Dose-related adaptive reconstruction of DMN in isoflurane administration: a study in the rat

Twenty adult male Sprague–Dawley rats (300-350g; Chengdu Dossy Experimental Animals Co., Ltd) were utilized following the National Institutes of Health guidelines, and experimental protocols were approved by the Ethical Committee on Animal Experimentation of the University of Electronic Science and Technology of China (UESTC).

Animals were housed five per cage in temperature-controlled IVC cages (Suzhou Fengshi Laboratory Equipment Co., Ltd) with food and water ad libitum before surgery. Animals were implanted with chronic homemade electrodes (insulated nichrome probes for eight deep brain regions, diameter: 200 m\(\mu\), resistance: 70–700 \(k\Omega\); stainless-steel screws for eight cortical brain regions, diameter: 500 m\(\mu\), resistance: 2–40 \(k\Omega\)) in DMN areas reported in an fMRI study [16] and a reference electrode (stainless-steel screw; diameter: 500 \(\mu m\), resistance: 2–40 \(k\Omega\)) in the cerebellum. During surgery, the animals were anesthetized with 1.5% isoflurane (4% for induction and 1.5–2% for maintenance, RWD Life Science) in the air (R540 anesthesia machine, RWD Life Science), positioned in a stereotaxic frame (68,028, RWD Life Science) using blunt ear bars, and maintained at a body temperature of 37 °C (RWD-69020, RWD Life Science). All stereotactic coordinates for each electrode were relative to the bregma and are shown in Table 1. After implantation, all the electrodes were assembled into the connectors and cemented to the skull of the rat with dental acrylic. After the surgery, all animals were fed in a clean cage while individually housed to recover for at least two weeks before local field potential (LFP) recording.

After recovery from surgery, each rat was habituated for at least 3–5 days before the experiments. For baseline data acquisition, LFPs in the free-moving state were obtained for 30 min. Thereafter, before anesthetic data recording, the animals were rapidly anesthetized with 4% isoflurane, air flow rate (0.8 L/min), and moved into a stereotaxic frame, and then isoflurane was discontinued. When animals’ senses and movement were restored, isoflurane was immediately switched to 0.75%. Animals’ body temperature was controlled and maintained. Three levels of isoflurane doses were tested, including 0.75%, 1.25%, and 1.75% isoflurane, in each animal, and spontaneous LFPs were monitored for 10 min after an equilibration time of 20 min during each dose, the data recording procedure and original data were partly shown in S1. The LFP signals were obtained by a Symtop amplifier device (UEA-FZ 41; Symtop Instrument Co., Ltd, Beijing, China) and custom-made LFP recording software. The signals were sampled at 1000 Hz and bandpass filtered for 0.1–30 Hz with a notch filter to reject the 50 Hz line noise.

We analyzed the inherent oscillatory properties of the DMN associated with QA and three-step increased isoflurane-dose states, e.g., 0.75%, 1.25%, and 1.75% isoflurane states (Fig. 1a). The spectrogram of DMN showed distinct changes as the rats received increasing doses of isoflurane. The LFP activity in the four frequency bands was increased during the lowest dose of isoflurane state, while the activity was pull-back with the increment of doses (Fig. 1b).

To find out the relative variation of each frequency band, we measured the relative power spectral densities (RP) of LFP in four frequency bands (delta: 1–4 Hz; theta: 4–8 Hz; alpha:8–13 Hz; beta: 13–30 Hz). The results of RP within DMN suggested that the variation tendency of the delta band was different from the other three higher-frequency bands. More specifically, the delta band activity was intensive under the lowest dose of isoflurane state (Fig. 1c, delta: QA vs. 0.75%, p < 0.01, one-way ANOVA with Bonferroni’s post hoc test, n = 20), but with the doses of isoflurane gain it was decreased (Fig. 1c, delta: 0.75% vs. 1.25% and 1.25% vs. 1.75%: p < 0.01, one-way ANOVA with Bonferroni’s post hoc test, n = 20). For theta, alpha, and beta frequency bands, the RP of DMN was decreased in 0.75% isoflurane-dose state, which seemed that the activity of the DMN was mainly in the delta frequency band. With the isoflurane gaining, alpha and beta activity were revered to be intense (alpha: QA vs. 0.75%, beta: QA vs. 0.75%: p < 0.01, one-way ANOVA with Bonferroni’s post hoc test, n = 20). On the contrary, RR, RFS, M, and BSR of the rats under different isoflurane doses showed that the vital sign, capability of response, and nervous system was suppressed by isoflurane (Stable 2).

We evaluated the topologic structure in 1–30 Hz and individual frequency bands (delta, theta, alpha, beta) to explore FC among DMN regions under different isoflurane-dose states. For this analysis, the DMN was a modularized organization (displayed in black boxes in Fig. 2), and the strong FC was found in the anterior part (prelimbic cortex (PrL), orbitofrontal cortex (OFC), and cingulate cortex (CG)), bilateral hippocampus (Hip), and the posterior part (secondary visual cortex (V2), temporal association cortex (TeA), and retrosplenial cortex (RSC)), which were common existed in the QA and three isoflurane-dose states in all frequency bands (Fig. 2a-d, S3). Under different isoflurane-dose states, DMN topological structure was adaptively reconstructed and displayed more modularized and segregated, especially in the posterior part of the DMN (Fig. 2b-d, S3).

Additionally, Fig. 2e shows the overall FCS of the DMN, which suggested that the FCS was increased under the lowest dose of isoflurane (0.75% vs QA: p < 0.01. One-way ANOVA with Bonferroni's post hoc test, n = 20). When the dose of isoflurane increased to a higher level, the FCS of the DMN decreased (Fig. 2e, 1.75% vs QA: p > 0.05. One-way ANOVA with Bonferroni's post hoc test, n = 20). However, the changes in FCS in different frequency bands were shown to be different with stepwise isoflurane doses. Figure S4 demonstrates that in the lowest isoflurane dose, the FCS of the DMN was increased in the delta band (Figure S4a delta: 0.75% vs QA: p < 0.05, one-way ANOVA with Bonferroni's post hoc test, n = 20), but decreased in higher frequency bands (Figure S4b-d, theta: 0.75% vs QA: p < 0.05, alpha: 0.75% vs QA: p < 0.05, and beta: 0.75% vs QA: p < 0.05. One-way ANOVA with Bonferroni's post hoc test, n = 20). As the isoflurane doses increased, the FCS decreased in the delta band (Figure S4a) and increased in higher frequency bands (Figure S4b-d).

For the QA state, the FC of long-range connectivity between anterior and posterior areas was more stable within 1–30 Hz (Fig. 3a), while the local connectivity was flexible (Fig. 3f). These stable and flexible patterns were broken under the 0.75% isoflurane dose, and adaptively built up with the increase of the isoflurane dose. Specifically, the stable patterns were more local during the 0.75% isoflurane dose, and tended to rebuild the long-range stable connection in 1.25% and 1.75% isoflurane-dose states (Fig. 3b-d). Furthermore, the long-range connections were performed predominantly with flexible patterns under the 0.75% isoflurane dose, and flexible local connections were rebuilt in 1.25% and 1.75% isoflurane doses (Fig. 3g-i). In addition, the FCS of the smallest fluctuation of the DMN showed that the FCS was decreased in the 0.75% isoflurane dose state, and increased in higher isoflurane doses (Fig. 3e, QA vs 0.75%, 1.75% vs 0.75%, 1.25% vs 1.75%, p < 0.05, One-way ANOVA with Bonferroni's post hoc test, n = 20). However, the FCS of the largest fluctuation of the DMN displayed the opposite trend of variations with isoflurane doses (Fig. 3j, QA vs 0.75%, 1.75% vs 0.75%, 1.25% vs 1.75%, p < 0.05, One-way ANOVA with Bonferroni's post hoc test, n = 20).

Along with 1–30 Hz, the effect of isoflurane doses on the time-series variation of network connection existed similar tendency in delta bands (Figure S4a, S5a, S6a, S7a). Furthermore, the local connections and long-range connections in theta, alpha, and beta bands were respectively stable and flexible in different doses of isoflurane, which were out of ‘broken up’ and ‘rebuilt up’ (Figure S4b-d, S5b-d, S7b-d, S7f-h).

Consisting with the FCS, the capability of information processing was increased, which was reflected by the increased Cluster (Fig. 4b, Cluster: 0.75% vs QA. p < 0.01. One-way ANOVA with Bonferroni’s post hoc test, n = 20). In the highest dose of isoflurane, the Cluster was decreased (Fig. 4a, Cluster: 1.75% vs 0.75%, 1.25% vs 0.75%: p < 0.01. One-way ANOVA with Bonferroni’s post hoc test, n = 20). However, the DMN segregation was aggravated with the increase of isoflurane doses (Fig. 4f, Mod: 1.25% vs 0.75%, 1.75% vs 0.75%, 1.25% vs 1.75%. p < 0.05. One-way ANOVA with Bonferroni’s post hoc test, n = 20).

As for the topological features within DMN shown in Fig. 4, the variation trend of the DMN structure in the delta band is in line with which in 1–30 Hz (Fig. 4b-e), but the trend of variations in the theta, alpha, and beta bands was opposite under isoflurane-dose states (Fig. 5g-j).

To assess the imposition of stepwise increased isoflurane doses on the DMN in different frequency bands, we entered PSD, FCS, Mod, Cluster, and FCS with the 20% smallest and 20% largest fluctuation FC into a linear regression model (Stable 3). The results revealed that all these parameters were significantly covaried with isoflurane doses in the theta and beta band. For the 1–30 Hz, delta, and alpha bands, some of the six parameters were also found to vary with isoflurane doses (In detail, for 1–30: FCS, FCS with the 20% smallest FC, Mod, and Cluster; for delta: FCS with the 20% smallest FC and Mod; for alpha: FCS with the 20% smallest and largest FC, PSD, and Mod).

For a healthy brain, topological features with a large Cluster and low Mod to maintain the balance between global integration and local segregation [8]. Disruption of topological features balance could be found in the disorder of consciousness, and anesthesia [27]. In this study, the increase of Cluster could be found in the 0.75% isoflurane-dose state (Fig. 4). With the isoflurane dose increased, this parameter was reversed to decrease and increase, independently. According to ‘the cognitive unbinding theory [36]’, our results hinted that the FC within DMN was stronger, and the capability of the information process was higher than QA in the lowest isoflurane dose. However, this ‘stronger’ and ‘higher’ does not mean the level of vigilance is higher than QA, which might reflect the isoflurane pushes the DMN into a state different from the resting state. In this state, the brain is still positively processing information even though awareness was fading away. With the increase of isoflurane dose, the topology with high Cluster was broken down, and the Mod in DMN was increased in isoflurane states (Fig. 4), which indicated that DMN has a lower integrative capacity and cannot maintain functional specialization.

The respiratory rate (RR) and burst-suppression ratio (BSR) (Stable 2), were independently decreased and increased in a dose-related fashion. Consistent with studies about patients [37‐40], rodents [41‐43], and other species [44‐46], indicating that there were deeper physical and cortical suppression changes associated with increasing doses of isoflurane.

In this study, the ‘static’ functional connectivity matrices showed that the DMN in rats under different states (QA and isoflurane-dose states) could be organized in subnetworks, which were identified anterior part, bilateral hippocampus (Hip), and the posterior part by the strength of FC between the brain regions of the DMN. These substructures of DMN generally could be found in rats in sleep [23] and in epilepsy [47]. While the modularized organization is reconstructed in spatial and temporal domains reflected by FC within and among subnetworks under isoflurane-dose states.

Spatially, the degree of modularization and segregation of FC in the anterior and posterior regions were getting higher during the isoflurane states. Even though the Hip-anterior and Hip-posterior FC was irresolute under different states, the anterior–posterior interaction was decreased under isoflurane-dose states. In the temporal domain described by fuzzy entropy, adaptive reconstruction was embodied in anterior–posterior stable connections changed to be more flexible as well as rebuilt along with isoflurane dose increase (Fig. 3b-c, Fig. 3f-g). Our results might suggest that under isoflurane-dose states, network integration and long-range information interaction were disrupted by adaptively reconstructing in DMN. Since the 'cognitive unbinding theory [36]’ proposed that the anesthetic effects disrupt the ability of effective information transformation across the cortex, our results supported this theory at least in part with DMN.

Surprisingly, along with the increase of isoflurane dose, adaptive reconstruction in time series was the paradoxical rebuilding of long-range stable interactions. We speculate that this ‘rebuilding’ might relate to the hyperexcitability in the burst-suppression pattern induced by a high dose of anesthetic [14], not relevant to the functional resumption. However, the trends of ‘rebuilding’ were covaried with isoflurane doses. Additionally combined with FCS and topological features under different isoflurane-dose states, these parameters reflect the properties of DMN and were also covaried with anesthetic doses (Fig. 4). These results might suggest that anesthetics might disrupt network optimization and adapt to other optimal constructions in the effect of different isoflurane doses to maintain the brain's intrinsic organization which is expected to prepare for arousal from anesthesia.

The Bispectral Index (BIS) and M-entropy, which have been widely used in clinical settings to monitor the depth of anesthesia in recent decades, their capabilities in predicting emergence from anesthesia, detecting intraoperative awareness, and improving early and intermediate-term survival are limited [6, 48‐51]. Our study suggests that the effect of isoflurane on vigilance may be reflected in various aspects of the brain network, such as the spectrum of different frequency bands, network connections and properties, and time-series dynamics of network connections. The results of simple linear-modeling regression demonstrated the Mod was significantly affected by isoflurane doses in several frequency bands. The parameters, FCS, 20% smallest FCS, 20% largest FCS, PSD, and Cluster, could be detected the significant impose of isoflurane doses in theta, alpha, and beta bands (Stable 3). This discovery could potentially address the limitations of current monitoring methods. Furthermore, a previous study on macaques using resting-state fMRI showed dose-dependent alterations in average connectivity and topological features [52]. Our article demonstrated similar alterations in network features in the electrophysiological DMN which could be detected more easily than fMRI. These dose-dependent alterations in parameters may serve as potential indicators of the level of anesthesia and offer new insights into anesthesia monitoring at the brain network level.