Animal studies demonstrate that anesthetic exposure during neurodevelopment can lead to persistent behavioral impairment. The changes in neuronal function underlying these effects are incompletely understood. Caenorhabditis elegans is well suited for functional imaging of postanesthetic effects on neuronal activity. This study aimed to examine such effects within the neurocircuitry underlying C. elegans locomotion.
C. elegans were exposed to 8% isoflurane for 3 h during the neurodevelopmentally critical L1 larval stage. Locomotion was assessed during early and late adulthood. Spontaneous activity was measured within the locomotion command interneuron circuitry using confocal and light-sheet microscopy of the calcium-sensitive fluorophore GCaMP6s.
C. elegans exposed to isoflurane demonstrated attenuation in spontaneous reversal behavior, persisting throughout the animal’s lifespan (reversals/min: untreated early adulthood, 1.14 ± 0.42, vs. isoflurane-exposed early adulthood, 0.83 ± 0.55; untreated late adulthood, 1.75 ± 0.64, vs. isoflurane-exposed late adulthood, 1.14 ± 0.68; P = 0.001 and 0.006, respectively; n > 50 animal tracks/condition). Likewise, isoflurane exposure altered activity dynamics in the command interneuron AVA, which mediates crawling reversals. The rate at which AVA transitions between activity states was found to be increased. These anesthetic-induced effects were more pronounced with age (off-to-on activity state transition time (s): untreated early adulthood, 2.5 ± 1.2, vs. isoflurane-exposed early adulthood, 1.9 ± 1.3; untreated late adulthood, 4.6 ± 3.0, vs. isoflurane-exposed late adulthood, 3.0 ± 2.4; P = 0.028 and 0.008, respectively; n > 35 traces acquired from more than 15 animals/condition). Comparable effects were observed throughout the command interneuron circuitry, indicating that isoflurane exposure alters transition rates between behavioral crawling states of the system overall. These effects were modulated by loss-of-function mutations within the FoxO transcription factor daf-16 and by rapamycin-mediated mechanistic Target of Rapamycin (mTOR) inhibition.
Altered locomotive behavior and activity dynamics indicate a persistent effect on interneuron dynamics and circuit function in C. elegansafter developmental exposure to isoflurane. These effects are modulated by a loss of daf-16 or mTOR activity, consistent with a pathologic activation of stress-response pathways.
Experimental data in laboratory animals demonstrate that early life exposure to anesthetics can induce lasting neurobehavioral and cognitive alterations
The neurobiological bases of these alterations are incompletely understood
Caenorhabditis elegans is a well suited experimental model for long-term functional imaging of neurons after anesthesia exposure
Exposure of Caenorhabditis elegans to isoflurane for 3 h during the first larval stage results in lifelong attenuation in spontaneous crawling reversal behavior
These effects correlate with persistently altered activity dynamics of command interneurons mediating crawling reversals
Genetic dissection of potential underlying mechanisms reveals that these effects are modulated by a loss of daf-16 or mechanistic Target of Rapamycin (mTOR) activity, consistent with a persistent pathologic activation of stress-response pathways
Rodent and primate studies suggest that early exposure to volatile anesthetics can be neurotoxic, producing lasting impairments in learning and memory acquisition.1–3 Primate studies further provide evidence for the impact of anesthesia on socioemotional development.4,5 Although persistent anesthesia-induced neurobehavioral alterations do not appear to be clinically prevalent among singly exposed patients,6–9 rare yet potentially high-risk populations cannot be identified without a mechanistic understanding of these effects.
The nematode Caenorhabditis elegans also exhibits persistent behavioral changes after volatile anesthetic exposure during development. Because C. elegans lacks a cardiovascular system, these changes cannot be related to perturbations of cerebral perfusion as might be implicated in mammals. Gentry et al.10 showed that larvae exposed to volatile anesthetics demonstrate deficits in their ability as adults to perform chemotaxis: the behavioral strategy for searching for food. This abnormality was exclusively observed in worms exposed during a narrow developmental window, the L1 larval stage. Chemotaxis is a complex behavior relying on chemosensation and modulation of behavioral crawling states. It is one of the subtler yet essential integrative tasks performed by the organism.11
C. elegans depends upon the stochastic nature of its locomotion to effectively chemotax. Its movement can be described as a random walk, facilitating dispersion throughout its environment.12,13 Locomotion consists of constant forward movement interrupted by sharp, randomized changes in its direction of movement (“pirouettes”). Chemotaxis is accomplished by modulating the frequency of these pirouettes as the animal nears a food source.14 Bouts of spontaneous backward movement (“reversals”), which interrupt the worm’s otherwise forward movement, initiate pirouettes and are critical to effective chemotaxis.15,16 The neurocircuitry underlying this behavior is well studied and exhibits a bimodal pattern of excitation, corresponding to either forward or backward movement states.17,18 Although sensory input modulates the state of this network, it also switches between states spontaneously, eliciting the observed intermittent reversal behavior. Here we examine persistent effects on this network function after developmental exposure to isoflurane, using functional neuronal imaging in C. elegans to demonstrate a mechanistic neuronal correlate to the previously described behavioral deficits.
Materials and Methods
C. elegans Strains
The animals were maintained at 20°C on nematode growth medium–agarose plates coated with Escherichia coli OP50. When specified, nematode growth medium plates contained 100 μM rapamycin (LC Laboratories, USA), initially dissolved in dimethyl sulfoxide to a concentration of 0.1%. The experiments were performed using hermaphrodites cultured to either early or late adulthood (2 and 10 days after the L1 larval stage, respectively). Imaging experiments were performed using the transgenic strain QW1574 (lite-1[ce314], zfis146[nmr-1::NLSwCherry::SL2::GCaMP6s, lim-4(−3328–2174)::NLSwCherry::SL2::GCaMP6s, lgc-55(−120–773)::NLSwCherry::SL2::GCaMP6s, npr-9::NLSwCherry::SL2::GCaMP6s] #18.9 [from zfex696]), which expresses cytoplasmic calcium-sensitive green fluorescent protein GCaMP6s and a nuclear-localized red fluorescent protein NLSwCherry in eight pairs of premotor interneurons (AVA, AVB, AVD, AVE, AIB, RIM, RIV, and PVC) on an otherwise unmodified wild-type N2 background. When specified, imaging experiments were performed using QW1574 worms that had been crossed with the following mutant strains: MT6347 (ced-3[n2433]), which harbors a missense mutation within ced-3, a key mediator of the apoptotic pathway; CB1370 (daf-2[e1370]), which harbors a temperature-sensitive mutation within the C. elegans insulin/insulin-like growth factor 1 receptor ortholog daf-2; or CF1038 (daf-16[mu86]), which harbors a suppressive mutation within the FoxO transcription factor daf-16. Behavioral experiments were performed with QW1574 as above. The QW1574 strain was generated in the laboratory of Dr. Mark Alkema (University of Massachusetts Medical School, Worcester, Massachusetts), and MT6347, CB1370, and CF1038 were obtained from the Caenorhabditis Genetics Center (University of Minnesota, Minneapolis, Minnesota). Invertebrate research, such as that involving C. elegans, is exempt from institutional animal care and use committee review, as specified by the Animal Welfare Act and the Health Research Extension Act.
Gravid adults were transferred to fresh nematode growth medium–agarose plates (with 100 μM rapamycin, when specified) and allowed to lay eggs for a period of 2 h. The adults were then removed, and the eggs were allowed to hatch. 15 h after the timed egg laying, half of the newly hatched L1-stage larvae were exposed to an atmosphere of 8% isoflurane in room air (19 to 22°C) for a period of 3 h as previously described,19 while the other half were maintained in an adjacent, matched environment under room air only. The anesthetized animals were recovered to room air. These colonies were then maintained in the laboratory for monitoring and assessment for either a further 2 days, (i.e., to early adulthood, 1 day after the larval stage) or for a further 10 days (i.e., to late adulthood, 9 days after the larval stage). The animals hatched on rapamycin-containing plates were maintained on these same plates throughout their development.
Anterior Touch Responsiveness
The worms were stroked with an eyelash across the anterior portions of their body while engaging in forward locomotion in a food-free environment. Individual animals were stroked 5 times, with an interval of at least 5 min between stimuli. The animals were scored as responsive or nonresponsive to each stimulus depending on whether they initiated a movement reversal or not, by an experimenter blind to condition. The anterior touch responsiveness assay was performed during late adulthood.
Spontaneous Reversal Rate
The worms were observed freely moving in an environment devoid of food. The animals were transferred to a clear agar dish and allowed to rest for 15 min, after which their crawling movement was video recorded. The movies were analyzed by an experimenter blind to condition using ImageJ20 to count the number of spontaneous reversals the animals initiated over a given period of time (between 2 and 10 min). The spontaneous reversal rate was assessed during early and late adulthood.
Before imaging, the worms were encased in a polyethylene glycol hydrogel after a brief exposure to 5 mM tetramisole. The hydrogel provides an effective method of restricting the mobility of the worms21 without exposing them to additional and potentially confounding pharmacologic agents. Within the completely mapped neurocircuitry of C. elegans, the AVA and AVB neurons are well characterized as mediating reverse and forward crawling, respectively, representing the two major behavioral states of the system.17 Spontaneous in vivo neuronal activity was assessed in the command interneuron AVA using confocal laser fluorescence microscopy and the calcium indicator GCaMP6s, as previously described.19 Image sequences were acquired for 10 min for each worm at a rate of 4 images/s. The acquired time-lapse images were analyzed by identifying and tracking the nucleus of the AVA neuron in the red fluorescent signal (as described above for the QW1574 strain) and then extracting the green GCaMP fluorescence intensity in the surrounding soma of the identified neuron. The animals were imaged during early and late adulthood.
As above, the worms were encased in a polyethylene glycol hydrogel after a brief exposure to 5 mM tetramisole and before imaging. The immobilized animals were then immersed in 1 × S-Basal solution (100 mM NaCl, 50 mM KPO4 buffer, 5 μg/ml cholesterol) with 5 mM tetramisole and imaged using a dual inverted selective plane illumination fluorescence microscope (Applied Scientific Instrumentation, USA). Three-dimensional volumetric time-lapse images of the head region were acquired for 10 min for each worm at a rate of 2 volumes/s. The calcium indicator GCaMP6s was used to capture spontaneous in vivo neuronal activity throughout the premotor command interneuron circuitry (neuron pairs: AVA, AVB, AVD, AVE, AIB, RIM, and RIV). Postprocessing of imaging data was performed with custom Python and MATLAB (Mathworks, USA) scripts to track the neuronal nuclei in three dimensions in the red fluorescent signal and then to extract the associated green GCaMP6s fluorescence intensities in the surrounding somas.22 The PVC neurons are located in the tail region and are therefore outside of the imaging volume. The animals were imaged exclusively during early adulthood, because tissue autofluorescence and reduced NLSwCherry expression prohibited reliable neuron tracking in worms aged to late adulthood.
Calcium transient onsets (i.e., on transitions) and offsets (i.e., off transitions) captured in AVA via confocal microscopy were manually identified by clear shifts in GCaMP6s fluorescence intensity (ΔF/F0; i.e., the difference in fluorescence above baseline divided by baseline fluorescence), from low to high and high to low, respectively. The duration of each transient was determined by measuring the time difference between the first frame of its on transition and the first frame of the following off transition. The duty ratio, representing the proportion of time that the neuron was in the on state versus the off state, was determined by summing all of the calcium transient durations for each recorded fluorescence signal and dividing by the total number of time points in that signal (i.e., 2,400 time points when acquired at 4 frames/s for 10 min).
To examine the manner in which AVA transitions between activity states, individual on and off transitions were manually isolated. Hyperbolic tangent functions fit closely to these activity state transitions and have been used previously to model ion channel currents across biologic membranes.23,24 Nonlinear fitting of these functions to the activity state transitions eliminates noise and allows for an unbiased assessment of transition properties. The fitted functions were used to calculate the time required to shift from 5% of the maximum value above baseline fluorescence to 95% (rise time) and from 95 to 5% (fall time), respectively.
Average AVA on transition plots were generated by overlaying individual transitions, such that the calculated 5% max(ΔF/F0) positions aligned at the same frame. Average off transition plots were generated in the same manner, with the 95% max(ΔF/F0) positions aligned. Alternatively, average time differential plots of AVA state transitions were generated by overlaying the time derivatives of individual transitions. The time differentials of GCaMP6s fluorescence intensity traces were calculated using total-variation regularization, which suppresses the noise amplification typical of alternative approaches.25
Calcium transient onsets (i.e., on transitions) and offsets (i.e., off transitions) captured within the premotor command interneuron circuitry via dual inverted selective plane illumination microscopy were identified automatically by convolving the activity trace with a windowing function. The windowing function consists of a square wave with a central null region. By comparing the mean signal intensity before the central null region (i.e., pre) with the mean signal intensity after the null region (i.e., post), a determination is made as to whether an on transition or an off transition occurred within the central region. If the ratio of post:pre is greater than 2, then an on transition is determined to have occurred. Conversely, if the ratio pre:post is greater than 2, then an off transition is determined to have occurred. Based on preliminary data obtained from confocal imaging, a 20-s window with a central null region of 4 s was used to detect on transitions, and a 60-s window with a central null region of 12 s was used to detect the slower off transitions. After regions that contained on and off transitions were identified, these transitions were fitted with hyperbolic tangent functions as described above, allowing their locations and rapidity to be determined precisely. Transitions whose duration exceeded the permitted maximum specified by the fitting parameters were rejected as outliers. No more than three outliers were found for any particular condition, and in total only 20 of 9,936 events considered were rejected as outliers (0.20%). All other data discussed in this paper were evaluated for outliers, but no further action was necessary. All other data generated/extracted were used, and no data were missing or reconstructed.
Comparisons between measurements were made using a 2 × 5 ANOVA, in which the groups were condition (control vs. isoflurane) and genotype/drug (wild type, daf-2, daf-16, ced-3, rapamycin). ANOVA was applied independently to the rise times and fall times to determine whether there was a detectable difference in means based upon the interaction of condition × genotype/drug. On rejecting the ANOVA null hypothesis that the measurements were samples of the same parent population, unpaired two-tailed t tests were then used to compare the means of the separate experimental configurations. Twelve such tests (m = 12) were performed separately on the rise time and fall time data, respectively. These tests were:
Tests 1 to 4: Comparison of the means of each of the genotypes/drug daf-2, daf-16, ced-3, and rapamycin dependent on condition (control vs. isoflurane).
Tests 5 to 8: Comparison of the control wild-type mean to the control mean of each of the genotypes/drug daf-2, daf-16, ced-3, and rapamycin.
Tests 9 to 12: Comparison of the isoflurane-exposed wild-type mean to the isoflurane-exposed mean of each of the genotypes/drug daf-2, daf-16, ced-3, and rapamycin.
The significance of the ANOVAs was assessed at α = 0.05. Although any individual test at P < 0.05 can be significant, performing multiple tests of significance raises the concern that some apparently significant outcomes may be generated by chance, and therefore the further tests under multiple comparisons were performed both at α = 0.05 and also using the Bonferroni correction26 at α = 0.05/m to identify a subset of results of extraordinarily robust significance. The Bonferroni correction is the most conservative test of significance and stipulates such a low α that the likelihood of even one significant value occurring by chance is less than 0.05. However, because the Bonferroni correction requires such a conservative α, it often leads adversely to frequent type II errors in which true discoveries are erroneously rejected. Nevertheless, 8 of the 13 significant results that were found at α = 0.05 remained so even under this stringent correction.
All data are reported in the text as the means ± SD. The figures display both the standard deviations of the data and also the 95% CI. We visualized and reviewed the histograms of the underlying distributions, and distributions consistent with the preconditions of the parametric t test were observed. All stated P values were calculated using the unpaired two-tailed t test. Whenever data asymmetries were observed, the nonparametric Mann–Whitney U test was further applied to confirm the validity of the t test derived P value. No gross disparities between such P values were observed. Sample sizes for all experiments were based on prior experience with the experimental designs rather than a priori statistical power calculations. All statistical methods were performed using custom scripts and/or functions supported by MATLAB R2017B (Mathworks).
Freely moving isoflurane-exposed animals locomoted via sinusoidal undulation, primarily in a forward direction with intermittent intervals of backward movement (reversals; fig. 1A), as is typical of wild-type behavior. No differences were observed in the ability of nematodes to respond to aversive mechanical stimuli, indicating that their sensory response remained grossly intact after isoflurane exposure. An eyelash drawn across the anterior portion of the body was similarly likely to yield a reversal response in both isoflurane-exposed and control worms (% response: untreated early adulthood, 94 ± 11, vs. isoflurane-exposed early adulthood, 91 ± 13; untreated late adulthood, 65 ± 29, vs. isoflurane-exposed late adulthood, 59 ± 28; P = 0.287 and 0.248, respectively; fig. 1B). However, the rate at which the worms spontaneously initiated backward movement while locomoting in a food-free environment was reduced in isoflurane-exposed animals. This behavioral shift persisted in both early and late adulthood (reversals/min: untreated early adulthood, 1.14 ± 0.42, vs. isoflurane-exposed early adulthood, 0.83 ± 0.55; untreated late adulthood, 1.75 ± 0.64, vs. isoflurane-exposed late adulthood, 1.14 ± 0.68; P = 0.001 and 0.006, respectively; fig. 1C).
Single Neuron Imaging
Employing the genetically encoded calcium indicator GCaMP6s expressed in a select neuronal subset, we measured the spontaneous activity of the AVA interneuron in both control and isoflurane-exposed animals by measuring its GCaMP6s fluorescence intensity using a confocal microscope (fig. 2A). The AVA interneuron is a key component of the circuitry that mediates the animal’s crawling state, displaying little activity during forward crawling but becoming activated during reversals.17,18
The activity of the AVA neuron displayed clear and rapid transitions between two discrete states in both the control animals and those that had been exposed to isoflurane at the L1 larval stage. The system rapidly transitions between the forward-moving state, in which AVA is “off” and the GCaMP signal is low, and the reversal state, in which AVA is “on” and the GCaMP signal is high (illustrated in fig. 2B). We found no change in the relative time spent in either state (i.e., the duty ratio was unchanged, 0.46 ± 0.22 vs. 0.42 ± 0.19 for controls; P = 0.517), nor the duration in which it stayed in a particular state. The average forward (i.e., off) state duration was 39 ± 77 vs. 40 ± 70 s (P = 0.987), and the average reversal (i.e., on) state duration was 23 ± 15 vs. 23 ± 17 s; P = 0.733. This neuronal behavior is consistent with that reported for unanesthetized C. elegans, as compared with animals under isoflurane anesthesia.19 These results demonstrate that the exposed animals had grossly recovered from their anesthetic as expected.
To examine more closely the manner in which AVA transitions between activity states, average traces were calculated for on and off transitions. Individual AVA on transitions (fig. 2C), indicated by clear changes from low to high fluorescence signals, were identified within the measured GCaMP trace for each trial, overlaid, and averaged (see Statistical Methods). Likewise, AVA off transitions (fig. 2D) were averaged at points indicated by clear changes from high to low fluorescence signals. The resultant plots for isoflurane-treated and untreated animals are divergent, with AVA transitions between states more rapid in animals with isoflurane exposure. Moreover, these effects become more pronounced with age: a larger divergence in AVA transition is seen in late adulthood in the 9-day-old animals.
The changes in AVA transition rates were quantified by fitting hyperbolic tangent functions to each state transition observed in the individual trials and calculating the average rise times and fall times for on and off transitions, respectively (see Statistical Methods). In isoflurane-exposed animals, the average AVA rise time was found to be shorter. This effect persisted in both early (1.9 ± 1.3 vs. 2.5 ± 1.2 s for controls; P = 0.028) and late adulthood (3.0 ± 2.4 vs. 4.6 ± 3.0 s; P = 0.008), with the difference between control and isoflurane-exposed animals becoming larger and more statistically significant later in life (fig. 2E). Comparably, the average fall time was found to be shorter during late adulthood (10.7 ± 4.3 vs. 14.1 ± 4.4 s; P = 0.001), following a nonstatistically significant trend in the same direction observed during early adulthood (10.9 ± 3.4 vs. 12.5 ± 2.9 s; P = 0.137; fig. 2F).
In further analysis, the time differentials of the individual AVA on and off state transitions observed during late adulthood were calculated (see Statistical Methods). These traces were overlaid and averaged, and the resultant plots are again clearly divergent. Traces derived from isoflurane-exposed animals reach a peak amplitude that is greater in magnitude (in both positive and negative transition rates) and also reach this peak more quickly than unexposed animals (fig. 2G). These traces demonstrate a faster and earlier rate of change of intracellular calcium concentration, as measured by GCaMP fluorescence, in animals exposed to isoflurane at the L1 larval stage compared with controls.
To determine whether the postexposure effects on AVA state transition speed are grossly observed throughout the circuitry underlying locomotion, we employed dual inverted selective plane illumination microscopy to capture spontaneous GCaMP fluorescence and, by proxy, the neuron activity within 7 pairs of premotor command interneurons. These 14 neurons (including the anatomical left and right AVA) are centrally involved in locomotive and reversal behavior17,27 (fig. 3, A and B; see Light-sheet Imaging). Activity traces were extracted from all 14 neurons within each movie, their transitions between states (i.e., rise: off to on; and fall: on to off) were identified, and the average rise and fall times were calculated (see Statistical Methods). In total, 40 movies per condition yielded 560 neuron traces each, from which ~1,050 on events and ~450 off events were extracted. As was the case with AVA alone, the average rise time observed across these interneurons was shorter in isoflurane-exposed young adult animals (3.2 ± 2.0 vs. 3.4 ± 2.5 s for controls; P = 0.030), whereas the fall time was unchanged (12.9 ± 6.0 vs. 12.6 ± 5.6 s for controls; P = 0.326; fig. 3, C and D).
In recent behavioral genetics studies, Morgan, Sedensky, and colleagues10,29 identified molecular modulators of the chemotaxis deficits that follow developmental anesthetic exposure in C. elegans. Key among these are genes within the apoptosis cell death pathway, the daf-2–dependent stress-response pathway, and the mechanistic Target of Rapamycin (mTOR). Command interneuron state transition speeds were therefore captured and analyzed, as above, in animals harboring specific loss-of-function mutations within ced-3, the primary executioner caspase within the C. elegans apoptosis pathway; daf-2, the C. elegans insulin/insulin-like growth factor 1 receptor ortholog; or daf-16, the FoxO transcription factor (nuclear translocation of which is negatively regulated by daf-2).30 Worms that had been maintained on the mTOR inhibitor rapamycin were also examined (see Anesthesia Methods). The results described below are summarized in figure 4 and table 1. The 2 × 5 ANOVA for the difference in means (as described under Statistical Methods) was significant for both the rise and fall times (P < 0.001 and P = 0.009, respectively). In ced-3 animals, the postexposure effects on interneuron rise and fall times mirrored those observed in worms with a wild-type background. The average rise time was shorter (3.3 ± 2.2 vs. 3.6 ± 2.5 s for controls; P = 0.010), whereas the fall time was unchanged (12.2 ± 4.3 vs. 12.5 ± 4.4 s for controls; P = 0.408). These effects were also mirrored in daf-2 worms, which displayed a shorter rise time (3.1 ± 3.3 vs. 3.5 ± 3.1 s for controls; P = 0.030) and unchanged fall time (10.7 ± 5.7 vs. 10.5 ± 7.1 s for controls; P = 0.762). However, in daf-16 animals, although the fall time remained similarly unchanged in the isoflurane-treated group (12.0 ± 4.3 vs. 12.0 ± 6.2 s for controls; P = 0.966), the rise time, conversely, lengthened (3.6 ± 2.5 vs. 2.9 ± 2.2 s for controls; P < 0.001). The postexposure effects on both rise and fall times were also modified in animals that had been maintained on rapamycin. In this case, the Rise Time remained unchanged (3.8 ± 2.1 vs. 3.7 ± 2.6 s for controls; P = 0.401), whereas the fall time lengthened (15.8 ± 8.5 vs. 13.8 ± 7.2 s for controls; P = 0.002). In the absence of anesthetic exposure, statistically significant differences in command interneuron state transition speeds were also observed between genotypes. Notably, daf-2 animals displayed shorter fall times (10.5 ± 7.1 vs. 12.6 ± 5.6 s for wild-type; P < 0.001), whereas daf-16 animals displayed shorter Rise Times (2.9 ± 2.2 vs. 3.4 ± 2.5 s for wild type; P < 0.001). Rapamycin-treated animals displayed both lengthened rise times (3.7 ± 2.6 vs. 3.4 ± 2.5 s for wild type; P = 0.044) and fall times (13.8 ± 7.2 vs. 12.6 ± 5.6 s for wild type; P = 0.009).
This study demonstrates that exposure of C. elegans to isoflurane during the L1 larval development stage fundamentally alters interneuron dynamics and circuit function within the nervous system throughout the animal’s lifetime. We measured both the attenuation of the nematode’s spontaneous reversal behavior (fig. 1) and altered dynamics of the interneurons mediating this behavior (figs. 2 and 3).
Our findings indicate a neuronal correlate to the chemotaxis deficits previously described by Gentry et al.10 in C. elegans after developmental anesthetic exposure. Efficient nematode chemotaxis is dependent upon the stochastic nature of the animal’s locomotion. Spontaneous randomized changes in the animal’s direction of movement (pirouettes) facilitate its dispersion throughout its environment. Spontaneous reversals constitute a key element of this behavioral strategy by initiating these pirouettes.15,16 A reduction in spontaneous reversal frequency, as observed in isoflurane-exposed worms, would therefore be expected to hinder the animal’s ability to efficiently chemotax, providing a causative link with the findings of Gentry et al.10
Furthermore, this study provides insight into the effects of isoflurane exposure during neurodevelopment on neuronal function. The command interneurons that we measure here constitute the key neuronal circuitry modulating crawling behavior. This system is comprised of two groups of premotor interneurons that mediate forward (AVB and PVC) and backward (AVA, AVD, and AVE) crawling movement, and an antecedent group of command premotor neurons (AIB, RIM, and RIV) involved in navigation and orientation of the head.17,27 Although worms will alter movement direction in response to sensory stimuli, they also do so spontaneously. These intermittent changes in movement are facilitated by the command interneuron circuitry that creates spontaneous bimodal activity patterns, consisting of either forward or reverse crawling states.18 The behavioral deficits in anesthetic-exposed animals will ultimately depend on the activity within this simple well characterized command circuitry. Measuring activity of the AVA command interneuron, we observe a distinct increase in the neuronal transition rate from an off to an on state, which corresponds to a transition from the forward to backward crawling state (fig. 2, C and E). Expanding these measurements to additional command premotor neurons in the animal’s head (AVA, AVB, AVD, AVE, AIB, RIM, and RIV) and automating the activity analysis, we confirm an increase in the off to the on state transition rate in isoflurane-exposed animals across the command motor neuron circuitry (fig. 3). It is further remarkable that the effect of isoflurane exposure on neuronal activity is an increase in transition speed rather than the decrease that one might expect from generalized degradation. Regardless, these alterations in neuronal activity indicate a fundamental shift in the behavioral state dynamics of the system at large and are congruent with the disruption of crawling behavior observed here and previously.18
To gain insight into the cellular mechanisms underlying the alteration in neuronal activity, we measured command neuron dynamics in animals with abolished functionality in specific proteins previously implicated in anesthesia-induced neurotoxicity. CED-3 is the primary executioner caspase within the conserved C. elegans apoptotic pathway.31 Although CED-3 activity appears to play a role in the chemotaxis deficit of isoflurane-exposed animals,10 the ced-3 null mutation had no effect on the activity dynamics of the command interneurons measured here (fig. 4). Likewise, the null mutation of the C. elegans daf-2 gene, the homolog of the mammalian insulin/insulin-like growth factor 1 receptor, did not alter the postexposure effects on interneuron activity (fig. 4). This is in contrast to previous behavioral measurements in which daf-2 null mutants did not display changes in chemotaxis index upon developmental anesthetic exposure.29 Thus, apoptotic cell death as well as insulin signaling, both of which have previously been implicated in the neurotoxic effects of anesthetic, appear to be independent of the subtle alterations in neuronal function measured here. Reconciliation of this discrepancy likely lies in the relative complexity of C. elegans chemotaxis, the efficient undertaking of which is dependent upon various behaviors that contribute to locomotive stochasticity and also upon neurophysiological processes, such as chemosensation. Although apoptotic cell death and insulin signaling appear to be independent of the postexposure effects we observe in interneuron physiology, these pathways may yet be involved in the manifestation of other neurobehavioral effects that hinder chemotaxis.
Importantly, both a null mutation within the FoxO transcription factor daf-16 and pharmacologic inhibition of mTOR with rapamycin alter the neurologic effects of anesthetic exposure. The daf-16 null mutation facilitated a novel increase in interneuron rise time after isoflurane exposure, whereas blockage of mTOR activity both abolished the isoflurane-induced effects on interneuron rise time and further facilitated a novel increase in interneuron fall time (fig. 4). These findings are reminiscent of prior work demonstrating blockage of anesthesia-induced chemotaxis deficiency in worms via mTOR inhibition.29 A growing body of literature further describes mTOR-dependent anesthesia-induced effects in neurodevelopmentally immature rodents. Mintz and colleagues32,33 have demonstrated in mouse models that isoflurane anesthesia is capable of eliciting abnormal dendrite arbor and spine development, pre- and postsynaptic marker expression, and also deficiencies in spatial learning and memory. These effects were all blocked by rapamycin administration. mTOR inhibition via rapamycin has also been shown to block other postexposure effects elicited by sevoflurane administration in postnatal mice. Notably, enhanced excitatory synaptic transmission in males, characterized by an upregulation of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor subunit GluA2 and an increased mEPSC frequency, was found to be dependent upon mTOR activation.34 Such changes to neuronal ultrastructure and synaptic transmission may well account for the behavioral and neurophysiological effects described here.
Our findings are also consistent with the activation of neuronal stress-response pathways in response to anesthetic exposure, as supported by recent work by Morgan, Sedensky, and colleagues.29 Indeed, whereas daf-16 facilitates a protective endoplasmic reticulum stress-response pathway,35,36 mTOR has also been shown to interact with this same stress-response pathway and others.37,38 Our results therefore add to a mounting body of evidence suggesting that anesthetization during critical periods of nematode development may impact normal neurodevelopment through pathologic activation of stress-response pathways and subsequently alter neuronal activity. Although the molecular and cellular mechanisms underlying these observations require further investigation, our results demonstrate that alterations in C. elegans neuronal function and signaling measured here are consistent with histological findings and behavioral defects found in higher organisms in that they are likely dependent on neuronal stress-response pathways.39–41
In totality, this study demonstrates the following:
Exposure to isoflurane at a critical developmental stage results in a defect in the spontaneous reversal rate in the adult animal that is congruent with defects in chemotaxis behavior observed previously.10,29
There is a pathologic alteration in the transition dynamics of the AVA neuron, a key command interneuron controlling the animal’s forward and reverse crawling states. This effect is inducible by exposure to isoflurane and absent in controls.
These effects on AVA transition dynamics are broadly observed throughout the command interneuron circuitry centrally involved in defining and determining the forward and reverse behavioral crawling states of the animal.
The mTOR and FoxO transcription factor daf-16 modulate these effects on transition dynamics, consonant with prior behavioral studies.29
Thus, we have observed a permanent defect in the function of neurons controlling behavior that is induced by anesthetic exposure in a developing animal. Moreover, we have linked these defects to the activation of stress-response pathways also implicated in the histological and behavioral effects seen in higher organisms.38–40 Given the tractability of C. elegans for functional neuronal imaging, its entirely mapped connectome, and suitability for genetic manipulation, this simple but powerful model system can serve as a platform for interrogating these molecular pathways underlying anesthesia-induced dysfunction of neuronal circuits after exposure early in life.
Supported by National Institutes of Health (Bethesda, Maryland) grant No. R01 GM121457 and by departmental sources.
Dr. Connor is a consultant for Teleflex, LLC (Wayne, Pennsylvania) on airway equipment design. This activity is unrelated to the material in this manuscript. The other authors declare no competing interests.