Oxytocin Selectively Gates Fear Responses Through Distinct Outputs from the Central Amygdala

See allHide authors and affiliations

Science  01 Jul 2011:
Vol. 333, Issue 6038, pp. 104-107
DOI: 10.1126/science.1201043


Central amygdala (CeA) projections to hypothalamic and brain stem nuclei regulate the behavioral and physiological expression of fear, but it is unknown whether these different aspects of the fear response can be separately regulated by the CeA. We combined fluorescent retrograde tracing of CeA projections to nuclei that modulate fear-related freezing or cardiovascular responses with in vitro electrophysiological recordings and with in vivo monitoring of related behavioral and physiological parameters. CeA projections emerged from separate neuronal populations with different electrophysiological characteristics and different response properties to oxytocin. In vivo, oxytocin decreased freezing responses in fear-conditioned rats without affecting the cardiovascular response. Thus, neuropeptidergic signaling can modulate the CeA outputs through separate neuronal circuits and thereby individually steer the various aspects of the fear response.

Fear can be severely immobilizing but can also be a major driving force for some of humans’ most heroic acts. In both cases, the internal emotional experience may be similar, although it may lead to substantially different behavioral outcomes (13). Studies on human emotions often use autonomic nervous system parameters to assess arousal, because of the role of our internal organs in the emotional state (4, 5). Projections from the central nucleus of the amygdala (CeA) to the hypothalamus and different brain stem nuclei coordinate behavioral and physiological fear expression (6). It has been postulated that different fear responses, characterized by more active or passive behavioral coping strategies, can be triggered by a neuronal switch within the CeA (7). The question thus arises whether fear responses only vary in intensity, or whether different qualities of fear responses exist, reflected in different associations between behavioral and physiological components. We investigated whether a neurophysiological basis for such a distinct regulation could be found in the CeA.

Most projections from the CeA to the hypothalamus and brain stem nuclei originate from the medial part of the CeA (CeM) (6). Oxytocin can inhibit neurons in the CeM through its excitatory effects on γ-aminobutyric acid (GABA) inhibitory (GABAergic) projections that originate from the lateral and capsular part of the CeA (8) (henceforth referred to as CeL). The CeL contains distinct neuronal populations (710) whose individual activation may differentially regulate active versus passive fear responses (7). Can a similar distinction of neuronal populations be found in the CeM (10)? Do projections from the CeM to selective targets in the hypothalamus and brain stem arise from distinct neuronal populations and, if so, are these under a specific inhibitory control from the CeL? Such specificity might provide a neurophysiological basis within the CeA to selectively regulate behavioral and physiological components of the fear response.

We first evaluated the target specificity of CeM projections by double fluorescent retrograde tracing of the ventrolateral column of the periaqueductal gray (PAG), which is implicated in the freezing response (6), and the dorsal vagal complex (DVC), which modulates cardiovascular responses (11). We injected green and red fluorescent latex microspheres, respectively, into the PAG and DVC of 3- to 4-week-old Sprague-Dawley rats (12, 13). After allowing 48 hours of retrograde transport, we killed the animals, verified the injection sites (Fig. 1, A and B), and assessed retrograde label in horizontal brain slices of the CeM. Both green and red microspheres were present throughout the CeM (Fig. 1, A and B), yet in separate neurons that were intermingled without any obvious clusters (Fig. 1C and fig. S1). Confocal quantification revealed 5.9% colabeling (n = 680 neurons, Fig. 1D, fig. S1, and table S1). Injecting a mixture of green and red microspheres in the DVC resulted in their copresence in all labeled CeM neurons (Fig. 1C), confirming sensitivity to detect colabeling.

Fig. 1

Distinct neuronal populations of the CeM project to PAG and DVC. (A and B) Coronal views of injection sites (A) in PAG of fluorescent green and (B) in DVC of red microspheres (left), corresponding CeM labeling (right). Scale bars: left, 1 mm; right, 500 μm. lPAG, lateral PAG; vlPAG, ventrolateral PAG; sp5, spinal trigeminal tract. (C) Separately labeled CeM neurons after coinjections of green and red microspheres in respective PAG and DVC (left) versus colabeled neurons after injections of both microspheres in DVC (right, scale bar, 50 μm). (D) Quantification of colabeled (coloc) CeM neurons (see also table S2). (E) Electrophysiological characteristics of CeM→PAGs, “PAG” and CeM→DVCs “DVC.” (*P < 0.05; **P < 0.01, n = 22 to 43 neurons) Error bars indicate SEM.

We next compared electrophysiological properties by whole-cell recordings from fluorescently labeled PAG- and DVC-projecting neurons (henceforth called CeM→PAGs and CeM→DVCs, respectively). CeM→PAGs (n = 42) were, on average, significantly more depolarized (–59.6 ± 1.6 versus –65.4 ± 1.2 mV) and had lower membrane resistance (358 ± 25 versus 640 ± 61 MΩ) and higher membrane capacitance (72.3 ± 5.6 versus 58.5 ± 4.0 pF) than CeM→DVCs (n = 43, Fig. 1E). In cell-attached configuration, average basic spiking frequencies of CeM→PAGs (n = 22) were significantly higher than CeM→DVCs (4.1 ± 0.5 versus 2.5 ± 0.4 Hz, n = 28).

Prompted by these anatomical and electrophysiological differences, we also tested their pharmacological characteristics. Although both projection neurons were similarly excited or inhibited by a range of neuropeptides (table S2), oxytocin—known to increase spontaneous inhibitory postsynaptic current (sIPSC) frequencies of CeM neurons (8) (fig. S2A)—only affected CeM→PAGs. Thus, bath perfusion of the oxytocin agonist [Thr4,Gly7]-oxytocin (TGOT) specifically and reversibly increased sIPSC frequencies in CeM→PAGs (from 2.2 ± 0.3 to 5.6 ± 0.7 Hz, n = 34) (Fig. 2A) but not in CeM→DVCs (from 1.5 ± 0.3 to 1.7 ± 0.2 Hz, n = 32) (Fig. 2B) (see also fig. S3). In cell-attached recordings, this translated into a selective inhibition of spontaneous spiking frequencies of CeM→PAGs (from 5.6 ± 0.7 to 3.1 ± 0.2 Hz, n = 10) (Fig. 2C) versus CeM→DVCs (from 3.5 ± 1.0 to 3.7 ± 0.9 Hz, n = 10) (Fig. 2D) (8).

Fig. 2

Distinct effects of TGOT on CeM→DVC and CeM→PAG neurons. (A and B) Representative traces of sIPSCs (top) and their average frequencies (bottom) recorded before (control) and after (TGOT) application of (A) CeM→PAGs and (B) CeM→DVCs. (C and D) The same for spontaneous spiking activity (*F2,36 = 8.4, P < 0.005; **F2,122 = 17.8, P < 0.0001, n = 20 to 34 neurons) Error bars indicate SEM.

To verify whether the failure of TGOT to inhibit CeM→DVCs was caused by the absence of inhibitory projections, we locally depolarized by puffing KCl (35 mM) from a patch pipette (13) (Fig. 3A) and simultaneously recorded sIPSCs from CeM→DVCs. Although bath-perfused TGOT did not affect sIPSC frequencies (from 1.2 ± 0.3 to 1.0 ± 0.2 Hz, n = 6), rapid increases occurred with KCl puffs at distinct locations in the CeL (from 0.9 ± 0.2 to 8.4 ± 1.6 Hz, n = 6) (Fig. 3B). In the same slice, TGOT remained capable of increasing sIPSC frequencies in CeM→PAGs (from 1.5 ± 0.3 to 3.8 ± 0.7 Hz, n = 6) (Fig. 3C). Puffing glutamate (Fig. 3C) or bombesin (fig. S4), for which receptors are expressed in the CeL and adjacent intercalated inhibitory neurons, gave similar results, which suggested an activation of inhibitory projections arising from either of these regions [discussed further in (13)].

Fig. 3

Local stimulation reveals inhibitory projections on TGOT-insensitive CeM→DVCs. (A) Local perfusion of neurons that project to CeM→DVCs (red) and CeM→PAGs (green) in a horizontal brain slice of the CeA with ACSF flow away from CeM. GABA-R, GABAA receptor; OT-R, oxytocin receptor; BLA, basolateral amygdala; LA, lateral amygdala; Ic, Intercalated neurons. (B) Average sIPSC frequencies in CeM→DVCs recorded with patch pipette “(1)” after bath-perfused TGOT followed by locally puffed KCl (n = 6), subsequently, in the same slice, in CeM→PAGs recorded with pipette “(2)” after bath-perfused TGOT (n = 6). (C) Action potential and sIPSC frequencies in respective CeL (n = 5) and CeM→DVC neurons (n = 5) after local perfusion with glutamate. (*P < 0.05, **P < 0.005) Error bars indicate SEM.

To investigate how these specific in vitro effects of oxytocin translated in vivo into fear-induced freezing behavior (mediated by the PAG) (6, 14) and cardiovascular changes (modulated by the DVC) (11, 15), we equipped rats with bilateral cannulae for local drug administration into the CeA and implanted radiotelemetric devices to monitor heart rate. To activate the amygdala, we applied a 2-day contextual fear-conditioning protocol (13) that resulted in >90% freezing after conditioning and >80% freezing responses upon reexposure to the context in all vehicle-injected animals (Fig. 4, A and C, and fig. S2B). Bilateral injection of TGOT or GABAA receptor agonist muscimol decreased these freezing responses to <50%) (Fig. 4C). In rats where TGOT had not decreased freezing, the injection sites, identified with fluorescent muscimol (Fig. 4B), were outside the CeA (fig. S5), which confirmed the CeA role in these contextual freezing responses [further discussed in (13)]. Although baseline heart rate before reexposure and the initial elevation in heart rate upon reexposure to the context were similar for all groups [baseline: artificial cerebrospinal fluid (ACSF) 400 ± 10; TGOT 394 ± 10; muscimol 394 ± 12 beats per min (bpm)] (Fig. 4D), the typical decrease that followed in the second 10-min period was inhibited by muscimol but not by TGOT. Bombesin reduced freezing and prevented the decrease in heart rate (Fig. 4C2 and 4D2) consistent with its inhibition of both CeM→PAGs and CeM→DVCs. Together, these data not only support the selective action of oxytocin on freezing behavior via CeM→PAGs, but also suggest a critical role of CeM→DVC neurons in the control of cardiovascular changes to fearful stimuli. Finally, heart rate variability (HRV) analysis (13) revealed an absence of increase in the high frequency band in the muscimol-treated rats (Fig. 4E), which indicated an inhibition of the parasympathetic activation (13) (table S3 and fig. S7). The failure of TGOT to affect this parasympathetic cardiovascular response is consistent with the absence of TGOT effects on CeM→DVCs (11, 15).

Fig. 4

Differential effects of drugs injected in the CeA on freezing and heart rate during a conditioned fear response. (A) Three-day fear-conditioning protocol: Inj. proc, sham injection procedure; yellow lightings, electric shocks. (B) Postmortem analysis showing fluorescent muscimol Bodipy diffusion inside the CeA. Scale bar, 1 mm. (C) Freezing responses measured on day 3 in the context alone. (D) The same for heart rate responses [D (left): F2,16 = 3.8, P < 0.05; D2: F1,14 = 5.3, P < 0.05]. (E) Power spectrum histograms of HRV –2 to 0 min before and 18 to 20 min after placing rats in conditioning context. LF, low-frequency component; HF, high-frequency component. Error bars indicate SEM (*P < 0.05). (F) Final model illustrating specific TGOT-inhibited freezing through selective inhibition of CeM→PAGs. Abbreviations as in (Fig. 3A).

The present findings provide evidence that specific behavioral and physiological components of the fear response are controlled by distinct neuronal populations in the CeM (Fig. 4F). These project to the PAG and the DVC; exhibit unique electrophysiological characteristics; and despite being spatially intermingled, are selectively modulated by oxytocin through inhibitory projections from the CeL. The functionality of this selectivity was further revealed at the behavioral and physiological level by oxytocin’s inhibition of freezing responses without affecting cardiovascular changes. Previous studies have distinguished neuronal populations in the CeL on the basis of expression of CRF or opioids (16) or on mutually inhibitory connections (10). These may play a specific role in the inhibition of CeM→PAGs, and, in combination with the distinct electrophysiological characteristics of CeM neurons, affect further local information processing [see e.g., (17)].

A specific regulation by the CeL of CeM neurons with different projections could have important implications for the mechanisms underlying the expression of fear. Distinct neuronal populations in the CeL are activated or inhibited during the expression of fear (9, 10) and this might represent a switch between active versus passive fear and associated coping strategies (7). Our present findings imply that the CeM differentially controls expressions of the fear response through separate projections to the brain stem. Our findings, instead of supporting a rigid association between behavioral and physiological expressions of fear (1, 2), suggest that these expressions may be specifically controlled by the CeL. This underlines, first, the importance of considering multiple parameters in the correct assessment of fear responses in animals, but it also opens the potential for new therapeutic applications (fig S8). In humans, panic disorder can manifest itself at the visceral level predominantly by increases in heart rate, respiratory rhythm, or gastrointestinal motility (18). While its onset appears to be triggered in the lateral and basal amygdala (19), its specific expression may result from a differential gating within the CeA. Although panic and other anxiety disorders are typically treated with benzodiazepines, future neuropeptide-based therapies might offer a more precise inhibition of their expression.

The amygdala orchestrates behavioral responses to both negative (fearful) and positive (rewarding) salient stimuli (20, 21), although the precise underlying circuits are still unclear. Distinct, intercalated CeL and CeM populations, targeted by projections from the basolateral amygdala (22) and brain stem (11), could play a role in regulating behavioral and physiological expressions associated with different emotions (35). Furthermore, the CeA expresses a multitude of neuropeptide receptors that can specifically affect local activity (8, 2325). Levels of oxytocin and its receptors can vary between individual rats according to genetic background (26), early life experience (27), internal state (28), or environment (24, 29) and have been associated with different degrees of anxiety and fear (27). As we found that oxytocin decreases freezing responses, yet leaves cardiovascular responses unaffected, this specific regulation could preserve the internal, visceral expression of fear, but alleviate the behavioral inhibition that leads to freezing. Such individual regulation may provide the most adequate reaction in circumstances when a proactive behavioral response is required, while preserving an internal, visceral adaptive response to fear.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S8

Tables S1 to S3


References and Notes

  1. Further information including materials and methods is available as supporting material on Science Online.
  2. Acknowledgments: We thank E. Welker for outstanding scientific advice; H. Markram and R. Perin for expert help with multiple patch recordings; R. Daniel and D. Rainnie for feedback on the manuscript; and O. Bussard, V. Martin, N. Pochon and L. Bonsignore for technical support. Supported by Swiss National Science Foundation grant 3100A0-116462.
View Abstract

Navigate This Article