The development and maintenance of neural circuits is highly sensitive to neural activity. General anesthetics, as befits their ability to produce unconsciousness, have profound effects on neural activity. As such, there is concern that these agents, by disrupting neural homeostasis, may alter cellular integrity and interfere with brain wiring, such as when exposure occurs during the vulnerable period of brain development (Perouansky and Hemmings, 2009; Lei et al., 2014). Thus, data from a variety of animal models including rats, mice, C. elegans, guinea pigs, pigs, sheep, and non-human primates indicate that anesthetic exposure during early neurodevelopment is associated with widespread apoptosis, changes in dendritic spine density, synaptic strength, and number, and long-lasting cognitive impairment (Rizzi et al., 2008; Liang et al., 2010; Gentry et al., 2013; Olutoye et al., 2015; Whitaker et al., 2016; Coleman et al., 2017; Walters and Paule, 2017). The implications of these data for use of anesthetics in children is a matter of debate (Sun et al., 2016; McCann et al., 2019) but concerns remain that long or multiple general anesthetics may have deleterious long-term consequences (Davidson and Sun, 2018).

One of the most striking and consistent observations of studies thus far is that the deleterious effects of anesthesia on the developing brain do not appear to be limited to specific agents or agents of a particular class. Most general anesthetics work, at least in part, through γ-Aminobutyric acid A (GABA_A) receptor agonism or NMDA receptor antagonism but agents that act on either or neither of these systems induce neurodevelopmental effects. For instance, agents with strong GABA_A receptor agonist properties such as isoflurane, sevoflurane, desflurane, propofol, etomidate, and midazolam cause neuroapoptosis and/or alterations in synapse number, morphology or strength in animals, but so too does ketamine, an agent with only minor effects on GABA_A receptors that acts mainly by antagonizing NMDA receptor-mediated neural excitation (Young et al., 2005; Slikker et al., 2007; Brambrink et al., 2010, 2012; Liang et al., 2010; Kodama et al., 2011; Creeley et al., 2013; Zheng et al., 2013; Coleman et al., 2017). While controversial, dexmedetomidine, an α2-adrenergic receptor agonist with little or no effect on either GABA_A or NMDA receptors, has also been shown to cause neuroapoptosis in developing neurons (Liu et al., 2016). The fact that these neurodevelopmental effects occur across a broad range of anesthetic/sedative agents with diverse receptor mechanisms of action implies they are mediated by a common, but unidentified, intracellular signaling pathway.

To test that hypothesis, we exposed cortical cells in vitro to anesthetics with differing mechanisms of action and quantified signaling pathway responses with a semi-hi-throughput protein quantification assay. The most consistent signaling event was a reduction in phosphorylation of the ribosomal protein S6, which we confirmed also occurs in the brain of neonatal mice anesthetized with sevoflurane. pS6 is a well-known member of the mTOR pathway, and mTOR signaling plays many important roles in brain development and neural homeostasis. Therefore, suppression of this pathway is a potential mechanism for the adverse effects of anesthetics on the developing brain.

After correcting for the false discovery rate, we identified signaling events in the in vitro experiments that were significantly altered by agent exposure; 6 of these involved changes in phosphorylation of the ribosomal protein S6 (pS6). Propofol and dexmedetomidine reduced pS6 at 6 h (p = 0.002 and 0.049, respectively; N = 4; Figures 2B, 3A). Likewise, etomidate (p = 0.027; N = 4) and APV/TTX (noted with two separate antibodies to distinct phosphorylation sites; p = 0.033 and 0.040; N = 4; Figures 2B, 3A), caused a significant reduction in pS6 at 24 h. In contrast, bicuculline, which increases neural activity (Arnold et al., 2005), increased both pS6 and pyruvate dehydrogenase after 1 and 24 h of treatment (p = 0.032 and 0.032, respectively; N = 4; Figures 2B, 3A). Given that anesthetics suppress pS6 similar to APV/TTX, whereas bicuculline increases it, these data suggest that anesthetics might be inhibiting pS6 levels by reducing neural activity.

We intentionally used supra-therapeutic anesthetic concentrations in these initial experiments to demonstrate proof-of-concept. Therefore, we next tested a range of clinically relevant concentrations of propofol and etomidate, ranging from 2 to 100 μM propofol or etomidate for 6 h. Propofol produced a significant reduction in pS6 at 2 μM (p = 0.005; N = 4), 10 μM (p = 0.006; N = 4), and 20 μM (p = 0.047; N = 4), and etomidate produced a significant reduction in pS6 at 5 μM (p = 0.002; N = 4), 10 μM (p = 0.002; N = 4), 20 μM (p = 0.01; N = 4), 50 μM (p = 0.002; N = 4), and 100 μM (p = 0.006; N = 4) indicating that the decrease in pS6 occurs at brain concentrations of these agents reported to cause clinical effect (Gredell et al., 2004; Benkwitz et al., 2007).

To determine if similar changes in pS6 occur in vivo during anesthesia, we anesthetized postnatal day 7 (P7) mice with 2% sevoflurane for 6 h and measured pS6 in the brain by Western blot. Sevoflurane is a commonly used volatile anesthetic in children and is reported to produce neuroapoptosis and synaptic abnormalities in neonatal rodents. Compared to control animals that received only oxygen, there was a significant decrease in pS6 in the brain of the sevoflurane anesthetized mice (p = 0.00015, N = 6 animals per group; Figure 4). Furthermore, there was a significant increase in cleaved caspase 3 in the brain of the anesthetized mice (p = 3.21 × 10⁻⁶, N = 6; Figure 4). These data indicate an anesthetic-induced decrease in pS6 occurs in the brain of neonatal animals at a concentration of sevoflurane that also triggers an increase in a marker of apoptosis.

Finally, we used a systems-based approach to further evaluate the data from the RPPA experiments. For each signaling event assayed using every time point and pharmacological treatment, we computed a Pearson correlation coefficient for every other assay in a pairwise manner. These correlation coefficients were then clustered using a pairwise clustering algorithm to develop a correlation matrix (Figure 5A), where only statistically significant correlations are shown (p < 0.05). This correlation matrix allows us to identify potential signaling modules that may mediate anesthetic-induced signaling. A highly correlated signaling module (identified in Figure 5A and highlighted in Figure 5B) includes pS6 and other cell mediators of the mTOR-signaling pathway (Figure 5C). A correlation matrix comparing how the module highlighted in Figures 5A,B is related across different pharmacological treatments, indicates that propofol, etomidate, dexmedetomidine, and APV/TTX all similarly modulate this pathway (Figure 5D).

Using a hi-throughput protein quantification assay, we demonstrate that clinically relevant concentrations of propofol, etomidate and dexmedetomidine, agents with different mechanisms of action, all suppress S6 phosphorylation in early stage cultured cortical cells. We also extended and validated these results in an vivo model and show that 6 h of sevoflurane anesthesia at clinically relevant concentrations leads to a reduction in pS6 as well as an increase in the apoptosis marker cleaved caspase 3. Phosphorylation of S6 is a well-known component of the mTOR-signaling pathway, and a systems-based approach to analyze the RPPA results further supports our hypothesis by showing that many signaling molecules involved in the mTOR pathway function in a highly correlated signaling module. The activation of this module was correlated between propofol, etomidate, dexmedetomidine and APV/TTX, suggesting that alterations in neural activity level may be responsible for mediating changes in the mTOR pathway. Consistent with this possibility, there was a small but statistically significant increase in expression of pyruvate dehydrogenase, a component of cellular respiration, in cells exposed for 24 h to bicuculline, a drug that increases neural activity. The fact that pyruvate dehydrogenase suppression was not observed with drugs that reduce neural activity suggests a floor in its expression, driven perhaps by ongoing basal, non-activity driven cellular metabolism.

The highly conserved mTOR pathway is involved in many aspects of neural development and homeostasis, making it an intriguing potential target for mediating neurodevelopmental effects of anesthetics (Lipton and Sahin, 2014). Interestingly, the mTOR pathway is implicated in isoflurane and propofol-induced synaptic and behavioral changes, albeit via an aberrant increase in mTOR signaling (Kang et al., 2017; Xu et al., 2018). In that work, the authors exposed older animals (P18) to 4 h of 1.5% isoflurane anesthesia and found 12 days after exposure (P30) that there was an increase in pS6 staining in the dentate gyrus. This increase in pS6 staining was associated with an increase in dendritic length, a reduction in the number of dendritic spines, and later spatial learning defects. Inhibiting the isoflurane-induced activation of the mTOR pathway with rapamycin rescued both the spine loss and behavioral defects, implicating the mTOR pathway in the neurodevelopmental effects of isoflurane, a prototypical volatile anesthetic. Further studies using primary cortical neurons harvested from E18 rats, and exposed to isoflurane between 3 and 7 DIV, showed a similar increase in the number of pS6 positive cells 3+ days after exposure to isoflurane (Kang et al., 2017; Xu et al., 2018).

Our experiments appear to be at odds with the results reported above, inasmuch as we report that propofol, etomidate, and dexmedetomidine reduce pS6 in cultured cells and that sevoflurane produces a similar reduction in the developing brain. However, others have reported that sevoflurane reduces pS6 in the developing brain (Tan et al., 2015; Xu et al., 2018) and that a decrease in mTor signaling, as induced by the mTor inhibitor rapamycin, is associated with apoptosis as identified by markers such as CC3 (Bowling et al., 2018). In addition, there are two important differences between our study and the one mentioned above. First, our anesthesia exposure occurred in DIV8 cortical cells (isolated from E16.5 mice) or P7 animals whereas the prior investigators exposed mice on P18. This is potentially important because the effects of anesthesia on the brain appear to be exquisitely sensitive to developmental age. For instance, apoptosis is first detectable in rodents when anesthesia exposure occurs around birth, reaches a peak when animals are exposed at P7, and then rapidly declines such that it is almost non-existent when exposure occurs on P14 (Yon et al., 2005). Further, others have reported that propofol anesthesia will lead to a loss of dendritic spines when exposure occurs at P5-P10 but that exposure a week later leads to an increase in spines, indicating that in rodents anesthesia has opposite effects depending on developmental age (Briner et al., 2011). Taken together, these data suggest that differences in the mTOR signaling response may underlie some of the age-dependent effects of anesthetic agents on neuronal viability and dendritic spines.

Another potential reason for inhibition of pS6 in our study versus an increase in another may be differences in the post-anesthetic interval, as we assayed pS6 immediately after anesthesia whereas other investigators did so 12 days after exposure in mice, or 3+ days after exposure in rat primary cortical cultures (Kang et al., 2017; Xu et al., 2018). Phosphorylation of S6 corresponds to the activity state of neurons; thus, like other activity-regulated immediate early genes such as Fos, Npas4, and Arc, a change in neuronal action potential firing leads to a corresponding change in phosphorylation of S6 (Kim et al., 2010). Indeed, pS6 is frequently used as an indirect assay for neural activity (Kim et al., 2010; Renier et al., 2016). As such, one would expect phosphorylation of S6 to be low at the end of a prolonged anesthetic, when neural activity levels are generally reduced. Using the same logic, a decrease in pS6 following sevoflurane anesthesia suggests the agent is inhibiting neural activity in the cortex of P7 animals, whereas others have suggested that it does so by increasing excitation (Zhao et al., 2011). On the other hand, increased pS6 days or weeks later might be secondary to, or a marker of, compensatory post-anesthetic neuronal hyperexcitability. Indeed, in the offspring of mice exposed to sevoflurane on post-natal D 16/17, there was an increase in mTor phosphorylation in the brain 24 h later (Ju, 2020). Homeostatic synaptic scaling and several other well-described mechanisms predict that perturbations in neural activity, such as the decrease that typically occurs during general anesthesia, will induce compensatory synaptic changes and a period of neuronal hyperexcitability subsequently (Turrigiano et al., 1998; Turrigiano, 2012; Hengen et al., 2016). Therefore, to the extent pS6 is an activity-regulated immediate early gene, an increase in cortical pS6 days after exposure to general anesthesia with sevoflurane may be a homeostatic response to an earlier period of anesthetic-induced neuronal depression.

Strengths of our study include use of in vitro and in vivo models, a multiplexed assay system, several anesthetic classes, and non-anesthetic in vitro controls. This work also has important limitations. While we identified a common cell-signaling module that is altered similarly across a wide variety of both intravenous and inhalational anesthetics, we have not directly linked this pathway to the central nervous system complications observed after early developmental exposure to these agents. As others have reported neurotoxicity in older animals with an increase in mTOR signaling, the conditions under which an acute change in pS6 is physiologic or detrimental remains to be determined. We did not investigate anesthetic-induced CC3 changes in the in vitro experiments because the reagent had not been validated for the RPPA platform, and thus we cannot confirm that propofol, etomidate and dexmedetomidine produced apoptosis-like changes in the in vitro model. However, there was an increase in CC3 in the brain of neonatal mice exposed to sevoflurane, suggesting that the pS6 changes are associated with apoptotic events. Finally, this study was largely performed using soluble agents in a tissue culture model. This limits generalizability and requires that results obtained in vitro for propofol, etomidate and dexmedetomidine be replicated in vivo. Notably, however, we confirmed the results in vivo with a widely used inhalational anesthetic (sevoflurane) at a clinically relevant dose.

We conducted these experiments in order to determine whether exposure of cortical cells to anesthetics of different classes alters similar signaling pathways, and thus to provide insight into how agents with different mechanisms of action may mediate similar changes within developing neurons. Here, we find that at clinically relevant concentrations, the intravenous anesthetics propofol, etomidate, and dexmedetomidine suppress pS6 levels in cultured cortical cells, and sevoflurane has a similar effect in the developing mouse brain. The fact that these effects are similar to those seen with the neural activity blockers APV/TTX and opposite those of the neural activity enhancer bicuculline suggest that pS6 is changing in response to alterations in neural activity. Because pS6 is a well-established member of the mTOR signaling pathway, our results suggest that some aspects of anesthetic actions on developing neurons are triggered by reductions in neural activity and subsequent perturbations in this important signaling pathway.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

MF, TG, DC, and GC conceived and designed the experiments. MF, TG, AP, and BH performed the experiments. MF analyzed the data and drafted the manuscript. All authors revised and approved the final manuscript.

This work was funded by the Harvard Anesthesia Training Program (GM007592), a Foundation for Anesthesia Education and Research GEMSSTAR Award, and NIA AG053280 to MF, NIA AG048522 to DC, and NIA AG051812 to GC.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.