Address correspondence to Dr Christopher W. Connor, MD, PhD, Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women's Hospital, 75 Francis Street, CWN L1, Boston, MA 02115. Address e-mail to [email protected]

The most longstanding question in anesthesiology is the mechanism of action of the volatile anesthetics. How do these agents, from ether through to modern fluorinated agents, induce an unconsciousness and unresponsiveness that is promptly reversible? Historically, the anesthesiology literature has focused on anesthetic mechanisms involving chemical synapses, such as γ-aminobutyric acid (GABA) and N-methyl-D-aspartate (NMDA), and on the disruption of either the cortex or thalamocortical loops. However, each of these explanations presents an incomplete account of the therapeutic and adverse effects of volatile anesthetics on the body. Although competitive infra-additive pharmacodynamic interactions of volatile anesthetics with particular channels have now been reported, all volatile anesthetics exhibit a linear and additive dose-response in terms of minimum alveolar concentration (MAC) rather than the sigmoidal Hill equation response that typifies protein interactions. Furthermore, some volatile agents that the Meyer-Overton correlation would predict are anesthetics that instead induce convulsions.

Several important aspects of general anesthesia remain unexplained. What is the pharmacologic basis of stage 2 of anesthesia or the hysteresis of induction versus emergence? How are general anesthetics able to affect the development of children and lead to cognitive dysfunction in the elderly? What processes account for all of the stereotypical changes in the electroencephalography (EEG) of anesthetized patients? Fundamentally, there is not a consensus on or a specific test that can be performed to confirm whether someone has lost consciousness under general anesthesia. Experiments utilizing the isolated forearm technique have questioned this assumption and the definition of general anesthesia. Other experiments have shown that wakefulness can be dissociated from the EEG signals induced by general anesthetics. Because of all of these inconveniently unsolved questions, novel hypotheses for additional pathways of anesthetic action can and should be proposed and investigated.

The prolonged mystery of the basic mechanism of action of volatile anesthetics suggests that we should consider the interaction between (i) a synapse type that has been difficult to test pharmacologically, (ii) a neurological structure that has been difficult to access, and (iii) a brain wave known to be a hallmark of general anesthesia. Several lines of evidence point to the involvement of gap junctions and the neuroanatomic structure of the claustrum as being necessary substrates for consciousness. Gap junctions, also known as electrical synapses, are transmembrane proteins that create a direct cytoplasmic connection between 2 cells. Gap junctions are challenging to assay and target pharmacologically. The claustrum is a thin, sheet-like structure lying between the cortex and subcortical structures, specifically between the striatum and the insula (Figure). It is hard to image with functional magnetic resonance imaging (fMRI) or localize with surface EEG. These are, in a sense, hidden structures whose importance may have been inadvertently overlooked in our search for the mechanism of action of the volatile anesthetics. It is worth considering whether these structures are viable targets of volatile anesthetics, and the induction of unconsciousness. We hypothesize that volatile anesthetics produce a loss of consciousness by disrupting gap junctions in the claustrum, which prevents the claustrum from synchronizing cortical gamma oscillations: a neural correlate of consciousness.

GAP JUNCTIONS

Gap junctions are distributed throughout the body, but are especially abundant in electrically excitable tissue such as brain, smooth muscle, and cardiac muscle. In the myocardium, gap junctions are responsible for propagation of electrical cardiac action potentials and their role is especially apparent when there is a fascicular block in the Bundle of His. Gap junctions independently arose as connexins (Cx) in vertebrates and innexins (Inx) in invertebrates. Pannexins (Panx) are homologous to innexins and are evolutionarily conserved from invertebrates, such as Caenorhabditiselegans, to humans. Pannexins predominantly form a channel between the intracellular and extracellular environment, rather than between 2 cells. Including connexins and pannexins, there are 21 types of gap junctions in the human genome. Eleven types of gap junctions are found in the nervous system, but the density of gap junctions varies regionally and over the course of neurological development. Cx45 is strongly expressed in the brain during the first 2 weeks of development but becomes absent in the adult brain except for in the CA3 region of the hippocampus, thalamus, and cerebellar granule cells. Cx43 is the most abundant connexin in the brain and is mainly expressed by astrocytes, oligodendrocytes, and microglia. In adults, neurons express 3 connexins and 2 pannexins (Cx26, Cx32, Cx36, Panx1, and Panx2) with Cx36 and Panx1 predominating.

Gap junctions are permeable to small molecules (smaller than 1.5 kDa) and to electrical current. This allows for adjacent cells to form a group of electrically coupled cells by which a depolarization in 1 cell can propagate directly to nearby cells. Gap junctions induce positive correlation between the cells that they connect. When a depolarization occurs, positive charge from one cell flows to another cell, making the first cell less positive and the second cell more positive. The result of this interaction is a coupling or synchronization rather than a directional excitation or inhibition. Networks of gap junctions allow for the production of oscillations, such as antiphase, phase-locked and bistable rhythms, which can act as timing synchronization for neuronal networks and are important for the generation of axon-axon fast ripples.

ARE GAP JUNCTIONS A TARGET OF VOLATILE ANESTHETICS?

Volatile anesthetics affect a wide variety of transmembrane proteins throughout the body. Although these effects have been better characterized for chemical synapses, electrical synapses are also affected by volatile anesthetics. At high (supratherapeutic) concentrations, volatile anesthetics cause an almost complete blockade of conductance through gap junctions. This near-total inhibition of gap junctions is likely not clinically relevant, except for special clinical cases such as the use of deep volatile anesthesia to terminate refractory status epilepticus. We contend that there is evidence of volatile anesthetic-mediated gap junction dysfunction at lower, clinically relevant concentrations, which may be the mechanism by which volatile anesthetics induce and maintain the state of general anesthesia.

Halothane inhibits gap junctions in cardiac myocytes and induces electrophysiological disturbances at clinically relevant concentrations (1-2 MAC). Isoflurane at 1 MAC strongly reduces ictal activity in slice preparations, which is at least in part due to gap junction blockade. Knockout of the gap junction, Cx36, in mice causes a lower EC50 for recovery from isoflurane (0.37%atm vs 0.49%atm for wild-type), which demonstrates an effect from gap junction binding in a specific subtype within a clinically usual range. Halothane decreases gap junction mean open time and has a greater inhibition of cells that coexpress Cx40 and Cx43 than those that express only Cx40 or Cx43. Although few gap junction-specific inhibitors are available, administration of carbenoxolone increases the sensitivity of mice to sevoflurane. On the other hand, modafinil, a wakefulness-promoting agent, keeps gap junctions open and decreases the resistance to current flow between electrically coupled cells.

Diffuse inhibition of neuronal and cardiac gap junctions is not required and is probably not desirable during general anesthesia. At concentrations compatible with clinical use, interference with gap junctions can cause impairment in the generation of gamma rhythms and cortical synchronization. This phase of anesthesia is distinct from the sort of isoelectric anesthesia that one would observe at, for example, very high isoflurane concentrations and complete blockade. Dickinson et al show a linear decline in gamma oscillation frequency with increasing isoflurane and halothane over the clinical range (0-2%atm) in a stimulated slice preparation. Furthermore, LeBeau et al show that gap junctions are necessary for the sustained production of these gamma oscillations, and that even partial blockade of gap junctions by volatile anesthetics is disruptive:

Also reported by Masaki et al, "Sevoflurane at clinically relevant concentrations (0.1-0.5 mM) suppressed the spontaneous oscillation in membrane current concentration-dependently." In addition to inhibiting gap junctions, volatile anesthetics have been shown to increase gap junction conductance in certain parts of the brain. Sevoflurane produces the greatest amount of locus coeruleus (LC) excitation via gap junctions, which corresponds to clinically observed agitation. The potency of LC neuron excitation parallels the stage 2 excitation seen in children exposed to various volatile anesthetics.

The binding site of anesthetics in gap junctions is not known, although a unique disulfide bond has been suggested. Volatile anesthetics and n-alcohols may disrupt gap junction conductance by decreasing the fluidity of cholesterol-rich domains. Cx43 contains a calmodulin (CaM) binding site that is not present in Cx40. Volatile anesthetics have been shown to bind to the hydrophobic core of CaM and inhibit its activity.

GAP JUNCTIONS AND GENERAL ANESTHESIA

Gap junctions are found in nearly all multicellular creatures and, indeed, all multicellular creatures, including plants, can be "anesthetized" with volatile anesthetics. Inhibition of gap junctions in the heart, by older inhalational anesthetics such as halothane, causes conduction abnormalities and other clinically observable disorders of the electrocardiogram. This effect is less pronounced with the newer halogenated ethers (eg, isoflurane, sevoflurane, desflurane) compared to halogenated alkanes such as halothane and chloroform. One might reasonably hypothesize that the structure of halogenated ethers increases selectivity for neural gap junctions compared to those found in cardiac myocytes. The fact that gap junctions are rich in cholesterol and associate with lipid rafts but are also proteins may help square the circle of specific protein binding versus lipid solubility as potency that is inherent in the Meyer-Overton hypothesis.

Gap junctions are expressed at high levels during infantile development. There has long been controversy about whether anesthetics negatively impact childhood development. Although exposure to general anesthetics can induce frank neuroapoptosis, an alternative and plausible hypothesis holds that the injury occurs to the neural connectome when anesthetic exposure coincides in space and time with normal developmental synaptic pruning. Basic animal research also supports this hypothesis, in which lifelong changes in behavior and neural function can be identified after infantile/larval exposure to isoflurane, even in the absence of cell death.

Gap junctions have a surprisingly high rate of turnover in the membrane, on the order of 1 to 5 hours with an estimated membrane half-life of only 2.6 hours. This may account for the observed neurological hysteresis of general anesthesia, in which induction of anesthesia is a different process than emergence from anesthesia, especially if some degree of irreversible inhibition of gap junctions were to result from induction and maintenance. Gap junctions are turned over by a process of active endocytosis, a process that has been shown to be inhibited by isoflurane, and particularly so in mice with abnormalities of mitochondrial complex I. This provides a possible connection to mitochondrial theories regarding the mechanism of action of the volatiles. Anesthetics may bind to various types of gap junctions and as a result produce different effects. The anesthetic binding site on gap junctions is not known to this level of granularity, but this may explain why some agents have convulsant properties in addition to anesthetic properties. Complications such as agitation during stage 2 of anesthesia and the development of postoperative cognitive dysfunction may be the result of off-target effects.

THE CLAUSTRUM AND GENERAL ANESTHESIA

Several neuroanatomical structures have been proposed as the effect site for general anesthesia, such as the reticular activating system in the brainstem, LC, raphe nuclei, basal forebrain, ventrolateral preoptic nucleus of the hypothalamus, thalamus, and several cortical regions. The thalamus and the connections that it makes with the cortex have been implicated in a loss of connectivity which leads to loss of consciousness during general anesthesia. However, the only case report of a patient being rendered unconscious by a precise electrical stimulation of the brain involves the claustrum. This occurred as a therapeutic accident during neurosurgical mapping for intractable epilepsy in a 54-year-old woman. Electrical stimulation of the claustrum (or, possibly, immediately adjacent in the anterior insula) produced immediate and reversible unresponsiveness in 10 of 10 trials, which occurred with stimulation of 1 particular electrode at a frequency of 50 Hz (ie, in the gamma band), resulting each time "in a complete arrest of volitional behavior, unresponsiveness, and amnesia without negative motor symptoms or mere aphasia" with immediate reversal on termination of the stimulation. Stimulation at other electrodes 5 mm away produced no such effect nor with other frequencies and currents. In other accidental human observations, combat casualties who sustained traumatic brain injury with damage to the claustrum had a prolonged period of unconsciousness. Isoflurane at 1% to 2% inhibits claustral neurons and reduces the connectivity between other structures, such as the medial prefrontal cortex and mediodorsal thalamus. Experimentally, electrical claustral stimulation in rats has recently been shown to deepen isoflurane-induced anesthesia.

Because the claustrum is a thin, deep neuroanatomical structure that is hard to image, it may be the perfect hiding place for a mechanism of anesthesia. Recent advances in ultrahigh field fMRI at 7 Tesla have allowed successful imaging of the claustrum and its connections. Anatomic studies of the claustrum have shown that it has projections throughout the cerebral cortex and, because of its location between the cortex and subcortical regions, may be critical for synchronizing the activity of the brain. The claustrum establishes reciprocal connections with essentially all cortical areas, including motor cortex, prefrontal cortex, cingulate cortex, occipital lobe, temporal and temporopolar cortices, parietooccipital and posterior parietal cortex, the frontoparietal operculum, somatosensory areas, prepiriform olfactory cortex, and the entorhinal cortex. The claustrum is mostly (85%) composed of Golgi type I (spiny glutamatergic excitatory) neurons, which project to the cortex and receive reciprocal projections from the cortex. The remaining aspiny claustral interneurons are densely connected by gap junctions, formed from connexin proteins. This allows for a low-resistance, bidirectional, electrical connection between claustral neurons and the synchronization of interneurons in the 30 to 70 Hz (gamma) range. This functionality likely exists to allow synchrony to be maintained between distant populations of cortical neurons, with the claustrum acting as an interactive gap junction syncytium of interneurons that has reciprocal connections with the rest of the brain, although this posited relationship between structure and function is not universally accepted.

INHIBITION OF CLAUSTRAL GAP JUNCTIONS AS A MECHANISM OF ANESTHESIA

For all the disparate mechanisms proposed for volatile anesthetic action, gap junctions, and the claustrum appear to be positioned to play a significant role. We offer evidence from the literature that clinically relevant concentrations of volatile anesthetics can cause an inhibition that is sufficient to throw sand in the gears of the neurological mechanisms of gamma oscillation generation, claustral synchronization, and hence impair consciousness. Isoflurane reduces the frequency of gamma oscillations by 41% in rat neocortex at 0.16 mM (which is approximately 0.6 MAC, a clinically relevant amount). This would also be sufficient to generate clinical cortical EEG changes that, for example, the BIS algorithms would numerically interpret as anesthetized. Only a partial inhibition of gap junctions is required for the machinery of conscious integration to fall apart. It is reasonable to conjecture that volatile anesthetics bind to gap junctions and disrupt the flow of electrical charge between cells, and that this loss of synchrony between neurons prevents the claustrum from synchronizing anatomically distant areas of the cortex. As a result, the cortex is unable to generate coherent gamma oscillations and the subject enters a state of unconsciousness. The loss of responsiveness under general anesthesia with volatile anesthetics is associated with a reduction in coherent gamma EEG power. While direct evidence of the action of volatile anesthetics here remains elusive, a wealth of circumstantial evidence suggests that this is where we should look more closely: a Supplemental Bibliography is provided (Supplemental Digital Content 1, https://links.lww.com/AA/E884). Other general anesthetics, and particularly ketamine, may produce unconsciousness via a different mechanism.

FUTURE DIRECTIONS FOR OPEN MINDS

Although anesthesiologists have long used volatile anesthetics without precise knowledge of their mechanism of action, this knowledge would be useful for improving the efficacy of these agents, reducing their side effects, and furthering the development of newer, more ideal agents or technologies. Progress on the clinical questions of developmental insult in pediatric patients and postoperative cognitive dysfunction in elderly patients remains impeded by gaps in our understanding of this mechanism of action. Several experiments could be performed to determine whether gap junctions, the claustrum, or its associated gamma oscillations are involved in the production of general anesthesia. Genetic mutation and optogenetics would help to clarify the role of gap junctions in the induction of general anesthesia. Another approach would be to make precise lesions in the claustrum while simultaneously monitoring gamma oscillations throughout the cortex and subcortical regions. The behavioral and electroencephalographic response to volatile anesthetic exposure in a claustral lesioned animal could then be quantified.

Finally, there is the overarching question of the relationship between general anesthesia and unconsciousness. Ethically, it would be helpful to know for certain whether humans or animals are conscious at any given time. Experiments using the isolated forearm technique have shown that patients can respond to commands while under general anesthesia. It is possible that consciousness is never wholly lost under general anesthesia, even if adequate inhibition of physical sensation and memory formation occur such that the patient does not experience terrible pain nor recollection. Is it possible to know for certain that the patient has lost consciousness? Without knowing the substrates or structures that allow for consciousness, it is impossible to directly and precisely monitor the loss and return of consciousness. This is an open and fundamental question for anesthesiologists.

ACKNOWLEDGMENTS

The authors thank Dr Samet Kapakin (Atatürk University, Erzurum, Türkiye) for gross anatomical images and 3D visualization of the claustrum. No generative AI tools were used.

DISCLOSURES

Conflicts of Interest: C. W. C. has consulted for Teleflex, LLC on issues regarding airway management and device design and General Biophysics, LLC on inhalational kinetics. These activities are unrelated to the material in this submission. Funding: Departmental support (R35 GM145319). This manuscript was handled by: Peter A. Goldstein, MD.