NeuroImage 86 (2014) 417–424

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Detection of functional connectivity in the resting mouse brain☆ Fatima A. Nasrallah a, Hui-Chien Tay a, Kai-Hsiang Chuang a,b,c,⁎ a b c

Magnetic Resonance Imaging Group, Singapore Bioimaging Consortium, Agency for Science Technology and Research, Singapore Clinical Imaging Research Centre, National University of Singapore, Singapore Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore

a r t i c l e

i n f o

Article history: Accepted 10 October 2013 Available online 22 October 2013 Keywords: Functional MRI Medetomidine Functional connectivity Resting state BOLD Translation

a b s t r a c t Resting-state functional connectivity, manifested as spontaneous synchronous activity in the brain, has been detected by functional MRI (fMRI) across species such as humans, monkeys, and rats. Yet, most networks, especially the classical bilateral connectivity between hemispheres, have not been reliably found in the mouse brain. This could be due to anesthetic effects on neural activity and difficulty in maintaining proper physiology and neurovascular coupling in anesthetized mouse. For example, α2 adrenoceptor agonist, medetomidine, is a sedative for longitudinal mouse fMRI. However, the higher dosage needed compared to rats may suppress the functional synchrony and lead to unilateral connectivity. In this study, we investigated the influence of medetomidine dosage on neural activation and resting-state networks in mouse brain. We show that mouse can be stabilized with dosage as low as 0.1 mg/kg/h. The stimulation-induced somatosensory activation was unchanged when medetomidine was increased from 0.1 to 6 and 10 folds. Especially, robust bilateral connectivity can be observed in the primary, secondary somatosensory and visual cortices, as well as the hippocampus, caudate putamen, and thalamus at low dose of medetomidine. Significant suppression of inter-hemispheric correlation was seen in the thalamus, where the receptor density is high, under 0.6 mg/kg/h, and in all regions except the caudate, where the receptor density is low, under 1.0 mg/kg/h. Furthermore, in mice whose activation was weaker or took longer time to detect, the bilateral connectivity was lower. This demonstrates that, with proper sedation and conservation of neurovascular coupling, similar bilateral networks like other species can be detected in the mouse brain. © 2013 Elsevier Inc. All rights reserved.

Introduction Resting-state functional connectivity MRI has emerged as a method for mapping intrinsic brain networks. These networks are based on coherent brain activities that are mostly detected by the blood oxygenation level dependent (BOLD) (Biswal et al., 1995) or perfusion (Chuang et al., 2008) functional MRI (fMRI). Similar and consistent brain networks, especially the bilateral connectivity in sensory and motor related areas, have been identified across species, from humans, monkeys to rats (Fox and Richle, 2007) (R. Hutchison et al., 2010; Vincent et al., 2007). As many transgenic models of diseases are available in mouse, it is enticing to be able to apply the same technique in the mouse brain. Recently, resting-state networks have been reported in mouse using fMRI (Guilfoyle et al., 2013; Jonckers et al., 2011) and optical imaging (Bero et al., 2012) (White et al., 2011). However, despite the structural and anatomical similarities between the mouse and the

☆ Part of the results has been presented in the Annual Meeting of International Society for Magnetic Resonance in Medicine, Salt Lake City, United States, 2013. ⁎ Corresponding author at: Singapore Bioimaging Consortium, 11 Biopolis Way, #02-02, Singapore 138667. fax: +65 64789957. E-mail address: [email protected] (K.-H. Chuang). 1053-8119/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.neuroimage.2013.10.025

rat brain, only unilateral connectivity in major cortical and subcortical areas was detected by fMRI (Guilfoyle et al., 2013; Jonckers et al., 2011). Functional MRI studies in the mouse brain have been challenging due to various technical and physiological issues. It suffers from lower detection sensitivity and more severe susceptibility-induced image distortions and signal losses from larger air-tissue interface in the smaller brain. Strategies such as using cryo-probe (Baltes et al., 2011) or susceptibility matching material (Adamczak et al., 2010) have been explored to overcome these challenges. Since resting-state BOLD signal is only a fraction of the activated signal, the lack of bilateral connectivity in the mouse brain may be due to limited sensitivity. Besides the technical issues, the need for proper maintenance of physiological conditions to preserve neurovascular coupling is more critical and largely dependent on the choice of anesthetics. So far, a few anesthesias have been demonstrated to allow robust BOLD activation to be measured in the mouse brain. Somatosensory BOLD activations using forepaw or hindpaw stimulation has been reported in isoflurane anesthetized mice with either free-breathing (Nair and Duong, 2004) or mechanical ventilation (Baltes et al., 2011), and in medetomidine sedated mice (Adamczak et al., 2010). In addition, anesthetics have been shown to have different impact on the restingstate networks. For example, we have shown that medetomidine can disrupt synchrony in the brain at high dose (Nasrallah et al., 2012).

418

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

The lack of bilateral connectivity in the mouse brain may be due to the medetomidine dosage used (Jonckers et al., 2011). Another study using high isoflurane level of 1.5% reported functional connectivity in areas related to the default mode network, but no other networks (Guilfoyle et al., 2013). Since isoflurane could cause bursting activity and neural suppression (Liu et al., 2011) and has variable effects on neurovascular coupling (Masamoto et al., 2009), it is crucial to ensure that proper anesthetic level is used and proper neurovascular relationship is maintained. In this study, we evaluated the influence of medetomidine dosage on neural activation and resting-state functional connectivity in the mouse brain using BOLD fMRI. Similar to what was found in the rat, somatosensory activation was not affected by medetomidine and bilateral connectivity can be detected in all the major brain areas at low dose of medetomidine. The dosage dependent suppression of bilateral connectivity was seen in the very high dose, which indicates different pharmacodynamics in mice compared to rats.

Materials and methods Experimental design To investigate the effect of medetomidine on functional connectivity in mice, resting state BOLD was assessed at two time points — 30 and 120 min after bolus injection of medetomidine – and under three dosages of medetomidine sedation – 0.1, 0.6, or 1 mg/kg/h (Fig. 1a). After 2 h of experiment, all mice were very lethargic and therefore atipamezole, an α2 adrenoceptor antagonist, was required for reversal of the effects. For such reason, the dosage effects were investigated in separate sets of animals (n = 11 for 0.1, n = 7 for 0.6, and n = 7 for 1.0 mg/kg/h dosage group) to avoid cumulative effects of the drug. Mice belonging to the 1.0 mg/kg/h dosage group were euthanized after the scan due to the poor recovery even after being antagonized. To evaluate the hemodynamic response under medetomidine, in between the two resting-state measures, somatosensory activation was studied using electrical stimulations of different currents – 0.5, 0.75, and 1 mA – delivered to the forepaw in different runs in a pseudo-randomized manner among animals. The stimulation was applied to either the left or the right forepaw using a pair of needle electrodes inserted under the skin between digits 2 and 4 and connected to a constant current stimulator (Isostim A320, World Precision Instruments, USA). The stimulus was given by a block design paradigm of 40 s resting and 20 s stimulation alternately repeated for

four times and adding 60 s of resting at the end (5 min total time). Another 5 min resting was allowed in between the runs. A pilot study was conducted to determine the optimal stimulation frequency by varying from 3, 6, to 9 Hz in a pseudo-randomized fashion in different fMRI runs with a pulse width of 0.3 ms and 1 mA. It was observed that the 6 Hz stimulus led to the strongest activation in the primary somatosensory (SI) region (data not shown), and was therefore used in the rest of the experiments. Besides, to rule out the effect of stimulation on the resting state network acquired at 120 min, restingstate fMRI was acquired at the same two time points but with no electrical stimulation applied in between in another sets of animals (n = 3 for the 0.1 and 0.6 dosage groups only). Animal preparation Animal study was approved by the local Institutional Animal Care and Use Committee (A*STAR, Singapore). C57BL/6 (n = 37 in total; 22 ± 2 g) female mice were used. Before anesthetizing the animals, they were carefully handled to minimize the stress and then quickly anesthetized with isoflurane (3% for induction) in a mixture of air and O2 gases (40% O2) and maintained at 2–3% during preparation through a nosecone with spontaneous respiration throughout the entire experiment. The animal was then secured on a MRI-compatible cradle (Rapid Biomedical GmbH, Germany) with ear bars and a bite bar to prevent head motion. A bolus of 0.3 mg/kg medetomidine (Dormitor®, Pfizer, USA) was injected intraperitoneally into the animal. At 15 min following bolus injection, continuous infusion of medetomidine was started through a PE50 catheter, inserted intraperitoneally, and isoflurane was discontinued. Respiration rate (RR), respiration pattern, and rectal temperature were monitored using a MRI-compatible physiological monitoring system (Model 1025, SA Instruments Inc, USA). The rectal temperature was maintained at ~37 °C by a feedback-controlled airheater (SA Instruments Inc, USA) during the experiments. Physiology To understand the physiological variation of mice at different dosages of medetomidine, blood oxygen saturation (spO2), respiration rate, and heart rate (HR) were measured from the thigh using MouseOx (STARR Life Sciences, USA) for up to 200 min on the bench in another sets of mice (n = 2 per dosage group). The same anesthetizing procedure and dosages were used. The room was kept dark and quiet throughout the experiments.

Fig. 1. Experiment design and physiology under different levels of medetomidine. (a) Schematic representation of the time frame of the experiment. At time 0, a bolus of 0.3 mg/kg medetomidine was injected intraperitoneally. At 15 min, isoflurane was switched off and medetomidine infusion started (red dashed line). Resting-state data was acquired at 30 and 120 min (blue dashed lines). FMRI of forepaw stimulation was conducted in between. (b) Traces of RR and HR recorded on the bench from bolus injection (time 0) up to 200 min at different dosages of medetomidine, where light and dark blue traces represent RR and HR at 0.1 mg/kg/h medetomidine (n = 2). Light and dark red traces represent RR and HR at 0.6 mg/kg/h medetomidine (n = 2). Light and dark green traces represent RR and HR at 1.0 mg/kg/h medetomidine (n = 2). The two black dashed lines indicate the timing corresponding to the start of resting-state fMRI scans. (c) The effect of medetomidine on the physiology in MRI. The blue bars indicate the time taken to detect robust and reliable BOLD activation and the red bars indicate the time the animal woke up (n = 8 for 0.1, and n = 6 for 0.6 and 1.0 mg/kg/h). Error bars represent SEM.

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

MRI

Results

All MRI measurements were performed on a 9.4 T/31 cm horizontal magnet (Agilent Technologies, USA). Two actively decoupled radiofrequency (RF) coils were used: a volume coil (7.2 cm inner diameter; Rapid Biomedical GmbH, Germany) was used for transmission and a custom-designed surface coil of 1 cm diameter positioned on the top of the animal's head as the receiver. The receiver coil was carefully positioned to ensure consistent B1 profile. Both the 1st and 2nd order shims were optimized in an 8 × 8 × 4 mm3 volume using the standard GE 3D shim protocol with the Agilent 205/120HD gradient shim coil. The shim volume covered the cortex and part of the subcortical area of the mouse brain. The typical line width achieved in the mouse brain was on the order of ~30 Hz. For functional imaging, 15 axial slices were acquired using a gradient-echo echo-planar imaging (GE EPI) with TR = 2 s TE = 15 ms, thickness = 0.5 mm, matrix size = 64 × 64, and FOV = 20 × 20 mm2. The resting-state fMRI was acquired for 10 min (300 volumes). The forepaw stimulation fMRI runs were acquired with a series of 150 repetitions and the same imaging parameters. High-resolution anatomical T2-weigted fast spin-echo was acquired at the same slice location with TR/TE = 2500/40 ms, echo train = 8, matrix = 256 × 256, FOV = 20 mm, and average = 2.

Physiological measurements

Data analysis Data was processed using custom-written software in Matlab (Mathworks, Natick, MA, USA) and in C. All data was carefully checked for head movement by displaying the time series images in movie mode and by calculating the center-of-mass. Motion correction using SPM (www.fil.ion.ucl.ac.uk/spm/) was also tested. No significant movement was detected but artifacts were introduced after correction. Therefore realignment was not applied. To detect forepaw activation, cross-correlation maps were computed by correlating the fMRI data with the box-car paradigm and thresholded at a correlation N0.2 with a cluster size of 4 pixels. Activation pixel number in the SI forepaw area was calculated and signal change was determined from all the pixels in the area activated by the 1 mA stimulation. Resting-state BOLD data was pre-processed by high-pass filtering at 0.01 Hz and low-pass filtering at 0.1 Hz. The average signals from the ventricles were regressed out to reduce contributions from physiological noise and the data was spatially smoothed by a Gaussian kernel with full width at half maximum = 1 pixel. Since correlation in the resting BOLD signal N 0.1 Hz has been reported in several studies in rats and human, we also inspected correlation spectra of the resting state signal without the low pass filter. The connectivity maps were calculated by correlation analysis based on the time-course of a 2 × 2 pixel region of interest (ROI) placed in the left hemisphere of each of the following regions: the primary and secondary somatosensory (SII), and visual cortices (VC), as well as subcortical areas, included the caudate putamen (CPu), thalamus (Thal), and hippocampus (Hip), defined with reference to the T2-weigted anatomical scans and the Paxinos and Watson mouse brain atlas (Paxinos and Watson, 1995). The seed regions for different dosages were chosen as close as possible in the same area. A correlation coefficient higher than 0.2 was considered significant and clusters smaller than 4 pixels were rejected. To determine the inter-hemispheric connectivity, ROIs were delineated on the entire functional regions in the left and right hemispheres based on the mouse brain atlas (Paxinos and Watson, 1995). The correlation coefficient between the timecourses from the ROIs were calculated. Statistical significance was tested using ANOVA with the Bonferroni test for multiple comparisons. p b 0.05 was regarded as significant. All results are presented as mean ± SEM.

419

To assess the physiological change, SpO2, HR, RR, and rectal temperature were recorded under constant infusion of 0.1, 0.6, and 1.0 mg/kg/h medetomidine on the bench. The RR quickly stabilized after 15 min likely due to the start of continuous infusion (Fig. 1b). The HR started high and gradually decreased. It took slightly longer time to stabilize and reached plateau between 30 to 60 min after the bolus injection. All physiological parameters were stable in the period between 40 and 120 min after initiation of medetomidine infusion. The RR of the spontaneously breathing animals was 100 ± 10 bpm (breaths per min; mean ± SEM) at 0.1, 100 ± 14 at 0.6, and 124 ± 17 at 1.0 mg/kg/h dosage. Temperature was stable in all animals investigated at 36.9 ± 0.1 °C. The SpO2 levels were very stable and no change was seen during the course of the infusion and was within the range of 98.0 ± 0.5% at 0.1, 97.6 ± 0.5% at 0.6, and 97.0 ± 2.0% at 1.0 mg/kg/h medetomidine, respectively. The heart rate was 200 ± 27 bpm (beats per min) at 0.1, 196 ± 26 at 0.6, and 206 ± 24 at 1 mg/kg/h within the 30–120 min intervals (Fig. 1b). The HR of the mice started to increase ~ 140 min in all the doses, but seemed to be able to be sedated to the end of measurement at 200 min. The stability of the sedation may be different in MRI environment especially with the noise and vibration of imaging. Heart rate, respiration rate, and movement were inspected with movement as the primary determinant for the wake-up time. Inside MRI, the RR of mice was slightly higher with 144 ± 12 bpm at 0.1, 141 ± 10 at 0.6, and 139 ± 10 at 1.0 mg/kg/h medetomidine from 30 to 140 min. The duration of sedation was 153 ± 21 min (n = 8) at 0.1 mg/kg/h dose. It increased slightly when the dosage of medetomidine was higher but the difference was not significant (Fig. 1c). In general, animals were stable within the 130 min imaging period. We also assessed the stability of cerebral physiology by determining the time taken to be able to detect robust BOLD activation in the somatosensory forepaw areas. Under all dosages of medetomidine, robust and consistent BOLD activations were detected at 60 ± 6 min post bolus injection (Fig. 1c).

Stimulation-induced BOLD activation Two mice, one from each of the 0.6 and 1.0 mg/kg/h groups, were excluded from the analysis because they tended to wake up within the first 30 min after medetomidine infusion. The reason may be due to the stress and anxiety of the animal not being habituated enough prior to the experiment. Three mice from the 0.1 mg/kg/h group were also excluded from the study because the forepaw stimulation at 1 mA induced high pain response which was evident from the activation in the cingulate cortex and thalamus of these animals. All 3 mice showed only unilateral connectivity in the resting state functional maps. This suggests that suboptimal sedation accompanied by pain response may lead to weak functional connectivity in mice. Robust and focal activations were detected in the contralateral SI forepaw area at different stimulating currents. No difference was observed in the activation maps (Fig. 2a) and time-courses (Fig. 2b) under the three dosages of medetomidine. There was a slight increase in the BOLD signal change with increased stimulus current from 0.5 to 1 mA (Fig. 2c). The highest BOLD signal change was detected at 1 mA current stimulation with a signal change of 1.80 ± 0.35% (n = 8), 1.71 ± 0.40% (n = 6), and 1.79 ± 0.30% (n = 6) detected under 0.1, 0.6, and 1.0 mg/kg/h, respectively. Both the signal change and activated area (Fig. 2d) did not have significant differences across dosages of medetomidine.

420

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

Fig. 2. Somatosensory activation under different levels of medetomidine. (a) T2-weighted anatomical image and activation maps of representative animals with 1 mA electrical stimulation to the left or right forepaws at 0.1, 0.6, and 1.0 mg/kg/h medetomidine. BOLD responses are consistently observed in the contralateral SI forepaw area. (b) Averaged time courses from the SI forepaw areas under 0.1 (n = 8), 0.6 (n = 6), and 1.0 (n = 6) mg/kg/h medetomidine (mean ± SEM). Time course under 0.5 mA is in light blue, 0.75 mA in green, and 1 mA in dark navy blue. Gray bars indicate the 20 s stimulation periods. (c) The percent signal change shows linear relationship with the stimulus current and is comparable across different dosages of medetomidine, where 0.1, 0.6, and 1.0 mg/kg/h medetomidine are represented in blue, red, and green, respectively. (d) The activated pixel number in the SI forepaw regions under 0.1, 0.6, and 1.0 mg/kg/h medetomidine increases similarly with stimulation current applied. No difference across medetomidine dosage is seen. Error-bars represent SEM.

Resting-state functional connectivity Representative functional connectivity maps at two time points, 30 and 120 min after the bolus injection, and under the three medetomidine dosages are represented in Fig. 3 and the Supplementary Material. Highly correlated resting BOLD signal were observed between the two hemispheres in the somatosensory (SI, SII) and visual cortices (VC), and strongly in the CPu, thalamus, and hippocampus at both time points measured under 0.1 mg/kg/h medetomidine (Fig. 3a). When 0.6 mg/kg/h medetomidine was used, no loss in bilateral correlation was detected in all networks detected at the 30 min timepoint (Figs. 3b and 4a). However, at the 120min time point, a significant reduction was seen in the thalamus (p b 0.05) under 0.6 mg/kg/h medetomidine. The correlation coefficient between the left and right thalamic nuclei was reduced from 0.72 ± 0.02 (0.1 dose) to 0.58 ± 0.04 (0.6 dose) (Fig. 4b). When medetomidine dosage was increased to 1.0 mg/kg/h, significant loss of the interhemispheric correlation was seen in almost all ROI's investigated and at both time points (Fig. 3c). This was not applicable to the CPu which showed insignificant change of the connectivity under this high drug dosage. The correlation coefficients remained unchanged: 0.69 ± 0.03 at 0.1, 0.68 ± 0.04 at 0.6, and 0.57 ± 0.07 at 1.0 mg/kg/h medetomidine (Fig. 4b). Despite no significant reduction in correlations, a decreasing trend was still obvious in the CPu. Correlation spectra of the resting-state signal from the left and right SI were calculated to determine the frequency that dominates the bilateral correlation. The peak frequencies identified were mostly

b0.1 Hz for all medetomidine dosages. The mean peak frequency was 0.07 ± 0.024 Hz at 0.1 dose, 0.08 ± 0.019 Hz at 0.6 dose and 0.10 ± 0.013 Hz at 1.0 dose. Although there is a tendency to correlate at higher frequency as dosage increased, the difference was insignificant (p = 0.15, 0.1 vs 1.0 dose). To assess the potential of forepaw stimulation on modulating the measured connectivity at the 120 min time point, another set of mice was studied without forepaw stimulation. The correlation coefficient between the left and right SI in these mice at the 120 min time points were 0.70 ± 0.05 (at 0.1 dose) and 0.66 ± 0.1 (at 0.6 dose), which are comparable with those measured after stimulations. Therefore, the connectivity measured was not altered by the short sensory stimulation used. To understand the impact of cerebral physiology on resting-state connectivity, we compared the time to detect activation and the activation signal change with respect to the bilateral correlation in SI under 0.1 and 0.6 dosages (Fig. 5). When the time it took to detect robust somatosensory activation is longer, the bilateral connectivity is lower (R2 = 0.56; p b 0.0015). When the activation signal change is lower, the bilateral connectivity is also lower (R2 = 0.81; p b 0.5 × 10−4). These indicate that the detection of bilateral connectivity would depend on animal physiology. Discussion The advances in resting state functional connectivity imaging have provided valuable insight into how brain networks are disrupted in

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

421

Fig. 3. Resting-state connectivity under different medetomidine dosages. Representative correlation maps under (a) 0.1, (b) 0.6 and (c) 1.0 mg/kg/h of medetomidine based on seed points in the SI, CPu, SII, Hip, Thal, and VC whereas the maps of the same dosage are from the same animal. The upper pane represents the connectivity maps at the 30 min time-point while the lower pane represents the maps of the same ROI's but at the 120 min time-point. Positive (red/yellow) and bilateral functional connectivity are consistently observed at 0.1 and 0.6 mg/kg/h but dramatically reduced at 1.0 mg/kg/h. Some correlated pixels seem to extend outside the brain, which are due to distortion near the surface of the brain.

diseases. Applying similar technique in mouse models of diseases can help to determine the underlying pathophysiological changes and to evaluate potential treatments. However, resting-state fMRI in the mouse brain has been challenging due to low detection sensitivity and large physiological variations. Here we demonstrated that robust functional activation can be detected under medetomidine sedation, indicating good sensitivity and proper physiology. Especially, bilateral connectivity in all the major functional areas can be reliably detected under suitable dosages of medetomidine. Such bilateralism is suppressed under high dosage of medetomidine, similar to that observed in the rat (Nasrallah et al., 2012). BOLD activation under medetomidine A critical aspect in animal fMRI studies is the tight physiological control to achieve proper neurovascular coupling. Anesthetics affect neuronal activity (Jenkins et al., 1999; Masamoto et al., 2007; Rojas et al., 2006) and the coupling of neural activity to hemodynamics and metabolism (Crosby et al., 1983; Franceschini et al., 2012; Lindauer et al., 1993). Several anesthetics used in rat fMRI studies, such as isoflurane and chloralose, require measuring arterial blood CO2, pH, etc. for controlling physiology. The small blood volume of the mouse

however restricts such procedures and therefore maintenance of physiology is a challenge. Anesthetics, such as medetomidine, that require minimal physiological control are desirable. Recently it is demonstrated that a 2 mA electrical stimulation can elicit 1.3% BOLD signal change in the mouse brain at 11.7 T under 0.6 mg/kg/h medetomidine without suppression of somatosensory evoked responses (Adamczak et al., 2010). By adopting similar protocol but with lower infusion dosage, we could achieve sedation for at least 150 min inside the MRI. Higher dosage extended the sedation but was not significant inside MRI (Fig. 1c). Medetomidine selectively activates the α2 adrenoceptor by which it induces sedation in a dose-dependent manner but only to an extent. Beyond certain dose, it does not deepen the intensity of sedation but prolongs its duration (Kuusela et al., 2000). Since the effectiveness of medetomidine sedation highly depends on the anxiety of the animal before being sedated, the longer sedation duration achieved in this study may be due to the handling and habituation of animals before induction of anesthesia. Previously, we showed that somatosensory BOLD activation in the rat brain is the same under 0.1 to 0.3 mg/kg/h medetomidine (Nasrallah et al., 2012). Further study showed that somatosensory evoked potential was also the same, indicating neurovascular coupling is not affected by medetomidine in the rat brain (Nasrallah et al.,

422

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

Baltes et al. showed that somatosensory BOLD activation of 3.5% following 1.5 mA stimulation could be detected using a cryoprobe at 9.4 T (Baltes et al., 2011). In that study, mice were anesthetized by isoflurane with mechanical ventilation to optimize the physiology. However, isoflurane alters the neurovascular coupling in a dose dependent manner (Masamoto et al., 2009) and could largely suppress neural activity (Liu et al., 2011) deeming it only useful for fMRI at very low levels.

Functional connectivity in the mouse brain

Fig. 4. Inter-hemispheric correlation coefficients in ROIs under different levels of medetomidine. Average correlation coefficients between the left and the right ROI's under 0.1 (blue), 0.6 (red) and 1.0 (green) mg/kg/h medetomidine at (a) 30 min and (b) 120 min time-points. Significant reduction in correlation coefficient was seen in thalamus only at 120 min under 0.6 mg/kg/h, and in all areas except the CPu under 1.0 mg/kg/h. Error bars represent SEM and *: p b 0.05, **: p b 0.01 (ANOVA with Bonferoni post hoc test of multiple comparison).

2014). To evaluate whether the same characteristics is also preserved in the mouse, we studied the stimulus current dependency under 0.1, 0.6 and 1.0 mg/kg/h medetomidine. We observed robust and significant activation in the somatosensory cortex on the order of 1.8% following 1 mA stimulation (Fig. 2). The BOLD signal increased linearly with stimulation current and was comparable across the different dosages of medetomidine. This implicates that the impact of medetomidine on neurovascular coupling may also be minimal.

Similar to other species, bilateral connectivity was observed in the mouse brain (Fig. 3 and Supplementary Material). These patterns resemble those previously found in rats (R.M. Hutchison et al., 2010) indicating that such connectivity is indeed preserved in rodents. The results are however different from those shown by Jonkers et al. where mostly only unilateral connectivity was found, except in the ventral hippocampus and piriform cortex (Jonckers et al., 2011). On the other hand, another study using optical intrinsic signal imaging, which measures hemodynamics of oxy-/deoxy-hemoglobin levels, found very high bilateral connectivity between the major cortical areas (White et al., 2011). Many factors, especially animal physiology and detection sensitivity, may account for the lack of bilateral connectivity in Jonkers et al. Although higher dose of medetomidine could suppress functional connectivity in mouse, the dosage needed is much higher (1.0 mg/kg/h) and hence would not be the issue. A plausible factor would be suboptimal sedation so that animal could be more conscious. We observed that in 3 mice that showed pain responses by sensory stimulation, which indicated insufficient sedation, their bilateral functional connectivity was largely depressed. Another factor may be whether the animal physiology is stable enough so that neurovascular coupling is well preserved. The RR of the animal was stabilized after the continuous infusion started, while the HR was stabilized 30 min after the bolus injection. Therefore we acquired the first resting-state scan at 30 min. To confirm that the neurovascular coupling was good, we started forepaw stimulation at 40–45 min and were typically able to obtain robust BOLD activation at the first or the second run. Furthermore, by comparing the restingstate connectivity acquired at 30 min with the one obtained at 120 min, the networks were almost the same, indicating that the neurovascular coupling was maintained well. One interesting observation was that the bilateral connectivity was negatively correlated with the time to detect activation. In animals taking longer time to detect robust BOLD activation, lower bilateral correlation in SI was seen at the resting state (Fig. 5a). In addition, the activated signal change also correlated with the bilateral connectivity at rest (Fig. 5b). Although no difference in

Fig. 5. Pearson's correlations between evoked activation and resting state. (a) Scatter plot showing significant correlation between the time to detect activation and the correlation coefficient measured from the bilateral SI regions at rest under 0.1 (blue) and 0.6 (red) mg/kg/h medetomidine dosages at 120 min (p = 0.0015). (b) Scatter plot showing high correlation between the activation signal change under 1 mA stimulation and the correlation coefficient measured from the bilateral SI regions at rest under 0.1 and 0.6 mg/kg/h medetomidine dosages at 120 min (p = 0.5 × 10−4).

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

activation detection time among dosages, a slight dosage dependency in the relationship between functional connectivity and the activation detection time was seen. The 0.6 dosage showed stronger impact on the strength of functional connectivity, reflected as a steeper slope compared to 0.1 dosage. On the contrary, the relationship between functional connectivity and the activation signal change was consistent between dosages. These indicate that the strength of connectivity is dependent on whether the animal physiology is maintained well enough to achieve strong neural activity and optimal neurovascular coupling. Therefore, proper physiology and avoiding pain responses by sensory stimulation are vital considerations when attempting to achieve robust resting BOLD connectivity measure in the mouse. It is recommended to conduct an activation experiment to serve as an indication of physiological state of the animal for proper interpretation of the resting-state signal. Effect of medetomidine on resting-state synchrony Previously we demonstrated that the α2 adrenoceptor agonist could suppress BOLD synchrony dosage and regional dependently in rats (Nasrallah et al., 2012). By employing much higher dosages of medetomidine, interhemispheric functional connectivity between regions was also significantly suppressed (Figs. 3c and 4). At low dose of 0.1 mg/kg/h, strong bilateral correlation was detected in the SI, SII, VC, Hip, Thal, and CPu, which agrees with previous studies in rats (Nasrallah et al., 2012; Pawela et al., 2008; Zhao et al., 2008). At a higher dose of 0.6 mg/kg/h, the bilateral connectivity was mostly the same, but the connectivity in the thalamus was reduced at 120 min. The thalamus is known to have very high density of α2 adrenoceptor (Wang et al., 1996) similar to that in the rat (King et al., 1995). This indicates that at 120 min, high enough medetomidine had accumulated to cause significant suppression effect. Going to extremely higher levels of medetomidine of 1.0 mg/kg/h, there was a dramatic suppression of resting-state correlation in all brain regions except the CPu, which is also known to have low α2 adrenoceptor density (Wang et al., 1996). The decrease in the functional correlation was similar to that seen in the rat, although the medetomidine concentrations required in the mouse differed. Moreover, the differential effect on brain activation and connectivity observed in the rats is also found in the mouse where no change in the BOLD activation profile was seen with any of the medetomidine dosages used (Fig. 2c). Technical issues of mouse resting-state fMRI Various technical issues have prevented the broad application of fMRI in the mouse brain which requires high signal-to-noise ratio (SNR) and spatial resolution (Van der Linden et al., 2007). Most fMRI studies are performed using GE EPI because of its greater sensitivity to the BOLD effect which also suffers from artifacts associated with magnetic susceptibility. The susceptibility difference at air-tissue and bone-tissue interfaces leads to large signal loss near the ear canals and geometric distortions in various brain areas including cortex. In addition, GE EPI is sensitive to physiological noise, which has shown to confound the resting-state connectivity measures in human (Birn, 2012) and rodent (Kalthoff et al., 2011). These artifacts are larger when using stronger magnetic field to achieve better SNR. Such susceptibility related artifacts could be reduced using spin-echo EPI but the sensitivity to BOLD also decreases. In our pilot study, we compared spin-echo and gradient-echo EPI. With a TE of 38 ms in spin-echo EPI, forepaw activation could be detected but not connectivity at the resting-state, even when more repetitions (up to 600) were used. Therefore GE EPI was used to achieve higher sensitivity at the price of geometric distortion. Further study will be needed to optimize methods for correcting the distortion, such as phase reversal or other techniques (Holland et al., 2010) or segmented EPI (Guilfoyle and Hrabe, 2006) so that registration can be conducted for group statistical mapping.

423

A limitation of medetomidine is the short sedative period. Although medetomidine has been shown to have dose-dependent effect on extending the sedation period, high dosage can suppress functional connectivity. In an attempt to prolong the duration of sedation without affecting functional connectivity, Pawela et al. (Pawela et al., 2009) showed that a sedation period of up to 6h could be achieved by stepping up the medetomidine dose from 0.1 to 0.3 mg/kg/h. Although higher medetomidine dosage can also increase the sedation period in mice, the extension is far less than that can be achieved in rats. Further studies will be needed to evaluate whether stepping up the dosage during the experiment can extend the duration of sedation in mice as it does in rats. Conclusion In this work, we demonstrated the existence of bilateral functional connectivity in major areas of the resting mouse brain. Therefore the resting-state functional connectivity seems to be a well preserved phenomenon in mammal. The potential to detect functional networks in the mouse with a simple and longitudinal sedative protocol will enable broader application of fMRI in the investigation of a wide variety of transgenic models available to further understand disease progression, therapy and their translation. Acknowledgments We would like to thank Mr. Krzysztof Pyka for his technical input. The work was supported by the Intramural Research program of the Biomedical Sciences Institutes, Agency for Science, Technology and Research (A*STAR), Singapore. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.neuroimage.2013.10.025. References Adamczak, J., Farr, T.D., Seehafer, J., Kalthoff, D., Hoehn, M., 2010. High field BOLD response to forepaw stimulation in the mouse. NeuroImage 51, 704–712. Baltes, C., Bosshard, S., Mueggler, T., Ratering, D., Rudin, M., 2011. Increased blood oxygen level-dependent (BOLD) sensitivity in the mouse somatosensory cortex during electrical forepaw stimulation using a cryogenic radiofrequency probe. NMR Biomed. 24, 439–446. Bero, A., Bauer, A., Stewart, F., White, B., Cirrito, J., Raichle, M., Culver, J., Holtzman, D., 2012. Bidirectional relationship between functional connectivity and amyloid-β deposition in mouse brain. J. Neurosci. 32, 4334–4340. Birn, R., 2012. The role of physiological noise in resting-state functional connectivity. NeuroImage 62, 864–870. Biswal, B., Yetkin, F.Z., Haughton, V.M., Hyde, J.S., 1995. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn. Reson. Med. 34, 537–541. Chuang, K.-H., van Gelderen, P., Merkle, H., Bodurka, J., Ikonomidou, V.N., Koretsky, A.P., Duyn, J.H., Talagala, S.L., 2008. Mapping resting-state functional connectivity using perfusion MRI. NeuroImage 40, 1595–1605. Crosby, G., Crane, A., Jehle, J., Sokoloff, L., 1983. The local metabolic effects of somatosensory stimulation in the central nervous system of rats given pentobarbital or nitrous oxide. Anesthesiology 58, 38–43. Fox, M., Richle, M., 2007. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat. Rev. Neurosci. 8. Franceschini, M., Radhakrishnan, H., Thakur, K., Wu, W., Ruvinskaya, S., Carp, S., Boas, D., 2012. The effect of different anesthetics on neurovascular coupling. NeuroImage 51, 1367–1377. Guilfoyle, D., Hrabe, J., 2006. Interleaved snapshot echo planar imaging of mouse brain at 7.0 T. NMR Biomed. 19, 108. Guilfoyle, D., Gerum, S., Sanchez, J., Balla, A., Sershen, H., Javitt, D., Hoptman, M., 2013. Functional connectivity fMRI in mouse brain at 7 T using isoflurane. J. Neurosci. Methods 214, 144–148. Holland, D., Kuperman, J., Dale, A., 2010. Efficient correction of inhomogeneous static magnetic field-induced distortion in Echo Planar Imaging. NeuroImage 50, 175–183. Hutchison, R., Mirsattari, S., Jones, C., Gati, J., Leung, L., 2010a. Functional networks in the anesthetized rat brain revealed by independent component analysis of resting-state fMRI. J. Neurophysiol. 103, 3398–3406. Hutchison, R.M., Mirsattari, S.M., Jones, C.K., Gati, J.S., Leung, L.S., 2010b. Functional networks in the anesthetized rat brain revealed by independent component analysis of resting-state fMRI. J. Neurophysiol. 103, 3398–3406.

424

F.A. Nasrallah et al. / NeuroImage 86 (2014) 417–424

Jenkins, A., Franks, N., Lieb, W., 1999. Effects of temperature and volatile anesthetics on GABAA receptors. Anesthesiology 90, 484–491. Jonckers, E., Van Audekerke, J., De Visscher, G., Van der Linden, A., Verhoye, M., 2011. Functional connectivity fMRI of the rodent brain: comparison of functional connectivity networks in rat and mouse. PLoS ONE 6, e18876. Kalthoff, D., Seehafer, J.U., Po, C., Wiedermann, D., Hoehn, M., 2011. Functional connectivity in the rat at 11.7 T: impact of physiological noise in resting state fMRI. NeuroImage 54, 2828–2839. King, P.R., Gundlach, A.L., Louis, W.J., 1995. Quantitative autoradiographic localization in rat brain of [alpha]2-adrenergic and non-adrenergic I-receptor binding sites labelled by [3H]rilmenidine. Brain Res. 675, 264–278. Kuusela, E., Raekallio, M., Anttila, M., Falck, L., Mölsä, S., Vainio, O., 2000. Clinical effects and pharmacokinetics of medetomidine and its enantiomers in dogs. J. Vet. Pharmacol. Ther. 23, 15–20. Lindauer, U., Villringer, A., Dirnagl, U., 1993. Characterization of CBF response to somatosensory stimulation: model and influence of anesthetics. Am. J. Physiol. Heart Circ. Physiol. 264, H1223–H1228. Liu, X., Zhu, X., Zhang, Y., Chen, W., 2011. Neural origin of spontaneous hemodynamic fluctuations in rats under burst-suppression anesthesia condition. Cereb. Cortex 21, 374–384. Masamoto, K., Kim, T., Fukuda, M., Wang, P., Kim, S.-G., 2007. Relationship between neural, vascular, and BOLD signals in isoflurane-anesthetized rat somatosensory cortex. Cereb. Cortex 17, 942–950. Masamoto, K., Fukuda, M., Vazquez, A., Kim, S., 2009. Dose-dependent effect of isoflurane on neurovascular coupling in rat cerebral cortex. Eur. J. Neurosci. 30, 242–250. Nair, G., Duong, T., 2004. Echo-planar BOLD fMRI of mice on a narrow-bore 9.4 T magnet. Magn. Reson. Med. 52, 430–434. Nasrallah, F., Tan, J., Chuang, K., 2012. Pharmacological modulation of functional connectivity: alpha-2-adrenergic receptor agonist alters synchrony but not neural activation. NeuroImage 60, 436–446.

Nasrallah, F.A., Lew, S.K., Low, A., Chuang, K.H., 2014. Neural correlate of resting-state functional connectivity under α2 adrenergic agonist, medetomidine. NeuroImage 84, 27–34. Pawela, C.P., Biswal, B.B., Cho, Y.R., Kao, D.S., Li, R., Jones, S.R., Schulte, M.L., Matloub, H.S., Hudetz, A.G., Hyde, J.S., 2008. Resting-state functional connectivity of the rat brain. Magn. Reson. Med. 59, 1021–1029. Pawela, C., Biswal, B., Hudetz, A., Schulte, M., Li, R., Jones, S., Cho, Y., Matloub, H., Hyde, J., 2009. A protocol for use of medetomidine anesthesia in rats for extended studies using task-induced BOLD contrast and resting-state functional connectivity. NeuroImage 46, 1137–1147. Paxinos, G., Watson, C., 1995. The rat nervous system, 2nd edition. Academic Press, San Diego. Rojas, M., Navas, J., Rector, D., 2006. Evoked response potential markers for anesthetic and behavioral states. Am. J. Physiol. Regul. Integr. Comp. Physiol. 291, R189–R196. Van der Linden, A., Van Camp, N., Ramos-Cabrer, P., Hoehn, M., 2007. Current status of functional MRI on small animals: application to physiology, pathophysiology, and cognition. NMR Biomed. 20, 522–545. Vincent, J., Patel, G., Fox, M., Snyder, A., Baker, J.T., Van Essen, D., Zempel, J., Snyder, L., Corbetta, M., Raichle, M., 2007. Intrinsic functional architecture in the anaesthetized monkey brain. Nature 447. Wang, R., Macmillan, L., Fremeau Jr., R., Magnuson, M., Lindner, J., Limbird, L., 1996. Expression of alpha 2-adrenergic receptor subtypes in the mouse brain: evaluation of spatial and temporal information imparted by 3 kb of 5' regulatory sequence for the alpha 2A AR-receptor gene in transgenic animals. Neuroscience 74, 199–218. White, B., Bauer, A., Snyder, A., Schlaggar, B., Lee, J., Culver, J., 2011. Imaging of functional connectivity in the mouse brain. PLoS ONE 6, e16322. Zhao, F., Zhao, T., Zhou, L., Wu, Q., Hu, X., 2008. BOLD study of stimulation-induced neural activity and resting-state connectivity in medetomidine-sedated rat. NeuroImage 39, 248–260.