Anesthesia analgesia in the amygdala

Abstract

General anesthetics during surgery are presumed to block pain by dampening brain activity and promoting loss-of-consciousness. A new study shows that anesthetics activate an endogenous analgesia neural ensemble in the central nucleus of the amygdala.

“Gentlemen, this is no humbug!” exclaimed the dentist William T. G. Morton, in 1846 Boston, ushering in the modern age of anesthesia. His new formulation of sulfuric ether vapors allowed—for the first time—the painless surgical removal of a tumor from a patient’s neck. Anesthesia has made surgical procedures in both humans and animals humane and pain-free, but the mechanism of action of general anesthesia has been a mystery¹. Now we are beginning to understand the neural circuit mechanisms underlying the remarkable analgesic effects of anesthesia.

In this issue of Nature Neuroscience, Hua and Chen and colleagues at Fan Wang’s lab uncovered three clusters of anesthesia-active neurons in the hypothalamus, the bed nucleus of the stria terminalis and the amygdala (Fig. 1a)². Wang’s group recently revealed the function of the hypothalamic supraoptic nucleus neurons in anesthesia-induced slow-wave sleep³. Here Hua and Chen et al. focus on the heterogeneous ensemble of inhibitory central nucleus of the amygdala (CeA) neurons that are activated by multiple general anesthetics. These CeA neurons appear to selectively influence pain by blocking nociception, the neural processes that generally lead to the perception of pain, globally throughout the brain².

Fig. 1 |. Anesthesia co-opts amygdalar analgesia ensembles.

a, Anesthesia induces sleep through activation of the supraoptic nucleus of the hypothalamus³ and inhibits pain through the CeA². b, Anesthesia-active neurons in the CeA lateral capsule region express enkephalin and PKCδ, but not CGRP-R. The anesthesia ensemble projects to several brain regions, notably pain-associated regions such as the insula, bed nucleus of the stria terminalis (BNST), PBN (parabrachial nucleus) and PAG, among others. c, Anesthesia neurons respond to various general anesthetics with dynamic and stable activities. d, Left: activation of the CeA anesthesia ensemble reduces reflexive and recuperative pain behaviors with no effect of sleep. Right: inhibition of the ensemble exacerbates reflexive and recuperative pain behaviors and blocks low-dose ketamine analgesia.

The authors used an elegant activity-dependent method termed CANE (capturing activated neuronal ensembles) to detect anesthesia responsive neural ensembles⁴. Hua and Chen et al. successfully identified a CeA anesthesia-active ensemble responsive to commonly used general anesthetics, such as ketamine and isoflurane. Using CANE to express the fluorescent calcium sensor GCaMP, the authors visualized the activity dynamics of individual isoflurane-captured neurons with in vivo imaging by miniature head-mounted microendoscopes (Fig. 1b). Isoflurane-active ensembles showed similar increased calcium activities and Fos expression upon later exposures to ketamine and dexmedetomidine, demonstrating that multiple general anesthetics engage a common set of neurons.

Intriguingly, a subset of the anesthesia ensemble displayed transient activity following the initial anesthetic exposure, while another subset demonstrated persistent activity throughout anesthesia, perhaps reflecting the maintenance of analgesia during the procedure. Although these activity dynamics fluctuated between different anesthetics, this suggests further functional subpopulations to be explored that may encode the internal state of the animal, recent experiences or simply anesthetic pharmacokinetics. As of this writing, it remains unresolved how input neural circuits, such as the reticular activating system¹, trigger this anesthesia ensemble or whether these compounds directly bind to molecular targets in CeA neurons.

The authors go on to causally link the CeA anesthesia ensemble to analgesia by optogenetically manipulating this neuronal population in the absence of anesthetics. Optogenetic activation of the channelrhodopsin-tagged anesthesia ensemble in freely behaving mice strongly decreased reflexive and recuperative responses to acute, persistent and neuropathic pain of multiple modalities (Fig. 1d). Conversely, ensemble inhibition was aversive and promoted exacerbated coping behaviors to noxious stimuli, including paw licking and face wiping (Fig. 1d). While a growing literature highlights a hemispheric lateralization phenomenon for CeA pain modulation⁵, the authors found that CeA anesthesia ensembles exist bilaterally and that unilateral activation of either the left or right CeA ensembles diminished pain behaviors.

In low doses, ketamine produces analgesia and diminishes arousal, without loss of consciousness⁶. The authors took advantage of this fact to validate their hypothesis that the analgesic and loss-of-consciousness effects of anesthesia are dissociable and act within unique brain circuits. To do so, Hua and Chen et al. tested the effect of optogenetic inhibition of the CeA anesthesia ensemble on the analgesic effects of low-dose ketamine in a model of capsaicin-induced pain in mice. Convincingly, the authors found that inhibition of anesthesia-activated CeA neurons eliminated low-dose ketamine analgesia, arguing that ketamine’s pain-relieving effect partly relies on supraspinal activation of GABAergic amygdalar neurons.

Optical activation of the ensemble had no effect on general motor or motivational behaviors, nor on the power spectrum of EEG. This critically demonstrates that activation of these neurons does not alter wakefulness states⁷. Additionally, the authors found that only ~5% of the anesthesia ensemble responded to certain noxious stimuli and largely did not overlap with formalin-evoked Fos-expressing amygdalar neurons. Thus, the substantial processing power of this network is not in facilitating but in dampening nociception.

The emerging picture of the CeA paints its broader function as a rheostat for a number of survival contingencies, from behavior selection to emotion regulation, including pain modulation⁸. Molecular identity and precise anatomic location are two common dimensions that partly characterize the function of CeA cell types. The anesthesia ensemble was largely confined to the CeA capsule region in the mid-posterior axis (Fig. 1b). These neurons form an ensemble of molecularly distinct cell types expressing markers for PKCδ and the endogenous opioid enkephalin. Prior work found that optogenetic activation of PKCδ neurons in the anterior CeA promotes defensive freezing⁹. However, the anterior and mid-anterior PKCδ neurons linked to aversive processing showed little overlap with the anesthesia ensemble. Enkephalinergic CeA neurons are anti-nociceptive¹⁰, suggesting that this subcircuit of the anesthesia ensemble may play a key role in pain modulation. Future work to test the role of endogenous opioids in this circuit upon optical analgesia could evaluate sensitivity to the opioid receptor antagonist naloxone. Indeed, optical excitation of the ensemble produced a reinforcing conditioned place preference in naïve uninjured animals. Further evaluation for potential reinforcement or addiction liabilities should be conducted before therapeutically targeting this ensemble for pain management.

The lateral capsule of the CeA integrates nociceptive and threat-related information from the basolateral amygdala (BLA) and parabrachial nucleus. This ‘nociceptive amygdala’ region, functionally identified by noxious-induced phosphorylated ERK but typically not Fos expression, is comprised of calcitonin gene related peptide receptor (CGRP-R) expressing neurons¹¹. However, the anesthesia ensemble was almost completely separate from the CGRP-R clusters in anterior and posterior regions. Whether there is a direct interaction between these microcircuits, where the anesthesia-ensemble inhibits the nociceptive CGRP-R circuit, remains an open question.

Furthermore, recent evidence¹² has found that activation of somatostatin-expressing CeA neurons induces pain-relief. Hua and Chen et al. suggest that these somatostatin neurons, which do not overlap with the anesthesia ensemble, mediate fear or threat responses and thus could relate to the phenomenon of stress-induced analgesia. However, the transient calcium activity of some anesthesia-ensemble neurons mentioned above could also reflect stress activation, i.e., the stress of going under general anesthesia. Indeed, stress has been shown to reduce nociceptive behaviors in rodents and people. However, further in vivo imaging revealed that CeA anesthesia ensembles are inhibited by stress and that restraint stress did not elicit Fos expression in the ensemble. This poses the perplexing question: if not stress nor threat nor noxious stimuli, then what ethological situations engage the anesthesia ensemble for anti-nociception? Could some of these purported analgesic effects be attributable to diminished attention and arousal?

A critical point of this work highlights that analgesia is not a property of anesthetics. Rather, anesthetics, with very distinct molecular mechanisms, interact with and take advantage of evolved neural circuits involved in endogenous control of nociception. Endogenous analgesia research, including placebo analgesia, typically focuses on descending modulation of pain in the periaqueductal gray (PAG) to brainstem circuits. The authors determined the axon projection topology of the anesthesia ensemble, revealing that these neurons project widely throughout pain-processing centers of the brain, including the PAG, insular cortex, striatum, intralaminar thalamus and bilateral BLA (Fig. 1b). Of note, most circuit-mapping studies do not observe CeA projections to the cortex or contralateral BLA, thereby necessitating future work to validate these connections. Nonetheless, activation of the CeA anesthesia ensemble potently suppressed noxious formalin-evoked Fos in all brain regions receiving apparent direct input. This highlights the strong impact of scaling nociceptive information across different functional domains of pain perception and the translational power that could come from leveraging future therapies that target the amygdalar pain-aversion ensembles¹³, as well as pain-suppression ensembles.

The discovery of this pain-suppressing neural population reinforces long-standing concerns of rodent pain research conducted under anesthesia, including functional MRI and some in vivo electrophysiology techniques. This work definitively shows that commonly-used rodent anesthetics potently suppress nociception, and thus prior conclusions might have missed the ground-truth processing and functional role of these circuits. This critical finding opens the door for the reevaluation of a number of pain-related hypotheses using neural activity measurements in awake, freely behaving animals that do not have the confounds of anesthetic anti-nociception.

Clearly a loss of consciousness blocks the conscious perception of pain; however, it has not been conclusively shown that the neural processes underlying loss-of-consciousness are anti-nociceptive, i.e., they actively repress pain processing in the CNS. While isoflurane and ketamine prevent the induction of noxious-induced Fos in nociceptive circuits, Hua and Chen et al. further demonstrate that anesthetic anti-nociception is a dissociable process from those inducing sleep and, likely, loss of consciousness. Intriguingly, fear conditioning has been demonstrated to occur during anesthesia¹⁴, suggesting that some nociceptive processes remain active and engage associative memory circuits. Approximately 10% of surgical procedures result in persistent postoperative pain that can transition to chronic pain, with incidence as high as 50–85% depending upon the surgical procedure¹⁵. Perhaps the CeA anesthesia ensemble is insufficiently engaged by anesthetics in certain surgical cases, which permits the formation of molecular pain memories and facilitates ‘learned’ persistent pain. While the authors demonstrated that several general anesthetics activate the ensemble, it is not clear whether different anesthetic dose-concentrations or combinations with other common perioperative drugs (for example, fentanyl induction, midazolam and a paralytic) differentially involve the ensemble. These parameters could induce relevant spatiotemporal patterns of CeA activity that might influence analgesic efficacy during surgery, which in turn could explain the terrifying, but rare, occurrence of anesthesia awareness, when patients experience pain and paralysis during surgery. Collectively, a full picture of this CeA ensemble’s dynamic properties could lead to the production of better and circuit-precise anesthetics, or even chronic pain relief therapies.