INTRODUCTION

The amygdala has a prominent role in mood and emotion. Intertwined with this role is the importance of the amygdala in social behavior. The amygdala is engaged by a range of social stimuli (Breiter et al, 1996; Fried et al, 1997, 2002; Davis et al, 2010), amygdala damage leads to impaired processing of social information (Adolphs et al, 1994, 2005), and patient populations with abnormal social behaviors often display abnormal amygdala activity (Baron-Cohen et al, 1999; Grady and Keightley, 2002; Stein et al, 2002; Wang et al, 2004; Williams et al, 2004; Phan et al, 2006; Coccaro et al, 2007; Rilling et al, 2007; Marsh et al, 2008; Jones et al, 2009; Kleinhans et al, 2009; Pinkham et al, 2011; Richey et al, 2014). For instance, patients with autism display hypoactivity of the amygdala during implicit processing of facial expressions (Critchley et al, 2000), and while detecting faces (Ashwin et al, 2006), and display hypoactivity or activity that does not parallel the task demands while making judgement of facial expressions (Baron-Cohen et al, 1999; Wang et al, 2004), although this may depend partly on attention towards faces and gaze direction (Weng et al, 2011; Tottenham et al, 2014). Furthermore, the activity of the amygdala is less coordinated with other brain regions involved in social behavior (Kleinhans et al, 2008; Rudie et al, 2012; von dem Hagen et al, 2013). The medial nucleus of the amygdala (MeA) has been particularly implicated in prosocial maternal and sexual behavior, as well as agonistic aggressive behaviors. In rodent models, the MeA is activated by socially relevant cues, such as maternal interaction, mating, pheromones of opposite-sex conspecifics, and predators (Fleming and Walsh, 1994; Fleming et al, 1994; Kollack-Walker and Newman, 1997; Meredith and Westberry, 2004; Choi et al, 2005; Samuelsen and Meredith, 2009b; Bergan et al, 2014), ablation of the MeA impairs recognition of opposite-sex odors (Petrulis, 2009; Maras and Petrulis, 2010), and activation of MeA afferents induces a gradient of social behavior from mating behavior to aggression (Hong et al, 2014; Unger et al, 2015).

Social experience during critical developmental periods is required for normal maturation of social behaviors (Einon et al, 1981; Potegal and Einon, 1989; Ikemoto and Panksepp, 1992; Hol et al, 1999; Pellis et al, 1999; van den Berg et al, 1999a,b; Von Frijtag et al, 2002). Rearing of rats in social isolation after weaning leads to a range of abnormal social behaviors including increased aggression, abnormal interactions with novel rodents, abnormal sexual behavior, and impaired social learning (Gerall et al, 1967; Meaney and Stewart, 1981; Hol et al, 1999; van den Berg et al, 1999a; Cooke et al, 2000; Von Frijtag et al, 2002; Melo et al, 2006; Agis-Balboa et al, 2007; Toth et al, 2012; Yusufishaq and Rosenkranz, 2013). Furthermore, postweaning social isolation causes reduced MeA volume (Cooke et al, 2000), reduced MeA neurogenesis (Lieberwirth et al, 2012), increased neural cell adhesion molecule (Gilabert-Juan et al, 2012), and morphological changes of MeA neurons (Ichikawa et al, 1993). However, it is not known whether social isolation after rearing leads to changes in the function of MeA neurons. The purpose of this study is to test whether postweaning social isolation perturbs the activity of MeA neurons.

The posterior MeA (MeAp) is particularly important in conspecific or socially relevant responses (Meredith and Westberry, 2004; Samuelsen and Meredith, 2009b), and was the focus of these studies. The MeAp has two functionally distinct subdivisions, the dorsal (MeApd) and ventral (MeApv) subdivisions. In this study, the in vivo activity of MeApd and MeApv neurons was measured from rats that were group housed or socially isolated in single housing beginning after weaning. Intrinsic neuronal excitability and excitatory synaptic drive are two key components that determine neuronal activity. Therefore, to determine whether the underlying proximal cause for altered MeAp neuronal activity was due to abnormal synaptic or intrinsic excitability, synaptic physiology and membrane excitability were measured in vitro. In addition, emergence of abnormal activity over the course of postweaning maturation could be due to interference with maturational processes or due to progressive impairment. Therefore, these alternatives were tested by measurement of the emergence of impairments over the course of social isolation.

MATERIALS AND METHODS

All procedures had prior approval from the Institutional Animal Care and Use Committee of Rosalind Franklin University, and followed the Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). Care was taken to minimize animal distress and reduce the number of animals used. Male Sprague–Dawley rats (Harlan Laboratories, Indianapolis, IN) were weaned at postnatal day (P) 18, and upon arrival (P21–22) were housed at the Rosalind Franklin University animal facility (polycarbonate solid bottom cage, 43.2 × 21.5 × 20.3 height, in cm; Teklad Sani-Chips bedding). The rats were provided water and food (Rodent Diet 2020X pelleted feed, Harlan Teklad) ad libitum. The housing rooms were set to a 12 h/12 h reverse light–dark cycle with lights off at 0700 hours. Temperature was maintained between 64 and 79 °F and the humidity was maintained between 30 and 70%. Rats were randomly assigned to control group housing (2–3/cage) or social isolation single housing (1/cage). All other aspects of husbandry were the same. Rats were used for experiments after 4–5 weeks (P50–60 postadolescents), or at intermediate time points, as described below. Several studies indicate that social behavior in males may be more sensitive to postweaning social isolation (Ferdman et al, 2007; Wall et al, 2012; Ahern et al, 2016). Therefore, males were the focus of the current study.

Experimental Outline

Rats were group housed our single housed on the day of arrival, and then tested at noted timepoints. To test the effects of social isolation on MeA activity, in vivo electrophysiological measures were obtained from young adults (after more than 4 weeks; P50–60). Both single-unit (group housing n=9 rats; social isolation single housing n=10 rats) and evoked field potentials were measured from MeApd and MeApv (MeApd group housing n=9 rats, social isolation single housing n=8 rats; MeApv group housing, n=9 rats, social isolation single housing, n=7 rats). To test the proximal causes for a change in MeA activity, in vitro whole-cell recordings were also obtained from young adults to measure synaptic activity (8 rats per group) and membrane responsiveness (8 rats per group). To determine the time course of the changes over the postweaning period, in vitro whole-cell recordings were obtained after 1 day (P22–23, n=8 rats per group), 1 week (P28–29, group housing, n=8 rats, single housing, n=7 rats), 2 weeks (P35–36, group housing, n=7 rats, single housing, n=7 rats), or as young adults (P50–60, n=8 rats per group).

In vivo Electrophysiology

Rats were anesthetized with urethane (1.5 g/kg, intraperitoneally) and 2% lidocaine jelly was infiltrated into their ears. Rats were placed in a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA or Stoelting Instruments, Wood Dale, IL), and core body temperature was maintained at ~37 °C (TC-1000 Temperature Controller; CWE, Ardmore, PA). Bore holes were drilled in the skull overlying the lateral nucleus of the amygdala (LAT; centered on −5.0 mm M–L, −3.0 mm A–P from bregma) and the MeA (centered on −3.2 mm M–L, −3.3 mm A–P from bregma, between 8.0 and 9.5 mm D–V). A concentric bipolar stimulation electrode (Rhodes Medical Instrument, Summerland, CA) was slowly lowered into the LAT to deliver stimulation and to record spontaneous local field potentials as a gauge of anesthesia state. After a minimum of 45 min, a glass microelectrode (2.0 mm outer diameter borosilicate; A–M Systems, Carlsborg, WA) was pulled (PE-2 microelectrode puller; Narishige Group, Tokyo, Japan) and filled with 2% Pontamine Sky Blue (Alfa Aesar, Ward Hill, MA) in 2 M NaCl (Thermo Fisher Scientific, Waltham, MA). The glass electrode was slowly lowered to the MeA to record neuronal activity. Signals were amplified (2400 Extracellular Preamplifier; Dagan, Minneapolis, MN) and digitized (5–10 kHz; InstruTECH ITC-18; HEKA Instruments, Bellmore, NY). Signals were monitored audially (AM10 amplifier; Grass Technologies, Warwick, RI) and visually (AxoGraph X software version 1.3.5; Axograph Scientific, Sydney, Australia) and saved for later analysis (Mac Pro; Apple, Cupertino, CA). Electrical stimulation was delivered (S88 Stimulator and PSIU6 Stimulation Isolation Unit; Grass Technologies, Warwick, RI) through the bipolar electrode with an intensity range of 0.1–0.9 mA, 0.2 ms duration, repeating at 0.2–40 Hz.

At the conclusion of experiments, Pontamine Sky Blue was iontophoresed through the recording electrode (−30 μA, >30 min; Constant Current Source, Fintronics, Orange, CT). The brain was removed and placed in 4% paraformaldehyde in 0.1 M phosphate buffer overnight, and then transferred to 25% sucrose (Thermo Fisher Scientific, Bannockburn, IL) in 0.1 M phosphate buffer. Brains were sectioned (60 μm thick) with a freezing microtome (Leica Microsystems, Buffalo Grove, IL) and stained with cresyl violet (Sigma-Aldrich, St Louis, MO). Recording and stimulation sites were verified by light microscopy.

In vitro Electrophysiology

Rats were deeply anesthetized with a mixture of ketamine (80–100 mg/kg; Webster Veterinary Supply, Devens, MA) and xylazine (10–20 mg/kg; Webster Veterinary Supply). Upon verification of deep anesthesia by the absence of response to foot pinch, rats were decapitated, and brains were rapidly immersed in ice-cold, aerated (95% O2/5% CO2) high sucrose artificial cerebrospinal fluid (ACSF) containing (in mM) 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 7 dextrose, 7 MgCl2, 0.5 CaCl2, 210 sucrose, 1.3 ascorbic acid, 3 sodium pyruvate, and ~290 mOsm. The brain was sectioned at 250–300 μm (Vibratome 1000; Ted Pella, Redding, CA) in the same ACSF. Brain sections were then placed in physiological ACSF saturated with 95% O2/5% CO2, containing (in mM) 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 10 dextrose, 1 MgCl2, and 2 CaCl2, with the addition of 1.3 mM ascorbic acid and 3 mM sodium pyruvate for 50–60 min at 34 °C. Recordings were performed at 32–34 °C in submerged slices in physiological ACSF (as above, without ascorbic acid or sodium pyruvate). When specified, (+)-bicuculline (10 μM; Ascent Scientific, Princeton, NJ; dissolved in dimethyl sulfoxide), picrotoxin (10 μM; Sigma-Aldrich; dissolved in ethanol), 6-cyano-7-nitroquinoxaline-2,3-dion (CNQX) disodium salt (10 μM; Ascent Scientific; dissolved in ddH2O), and DL-2-amino-5-phosphonopentanoic acid sodium salt (50 μM; Abcam Biochemicals, Cambridge, MA; dissolved in 100 mM NaOH) were added to the ACSF to block GABAA-, AMPA- and NMDA receptor-mediated currents. Final solvent concentrations were <0.1% of the total ACSF volume.

Recording pipettes were pulled (borosilicate glass, 2.0 mm outer diameter, 2–8 MΩ open tip resistance; Sutter Instruments, Novato, CA). Electrodes were filled with (in mM) 120 K-gluconate, 20 KCl, 0.2 EGTA, 10 HEPES, 2 NaCl, 4 ATP-Mg, 0.3 GTP-Tris, 7 Tris-phosphocreatine, and 0.2% neurobiotin (Vector Laboratories, Burlingame, CA), with a pH of 7.3. Whole-cell recordings were performed from visualized neurons in the MeAp under IR-DIC conditions. Signals were amplified (AxoClamp 2B; Molecular Devices, Sunnyvale, CA) and low-pass-filtered at 3–5 kHz and digitalized at 10 kHz (InstruTECH ITC-18; HEKA Instruments). Mean series resistance for each group was below 25 MΩ. Data were monitored and saved for later analysis (AxoGraph X software v.1.3.5 (Axograph Scientific), on Mac Pro (Apple)). Upon completion of recordings, slices were fixed overnight and stored for up to 3 weeks at 4 °C (4% paraformaldehyde in 0.1 M phosphate-buffered saline, PBS). Sections were rinsed three times with PBS, treated with Triton X-100 (VWR International, Radnor, PA; 1% in PBS) for 6 to 8 h, and then incubated in the Vectastain ABC Reagent (Vector Laboratories) in PBS at room temperature overnight. After three rinses with PBS, sections were reacted with diaminobenzidine (DAB) and H2O2 (Peroxidase Substrate Kit DAB; Vector Laboratories) in water to visualize the neurobiotin-filled neurons. Sections were washed in PBS repeatedly to stop the reaction. Sections were mounted, dried, and coverslipped. Stained sections were used to localize the recording sites.

Data Analysis and Statistics

Recording sites in vivo and in vitro that were confirmed to lie within the MeAp were analyzed. MeApd and MeApv subdivisions were delineated based on the atlas of Paxinos and Watson (Paxinos, 2009). In vivo firing rate was quantified as the number of extracellular action potentials per s over at least 5 min time epochs. Extracellular action potentials required a signal-to-noise ratio of at least 3:1, consistent amplitude, and half-width (<15% change over 5 min) for inclusion in analysis. Evoked field potentials were measured upon stimulation of the LAT (0.2 Hz, 0.1 ms duration). When LAT train stimulation was examined (10–40 Hz), a stimulation intensity to evoke a half-maximal response was used, with an intertrain interval of 30 s. A minimum of 10 sweeps was averaged and the slope of the field potential was measured. Excitatory postsynaptic currents (EPSCs) were measured in vitro with whole-cell recordings in the presence of GABA and NMDA receptor blockers (above). A bipolar stimulation electrode was placed ~100 μm from the soma of the recorded neuron to evoke excitatory postsynaptic responses (0.2 Hz, 0.1 ms duration, intensity to evoke 30–80 pA EPSC). A minimum of 20 sweeps was averaged and the peak amplitude of the EPSC was measured. Miniature excitatory postsynaptic currents (mEPSCs) were measured in the presence of GABA blockers (above) and TTX (1 μM) and quantified using a variable amplitude template (Axograph X), with a required signal-to-noise ratio of three times the standard deviation of the noise. Frequency, amplitude, half-width, and decay were quantified. Resting membrane potential (Vrest) was measured as the membrane voltage in the absence of holding current in the voltage clamp. In separate experiments, GABA receptor blockers (above) and glutamate receptor (NMDA and AMPA) blockers were applied and subsequent whole-cell voltage measures were obtained in bridge mode. In these experiments, Vrest was measured as the membrane voltage in the absence of current. The remainder of the voltage recordings were obtained while holding the membrane potential close to −70 mV. Input resistance (Rn) was measured as the linear fit of the voltage response to hyperpolarizing current steps (−10 to −40 pA). Membrane excitability was measured as the number of action potentials evoked by depolarizing current steps (20–80 pA). This measure was best fit to a second-order polynomial to derive the current amplitude that evokes a 50% response (EC50) and the extrapolated maximal number of action potentials (Bmax). Neurons that displayed electrophysiological parameters that match GABAergic interneurons of the MeA (Keshavarzi et al, 2014) or displayed >15% variability across time in holding current, Vrest, input resistance or access resistance were not included for analysis. Data were analyzed using Prism (version 6.0; GraphPad Software, La Jolla, CA), with a p<0.05 considered statistically significant. Data were tested for normal distribution (D’Agostino and Pearson’s normality test) and homogeneity of variance (Bartlett’s test). Planned comparisons across one factor were performed with a two-tailed unpaired t-test. If more than two groups were compared, a one-way ANOVA was used instead. When multiple factors were compared between groups, a two-way ANOVA (or if applicable, two-way repeated-measures (RM)-ANOVA) was used. ANOVAs were followed with comparisons between individual groups using Holm–Sidak’s multiple comparisons tests. Data are presented as mean±SEM, unless noted otherwise.

RESULTS

The firing rate of MeA neurons was measured after >4 weeks of control group housing or social isolation single housing (N=9–10 rats per group, see Materials and Methods section). Neurons were histologically confirmed to lie within the MeApd or MeApv (Figure 1a). Social isolation significantly decreased the in vivo firing rate of MeA neurons (Figure 1b; control 3.0±0.4 Hz, n=33 neurons, social isolation 1.3±0.2 Hz, n=35 neurons; p=0.0004, t=3.74, two-tailed unpaired t-test). This decreased neuronal firing was observed in both MeApd (Figure 1c; p=0.048, t=2.07, d.f.=30; control 3.6±0.7 Hz, n=17 neurons, social isolation 1.9±0.4 Hz, n=15 neurons) and MeApv (Figure 1c; p=0.0008, t=3.70; control 2.7±0.5 Hz, n=16 neurons, social isolation: 0.9±0.2 Hz, n=20 neurons). The firing of MeA neurons can be driven by their response to excitatory inputs. MeA receives excitatory glutamatergic inputs from several regions, including the LAT (Pitkanen et al, 1995). LAT was chosen for stimulation because its output is glutamatergic (McDonald, 1996) and it has functional importance in guidance of affective behavior, including some social behaviors (Hoffman et al, 2007; Davis et al, 2010; Jeon et al, 2010; Gadziola et al, 2012). LAT stimulation evoked a stimulation intensity-dependent local field potential in the MeA (Figure 1d; slope of the local field potential across intensity, main effect of intensity p<0.0001, F(4,124)=360.4, two-way RM-ANOVA). Social isolation significantly decreased the response of the MeA to LAT stimulation (Figure 1d; main effect of social isolation p=0.001, F(1,31)=13.36, social isolation × intensity interaction p<0.0001, F(4,124)=15.55, two-way RM-ANOVA, N=9 rats per group; p<0.05 at 0.5–0.9 mA, Holm–Sidak’s multiple comparisons test). The effect of social isolation on evoked excitatory input was specific for MeApd (Figure 1e; social isolation × intensity interaction p<0.0001, F(4,60)=18.47, control n=9 rats, social isolation n=8 rats) and not statistically significant in MeApv (Figure 1e; social isolation × intensity interaction p=0.081, F(4,56)=2.19, two-way RM-ANOVA, control n=9 rats, social isolation n=7 rats). This indicates that, while social isolation has similar overall effects on the firing of MeApd and MeApv neurons, different mechanisms may underlie these effects in MeApd and MeApv.

Figure 1
figure 1

Postweaning social isolation decreases the in vivo activity of medial nucleus of the amygdala (MeA) neurons. (a) Recording sites in MeApd (MeA dorsal) and MeApv (MeA ventral) neurons. MeApd and MeApv were histologically verified by the deposit of Pontamine Sky Blue (white arrow). The section displayed here corresponds to −3.30 mm caudal from bregma in the atlas of Paxinos, 2009. (b) In vivo extracellular recordings (left) were obtained from control group-housed or age-matched isolation-reared rats. The firing rate of MeA neurons from isolation-reared rats was significantly less than group-housed rats (box plot±Tukey). *p<0.05 two-tailed unpaired t-test. (c) The firing rate of MeApd and MeApv neurons from isolation-reared rats was significantly less than group-housed rats. *P<0.05 in post hoc Holm–Sidak’s multiple comparisons test after two-way repeated-measures analysis of variance (ANOVA). (d) Field potentials evoked by stimulation of lateral nucleus of the amygdala (LAT) displayed stimulation intensity-dependent amplitude (left). (e) The slope of the evoked field potentials was significantly lower in the isolation-reared group in MeApd (left) but not MeApv (right). *P<0.05 two-way repeated-measures ANOVA.

PowerPoint slide

Full size image

Several mechanisms can cause decreased firing and response to excitatory afferents, including presynaptic and postsynaptic factors. To test the proximal cause of the effect of social isolation, we first measured postsynaptic aspects that influence neuronal responsiveness while synaptic inputs were blocked in vitro (see Materials and Methods section, n=8 rats per group). There was no significant difference in the resting membrane potential between control and social isolation groups (Figure 2a; Vrest control −60.6±0.6 mV, n=24 neurons, social isolation −61.9±0.7 mV, n=27 neurons, p=0.182, t=1.364, d.f.=34, two-tailed unpaired t-test). Input resistance was measured to gauge neuronal membrane responsiveness (Figure 2b). Social isolation significantly decreased input resistance (control 206.9±15.1 MΩ, n=18 neurons, social isolation 151.9±13.1 MΩ, n=18 neurons, p=0.009, t=2.76, d.f.=34, two-tailed unpaired t-test). However, these effects of social isolation were nucleus-specific, with reduced input resistance of MeApv neurons but not MeApd neurons (Figure 2c; social isolation × nucleus interaction p=0.037, F(1,32)=4.738; Holm–Sidak’s multiple comparisons test MeApd p>0.05, control n=9 neurons, social isolation n=7 neurons, MeApv p<0.05 (control n=9 neurons, social isolation n=11 neurons)).

Figure 2
figure 2

Postweaning social isolation decreases membrane excitability of medial nucleus of the amygdala (MeA) neurons. MeA neurons were recorded in vitro using whole-cell recordings. Blockers of γ-aminobutyric acid (GABA) and glutamate were present during recordings. (a) The resting membrane potential was not significantly different between group-housed and isolation-reared rats. (b) Input resistance was measured as the voltage response to hyperpolarizing current steps (left). (c) Social isolation decreased the input resistance in neurons from the MeApv (MeA ventral) but not the MeApd (MeA dorsal) neurons (box plot±Tukey). *P<0.05 in post hoc Holm–Sidak’s multiple comparisons test after two-way repeated-measures analysis of variance (ANOVA). (d) Membrane excitability was quantified as the number of action potentials in response to depolarizing current steps (left). (e) Social isolation decreased membrane excitability in neurons from the MeApv but not the MeApd. *P<0.05 two-way repeated-measures ANOVA.

PowerPoint slide

Full size image

Membrane excitability, measured as action potential firing in response to depolarization (from −70 mV; Figure 2d), was significantly decreased after social isolation (main effect of social isolation p<0.0001, F(4,112)=60.21; social isolation × current step amplitude interaction p<0.0001, F(4,112)=10.55, two-way RM-ANOVA; significant differences at 40–80 pA current steps, p<0.05 Holm–Sidak’s multiple comparisons test). These effects of social isolation were similarly nucleus-specific, with reduced membrane excitability only in MeApv neurons (Figure 2e; main effect of social isolation p=0.0049, F(1,25)=9.516; social isolation × current step amplitude interaction p<0.0001, F(4,100)=6.676; Holm–Sidak’s multiple comparisons test p<0.05, 40–80 pA, n=13 control neurons, n=14 social isolation neurons), and not MeApd neurons (main effect of social isolation p=0.47, F(1,22)=0.551; social isolation × current step amplitude interaction p=0.523, F(4,88)=0.809; n=11 control neurons, n=13 social isolation neurons). This demonstrates a postsynaptic reduction of neuronal excitability that can contribute to decreased firing of MeApv neurons.

Neuronal activity can be driven by excitatory synaptic inputs. Therefore, presynaptic factors that influence the release of glutamate can also significantly impact neuronal activity. To test if social isolation decreases the strength of glutamatergic inputs, glutamatergic synaptic currents were measured in vitro (Figure 3a; n=9 rats per group, see Materials and Methods section). The frequency of mEPSCs was significantly lower after social isolation (Figure 3a; control 3.9±0.5 Hz, n=22 neurons, social isolation 2.0±0.3 Hz, n=20 neurons, p=0.0069, t=2.847, two-tailed unpaired t-test). Social isolation decreased mEPSC frequency recorded from MeApd but not MeApv neurons (Figure 3b; nucleus × social isolation interaction, p=0.049, F(1,38)=4.133, two-way ANOVA; Holm–Sidak’s multiple comparisons test, MeApd p<0.05, control n=12 neurons, social isolation n=10 neurons; Holm–Sidak’s multiple comparisons test, MeApv p>0.05, control n=10 neurons, social isolation n=10 neurons). There was no significant effect of social isolation on mEPSC amplitude (Figure 3c; control 10.5±0.4 pA, n=22 neurons, social isolation 9.7±0.6 pA, n=20 neurons, p=0.246, t=1.177, two-tailed unpaired t-test) and no significant difference in mEPSC decay time (control 4.3±0.3 ms, n=22 neurons, social isolation 4.1±0.3 ms, n=20 neurons; p=0.81, t=0.237, two-tailed unpaired t-test; control 4.3±0.3 ms, n=22 neurons, social isolation 4.2±0.3 ms, n=20 neurons). Paired-pulse ratio was measured as an additional index of presynaptic function. Social isolation caused a change in paired-pulse response, from slight facilitation towards depression (Figure 3d; p=0.006, t=2.944, d.f.=35, two-tailed unpaired t-test; control n=18, social isolation n=19), but this was specific for the paired-pulse response of MeApd neurons but not MeApv neurons (Figure 3d; nucleus × social isolation interaction, p=0.032, F(1,34)=5.035, two-way ANOVA; Holm–Sidak’s multiple comparisons test, MeApd p<0.05, control n=8 neurons, social isolation n=9 neurons; Holm–Sidak’s multiple comparisons test, MeApv p>0.05, control n=10 neurons, social isolation n=11 neurons). This demonstrates that social isolation modifies presynaptic factors in MeApd but not MeApv neurons.

Figure 3
figure 3

Postweaning social isolation decreases excitatory synaptic input to medial nucleus of the amygdala (MeA) neurons. MeA neurons were recorded in vitro using whole-cell recordings in the presence of γ-aminobutyric acid (GABA) blockers to isolate glutamatergic excitatory postsynaptic current (EPSC) inputs. (a) Miniature EPSCs (mEPSCs) were measured (+1 μM tetrodotoxin (TTX)) from group-housed and isolation-reared rats. The frequency of mEPSCs was significantly less in isolation-reared rats, demonstrated in cumulative frequency plots from exemplar MeA neurons from group-housed and isolation-reared rats. (b) Social isolation decreased the frequency of mEPSC MeApd (MeA dorsal) neurons but not MeApv (MeA ventral) neurons (*p<0.05 in post hoc Holm–Sidak’s multiple comparisons test after two-way repeated-measures analysis of variance (ANOVA)). (c) The amplitude of mEPSCs was not significantly different between group-housed and isolation-reared rats, in cumulative frequency histogram (left) and grouped average data (right). (d) The ratio of the amplitude of pairs of evoked EPSCs (paired-pulse ratio, 50 ms interval) was measured as an index of presynaptic differences (left). There was a significant shift towards depression of excitatory synaptic input in MeApd neurons from isolation-reared rats but not MeApv neurons (*p<0.05 in post hoc Holm–Sidak’s multiple comparisons test after two-way repeated-measures ANOVA).

PowerPoint slide

Full size image

These data indicate that presynaptic and postsynaptic factors can contribute to in vivo impairments of the MeA after social isolation, but in a manner that depends upon the subdivision of the MeA. Presynaptic and postsynaptic factors contribute to different aspects of neuronal integration of synaptic inputs. Therefore, it is expected that presynaptic and postsynaptic abnormalities will lead to different abnormalities of synaptic integration, and would therefore lead to different impairments of synaptic integration in the MeApd and MeApv. To test whether these two types of impairments lead to different in vivo impairments, integration of synaptic inputs in MeA was measured during stimulation of LAT at 10, 20, or 40 Hz (Figure 4a). The ratio of the last and first evoked responses was used as an index of integration (Figure 4b; see Materials and Methods section). Synaptic integration was maximal at 20 Hz (integration index, main effect of frequency p<0.0001, F(2,93)=25.63; integration index at 20 Hz significantly >10 and 40 Hz, p<0.05 Holm–Sidak’s multiple comparisons tests). Social isolation reduced the responsiveness of the MeA when measured at 10 Hz (control n=16, social isolation n=22 rats; social isolation × pulse number interaction p<0.0001, F(9,324)=9.01), 20 Hz (control n=15 rats, social isolation n=16 rats; social isolation × pulse number interaction p<0.0001, F(9,261)=9.77), and 40 Hz (control n=14 rats, social isolation n=16 rats; social isolation × pulse number interaction p=0.009, F(9,252)=2.51). When analyzed by the nucleus, differences in the pattern of the effects of social isolation emerged (Figure 4c, dotted line). Social isolation similarly reduced integration in MeApv across stimulation frequencies (Figure 4e; main effect of frequency p=0.861, F(2,24)=0.1511; frequency × pulse number interaction p=0.328, F(18,216)=1.126, two-way RM-ANOVA), a pattern expected with generally decreased neuronal membrane responsiveness. In contrast, the effects of social isolation on integration in the MeApd depended on the stimulation frequency (Figure 4d; main effect of frequency p=0.002, F(2,24)=8.082; frequency × pulse number interaction p<0.0001, F(18,216)=11.62, two-way RM-ANOVA), a pattern expected with frequency-dependent synaptic deficiency.

Figure 4
figure 4

Postweaning social isolation impairs integration of synaptic inputs in vivo. (a) The lateral nucleus of the amygdala (LAT) was stimulated in train of 10 pulses (arrows) to evoke 10 field potentials at 10, 20, or 40 Hz. The slope of the evoked field potentials was measured. (b) An integration index can be quickly gauged by assessing the final field potential of the train (normalized to the first field potential; s10/s1). A larger integration index indicates better summation of inputs. Social isolation decreases the integration index. (c) The integration index is maximal at 20 Hz, indicative of optimal summation at this frequency. The pattern of the effects of social isolation on integration index differed between MeApd (MeA dorsal) (reduction dependent on frequency; significant social isolation × frequency interaction, two-way repeated-measures analysis of variance (RM-ANOVA)) and MeApv (MeA ventral) neurons (general reduction across frequencies; main effect of social isolation, no significant social isolation × frequency interaction, two-way RM-ANOVA). (d) The frequency-dependent effects of social isolation on integration in MeApd is verified by measuring the facilitation ratio across the stimulation trains (left; *p<0.05 two-way RM-ANOVA)). The effects of social isolation in MeApd (social isolation—control) is different at 10, 20, and 40 Hz (right; *p<0.05 two-way RM-ANOVA). (e) In contrast, while social isolation decreased the integration in MeApv at each frequency (*p<0.05 two-way RM-ANOVA), the pattern of effects of social isolation in the MeApv is similar at 10, 20, and 40 Hz (left). The effect of social isolation in MeApv (social isolation—control) is not different at 10, 20, and 40 Hz (right).

PowerPoint slide

Full size image

The MeA is still maturing during the postweaning period (Schulz et al, 2009; Cooke, 2011; De Lorme et al, 2012; Bergan et al, 2014). Social isolation can stunt the maturation of other brain regions (Sanchez et al, 1995; Helmeke et al, 2001; Lapiz et al, 2003; Gos et al, 2006; Agis-Balboa et al, 2007; Gilabert-Juan et al, 2012; Makinodan et al, 2012). The MeA abnormalities caused by social isolation may also be due to a stunting of maturation, such that MeA neurons display an immature phenotype. Alternatively, the abnormalities caused by social isolation may be due to an increasing impairment of function that arises over time. To test these possibilities, MeA neurons were recorded at several ages over the course of postweaning social isolation (Figure 5a). There was little evidence for postweaning developmental change in membrane responsiveness (Figure 5b; input resistance, main effect of age p=0.732, F(3,95)=0.430, one-way ANOVA; adult n=23 neurons per 9 rats, P22–23 n=26 neurons per 8 rats, P28–29 n=25 neurons per 7 rats, P35–36 n=25 neurons per 7 rats) or excitability across measured ages, with only a small, not significant, shift in group-housed rats across age (Figure 5c; main effect of age p=0.605, F(3,88)=0.618; age × current step interaction p=0.861, F(12,352)=0.576, two-way RM-ANOVA; adult n=24 neurons per 9 rats, P22–23 n=25 neurons per 8 rats, P28–29 n=19 neurons per 7 rats, P35–36 n=24 neurons per 7 rats). This was further verified by measurement of EC50 and Bmax across age in group-housed rats (Figure 5d and e; EC50 p=0.926, F(3,80)=0.156, one-way ANOVA; Bmax p=0.955, F(3,73)=0.108, one-way ANOVA). Thus, there did not appear to be ongoing maturation of membrane excitability across this age range. In contrast, there was a clear shift in measures of presynaptic function between puberty and adulthood, with an increased frequency of mEPSCs (Figure 6a; p=0.030, F(3,103)=3.091, one-way ANOVA) and a shift in paired-pulse ratio (Figure 6b; p=0.029, F(3,55)=3.247, one-way ANOVA). On closer examination, this maturation of excitatory input was localized to MeApd (Figure 6a and b; mEPSC frequency, p=0.043, F(3,56)=2.894; paired-pulse ratio, p=0.034, F(3,24)=3.404, one-way ANOVA), with no significant postweaning change in presynaptic function measured in MeApv (mEPSC frequency, p=0.374, F(3,43)=0.374; paired-pulse ratio, p=0.449, F(3,28)=0.9101, one-way ANOVA). In summary, postweaning maturation was observed only in MeApd, and it was specific for maturation of excitatory synaptic input.

Figure 5
figure 5

The time course of the isolation-rearing-induced deficits favors progressive impairment of MeApv (medial amygdala ventral) membrane excitability. (a) In vitro measures of excitability were obtained at P22–23, P28–29, P35–36, and >P50. (b) Input resistance of MeApd and MeApv neurons at different time points from weaning were similar across age in the group-housed rats. However, social isolation reduced membrane responsiveness in neurons of the MeApv but not the MeApd (medial amygdala dorsal) (*p<0.05 two-way repeated-measures ANOVA). (c) Membrane excitability was also measured and did not change across age in group-housed rats. However, social isolation decreased membrane excitability in MeApv neurons, but not MeApd neurons. To facilitate comparison, membrane excitability was assessed as Bmax (d) and EC50 (e). There was no significant effect of age on MeA neuron excitability in group-housed rats. However, Bmax and EC50 were both reduced in MeApv neurons from isolation-reared rats compared with group-housed rats, but not in MeApd neurons (*p<0.05 two-way repeated-measures ANOVA). Bmax, extrapolated maximal number of action potentials; EC50, current amplitude that evokes a 50% response.

PowerPoint slide

Full size image
Figure 6
figure 6

The time course of the isolation-rearing-induced deficits indicates stunted maturation of excitatory inputs to MeApd (medial amygdala dorsal). In vitro measures of presynaptic function were obtained at P22–23, P28–29, P35–36, and >P50. (a) There was a significant increase in excitatory input (miniature excitatory postsynaptic currents (mEPSC) frequency) to MeApd neurons across age in group-housed rats, consistent with periadolescent maturation. This increase was not observed in isolation-reared rats (*p<0.05 Holm–Sidak’s after significance in two-way analysis of variance (ANOVA)). In contrast, there was no evidence of periadolescent changes in mEPSC frequency to MeApv (medial amygdala ventral) neurons of group-housed or isolation-reared rats (right). (b) To further test the presynaptic locus of these changes, paired-pulse ratio was measured. There was a significant shift of paired-pulse ratio of excitatory input to MeApd neurons across age in group-housed rats (left), consistent with presynaptic periadolescent maturation (*p<0.05 two-way ANOVA). This increase was not observed in isolation-reared rats. No effect of age or isolation was observed on paired-pulse ratio of excitatory inputs to MeApv neurons (right).

PowerPoint slide

Full size image

Social isolation caused a progressive decrease of postsynaptic membrane responsiveness (input resistance, main effect of social isolation p=0.041, F(1,178)=4.237, two-way ANOVA) and a parallel decrease of membrane excitability across age, measured as EC50 (main effect of social isolation p=0.033, F(1,163)=4.602, two-way ANOVA). This effect of social isolation on postsynaptic neuronal input resistance and excitability was observed in MeApv neurons (Figure 5b and d; input resistance main effect of social isolation p=0.023, F(1,90)=5.322, two-way ANOVA; EC50 main effect of social isolation p=0.050, F(1,82)=3.910, two-way ANOVA) but not MeApd neurons (Figure 5b and d; input resistance main effect of social isolation p=0.907, F(1,80)=0.0139, two-way ANOVA; EC50 main effect of social isolation p=0.465, F(1,74)=0.5407, two-way ANOVA). Because these postsynaptic properties did not change in control group-housed rats across the postweaning period, this effect of social isolation is consistent with a progressive impairment of postsynaptic MeApv neuronal excitability that emerges over the course of social isolation, instead of a stunting of maturation.

In contrast, while control rats displayed maturation of presynaptic function across the postweaning period, social isolation rats did not display this maturation (mEPSC frequency p=0.899, F(3,95)=0.1954, one-way ANOVA; paired-pulse ratio p=0.992, F(2,39)=0.0081, one-way ANOVA). There was a significant difference between group-housed and isolation-housed rats in presynaptic function, but this was observed only in MeApd neurons, with differences in mEPSC frequency (Figure 6a; main effect of social isolation p=0.047, F(1,103)=4.0418, two-way ANOVA) and paired-pulse ratio (Figure 6b; main effect of social isolation p=0.023, F(1,48)=5.533, two-way ANOVA). No effect of social isolation on mEPSC frequency or paired-pulse ratio was observed in MeApv neurons (Figure 6b and d; mEPSC frequency p=0.662, F(1,87)=0.1930, two-way ANOVA; paired-pulse ratio main effect of social isolation p=0.403, F(1,58)=0.7084, two-way ANOVA). This indicates that a maturational stunting of excitatory drive occurs in the MeApd of rats socially isolated during the postweaning period.

DISCUSSION

The MeA has a fundamental role in a range of social behaviors. Postweaning social isolation causes impairments in many of these social behaviors. Here we found that postweaning social isolation decreased the activity of neurons in both the MeApd and MeApv. However, the underlying pathophysiology differs between these nuclei, with impaired excitatory input in the MeApd but reduced membrane excitability in the MeApv.

The postweaning period may be especially sensitive to disruptions of the social environment because social experience during this critical period is required for normal development. Postweaning social isolation within a critical preadolescent time window leads to long-term effects on social behaviors, including behaviors that rely on the MeA, such as sexual mounts/intromissions and non-contact penile erections, social aggression and avoidance, defensive response patterns during social aggression, and reduced social vocalizations (Ward and Reed, 1985; Bakker et al, 1995; Wongwitdecha and Marsden, 1996; Pellis et al, 1999; Cooke et al, 2000; Lukkes et al, 2009; Shoji and Mizoguchi, 2011; Seffer et al, 2015), and other behaviors (Hall et al, 1997, 1998; Varty and Geyer, 1998; Bakshi and Geyer, 1999; Makinodan et al, 2012). Effects of postweaning social isolation on these social behaviors may be mediated by the MeA. By analogy to other brain regions, MeA development during the postweaning period may require critical input to mature normally, such as that offered by social experience. The absence of social experience during this critical window may lead to the absence of social experience-dependent plasticity. The MeA undergoes prepubertal maturation and peripubertal refinement in response to gonadal hormones that parallels the refinement of these social behaviors (Cooke, 2006). There is enlargement of the MeApd during puberty (Romeo and Sisk, 2001; Chareyron et al, 2012; De Lorme et al, 2012) that is coupled with increased synaptophysin (Zehr et al, 2006), PSD-95, and vesicular glutamate transporter 2, consistent with increased number of excitatory synaptic inputs (Cooke, 2011). Disruption of synaptic refinement in MeApd may contribute to the effects of social isolation on social behaviors. Indeed, the normal peripubertal maturation of excitatory synaptic input, as reflected in the frequency of EPSCs and paired-pulse ratio, is stunted in the MeApd of socially reared rats (Figure 6). Further supporting this interpretation, previous findings indicate that social isolation impedes normal maturation of MeA volume (Cooke et al, 2000). In addition to interference with maturation, social isolation caused de novo impairments in MeApv. Progressively reduced membrane excitability of MeApv neurons during social isolation is expected to also contribute to behavioral impairments, and could impede the normal refinement of MeA-dependent social behaviors during development. The driving force behind these impairments caused by social isolation may be similar to long-term changes caused by sensory deprivation during critical developmental periods. Monocular deprivation and dark rearing during critical developmental windows cause transient changes followed by long-term decreases in the intrinsic and synaptic responsiveness of visual cortex (Hubel et al, 1977; Fagiolini et al, 1994; Gordon and Stryker, 1996; Smith et al, 2009; Nataraj et al, 2010). The net result of these effects of social isolation on pre- and postsynaptic function is decreased ability of MeAp neurons to become activated. This decreased responsiveness may lead to decreased guidance of social behaviors by the MeAp in adults. In particular, inputs from the basolateral amygdala, including the lateral nucleus, guide behaviors in response to conditioned social cues (Martinez et al, 2011; Takahashi et al, 2013), and inputs from olfactory regions guide social behaviors in response to pheromones and other odor cues (Kang et al, 2009; Samuelsen and Meredith, 2009a; DiBenedictis et al, 2015). Diminished responses to these inputs is expected to lead to reduced production of social behaviors (DiBenedictis et al, 2015; Noack et al, 2015). However, the outcome of these impairments on MeApd and MeApv is different, with a general decrease of responsiveness in MeApv but a more specific deficit in MeApd at higher frequency inputs. This may lead to continuous impairment of MeApv function, but impaired MeApd function only when the MeA is challenged by novel or complex social situations.

There are two primary types of neurons in the MeA, principle projection neurons and interneurons. Based on the degree of spike frequency adaptation, small hyperpolarization-activated voltage sag, and shape of fast afterhyperpolarization potentials, the neurons recorded in this study are most likely principle projection neurons of the MeA (Bian et al, 2008; Keshavarzi et al, 2014). Although there is a degree of overlap, many neurons of the MeApd project to the ventrolateral part of the ventral medial hypothalamus (VMH) region important in mating behavior, whereas many neurons of the MeApv project to the dorsomedial VMH region important in agonistic behavior. Therefore, it is tempting to speculate that deficits in MeApd and MeApv function are expected to lead to inability of MeApv to direct VMH-dependent agonistic behavior, and deficiency in MeApd ability to drive VMH-dependent sexual behavior. Indeed, deficits observed after social isolation are consistent with abnormal modulation of the vigor and context of VMH-dependent sexual (Bakker et al, 1995; Cooke et al, 2000) and agonistic behavior (Potegal and Einon, 1989; Wongwitdecha and Marsden, 1996; Toth et al, 2008). Our results provide a specific neural intermediary for the effects of social isolation on the development of MeA-modulated social behavior. These results also hint at the amygdala pathology that could contribute to abnormal social behavior in those that experienced impoverished social environment during childhood.

FUNDING AND DISCLOSURE

The authors declare no conflict of interest.