Limited prefrontal cortical regulation over the basolateral amygdala in adolescent rats

Cognitive regulation of emotion develops from childhood into adulthood. This occurs in parallel with maturation of prefrontal cortical (PFC) regulation over the amygdala. The cellular substrates for this regulation may include PFC activation of inhibitory GABAergic elements in the amygdala. The purpose of this study was to determine whether PFC regulation over basolateral amygdala area (BLA) in vivo is immature in adolescence, and if this is due to immaturity of GABAergic elements or PFC excitatory inputs. Using in vivo extracellular electrophysiological recordings from anesthetized male rats we found that in vivo summation of PFC inputs to the BLA was less regulated by GABAergic inhibition in adolescents (postnatal day 39) than adults (postnatal day 72–75). In addition, stimulation of either prelimbic or infralimbic PFC evokes weaker inhibition over basal (BA) and lateral (LAT) nuclei of the BLA in adolescents. This was dictated by both weak recruitment of inhibition in LAT and weak excitatory effects of PFC in BA. The current results may contribute to differences in adolescent cognitive regulation of emotion. These findings identify specific elements that undergo adolescent maturation and may therefore be sensitive to environmental disruptions that increase risk for psychiatric disorders.


Cognitive regulation of emotion develops from childhood into adulthood. This occurs in parallel with maturation of prefrontal cortical (PFC) regulation over the amygdala. The cellular substrates for this regulation may include PFC activation of inhibitory GABAergic elements in the amygdala. The purpose of this study was to determine whether PFC regulation over basolateral amygdala area (BLA) in vivo is immature in adolescence, and if this is due to immaturity of GABAergic elements or PFC excitatory inputs. Using in vivo extracellular electrophysiological recordings from anesthetized male rats we found that in vivo summation of PFC inputs to the BLA was less regulated by GABAergic inhibition in adolescents (postnatal day 39) than adults (postnatal day 72-75). In addition, stimulation of either prelimbic or infralimbic PFC evokes weaker inhibition over basal (BA) and lateral (LAT) nuclei of the BLA in adolescents. This was dictated by both weak recruitment of inhibition in LAT and weak excitatory
effects of PFC in BA. The current results may contribute to differences in adolescent cognitive regulation of emotion. These findings identify specific elements that undergo adolescent maturation and may therefore be sensitive to environmental disruptions that increase risk for psychiatric disorders.
The PFC sends robust glutamatergic projections to the basolateral amygdala (BLA) that can drive neuronal activity. But these projections can also regulate the amygdala through several inhibitory intermediates, such as activation of GABAergic interneurons in the BLA and in the intercalated cell masses [29][30][31][32][33][34][35][36][37] . This restrictive relationship between PFC and BLA activity can guide or regulate the expression of anxiety and fear. However, these PFC projections still undergo refinement in late adolescence 31,38 . Furthermore, weaker regulation of emotion in adolescents implies that the PFC does not exert a potent influence over the adolescent amygdala. The neurobiological substrates for this functional immaturity are not entirely clear, but might include weaker excitatory effects of PFC onto BLA principal output neurons or onto BLA inhibitory interneurons, or weaker inhibitory influences of these interneurons on BLA principal neurons. The only study to date that examined this 31 found that PFC inputs to BA exert weaker direct excitatory effects in immature mice. While this was a very important finding, it focused only on a part of the BLA, and by necessity had technical limitations of the in vitro approach and optogenetic viral infections at different ages. The purpose of the current study was to measure in vivo excitatory and inhibitory effects of PFC inputs to BLA, and to test (1) whether the adolescent BLA in vivo is under less inhibitory regulation by the PFC than the adult BLA, and (2) if this weaker impact of PFC on BLA can be ascribed to differences in the excitatory or the inhibitory effects of PFC inputs.

Impact of PFC inputs to the BLA.
To gauge the BLA response to PFC inputs, LFPs in the BLA were measured during train stimulation of PFC and compared between adults (PND 72-75) and adolescents (PND 39). The LFP response in both LAT and BA were examined (n = 11 hemispheres from 11 rats from each age; Supplemental Figs 1 and 2). Three frequencies were chosen to roughly mirror three primary bands of cortical activity in awake animals (10 Hz, alpha/theta; 20 Hz, beta; 40 Hz, gamma; Fig. 1A). These same frequencies of interneuron activity produce contrasting recruitment of inhibition and short term dynamics of GABAergic systems in the BLA 51,52 . LFPs can increase across the train (facilitating) or can decrease across the train (depressing) depending on several factors, including interaction between evoked inhibitory and excitatory synaptic inputs. Facilitation was dependent upon stimulation frequency in LAT (PND 72-75: frequency x pulse interaction, p < 0.0001, F(18,261) = 5.332, n = 11 rats; PND 39: frequency x pulse interaction, p < 0.0001, F(18,270) = 3.899, n = 11 rats, two-way RM-ANOVA) and in BA (PND 72-75: frequency x pulse interaction, p < 0.0001, F(18,270) = 3.837, n = 11 rats; PND 39: frequency x pulse interaction, p = 0.0026, F(18,270) = 2.282, n = 11 rats, two-way RM-ANOVA). Facilitation (normalized LFP slope >1) was generally observed at 10 Hz (Fig. 1B,D, left) while depression (normalized LFP slope <1) was observed at 40 Hz (Fig. 1B,D, right). Depression at 40 Hz suggests that higher stimulation frequencies recruit activation of BLA inhibitory elements that suppress summation 53 .
To determine whether LFP facilitation shifted from adolescence to adulthood, this was compared between PND 72-75 and PND 39 rats at each frequency. In the LAT, LFP facilitation was significantly greater in PND 39 rats compared to PND 72-75 rats at 20 Hz (Fig. 1B, middle; age x pulse interaction p = 0.0001, F(9,180) = 4.019, main effect of age p = 0.1946, F(1,20) = 1.801, two-way RM-ANOVA, n = 11 PND 72-75 rats, n = 11 PND 39 rats), but was similar at 10 Hz or 40 Hz (10 Hz: age x pulse interaction p = 0.9651, F(9,180) = 0.3274, main effect In the LAT, facilitation of LFPs at 20 Hz was suppressed in PND 72-75 rats (normalized value < 1.0) across the train, and significantly different than PND 39 rats (*p < 0.05, two-way ANOVA). In contrast LFPs showed similar facilitation between PND 72-75 and PND 39 at 10 Hz and similar depression at 40 Hz (p > 0.05, two-way ANOVA). (C) The summation ratio (last LFP/first LFP) was significantly lower in PND 72-75 rats at 20 Hz (*p < 0.05, post-hoc Holm-Sidak's multiple comparisons test after two-way ANOVA). (D) In the BA, significant LFP facilitation was observed at 10 and 20 Hz, and this was greater in PND 72-75 compared to PND 39 rats (*p < 0.05, two-way ANOVA), while a similar depression was observed at 40 Hz (p > 0.05, two-way ANOVA). (E) The summation ratio was greater in PND 72-75 rats at 10 Hz and at 20 Hz compared to PND 39 rats (*p < 0.05, post-hoc Holm-Sidak's multiple comparisons test after two-way ANOVA). To compare across both age and frequency, the summation ratio was quantified (ratio >1 indicates facilitation, ratio <1 indicates depression). Consistent with facilitation measures, summation ratio was greater in the PND 39 LAT at 20 Hz compared to PND 72-75 LAT ( Fig. 1C; p < 0.05 post-poc Holm-Sidak's multiple comparisons test), but lower in the PND 39 BA at 10 and 20 Hz compared to PND 72-75 ( Fig. 1E; p < 0.05 post-poc Holm-Sidak's multiple comparisons test). This opposite pattern of age differences across LAT and BA makes it unlikely that the same cellular changes account for age difference in these nuclei.
Age differences in the facilitation of LFPs can be due to many factors. Two prominent factors could be immaturity of PFC glutamatergic excitatory drive to BLA in adolescence or immaturity of GABAergic inhibitory circuits recruited by PFC inputs to BLA in adolescence. If the substantial suppression at 40 Hz reflects greater recruitment of GABAergic circuits, the inverse is that low frequencies (10 Hz) may be a better gauge of excitatory inputs. Mid-range frequencies (20 Hz) may reflect a balance between glutamatergic and GABAergic influences. The pattern observed in the LAT, of less suppression at 20 Hz, might hint toward weaker inhibition of LAT in adolescents that has not yet reached the mature adult level (with similar degrees of maximal inhibition when pushed at 40 Hz), while the pattern observed in the BA, of less facilitation at 10 Hz and 20 Hz, might hint toward immature excitatory drive of BA in adolescents. To begin to parse glutamatergic and GABAergic causes for age differences, the effects of PFC train stimulation was measured after local blockade of GABA A receptors (PTX, 10 pmol/100 nL in ACSF) or vehicle (n = 20 hemispheres from 16 rats at each age, divided in a counterbalanced 2 × 2 design between (LAT or BA recordings) x (PTX or vehicle), Supplementary Figs 1 and 2; rats from the study above were examined here after local administration (n = 11 rats/age) in addition to another 5 rats/age). The expectation is that if age differences are caused by weaker GABAergic systems in adolescence, the age differences will be diminished by PTX. In contrast, if age differences are due to weaker excitatory drive in adolescence, the age differences will still exist after PTX.
Consistent with low frequency reflecting primarily excitatory inputs, and not strongly reflecting GABAergic factors that suppress facilitation, we found that PTX did not significantly impact LAT LFP facilitation at 10 Hz in  While this capacity is highly important, it is ultimately the firing of BLA neurons that will determine BLA output. To more directly test age differences in PFC influences on BLA, the response of single BLA neurons upon PFC stimulation was measured before and after vehicle or PTX administration (using the rats from above (n = 20 rats/ age) and additional rats (additional n = 29 hemispheres from 16 P72-75 rats for a total of 49 hemispheres from 36 adult rats: control n = 27 hemispheres, vehicle n = 11 hemispheres, PTX n = 11 hemispheres; additional n = 33 hemispheres from 18 P39 rats for a total of 53 hemispheres from 38 adolescent rats: control n = 29 hemispheres, vehicle n = 12 hemispheres, PTX n = 12 hemispheres); Control recordings were performed with no local administration; Supplementary Fig. 5). These same rats were used to obtain the data for all subsequent results. The response of BLA principal neurons to PFC stimulation was categorized as Excitatory, Inhibitory, or No response (  All these analyses indicate a heavily Inhibitory response irrespective of age. While the heavily Inhibitory response profile may be similar between age groups, the underlying causes for an abundance of Inhibitory types may be different. Indeed, the LFP data suggest that there may be different causes for a skew towards Inhibitory outcomes across age, with weaker PFC excitatory drive in adolescents but greater PFC inhibitory effects in adults. Similarly, a skew towards an Inhibitory response phenotype may be caused by either (a) weaker PFC excitatory drive to BLA projection neurons so that Inhibitory response types are favored over Excitatory response types or (b) greater recruitment of inhibitory circuits. Put operationally, in the balance between response types, Inhibitory is more prominent than Excitatory because (a) Excitatory responses are subthreshold or absent and measured as Non-Responders or (b) Excitatory responses are inhibited and measured as Inhibitory.
If PFC inputs are weaker in one group, it is expected that a higher proportion of neurons would be No response due to a weaker, subthreshold impact on firing. To examine this further, the proportion of Non-responding neurons relative to Excited neurons was examined. The ratio was higher in PND 39 rats (Fig. 4D,E, left; main effect of age p = 0.0318, F(1,3) = 14.50, two-way ANOVA). This may hint towards overall weaker PFC inputs to the BLA in PND 39 rats.
While this is suggestive of a mechanistic difference between adults and adolescents, the Inhibitory response may mask accurate comparison of Excitatory and Non-responses. Therefore, this ratio was assessed again upon local intra-BLA blockade of GABA A receptors with PTX via cannula infusion compared to vehicle. This blocks the Inhibitory phenotype, but the neurons that normally show an Inhibitory phenotype may still respond to PFC in a different manner (Excitatory) or now show No response. After blockade of the Inhibitory phenotype, a shift in the relative distribution of response types towards No response would indicate that the dearth of Excitatory phenotypes is due to weak excitation of projection neurons (a, above; Fig. 4C). In contrast, a shift toward Excitatory responses would indicate that inhibition actively suppressed excitation and decreased the incidence of Excitatory response types (b, above;   . Thus, blockade of GABA A -mediated inhibition in PND 39 rats does not uncover significant excitation. These data indicate that the relatively low incidence of Excitatory responses to PFC stimulation in PND 72-75 rats is partly due to GABAergic inhibition, but the low incidence of Excitatory responses in adolescents is partly due to weaker excitatory effects of PFC inputs to LAT and BA.
Limited PFC regulation over adolescent amygdala. The neuronal response profile is a rough indication of the overall influence of an input. This can yield important clues about a shift in excitation and inhibition across age. But this measure does not capture the magnitude of the inhibitory response, nor its temporal aspects.
To determine if the strength of inhibition varies across age, inhibition evoked by PFC stimulation (Fig. 5A) was quantified as the area under PSTHs (Inhibition AUC ; Fig. 5B). Only Inhibited neurons with a full input-output curve were included in this analysis. Inhibition AUC was dependent upon the stimulation intensity (Fig. 5C,D) and was weaker in PND 39 rats (Fig. 5A,C,D). Inhibition AUC was different across age at all PFC inputs to the BLA (Fig. 5C,D;  Maturation of Excitatory inputs. The results above demonstrate a shift in the effects of PFC inputs to BLA between adolescence and adulthood, and that this can be accounted for by immaturity of GABAergic influences in LAT with a potential additional component of immaturity of glutamatergic drive in BA. The contribution of the glutamatergic element can be tested directly by measurement of the BLA neuron excitatory response to PFC stimulation. The probability of a monosynaptic response to stimulation was quantified across PFC inputs to BLA principal neurons that showed an Excitatory response profile (Fig. 6A), with the hypothesis that immature innervation of PFC → BLA would produce weaker response in adolescence.
The LAT response to inputs from PFC was only weaker in PND 39 rats in the IL → LAT input (Fig. 6B,  To test the excitatory drive in the absence of GABA A influences, experiments were performed after local intra-BLA infusion of PTX or vehicle (as above), and excitatory drive was measured (n = 5 rats/group). When GABA A -mediated inhibition was blocked, IL → LAT inputs remained weaker in PND 39 confirming weaker excitatory strength of this input (age x stimulation interaction, p = 0.0001, F(4,56) Despite weaker overall excitatory strength of PFC → BLA in adolescents, the measured responses to PFC before blockade of GABA A influences were not dramatically different across age at the highest stimulation intensity. This could be caused by heavier GABA A influences over the PFC → BLA response in adults that countermand the increased excitatory strength of PFC → BLA. To test if there is greater GABA A -mediated regulation PFC → BLA in adults, the effect of PTX on PFC → BLA excitatory responses was compared across groups (PTX -ACSF mean) in the same rats examined above.

Discussion
The BLA response to extrinsic inputs includes a heavily inhibitory component 32,[54][55][56] . The current findings demonstrate a common theme, whether measured by evoked LFPs or single neuron responses, that PFC inputs to BLA do not exert as potent a response in adolescents compared to adults. While evidence emerges for a weaker inhibitory regulation in LAT and BA during adolescence, the primary contributing mechanism for this weakness is region specific. In the LAT there is weaker evoked inhibition in adolescents, whereas in the BA there is a weaker excitatory influence of PFC inputs. This is demonstrated by differences in the summation of LFPs, their sensitivity to PTX, and differences in the proportion of response types. The resultant inhibitory regulation of PFC over LAT and BA is weaker in adolescents, demonstrated by less PFC-evoked inhibition of neuronal firing. This is consistent with recent findings that the potency of PFC inputs to BLA changes across development 31 and in vivo and in vitro studies that indicate weaker GABAergic regulation in prepubertal rats 57,58 .
There are some limitations of the current findings. While it is important to examine these circuits in vivo, anesthesia limits the generalizability of these findings. In addition, PFC was activated by electrical stimulation that can have some spatial spread. The stimulation intensities were kept below 1.0 mA to limit current spread but there may be a small degree of overlap between effects of PrL and IL stimulation. However, we do not believe that this overlap is near to complete because experiments where both PrL and IL were stimulated in series using two aligned stimulation probes still produced different effects. Additional confounds can arise, such as recruitment of multisynaptic paths to the BLA during train stimulation or antidromic activation of BLA circuits. While these potential confounds cannot be ignored, the overall conclusions would still point to less regulation of the adolescent BLA.
The relationships between maturation of the excitatory and inhibitory influences of PFC inputs to BLA may be inter-related. PFC → BLA can evoke EPSPs in BLA projection neurons that drive action potential firing 34,44,59 . Weaker PFC → BLA may produce smaller EPSPs and produce lower probability of action potential firing. PFC → BLA also excites BLA interneurons 31,32,34,44 that exert an inhibitory effect on BLA projection neurons that is rapid enough to curtail the direct PFC → BLA excitatory effects on projection neurons 34,44 . Weaker PFC → BLA interneurons therefore may also produce less inhibition. This co-occurring shift in excitatory and inhibitory effects produced by weaker PFC inputs may account for a similar distribution of Excitatory/Inhibitory response profiles across age.
The nature of the frequency dependence of the PFC → BLA supports a mechanism that includes a shift in the balance between excitatory and inhibitory inputs. Single pulse stimulation of PFC evoked a rapid inhibition that peaked 40-50 ms after onset, and then decayed. Our data from single units and after PTX suggests that this occurs overlayed on a background of weaker excitatory inputs with a rapid onset and offset within a 10 ms window. When a second stimulation occurs at an interval shorter than 40-50 ms (before the inhibition caused by the first stimulation has peaked), one might expect that the inhibition caused by the first will have minimal impact on the response to the second stimulation, but the inhibitory component of the second stimulation can begin to overlap with the decay of the inhibitory component of the first stimulation. Stimulation trains at higher frequencies (e.g. with intervals of 10-25 ms, 100-40 Hz) would allow subsequent inhibitory events to add onto the decay of previous inhibitory events, with a result of increased inhibition across the train. Stimulation trains at lower frequencies (e.g. with intervals 50-100 ms, 20-10 Hz) would result in subsequent inhibitory events occurring during later phases of decay of the previous event, and less inhibitory impact. This fits well with the sensitivity of high frequency LFPs to PTX, and less sensitivity at lower frequencies, as well as the initial facilitation of LFPs observed during the first few pulses of 40 Hz trains. This concept has been documented in other regions, including cortex and hippocampus, where higher frequency stimulation (40 Hz) recruits substantial GABAergic inhibition 60 , and high frequency rhythms (gamma) are linked to the firing of fast-spiking interneurons [61][62][63] . The architecture of the BLA produces a similar outcome. There are different types of BLA GABAergic interneurons defined chemically, for instance cholecystokinin-containing (CCK) and parvalbumin-containing (PV) interneurons [64][65][66] , or electrophysiologically, for instance stuttering and fast-firing interneurons 67,68 . CCK interneurons have a maximal firing rate of ~20-30 Hz, while PV interneurons have a maximal firing rate >40 Hz 64,65 . BLA gamma frequency rhythmicity requires BLA PV fast-firing interneurons 69 . The importance of PV interneurons in gamma frequency oscillation is further guaranteed due to difference in short-term synaptic dynamics and frequency-dependence (between 10-40 Hz) of PV, CCK and axo-axonic interneurons 51,67 , with all types of BLA interneurons showing little frequency dependence of effects on BLA projection neurons at 1 Hz, and a near maximal effect when PV interneuron firing approaches 40 Hz 51,52 . Based on this, PV interneurons are expected to be increasingly recruited by excitatory inputs across a wide range of frequencies.
PFC excitatory influences may guide appropriate BLA-mediated behaviors while inhibitory influences may impose regulation over inappropriate BLA-mediated behaviors or control the magnitude of responses. The slow maturation of excitatory influences and resultant smaller afferent evoked inhibition 31 is expected to produce a two-fold effect, weaker guidance of BLA-mediated behaviors and weaker regulation of these behaviors. This may produce age differences in behaviors that are sensitive to the balance between glutamate and GABAergic in the BLA, including generalization of conditioned fear and fear extinction [70][71][72][73][74][75][76][77][78][79][80][81][82][83][84][85] . Indeed, measures of reduced GABAergic influences, or a shift in the inhibitory-excitatory effects of PFC inputs to favor excitation have been observed after fear conditioning 30 , with opposite changes after extinction, such as decreased summation of LFPs 86 , and reduced excitatory impact of PFC inputs 44 . These behaviors are also linked to rhythmic activity in the BLA. Different types of BLA activity become dominant during specific BLA-mediated behaviors, and are an important component of BLA functions. BLA interneurons are recruited to theta synchrony in the presence of noxious stimuli 87 , BLA theta synchronization with sensory cortical regions occurs in conditions that require a BLA-mediated response during fear recall BLA 88 , and PFC-BLA synchrony at theta is observed during fear recall and expression 89,90 . Our results would indicate that theta frequency of inputs is expected to exert less inhibition over BLA, permitting greater BLA-mediated responses. In contrast, PFC-BLA (specifically IL-LAT) phase coherence shifts Top-down regulation of emotion matures across juvenile and adolescent years 17,[92][93][94] . In parallel, there is structural maturation of PFC that lags behind amygdala maturation 11 , maturation of the connections and functional interactions between corticolimbic structures 12,92 , development of a reciprocal relationship between the activity of the PFC and amygdala 13,95,96 , and a shift in the PFC regions that exert the strongest influence over amygdala 16,97 .
PFC input density in the BLA reaches adult levels by early adolescence 38 , as does the amount of PFC neurons that project to the BLA 98 . The spine number 99 and synaptic protein synaptophysin is stable throughout the adolescent to adult period 100 , consistent with maturity of the structural components of PFC excitatory inputs during adolescence. However, the response to excitatory synaptic inputs or synaptic function may continue to mature throughout adolescence. Adult patterns of BLA interneuron chemical phenotype appears by P30 101,102 , and basic GABAergic synaptic properties are mature by this same time 103 . However, there is substantial evidence that GABAergic systems in the BLA are still not fully matured until later ages. For instance, perineuronal nets (PNNs) in the BLA continue to mature throughout adolescence 104 . Immature elaboration of PNN is considered a hallmark of juvenile plasticity in cortex and related to GABAergic regulation that has not reached maturity 105,106 . This slower maturation is accompanied by increased GABA synthesis between 1-2 months postnatal 107 , and increased intra-BLA inhibition evoked by excitatory inputs to the BLA between juvenile and adult ages 57 , and a shift in the functional impact of GABAergic systems in vivo 58 . Adolescent state of these BLA GABAergic systems may contribute to the weaker GABAergic regulation of the adolescent BLA in the current study.

Methods
All experiments were approved by the Institutional Animal Care and Use Committee of Rosalind Franklin University, and experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). Care was taken to reduce any unnecessary distress to the animals and to limit the total number of animals used. In vivo extracellular recording. In vivo extracellular recordings were performed in anesthetized rats (urethane, Sigma-Aldrich, St. Louis, MO; 1.5 g/kg dissolved in 0.9% saline, intraperitoneally) as described 118 . Rats were placed in a stereotaxic device (Stoelting, Wood Dale, IL) after deep anesthesia was confirmed. Their body temperature was monitored via a rectal temperature probe, and maintained at 36-37 °C using a heating pad with a temperature controller (Model TC-1000, CWE Inc, Ardmore, PA). The BLA and PFC were localized using a stereotaxic atlas 119 . For adult rats, the coordinates used for the BLA centered on 4.8 mm-5.2 mm lateral, 2.5 mm-3.8 mm caudal from bregma. The coordinates used for PFC were 0.7 mm lateral, 2.7 anterior from bregma, and 3.9 ventral (PrL) or 5.1 ventral (IL) from the brain surface. For adolescent rats, coordinates were scaled according to the measured distance between the bregma and interaural skull landmarks. Burr holes were drilled in the skull at locations overlying the BLA and the PFC. A bipolar concentric stimulation electrode (Rhodes Medical Instruments CA, USA or MicroProbes, Gaithersburg, MD, USA) was lowered into the IL or PrL. Single-barrel recording electrodes were constructed from glass pipettes (2.0 mm outer diameter, World Precision Instruments, Sarasota, FL), pulled using a vertical microelectrode puller (PE-2; Narishige, Tokyo, Japan), and broken under microscopic view to produce a 1-2 µm diameter tip. The recording electrode was filled with 2% Pontamine Sky Blue (Alfa Aesar, Ward Hill, MA) in 2 M NaCl (Fisher Scientific, Pittsburgh, PA) and then slowly lowered into the amygdala via a hydraulic microdrive (Model MO-10, Narishige).

Subjects. Male
During extracellular recording, signals were amplified with a headstage connected to a preamplifier (2400 Amplifier, Dagan Corporation, Minneapolis, MN or 1800 Amplifier, A-M Systems, Sequim, WA), filtered at 0.1-0.3 Hz (low cut-off frequency) and 3 kHz (high cut-off frequency), and outputted simultaneously to an oscilloscope (Model 2532 BK Precision, Yorba Linda, CA) and an audio monitor (Model AM8, Grass Instruments, West Warwick, RI). In addition, amplified outputs were digitized (10 kHz; Model ITC-18, HEKA, Bellmore, NY) and fed to a personal computer (Mac Pro/2.8 Apple, Cupertino, CA), monitored using Axograph X software (Sydney, Australia) and stored on a hard disk for off-line analysis. The anesthesia state of the animal was monitored by cortical local field potential oscillations recorded from the concentric electrode in the PFC throughout recordings of neuronal firing. Animals were considered under deep anesthesia when the cortical field oscillations displayed a predominant slow (~1 Hz) rhythmic waveform. Local field potential recordings. The approach for recording LFPs was modeled after approaches used to probe in vivo GABAergic regulation in cortical recordings with much success 60 . PFC was stimulated (as above; 0.1-0.9 mA) while the evoked local field potential was recorded. A stimulation intensity and duration was selected that evoked a field potential of approximately 50% the maximal amplitude. The PFC was stimulated at 10, 20, and 40 Hz (10 pulses/train, 10 s inter-stimulus interval). The initial slope of each evoked local field potential in the stimulus train was measured. In a set of experiments, the recording electrode was filled with recording solution and drug/vehicle for local application during recordings, as described below.
Single unit recordings. Single neurons were recorded throughout the BLA. Upon isolation of a single unit, baseline firing rates in spontaneously firing BLA projection neurons were recorded for 5 min before the PFC was electrically stimulated (as above, 0.1-0.9 mA, single pulses). Null stimulation (0 mA) data in the same pattern was obtained for comparison. A minimum of 40 sweeps at each stimulation intensity was acquired. The response of BLA neurons to PFC stimulation was measured as changes in the number of action potentials in peri-stimulus time analysis (see Data Analysis). In a set of experiments, drug/vehicle was delivered locally by cannula, as described below.
Microiontophoretic application of glutamate. Neurons of the BLA fire slowly under anesthesia, presenting difficulty for assessing inhibition of firing. Therefore, upon identification of a BLA neuron with an inhibitory response to PFC stimulation, slowly firing BLA neurons were induced to fire at 4-8 Hz with microiontophoretic application of glutamate. Multibarrel microelectrodes (4 barrels; A-M Systems) were constructed using a vertical microelectrode puller (PE-2; Narishige), and the tip was broken back under microscopic guidance. One barrel of the microelectrode was filled with 2% Pontamine Sky Blue (Alfa Aesar) in 2 M NaCl (Fisher Scientific) for electrophysiological recordings and a second barrel was filled with 1 M NaCl for automatic current balancing. One of the remaining barrels was filled with 50 mM glutamate (pH 8.0; Alfa Aesar) dissolved in 20 mM NaCl solution.
The last barrel was empty. Glutamate was ejected with anodal iontophoretic current (E104B; Fintronics, Orange, CT). Retaining currents of the opposite polarity were used (10 nA) before and after ejection. The glutamate ejection current was adjusted to maintain a stable firing rate of BLA projection neurons. After a stable firing rate was achieved with glutamate iontophoresis, PFC was stimulated as above.
Local application of drugs. In specified experiments, intra-BLA local drug infusions were performed during electrophysiological recordings by one of two means 118 : (1) a cannula for infusions (pulled glass pipette with shank diameter 50-100 µm) was lowered into the BLA (15 degree angle off the rostral-caudal axis) and drugs were applied through this cannula while a different higher impedence electrode was used to measure the firing of BLA neurons; or (2) a low impedence single barrel recording electrode (>50 µm tip) was lowered into the BLA and LFPs were recorded with this electrode before and after drugs were applied through this electrode. Data analysis. When possible, an individual rat was used to measure both LFPs and single units. When drug was applied locally by pressure ejection during LFP recordings, PTX and ACSF vehicle were applied to opposite hemispheres of the same rat in a counterbalanced manner. Multiple different manipulations were not performed in the same hemisphere.
Analysis of extracellular recordings. BLA contains projection neurons and interneurons. Putative projection neurons and interneurons were separated based on previously-established criteria that utilize action potential width 58,118,120 . However, because these characteristics can vary depending on electrode and filter settings, these criteria were retested by measuring the half-width of action potentials and plotting a frequency distribution of half-widths. The best-fit for this distribution was tested between one 2 nd order polynomial (indicative of a single population) and two 2 nd order polynomials (indicative of two populations). When data fits are consistent with two populations, a cut-off of 0.225 ms appropriately separates the populations under these recording and filtering conditions 58 . To reduce uncertainty, a buffer on either side of this intersection (0.205-0.240 ms) was added, and neurons within this window were excluded. Neurons above this cut-off were classified as projection neurons, while neurons below this cut-off were classified as putative interneurons. To increase reliability, only neurons that showed a biphasic action potential waveform were included. Projection neurons were included in analysis if they met the following criteria: they were located within the BLA as determined by reconstruction based on histological staining, action potentials had a clearly visible signal to noise ratio (>3:1), the firing rate was stable, and the action half-width was greater than 0.240 ms. The spontaneous firing rate was measured as the number of action potentials/s (Hz) over a minimum of 4 minutes.
Single unit response to PFC stimulation. For analysis of response to PFC stimulation, the action potential firing over a 3 sec peri-stimulation period (40-100 repetitions) at the same stimulation intensities were organized into peri-stimulus time histograms (PSTHs), divided into 300 bins (10 msec bin width) and the number of action potentials in each bin was tabulated. The response of BLA projection neurons to PFC stimulation was then classified into one of 3 qualitative types, based on the firing changes after stimulation: Inhibitory response, Excitatory response and No response. Each neuron was characterized and grouped by response type. After categorization by type, the response to PFC stimulation was averaged within each group and compared between adolescent and adults or between intra-BLA infusion groups. Inhibitory response: Neurons were grouped in the Inhibitory response type if they displayed suppressed firing after PFC stimulation compared to null stimulation (0 mA). This was defined as 3 consecutive bins with a firing rate that was ≥2 standard deviations lower than the average null stimulation firing or 5 consecutive bins with 0 action potentials during a 1 s time window after stimulation. The first of these bins was marked as the onset of inhibition. Area under PSTH was used to quantify the degree of inhibition. Area under the PSTH produces a measure of inhibition that incorporates a temporal component. The response was normalized to the baseline firing rate [(AP post − AP pre ) ÷ AP pre ], where AP pre = mean firing rate of the pre-stimulation epoch and AP post = mean firing rate of each 10 ms bin during a 300 ms window immediately following stimulation. This normalized response at each bin was summated across the 300 ms window. The duration of inhibition was also measured. The time from the inhibition onset was first determined (as above, derived from averaged data). The time from this onset of inhibition to the time of the first action potential after inhibition onset was measured during each PFC stimulation sweep. These time epochs were averaged for each neuron as the measure for duration of inhibition.
Excitatory response: Neurons were grouped in the Excitatory response type if they displayed an increase greater than 5 times baseline firing rate within a 10-30 ms window after the stimulation, and the response was consistent with a monosynaptic input. Thus, neurons were excluded if the apparent excitatory effect fit criteria for antidromic responses (<1 ms variability of latency and reliably followed 300 Hz stimulation) or a polysynaptic response (latency >30 ms, and variability of response >5 ms). The excitatory response was measured as the action potential firing probability in the 10-30 msec window after stimulation after mean baseline firing rate was subtracted.
No response: Neurons were grouped in the No response type if they fell short of the criteria described above.
LFP response to PFC stimulation. Traces were filtered (200 Hz) and averaged (at least 10 consecutive traces). The slope of the evoked LFP was determined by measurement of the rise/run using Axograph X software (Sydney, Australia). The slope of each LFP evoked during a train (10 stimuli) was measured. LFP facilitation/depression was normalized and quantified as [slope LFP x ÷ slope LFP 1 ], where LFP x = the slope of each LFP during a train, and LFP 1 = slope of the 1 st LFP of the train. The summation ratio was quantified as the slope of the last LFP (10 th LFP) normalized to the first LFP of the train (LFP 10 ÷ LFP 1 ). A ratio >1 indicates summation, a ratio <1 indicates suppression.
The data analysed during the current study are available from the corresponding author on reasonable request.

Statistical analysis.
Statistical tests were performed using Prism 5 software (GraphPad, La Jolla, CA). A p value < 0.05 was considered statistically significant. The proportion of neurons exhibiting each of the 3 types of responses to PFC stimulation was compared between groups using a Chi-square test. Data were tested for normal distribution (D' Agostino and Pearson normality test). When two groups were subjected to a planned comparison they were compared with a two-tailed unpaired t-test. When multiple factors were analyzed, measures were compared using a two-way ANOVA with age (adolescent or adult) as a main factor and stimulation intensity or stimulation pulse number as a repeated measures factors when appropriate. Holm-Sidak's multiple comparisons test was used for further comparison when a significant difference was found in ANOVA. All data were presented as mean ± SEM, unless otherwise specified.