Synopsis

Introduction: The stringent motion requirement coupled with relatively long experimental time required for an imaging session imposes special challenges for rodent fMRI. Although it is possible to minimize stress induced by the imaging environment and to immobilize a rodent for a limited period of time after substantial behavioral training (1, 2), most rodent MRI studies have been carried out under anesthesia. In neuroimaging of disease models, it is often desirable to image the same animals across multiple time points, and to monitor disease progression and treatment effect longitudinally. Therefore, it is of critical importance to develop an fMRI protocol that maintains neurovascular coupling, permits longitudinal experiments with minimal invasiveness, and is easy to implement. Recently, using an anesthetic regime that combines low doses of dexmedetomidine and isoflurane, we have successfully identified the default mode network in rat brain, suggesting that this preparation causes minimal suppression of brain network functions (3). The goal of this study is to systematically characterize and to optimize physiological conditions for fMRI experiments under this anesthetic regime.

Materials and methods: A total of 26 male SD rats were used in this study. The fMRI experiment followed a protocol in (3). Briefly, rats were initially anesthetized with 2% isoflurane followed by a loading dose of dexmedetomidine (0.015 mg/kg, i.p.). Continuous subcutaneous dexmedetomidine was initiated (0.015 mg/kg/hr). Isoflurane was gradually tapered to 0.5%. Respiration rate, cardiac rate and oxygenation level were non-invasively monitored (Model 1030, SA instruments). Task-evoked fMRI using an electrical forepaw stimulation model (N=7) and resting state MRI scans (N=9) were both applied to empirically evaluate the functional state of the animals. The stimulation paradigm: 3 cycles of 20 sec ON and 20 sec OFF, plus 20 sec pre-stimulus baseline. Scan parameters: Bruker 9.4T scanner, TR/TE=1000/15 ms, FOV=32 mm, matrix size = 64×64. Seed-based correlation analysis was applied to the resting state data. Blood gas analysis experiment was perform on bench that followed the identical protocol above except that a femoral artery was catheterized for blood gas measurement (GEM Premier 3000).

Since the purpose of the somatosensory stimulation experiment was to evaluate the fMRI signal consistency across the 3.5-hour experimental period, we calculated fractional signal changes between the ON and OFF periods and parsed the experimental duration into 3 time windows following dexmedetomidine induction: I: 0 to 30 min; II: 30-90 min; III: 90-210 min. fMRI response and resting state fMRI scan data are analyzed across these 3 windows. Blood gas data were analyzed using linear mixed-effects modeling (R package) with random slopes and intercepts for individual animals. The Kenward-Roger approach was used to estimate the degrees of freedom.

Results: Both BOLD response and resting state functional connectivity showed time-dependence, and were sensitive to the physiological state of the animals. BOLD response in time window I was significantly lower than windows II and III. (Fig. 1). Complex connectivity patterns emerged during windows III when physiological state reached a stable optimal condition (Fig. 2).

Discussion: These data suggest that physiological parameters reached optimal condition 90 min post dexmedetomidine initiation. Both evoked BOLD response and resting state fMRI signal were stable during this time window. Somewhat surprisingly, even with respiration rate in the range of 85 BPM, arterial PCO2 was about 45 mmHg, indicating that the animals were still under a slightly hypercapnic condition.

Acknowledgements

References

1) King, J.A. et al. J Neurosci Method 2005; 148:154-160.

2) Martin C. et al., J Neurosci Method 2002; 120:25-34.

3) Lu H et al., PNAS 2012; 109: 3979-3984.