Introduction

Mouse fMRI has been increasingly interested in the MRI community due to abundant transgenic models and available animal-dedicated MRI scanners and has been successfully conducted under anesthesia^1,2,3. In our fMRI study (ISMRM 2019, program #362), ketamine/xylazine mixture (glutamate receptor antagonist) was successfully adopted for visual stimulation. However, anesthetic agents affect neural activity and hemodynamic responses, consequently, the choice of anesthesia will make a great impact on fMRI responses^1,2. Therefore, it’s ideal to perform experiments under an awake condition. Our specific goals are i) to develop an awake mouse fMRI protocol, and ii) to compare fMRI responses obtained during wakefulness vs. ketamine/xylazine anesthesia.

Methods

Animal Preparation

Fourteen male C57BL/6N mice (8-10 weeks old) were used with approval; seven for awake fMRI, two for corticosterone measurements during habituation and five for anesthetized fMRI. MR-compatible head fixation and body restraint apparatus were designed for minimizing head and body motions. The head fixation (Fig. 1) was bonded to the dorsal skull after skin removal under ketamine/xylazine anesthesia. After 7-day recovery, mice were habituated for 10 days: 7 days in the mock scanner with 110–120dB EPI sound and 3 remaining days in the 9.4T scanner. Seven mice were performed for 2-hour fMRI scanning in wakefulness. In another two mice, 40µl tail vein blood was collected from each mouse at 4 days before, immediately after the 1^st day, 5t^h day and 10^th day of habituation. For anesthetized fMRI, ketamine/xylazine³ was used.

Functional Imaging Data Acquisition

fMRI was conducted on 9.4T/30cm Bruker scanner using single-shot GE-EPI sequence with TR/TE=1000/20ms, flip-angle=50°, spatial resolution=156×156×500μm³, and 9 coronal contiguous slices without gap. For visual stimulation, two 0.5mm diameter optic fibers were placed bilaterally 2cm away from both eyes of the animal, and connected to white cold LED driver (Thorlabs DC2000). Stimulus parameters were illuminance of ~10lux, pulse duration of 10ms, and two different frequencies of 5Hz and 10Hz. Each fMRI trial consisted of 30s offs–10s on–40s off–10s on–30s off (Fig. 1). Six to eight trials were obtained for each frequency.

Data Analysis

Data were processed with Matlab and AFNI. Frame-wise displacement (FD) was calculated to evaluate animal motions before preprocessing. If the FD value exceeded the threshold of 1 voxel size or changed synchronously with the stimulus period, the trial was excluded from further analysis. The preprocessing included slice timing correction, motion correction, temporal detrending, temporal normalization from baseline and trial averaging. Then data were normalized to an EPI template, standard GLM analysis was applied to identify significant BOLD responses. Regions of interest (ROI) were defined on EPI images based on Allen Mouse Brain Atlas (Fig. 1), which were the dorsal lateral geniculate nucleus (LGd), superior colliculus (SCs), lateral posterior nucleus of the thalamus (LP), primary visual cortex (V1), and higher-order visual area (V2). The group activation maps were generated using a one-sample t-test with 0.2mm FWHM Gaussian kernel spatial smoothing (p<0.05; FWE corrected).

Results

Since awake fMRI induces stress, habituation is critical for reducing stress. Corticosterone levels were ~10ng/ml before started, 30ng/ml on the 5th day and 15ng/ml after the 10th day of habituation (n=2), suggesting the stress level is reduced by habituation, but still modest. Among seven mice, one was excluded because of large motion. 77% of trials (65/84 total trials in 6 animals) passed the motion criteria. Fig. 2A shows the group-level BOLD-fMRI activation maps of 5Hz stimulus for awake and anesthetized mice. Most ROIs (Fig. 1) were robustly activated in both conditions except V1. Time courses from ROIs show detailed temporal characteristics (Fig. 2B). LGd, LP, and SCs for both conditions had similar responses. Unlike the anesthetized, the response in V1 for the wakefulness increased sharply, followed by a quick decay, leading to negative BOLD change during the stimulation period. When stimulus frequency increased to 10Hz (Fig. 2C), activities in awake mice were increased in subcortical areas compared to the anesthetized, but fully suppressed in cortical areas. This observation in V1 is quite unexpected. BOLD time courses of all ROIs were normalized to compare dynamic response between two conditions (Fig. 3). The peak response of awake mice was apparently 1–3s earlier in all regions compared to the anesthetized (Fig. 3).

Discussion & Conclusion

We have successfully developed an awake fMRI protocol by designing the restraint apparatus and habituation steps. For both awake and anesthetized conditions, BOLD-fMRI activities were observed in all visual pathway-related areas. The most interesting observation is V1 activity; under anesthesia, the BOLD activity is prolonged and spreads a larger area, while the awake BOLD response is biphasic/negative. This observation can be explained by electrophysical studies^4,5. During wakefulness, inhibition is much stronger than excitation and has extremely broad spatial selectivity, resulting in a brief and spatially selective BOLD response. Under the wakefulness, the temporal frequency tuning of LGd, LP and SCs were shifted toward higher frequency, while V1 had no significant differences. Since our fMRI observations concur with the published electrophysiology data^4,5,6,7,8, our awake fMRI system is well-acceptable for neuroscience research. Based on this preliminary result, further studies will be conducted with the awake mouse model for filling the gap between macroscopic fMRI and microscopic electrophysiology studies.

Acknowledgements

This work was supported by IBS-R015-D1.