Anesthesia is used in many neuroscience studies. However, its effect on brain network properties is not fully understood. Previous studies reported that increasing levels of anesthesia can induce neural suppression or loss of neural responsiveness to sensory stimulation^{1, 2}, yet these studies were mainly focused on a single sensory modality. Crossmodal responses to sensory stimulation, which can be important for the brain to form a collective representation of the environment and facilitate behavioral responses, have been commonly observed in awake subjects³. However, how anesthesia modulates crossmodal responses remains largely unknown. In this study, we investigated the brain responses to auditory stimulation at different isoflurane levels using large-view BOLD fMRI. Crossmodal responses to pure tone stimuli were observed in the visual cortex (VC) at a high isoflurane level.

Animal preparation Adult male SD rats (n=5, ~300g) were anesthetized using isoflurane during fMRI experiment. For each animal, the experiment was first performed at 1.0% and subsequently at 2.5% isoflurane. Note that after increasing the isoflurane level, the animals were allowed to rest for 15min in order for the physiological conditions to stabilize.

Auditory stimulation Auditory stimulation was generated by a magnetic speaker and delivered through custom-made tubes into the right ear of animals. A block-design paradigm (20s on and 40s off, 4 blocks, see Figure 1) was used to present pure tone (1.3, 7 and 20kHz) stimulation to the animals. Sound pressure level at the end of the tube was ~100dB.

fMRI data acquisition and analysis All fMRI data was acquired on a 7T Bruker scanner using GE-EPI (FOV=32×32mm², matrix=64×64, α=56°, TE/TR=20/1000ms, sixteen 1.0 mm slices without gap). Data were first realigned, co-registered, in-plane smoothed and high-pass filtered before the standard GLM analysis was applied to identify significant BOLD responses.

Figures 2 presents the BOLD activation maps for 1.3kHz and 7kHz sound stimulation from the average of five animals. At 1.0% isoflurane, responses to both stimuli were observed within major brainstem auditory nuclei, i.e. the inferior colliculus (IC), lateral lemniscus (LL) and superior olivary complex (SOC). No BOLD response was observed in the VC. In contrast, at 2.5% isoflurane, while the IC, LL and SOC BOLD responses to both stimuli clearly decreased, strong BOLD responses were observed in the bilateral VC.

Figure 3 shows the BOLD signal profiles in the IC and VC at different isoflurane levels. The IC BOLD responses decreased when isoflurane was increased, while the VC BOLD responses were only observable at 2.5% isoflurane. More interestingly, clear differences were seen between the BOLD signal profiles in these two structures. In both structures, the BOLD signal reached the first peak 5-7s after the onset of stimulation. However, after this peak, the IC BOLD signal exhibited an increasing trend throughout the period of stimulation, while the VC BOLD signal didn't.

Figure 4 shows the BOLD responses to 20kHz sound stimulation from a single animal. Similar to 1.3kHz and 7kHz stimulation, IC and LL responses decreased when isoflurane was increased from 1.0% to 2.5%, while bilateral VC responses were only observed at 2.5% isoflurane.

Our fMRI results showed that within the auditory system, responses to pure tone stimulation decreased at higher isoflurane level, as expected from previous studies^{1, 2}. Surprisingly, crossmodal responses to the auditory stimulation in the VC, were absent at 1.0% but robustly observed at 2.5% isoflurane. Despite the close proximity of the IC and VC, the BOLD signal profiles in these structures exhibited different patterns during the 20s period of stimulation (Figure 3). This indicated that the VC responses were not caused by partial-volume effects from the IC. Furthermore, our fMRI results were supported by a recent electrophysiology study, which similarly found VC responses to 1.3kHz pure tone stimulation at 2.5% isoflurane using local field potential (LFP) recordings⁴. Compared to LFP, the large-view fMRI data here measured the signals in bilateral VC under a broad frequency range of sound stimulation (1.3-20kHz). These results demonstrated that anesthesia could broadly and profoundly modulate the brain crossmodal response properties. While the origin of the auditory evoked VC responses remains unclear, the present results could be associated with several lines of studies in literature. For example, they may be related to studies on the anatomical connections between auditory regions (e.g. the IC and the auditory cortex) and the VC^5-7. Such connections might serve as the neural substrates of the auditory evoked VC responses. The present results may also be related to recent resting-state fMRI studies showing that the spontaneous BOLD signal fluctuations during deep anesthesia are highly synchronized across large-scale cortical regions⁸.