhalation and performed stereotaxic surgery. Each injection contained 400 nl of AAV infused at a rate of 0.4 nl/s using a microinjector. We recovered the mice for 7 d after the AAV injection and anesthetized mice again to implant the wireless optogenetic LED probe into the virus-expressing OFC area. Because of the thickness of the probe (0.1 mm) and the location of the LED on the probe (+0.22 mm from the tip), we implanted the probe into the anterior and ventral part of the OFC (AP, +2.45; ML, ±1.3; DV, −2.77) to target the LED facing the virus-expressing area without any lesion in it.
Behavior test.
To connect the small module part to the implanted probe on the mouse head, we briefly anesthetized the mice by isoflurane inhalation. After the module was connected, the mice had 10 min recovery phase in their home cage. On Day 0, we habituated mice wearing a module for 15 min in the open field chamber. On Day 1, we habituated the mice wearing a module in the sociability chamber for 5 min. We then presented a novel social target and a novel object (O) for 5 min and recorded their approaching behaviors to measure the sociability of the mice. On Day 2, we performed a sociability test with blue or green LED stimulation for optogenetic modulation. We gave 20 Hz blue light stimulation by 473 nm LED for activation or continuous green light stimulation by 532 nm LED for inactivation. The LED connected to the probe during the 5 min sociability test session using the custom-designed flexible optoelectronic neural probes. Interaction times of mice were analyzed on Day 1 and Day 2, as described previously. On Day 3, we performed an open field test for 5 min with optogenetic stimulation and 5 min without stimulation to measure the effect of light activation on the locomotion behavior. After the behavior test, we performed the histology of the brain samples from the mice to check the virus expression and probe the implanted site. We collected data only if the virus expressed >65% of the total volume of the OFC and the probe was targeted at the virus-expressed area.
Experimental design and statistical analysis
All statistical analyses were performed using custom codes in IBM SPSS statistics 21 (IBM) and OriginPro 2019 (Origin). The normal distribution of data were verified using the Shapiro–Wilk normality test. For parametric tests, we used the paired Student’s t test to compare matched pairs, unpaired Student’s t test to compare two groups, one-way ANOVA with Fisher’s least significant difference post hoc test to compare multiple groups, and two-way ANOVA to compare I–F curves of PV+ neurons. For nonparametric tests, we used the Wilcoxon signed-rank test to compare matched pairs, the Mann–Whitney U test to compare between two independent groups, and the Kolmogorov–Smirnov test to compare the distributions of two groups. The number of animals analyzed by statistical tests and the exact p values from each experiment were reported within each figure legend.
Results
Postweaning social isolation enhances sociability in female mice
Previous reports have shown that both short-term and long-term social isolation affects mouse social behaviors (Makinodan et al., 2012; Matthews et al., 2016). However, the effects differ depending on the duration of isolation, and the data were collected only in male mice. To examine whether short-term and long-term social isolation effects on mouse sociability differ between the sexes, we performed the sociability test in both male and female mice. For the sociability test, we put two small chambers at diagonal corners in the maze, each of which contained a novel object (O) or a novel and sex-matched mouse (M), and measured the relative times of mice sniffing or spending at or nearby the novel mouse compared with those times at the novel object as indices of the mouse sociability. We quantified the sociability in mice that had experienced three different postweaning social conditions at P80: GH, SI for 1 d (1dSI), or SI for 60 d after weaning (i.e., PWSI; Fig. 1A,B). Both the sniffing time of 1dSI mice when they came into contact with a novel mouse and their time spent in the social interaction zone did not differ significantly from those of GH mice in both males and females (Fig. 1C,D).
Notably, PWSI induced changes in sociability in a sex-specific manner: only female mice showed a significant increase in social-sniffing time and social interaction time, while PWSI male mice showed normal sociability, the same as the GH mice (Fig. 1C,D). We also confirmed that the effect of PWSI on female mice was not simply because of hormonal changes that occur during the estrous cycle. First, we confirmed that the estrous cycle was not different between GH and PWSI female mice (Fig. 1E). We next confirmed that the mouse sociability (both social-sniffing index and social interaction index) that we measured was consistent across the estrous cycle in female mice (Fig. 1F–H). Therefore, our data indicate that long-term social isolation during adolescence affects the social behavior of females more than males, causing hypersociability in females.
PWSI reduces PV expression in the OFCL of female mice
We next examined whether the social experience during adolescence is required to mature PV+ inhibitory circuits in the OFC. We quantified the number of PV+ or PNN+ neurons in the OFC of the GH and the PWSI mice at P80. We collected data from male and female mice separately. We first immunostained both for the PV, a marker for fast-spiking GABAergic neurons, and the NeuN, a marker for all neurons (Fig. 2A). We counted the number of immunostained neurons automatically in the three coronal sections that were collected across the center of the OFC (+2.1 to approximately +2.34 mm from the bregma) per each hemisphere in each mouse (see Materials and Methods; Fig. 2B,C). Interestingly, the number of PV+ and NeuN+ neurons in the OFC was significantly lower in the brains of PWSI female mice compared with the GH female mice, especially in the left hemisphere (Fig. 2D–F). In male mice, however, the number of PV+ and NeuN+ neurons was not significantly different between GH and PWSI groups (Fig. 2D–F). When we counted the number of PV– and NeuN+ neurons by subtracting the number of PV+ neurons from the total number of NeuN+ neurons, there was no difference between groups (Fig. 2F, filled bars; female, OFCL, p = 1, Mann–Whitney U test). These data indicate that the reduction of NeuN+ neurons preferentially occurs in the PV+ neurons rather than in overall PV– neurons in the OFCL (Fig. 2F). We next quantified the PNN+ neurons in the OFC by staining for WFA, a marker for PNNs (Fig. 2G). Both PV+ and PV– neurons were surrounded by the PNNs in the OFC (Fig. 2H). We found that the numbers of WFA+ neurons in the OFC were not different between GH and PWSI in both male and female mice (Fig. 2I). Collectively, our data indicate that social experience during adolescence is crucial for the maturation of PV+ neurons, especially the PV expression, in the OFCL of female mice.
Late adolescence is a critical window for maturing sociability in female mice
PWSI mice experienced long-term social isolation from when they were weaned at P21 to when they became adults at P80. This period includes both early juvenile periods before the onset of puberty and late adolescence after the onset of puberty. Since puberty leads to sexually dimorphic changes in behaviors, we wondered how mouse sociability matures after the onset of puberty in males and females. We thus tracked sociability changes from P50 (late adolescence; Laviola et al., 2003) to P80 (adult) in both sexes of GH mice by measuring the social-sniffing index at P50 and at P80. Interestingly, we found a significant decrease in sociability at P80 in female mice but not in male mice (Fig. 3A). We further found that during this period, PV+ neurons increased significantly in female OFCL. In contrast, OFCR did not show such an increase between P50 and P80 (Fig. 3B). These data suggest that late adolescence from P50 to P80 is a critical window for the maturation of sociability with an increase in PV+ neurons in the OFCL in female mice.
We next examined the effect of social isolation during late adolescence in both female and male mice. We directly isolated the mice after P50 until P80 (P50SI; Fig. 3C). In addition, we re-group housed (re-GH) a group of PWSI female mice at P50 to make them re-experience social interaction during late adolescence (re-GH; Fig. 3C). The P50SI female mice showed an increase in sociability like the PWSI female mice, while neither re-GH female mice nor P50SI male mice showed any changes in sociability compared with control GH mice (Fig. 3D). Not only our modified three chamber-like test, but a standard three-chamber test also showed an increase in sociability in P50SI female mice (data not shown). In addition, the number of PV+ neurons in re-GH female mice was not significantly different from that in GH female mice in both hemispheres of the OFC (Fig. 3E), indicating that social experience during late adolescence was sufficient to mature the PV+ neurons in the OFCL.
Finally, we wondered whether the enhanced sociability in P50SI female mice, measured by our sociability test, was related with changes in aggression. To examine this, we performed the resident-intruder test (Tan et al., 2021) on P50SI male and female mice. We counted both offensive behaviors, such as threatening and attacking behaviors, and social behaviors, such as nose sniffing and anogenital sniffing, separately during the test (Fig. 3F). In female mice, the aggression level did not increase by P50SI. Still, the social-approaching behaviors significantly increased (Fig. 3G). On the other hand, P50SI male mice showed a significant increase in aggression without any changes in social behaviors (Fig. 3H). Collectively, our data indicate that social isolation during late adolescence induced hypersociability in female mice without increasing aggression, whereas it induced hyperaggression in male mice.
PWSI disrupts social sniffing-induced activity in the OFCL of female mice
Because fewer PV+ neurons in the OFCL were found following PWSI in female mice, we investigated whether the OFCL neural response to social contact has been altered in head-fixed female mice by in vivo multichannel recordings (Fig. 4A,B). During the recording, both GH and PWSI female mice showed sniffing behaviors toward an object or another mouse under the head-fixed condition, and the number and duration of sniffing times did not differ between groups (Fig. 4C,D). We identified neurons that showed significant sniffing-induced activity in the OFC (Fig. 4E–H). Interestingly, in GH females, more neurons showed sniffing-induced activity in the OFCL than in the OFCR responses, and their responses were stronger when a mouse sniffed another mouse than when it sniffed an object (Fig. 4E,F). More importantly, these socially responsive neurons in the OFCL of female mice largely disappeared after PWSI (Fig. 4E,G,H). On the other hand, the sniffing-induced activity of the OFCR did not differ between GH and PWSI female mice (Fig. 4F). Among the responsive neurons in the OFCL, the fraction of the fast-spiking neurons was smaller in the PWSI mice than in the GH mice. These data indicate that PWSI decreased social sniffing-induced activity in the OFCL neurons, particularly including fast-spiking neurons, and reduced overall social representation in the OFCL.
PV+ expression, not PNNs in the OFCL, shapes social behavior in female mice
As we found that PWSI impaired the PV expression in the OFCL and increased social behavior, we next examined the causal relationship between the reduction of PV expression in the OFCL and the increase in the sociability of female mice. To reduce the PV expression in vivo, we used the CRISPR/Cas9 system (Swiech et al., 2015) and designed two pairs of sgRNAs to target the fourth and fifth exons of the PV gene (sgPV). We first confirmed that those sgRNAs knocked out the PV gene by either deletion or insertion of the targeted sequences in >85% of the Cas9-transfected Neuro2a cells (Fig. 5A). We then constructed the AAV vectors to deliver those sgRNAs together with the tdTomato gene to neurons in vivo (Fig. 5A) and injected the AAVs into the cortex of PV::Cas9-eGFP mice, which were created by crossing PV-Cre with Cas9fl/fl-eGFP mice (sgPV; Fig. 5B,C). As a control, we injected the AAV that expressed only the tdTomato without sgRNAs into the crossed mice (tdTom). In the sgPV mice, originally PV+ neurons were expressed with both Cas9-eGFP and tdTomato, and their PV gene was knocked out by Cas9 in the PV+ neurons selectively (Fig. 5B,C).
To knock down PV in the OFCL during late adolescence until mice became adults, we injected the AAVs into the OFCL when mice were at approximately P50 and examined changes in the sociability of the injected mice at P80. We confirmed the reduction of PV expression in the eGFP+/tdTomato+ neurons without any decrease in PNN expression by immunostaining the PV and the PNNs in the OFCL (Fig. 5D,E). The female mice with PV knockdown in the OFCL showed a significant increase in sociability (Fig. 5H), similar to the PWSI or P50SI females (Figs. 1B–D, 3D). When we knocked down the PV in the OFCR of female mice or the OFCL of male mice, we did not see such an increase in sociability compared with the control mice (Fig. 5H). We further examined whether a reduction in PNNs surrounding PV+ neurons affects female sociability. We infused chABC, an enzyme that degrades PNNs by removing sulfate side chains of proteoglycans, into the OFCL (Fig. 5F). Treatment with chABC significantly decreased the number of WFA+ neurons in the OFCL without affecting PV expression (Fig. 5G). Not like the female mice knocked down with PV, the female mice treated with chABC showed normal sociability (Fig. 5I). Together, our data indicate that the PV expression, but not the PNN maturation, in the OFCL was a key factor shaping the social behavior of female mice.
Decreased PV expression reduces social sniffing-induced activity via hyperexcitation of OFCL neurons
We next investigated whether the decrease of PV expression in the OFCL was responsible for the reduced sniffing-induced activity in the OFCL of the PWSI female mice. We performed in vivo recordings in female mice injected with sgPV AAVs (sgPV) or injected with tdTomato AAVs (tdTom) in their OFCL (Fig. 6A–G). First, we found that the average baseline firing rates were significantly higher in the spPV group compared with the tdTom group (Fig. 6D,E). Furthermore, similar to PWSI, knocking down PV by sgPV decreased social sniffing-induced activity in the OFCL of female mice (sgPV; Fig. 6F,G). Conversely, the expression of tdTomato in the OFCL did not alter the sniffing-induced activity of female mice (tdTom; Fig. 6F,G). Knocking down PV in the OFCL did not affect the sniffing behavior itself, as the number and duration of sniffs in the altered mice were similar to those in the tdTom control mice (Fig. 6C). These data indicate that the knockdown of PV increased overall firing activity while reducing the sniffing-induced activity in the OFCL.
As we observed an increase in the baseline activity of OFCL neurons by knocking down the PV, we next examined whether the knockdown of PV has altered the excitation–inhibition balance in the OFCL of female mice. We first performed patch-clamp recordings of the PV+ neurons in the OFCL by targeting tdTomato+ neurons of PV::Cas9 mice injected with sgPV or tdTom AAVs into the OFCL (Fig. 6H–L). We found that the knockout of the PV gene by sgPV in PV+ neurons reduced the intrinsic excitability of those neurons (Fig. 6I) without any changes in sEPSCs (Fig. 6L). Other membrane properties, such as the half-width of spike waveforms and the input resistance, were normal in those neurons (Fig. 6J,K). We next measured the sIPSCs in the OFCL neurons, which were not labeled with the tdTomato in the sgPV and the tdTom mice. These neurons were regular spiking and thus putative excitatory neurons in the OFCL (Fig. 6H,M). The OFCL excitatory neurons in the sgPV mice showed a significant decrease in both amplitudes and frequencies of spontaneous IPSCs compared with the tdTom control mice (Fig. 6M). These data indicate that the decreased expression of PV in the OFCL reduced the excitability of the PV+ neurons and, in turn, reduced inhibition of the neighboring neurons in the OFCL of female mice.
Our recording data suggested that the decrease in inhibition on knocking down the PV gene caused an increase in excitation, which led to a rise in the baseline activity of OFCL and disruption in sniffing-induced activities (Fig. 6D–G). We thus examined whether the abnormal increase of OFCL activity caused hypersociability in female mice. We expressed the ChR2 in the excitatory neurons of the OFCL by injecting the AAV virus, which expresses the ChR2 fused to the eYFP under the CaMKIIα promotor, to the wild-type female mice (Fig. 7A,C). We then delivered the light stimulation to the OFCL by inserting a wireless optogenetic device that we developed (Kim et al., 2021; Qazi et al., 2022) to provide the 20 Hz blue lights during the sociability tests (Fig. 7B; see Materials and Methods). The stimulation was given on the second day of the sociability tests. We measured the first-day sociability (Day 1) without the light stimulation and did not observe any differences in the sociability between the ChR2-eYFP and the eYFP groups. We tested their sociability again on the second day with the blue-light stimulation on the OFCL (Day 2). The eYFP control mice showed a significant decrease in their sociability on Day 2 compared with Day 1, potentially because of habituation (Fig. 7D,E). On the other hand, the ChR2-eYFP-expressed mice showed higher levels of social-sniffing behaviors with light stimulation than the control mice on Day 2 (Fig. 7D). These data indicate that the optogenetic activation of the OFCL induced higher sociability and almost no habituation in social approaching behaviors on Day 2 (Fig. 7D,E). We next optogenetically inactivated the PV+ neurons in OFCL directly during the sociability test on Day 2 by expressing the archaerhodopsin-3 (eArch3.0) in PV+ neurons and delivering 532 nm light through our wireless probe (Fig. 7G). Similar to the optogenetic activation of excitatory neurons in the OFCL, the inhibition of PV+ neurons induced hypersociability in female mice without habituation on Day 2, when we gave the light stimulation in the OFCL (Fig. 7H,I). The light stimulation did not affect the locomotion in both groups (Fig. 7F,J). Collectively, our data demonstrate that hyperexcitation of the OFCL by reducing inhibition caused hypersociability in female mice.
Discussion
Our study found that adolescent social experience is critical for shaping sociability in female mice. Social isolation during late adolescence enhanced sociability in female mice while impairing the maturation of PV+ neurons and decreasing social sniffing-induced activity in the OFCL. The knockdown of PV in the OFCL of female mice induced the same phenotype as the PWSI mice: enhanced sociability and decreased sniffing-induced activity in the OFCL. These effects were because of the imbalance between excitation and inhibition, as the PV knockdown caused hyperexcitation in the OFCL. We further found that optogenetic activation of excitatory neurons or suppression of PV+ inhibitory neurons in the OFCL led to enhanced sociability in female mice. This finding supports the idea that the excitation–inhibition balance in the prefrontal cortex is critical for shaping social behaviors in mice (Yizhar et al., 2011). Our data also suggest that adolescent social isolation influences male and female mice differently, both in how the prefrontal cortex matures and social behavior matures.
Importance of social experience during adolescence in shaping female social behavior
Social experience during adolescence is important for the maturation of the mammalian brain (Spear, 2000; Fuhrmann et al., 2015). In humans, social isolation during adolescence can increase the occurrence of neurologic disorders and cause defects in brain function (Almeida et al., 2021). Similarly, depriving rodents of social experience during adolescence caused defects in the maturation of their brains and alterations in social and cognitive behaviors (Cooke et al., 2000; Schubert et al., 2009; Makinodan et al., 2012; Hinton et al., 2019; Potrebic et al., 2022). Furthermore, behavioral and physiological changes caused by adolescent social isolation were often sex specific (Pisu et al., 2016; Bicks et al., 2020; Tan et al., 2021). In this study, we found that PWSI, particularly after puberty (from P50 to P80), impairs the maturation of inhibitory circuits in the OFCL and causes maladaptation in social exploration behavior in female mice but not in male mice. Our study is unique compared with recent studies that examined the effects of social isolation during early adolescence (Bicks et al., 2020; Tan et al., 2021). Our findings emphasize the importance of later adolescence when females adjust and shape their social-approaching behaviors with the maturation of PV+ neurons in the OFCL.
A recent study has shown that the 1dSI increased sociability in male mice (Matthews et al., 2016). We also found that male mice that had experienced 1dSI spent more time interacting with a novel mouse than did GH male mice (GH, 221.11 ± 17.93 s; 1dSI, 317.97 ± 31.46 s; p = 0.023, Student’s t test). However, the 1dSI male mice also showed a tendency of increased interaction with an object in the other container (GH, 107.29 ± 9.88 s; 1dSI, 138.28 ± 23.36 s; p = 0.256, Student’s t test). As a result, their social-sniffing and social interaction indices, which indicate the ratio between social interaction and object interaction times, did not change in 1dSI male mice (Fig. 1C,D). Our data suggest that 1dSI may enhance object exploration behaviors in male mice. This possibility was not tested in the previous report, as they used an empty chamber without any object as opposed to the social chamber. Future studies are required to understand the effect of 1dSI in male mice fully. Regardless, our study focuses on the social isolation effect on female mice, and we found no effects of 1dSI on social and object interaction times in female mice. Our data demonstrate that long-term social isolation during late adolescence affects social behaviors more effectively than the 1dSI in females.
PWSI not only deprives sensory experience from social interaction but also enhances stress in social mammals like mice (Fone and Porkess, 2008). It has been shown that both sensory deprivation and stress can impair the maturation of PV+ neurons in the sensory cortex as well as in the hippocampus (Huang et al., 1999; Czeh et al., 2005; Hu et al., 2010). The OFC, which is classified as a secondary olfactory cortex, is one of the key areas involved in olfactory processing (Lipton et al., 1999) and sniffing behaviors (Kareken et al., 2004). Furthermore, the olfactory system is the most important way through which mice detect other mice (Radyushkin et al., 2009). The OFC has a strong connection with the olfactory system and the hippocampus (Wikenheiser and Schoenbaum, 2016). An fMRI study in humans showed that in response to odorant cues with emotional valence, only females showed increased activity in their OFCL (Royet et al., 2003). Future studies are required to unravel which types of social olfactory cues and stress-related factors affect the maturation of PV+ neurons in the OFCL and social behavior.
Our data suggest that socially evoked activity in the OFCL is important for reducing social-approaching behaviors when the animal becomes an adult. Excessive sociability is a symptom of patients with Williams–Beuren syndrome (Mimura et al., 2010; Sakurai et al., 2011; Segura-Puimedon et al., 2014; Stoppel and Anderson, 2017) or Angelman syndrome (Stoppel and Anderson, 2017; Simchi and Kaphzan, 2021), rare neurodevelopmental genetic disorders. Interestingly, these patients show abnormal activity patterns in the OFC. Hypersociability is also found in some mouse models of autism (Yoo et al., 2019). Our data demonstrate that PWSI induced similar behavioral abnormalities in female mice, suggesting that the normal maturation and activity of the OFCL, at least in females, is important for maintaining an adequate level of sociability in such disorders.
PV expression in the OFCL is important for female social behavior
Recent studies found important roles of PV expression in the PFC in shaping social and cognitive behaviors. For example, several mouse models of autism that show impaired sociability have shown reduced PV expression in the mPFC (Filice et al., 2016; Kobayashi et al., 2018). Furthermore, maternal separation before weaning (Goodwill et al., 2018) or social isolation during early adolescence (Hinton et al., 2019) reduced PV expression in the OFC and impaired flexibility in behavioral decisions in rodents. Whole-brain knockout of PV caused the mice to show autism-like behavior (Wöhr et al., 2015), but it is still unclear whether the reduced PV expression in subregions of the PFC caused such behavioral deficits. Our study newly applied the CRISPR/Cas9 system to reduce PV expression locally in the OFCL. Using this method, we confirmed that PV expression in the OFCL during adolescence is critical for shaping social behaviors in female mice (Fig. 5). This technique can be widely used for studying area-specific PV protein function in various brain areas, although it lacks the reversibility of the loss of function by permanent gene deletion in vivo.
PV is a calcium-binding protein that plays a key role in regulating the physiological properties of PV+ neurons. In our study, knocking down PV decreased the excitability of the PV+ inhibitory neurons, which in turn decreased synaptic inhibition on nearby excitatory neurons (Fig. 6H–M). The relative reduction in inhibition compared with excitation in the OFCL potentially disrupted overall social sniffing-induced activity in the OFCL (Fig. 6F,G) and hypersociability (Fig. 5H). We further proved this idea by the optogenetic activation of excitatory neurons in the OFCL that led to hypersociability (Fig. 7). Moreover, a reduced number of PV+ neurons can also disrupt the network property of the OFCL, similar to the previous report showing that the loss of PV+ neurons correlated with a decrease in the network oscillation in the PFC (Lodge et al., 2009). Maturation of network activity during adolescence may be critical for shaping social sniffing-induced activity in the OFCL.
In a study of virgin female mice, PV and PNN expression in the auditory cortex was reported to be correlated with learning to retrieve pups (Krishnan et al., 2017); this activity, which occurs in the left auditory cortex, is important for shaping maternal behaviors in female mice (Marlin et al., 2015). Although several studies have shown that the expression of PNN is important for the maturation of PV+ neurons (Gogolla et al., 2009), we found that expression of the PV but not the PNN is important for the maturation of the female OFCL during late adolescence. Our data suggest that transcription or translation of PV and PNN may be separately regulated, and adolescent social isolation specifically disrupted PV expression only in a subset of neurons in the OFCL. Our data further support the idea that plasticity in PV+ neurons in the OFCL is important for shaping female sociability and is also characterized by hemispheric asymmetry. Both genetic and epigenetic mechanisms have been identified as being critical for regulating normal social behavior (Hensch, 2005), and thus social behaviors can be easily disrupted in a wide range of brain psychiatric disorders (Blakemore, 2008). Our study further provides insight into how the disruption of experience-dependent gene expression in the PFC inhibitory neurons, such as reduced PV expression in the OFCL, may underlie abnormal hypersociability that characterizes certain psychiatric disorders.
Footnotes
This work was supported by grants from the National Research Foundation of Korea (Grants 2021R1A2C3012159 and 2021R1A4A2001803, to S.-H.L.; and 2021R1A2C4001483, to J.-W.J.) and the KAIST Global Singularity Program for 2020 (to S.-H.L.). This work was also supported by the Rural Development Administration (Grant PJ016403, to S.K.) and by the Institute for Basic Science (Grants IBS-002-D1, to E.K.; and IBS-002-D3, to S.-H.L.). We thank all members of the Lee laboratory for helpful discussion, Dr. Soo Young Kim for discussion on postweaning social isolation, and Emily Wheeler (Boston, MA) for editorial assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Seung-Hee Lee at shlee1{at}kaist.ac.kr