|Home | About | Journals | Submit | Contact Us | Français|
In vivo localized high-resolution 1H MR spectroscopy was performed in multiple brain regions without the use of anesthetic or paralytic agents in awake head-restrained rats that were previously trained in a simulated MRI environment using a 7 Tesla MR system. Spectra were obtained using a short echo time single-voxel point-resolved spectroscopy technique with voxel size ranging from 27–32.4 mm3 in the regions of anterior cingulate cortex, somatosensory cortex, hippocampus, and thalamus. Quantifiable spectra, without the need for any additional post-processing to correct for possible motion were reliably detected including the metabolites of interest such as γ-aminobutyric acid, glutamine, glutamate, myo-inositol, N-acetylaspartate, taurine, glycerophosphorylcholine/phosphorylcholine, creatine/phosphocreatine, and N-acetylaspartate/N-acetylaspartylglutamate. The spectral quality was comparable to spectra from anesthetized animals with sufficient spectral dispersion to separate metabolites such as glutamine and glutamate. Results from this study suggest that reliable information on major metabolites can be obtained without the confounding effects of anesthesia or paralytic agents in rodents.
In vivo high-resolution proton magnetic resonance spectroscopy (1H MRS) has been employed extensively to simultaneously measure the concentration of a large number of neuro-metabolites in normal and diseased brains of human and animals (1–9). Unlike human MRS studies that can be readily performed in conscious participants, in vivo animal MRS is typically performed under anesthesia in order to keep the animal from moving during data acquisition.
It is well documented that neuronal activity, basal cerebral blood flow (CBF), hemodynamic coupling, and blood-oxygenation-level-dependent (BOLD) based signals can be altered by anesthetics or paralytic agents (10–15). Several functional magnetic resonance imaging (fMRI) protocols using awake animals have been developed to circumvent these problems in primates (16–19), rats (14,15, 20–26), mice (27), and rabbits (28).
A number of in vivo MRS studies demonstrated that anesthesia affects cerebral metabolic profiles in rats (29–32). These studies demonstrated that different anesthetics (isoflurone, propofol, pentobarbital) result in different brain metabolic profiles of glutamate (29), lactate and glucose (29–32). To the best of our knowledge, only one study investigated neuro-metabolic profiles using high-resolution localized 1H MRS in awake non-human primates (33). An awake MRS preparation that utilizes rodents is still lacking despite the extensive use of rodents in basic and translational neuroscience research. The goals of this study were, (a) to develop a protocol to perform MRS experiments in awake rats, and (b) to determine whether metabolic profiles in various regions of the rat brain can be measured reliably at 7 Tesla in awake, head-restrained preparations using high-resolution 1H MRS. To our knowledge, this is the first report of high-resolution 1H MRS in awake rats.
Five female Sprague-Dawley rats (22 days old) were obtained from Harlan Laboratories (Raleigh, NC) and were group-housed in standard cages in a 12:12 light/dark cycle with access to food and water ad libitum. All procedures performed on the animals were in strict accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals and approved by the University of Maryland Baltimore College of Dental Surgery, Institutional Animal Care and Use Committee.
To train and acclimate the animals to the MRI environment, we used methods similar to those described in King et al., (23) with some modifications. The animals were acclimated and trained daily to go into a loose-fitting restraining cloth and placed in a custom made plastic replica of the animal bed/holder used in the scanner. During handling and acclimation, the animals initially were restrained for a period of 5 minutes. The period of restraint was increased gradually and reached a maximum of 30 minutes towards the end of the training period. During the training, the animals were exposed to tape recordings of the sound bursts generated by gradient switching in the magnet during MRS experiments. The animals were trained daily until they reached adulthood (approximately 2 months) and by the end of the training period, the rats learned to remain calm in the replica of the scanner holder. After training, the animals were prepared for surgery to attach a custom designed plastic head holder to allow for further head-restraint.
The animals were anesthetized with 2% isoflurane, attached to a stereotaxic frame and placed on a thermo-regulated heating pad. Depth of anesthesia was determined every 15 minutes by monitoring pinch withdrawal, eyelid reflex, corneal reflex, respiration rate, and vibrissae movements. Incision sites were injected with local anesthetic (0.5% marcaine) to further reduce the possibility that animals would experience pain. A midline incision overlying the skull extending from the inion to lambda (10 mm) was made using a #11 scalpel blade. The scalp was reflected laterally and six predetermined atraumatic openings were prepared using a pin vise and #66 drill bit (Plastic One Inc., VA). The locations of these openings were as follows: 2 anterior in the frontal bone, 2 posterior in the occipital bone, and 2 bilateral in the parietal bone. Six polyetheretherketone screws (PEEK, 1.7 × 4 mm, NetMotion, CA) were used to attach a custom made acrylic resin plastic head holder (Jet Acrylic Resin, Dentsply, PA) to the skull. The acrylic head holder was oval in shape (3 mm thick) with two lateral extensions (15 mm long, 4mm thick) and fit the calvarium of the rat intimately. The lateral extensions were used to restrain the animal during MRS experiments (Fig. 1a). At the end of surgery, the skin edges were approximated and sutured. During recovery, the animals were placed on a thermo-regulated heat pad and observed every 15 minutes until they recovered completely and monitored daily thereafter. After surgery, buprenorphine HCl (0.01–0.05 mg/kg s.c.) was administered every 12 hours for 48 hours to reduce pain and discomfort.
One week after surgery the animals were reintroduced to the replica of the scanner holder. For a period of 7 days, the animals were trained to be restrained using the implanted head mount by attaching the lateral extensions to the scanner replica using plastic screws. The rats cooperated and there was no difficulty positioning them in the restrainer. No food and water were given to the animals during the periods of restraint. However, they were rewarded with a treat (Cheerios® and Froot Loops®) after every restraining session.
In vivo MRS experiments were performed on a Bruker BioSpec 70/30USR Avance III 7T horizontal bore MR scanner (Bruker Biospin MRI GmbH, Germany) equipped with a BGA12S gradient system capable of producing pulse gradients of 400 mT/m in each of the three orthogonal axes and interfaced to a Bruker Paravision 5.1 console. A Bruker four-element 1H surface coil array was used as the receiver and a Bruker 72 mm linear-volume coil as the transmitter. During the spectroscopy experiments, the animals were placed and restrained in the scanner bed in a manner similar to the one in which they were trained. An MR-compatible small-animal monitoring system was used to monitor animal respiration rate continuously through the entire experiment (SA Instruments, Inc., New York, USA). Screen snapshot were taken approximately every 30 minutes to record both the breathing waveforms and the respiration rate. A three-slice (axial, mid-sagittal, and coronal) scout using fast low angle shot MRI (FLASH) was obtained to localize the rat brain (31, 32). A fast shimming procedure (FASTMAP) was used to improve the B0 homogeneity for each voxel (36). T2-weighted MR images covering the entire brain were obtained using a 2D rapid acquisition with relaxation enhancement (RARE) sequence (37) with repetition time/effective echo time (TR/TEeff) = 4500/28 ms, echo train length = 4, field of view (FOV) = 3.5 × 3.5 cm2, matrix size = 256 × 256, slice thickness = 1 mm, number of averages = 2 for anatomic reference. Four spectroscopy voxels representing regions of interest were selected: anterior cingulate cortex (3 × 3 × 3 mm3), somatosensory cortex (1.8 × 4.5 × 4.0 mm3), hippocampus (2.5 × 4.0 × 3.0 mm3), and the thalamus (3 × 3 × 3 mm3). For 1H MRS, adjustments of all first- and second-order shims over the voxel of interest were accomplished with the FASTMAP procedure. Typically, the in vivo shimming procedure resulted in approximately 13 Hz full-width half maximum line-width of the unsuppressed water peak over the spectroscopy voxel. The PRESS (Point-RESolved Spectroscopy) pulse sequence was customized using a sinc7H pulse with a hamming window modulation providing an excitation bandwidth of 10 kHz. Two asymmetric pulses with a bandwidth of 2823 Hz were used for refocusing (38) which corresponds to 32% chemical shift displacement error in two directions for the chemical shift range of 3 ppm. The TE was shortened to 10 ms in order to alleviate signal attenuations caused by J modulation and T2 relaxation. The center frequencies of the above three localization pulses were placed at 3.0 ppm to reduce the lipid contamination caused by the chemical shift displacement. This custom made PRESS sequence (TR/TE = 2500/10 ms, number of average = 356) was used for MRS data acquisition from various regions including anterior cingulate cortex, somatosensory cortex, hippocampus, and thalamus. The acquisition time for each voxel was 15 minutes. In one animal, the MRS scan was repeated using the same parameters on two consecutives days to test for reproducibility of the MRS results.
Chronic restraint is stressful and the stress has been shown to be associated with reduced mechanical withdrawal thresholds and hyperalgesia in rodents (39, 40). Therefore, to test if our acclimation, head-restraint and MRS procedures produced chronic stress and hyperalgesia we assessed hindpaw mechanical withdrawal thresholds in all the animals. Mechanical withdrawal thresholds of both hindpaws were assessed in all the animals at: baseline (3 consecutive days before implantation of the head holder); 1 week after the implantation of the head holder; and immediately after each MRS scan. A dynamic planter aesthesiometer (Ugo Basile, PA) was used to determine hindpaw mechanical withdrawal thresholds as described by Dolan and Nolan (41). The non parametric Friedman test was used to assess if mechanical withdrawal thresholds changed significantly over time. A p value of less than 0.05 was considered significant.
Quantification of the MRS was based on frequency domain analysis using “Linear Combination of Model spectra” (LCModel) (42). In vivo spectra were analyzed by a superposition of a set of simulated basis set provided by the LCModel software. The reference for determining metabolite concentration was the water signal, which was acquired from the same voxel. The metabolic profile was measured with the same parameters except the number of averages was set at 8. The results were normalized by LCModel package to the metabolite concentrations and expressed as micromoles per gram wet weight (mM). Cramér-Rao lower bounds (CRLB) as reported from the LCModel analysis was used for assessing the reliability of the major metabolites. To provide an indicator of reproducibility, the coefficient of variability (CV) was calculated for the individual metabolites across individual animals as the ratio of the square root of the standard deviation to the mean.
In three of the animals MRS data were obtained from all four regions. In one animal, MRS data were obtained from three regions (somatosensory cortex, hippocampus, and thalamus). In the fifth animal; MRS data were obtained from anterior cingulate cortex as this was the first trained animal. Therefore, for the quantitative analysis, four data sets (n=4) were used in each region although a total of five animals were used in this study.
In all animals, we assessed mechanical withdrawal thresholds to test if chronic restraint resulted in stress and hyperalgesia (39, 40). Our training and restraining procedures did not result in any significant change over time in mechanical withdrawal thresholds (p = 0.31, ANOVA on Ranks; Fig. 1b).
Axial anatomic images along with the four spectroscopic voxel locations including the anterior cingulate cortex, somatosensory cortex, hippocampus, and the thalamus are shown in Fig. 1c. High-resolution localized 1H MR spectra from the four voxels are shown in Fig. 2. In these spectra, a number of metabolites can be reliably detected. N-acetylaspartate (NAA) methyl singlet at 2.01 ppm was assigned as the reference. The line width of the singlet 1H metabolite resonances (as measured on tCr at 3.0 ppm; Fig. 2) was typically 11–15 Hz (0.037–0.050 ppm) in awake animals and was comparable to the spectra from anesthetized animals (9). The concentrations of the metabolites measured by LCModel from these four voxels are listed in Table 1. The reliability of the major metabolites was estimated by the LCModel using the CRLB are also listed in Table 1. In general, the CRLB values of Glu (3–7%), Ins (5–10%), Tau (6–8%), GPC+PCh (4–7%), NAA+NAAG (3–5%), Cr+PCr (3–5%) and Glx (4–8%) were not more than 10 %. For the low concentration metabolites GABA (9–18%) and Gln (9–15%), the highest CRLB value was 18%. The exceptions are GABA (13–21%), Gln (17–27%) and Tau (15–29%) levels in the thalamus.
To test the reproducibility of MRS, we repeated the MRS scan in one rat on two consecutive days. Fig. 2a shows spectra acquired on day-1 from the 4 voxels and Fig. 2b shows spectra acquired from the same rat on day-2. Spectra obtained from the 4 regions appear to be similar at the two time points. In Table 2, we quantitatively analyze the spectra acquired in the somatosensory cortex (Fig. 2). The coefficient of variation was not more than 11% on all the metabolites considered. These findings suggest that the awake animal MRS is reproducible in the same animal. However, it is important to note that the CV values obtained from only 2 experiments may not be reliable. In addition, the breathing waveforms and respiration rates of this rat exhibited stable breathing pattern and rate (53–76 breath/min) suggesting minimal stress on the animal during both scans.
In this study, a novel head-restraining system was successfully implemented for the first time to obtain metabolic information using MRS in an awake rodent model for the first time. Using this preparation, it is possible to obtain localized 1H MR spectra in several brain regions of interest (anterior cingulate cortex, somatosensory cortex, hippocampus and thalamus) without the need for any post-processing scheme to correct for possible motion during data acquisition. In all four regions, most brain metabolites or the combinations were reliably detected (CRLB < 18 %), except for some metabolites in the thalamus. Because of the use of a surface coil, the rostral regions of the anterior cingulate cortex, somatosensory cortex, and hippocampus, exhibited high signal-to-noise ratio (~22) and the spectra obtained were highly reliable (CRLB < 18%). The overall signal to noise of the spectra for thalamus was lower (~13) as this region was at the periphery of the reception field of the four-element surface coil. In this region, metabolites with low concentration provided less reliable spectra (GABA, Gln, and Tau; CRLB = 21–29 %; Table 1).
The concentrations of the metabolites quantified in this study are within the range specified in the rat brain (see Table 3 in 43). It is important to note however that the GABA concentration may be overestimated in this study because there can be a significant overlap with the macromolecules in the GABA region of the spectrum. In addition, the Tau concentration may be underestimated because of its long spin-lattice relaxation time and short TR (44).
However, the concentrations of metabolites appear to be different from those reported in Xin et al. and Rao et al. (45,46). In our study, MRS experiments were performed in 14-week old female rats to assess metabolic profiles in the somatosensory cortex, anterior cingulate cortex, hippocampus and the thalamus. In Xin et al., (45) the animal gender was different (adult male rats) and although the regions of interest studied included the somatosensory cortex and hippocampus, the voxel sizes used were significantly different [somatosensory cortex: 6 × 1.5 × 2 mm3 vs. 1.8 × 4.5 × 4.0 mm3 in this study; hippocampus: 3 × 2 × 2 mm3 vs. 2.5 × 4.0 × 3.0 mm3] (45). In Rao et al., (46) the age and gender of the animals (8-week old, male) were also different. The methodological differences between the current study and the studies referenced above may explain the differences in metabolites’ concentrations. It is well documented that metabolic profiles depend on the age and gender of the animals as well as the region studied (46–48).
A major concern when performing experiments in restrained animals is the stress induced (39,40,49) and its effect on MRS results. To minimize this concern, we adopted similar methods to those described in King et al. (23) and acclimated and handled the animals from a young age before performing the experiments. We also assessed changes in mechanical withdrawal thresholds over time and monitored breathing during the MRS experiments. The acclimation and training procedure proved effective in reducing animal stress as evident from the lack of change in mechanical withdrawal thresholds and the stable and normal breathing rate. Future experiments should test if awake MRS experiments can be performed using shorter training paradigms as those described in Becerra et al. (50). In addition, the experiments should assess the state of the animal and investigate whether stress was induced or not. Such experiments will benefit from quantifying serum cortisone levels during the imaging procedures, and monitoring the heart rate (23, 24).
MRS acquisitions require that the main magnetic field, B0, remain very stable during data acquisition. The movement of the animal during the data acquisition can significantly change the value of the local B0 over time. B0 fluctuations can broaden the line width of the observed metabolites in the MR spectrum thereby losing fidelity in resolving two closely resonating spins. In a 1H MRS study on awake non-human primates at 7 T (33), line-width changed from 15.6 Hz at the beginning of the study to 19.5 Hz for the averaged spectrum. These changes in the spectral line width were due to major body movement of the awake animal during data acquisition. However, respiration and small body movement did not show a significant spatial B0 dependence and shim effect. To correct for movement, the authors applied a scheme which rejected periods of major body motion and corrected frequency/phase of each of the single acquisition of the spectra before averaging (33). Using this correction scheme the line width of the spectrum was reduced to 15.8 Hz. In our experimental design a motion correction scheme was not performed. The shimming procedure routinely resulted in line widths of 11–15 Hz of the single 1H metabolite resonance (0.037–0.050 ppm) and a good separation of peaks that resonate closely was possible (e.g; glutamate (2.35 ppm) and glutamine (2.45 ppm). Furthermore, repeated acquisition of the spectra over two different days were highly reproducible (CV <11%) suggesting that preparing the animal in this manner leads to negligible motion and provides a stable environment for obtaining quality MRS. However, the quality of the raw spectra acquired can be further improved by performing additional data preprocessing steps as those described in Pfeuffer et al., (33). Correcting for animal movement and frequency/phase of single acquisitions will reduce the minimum detectable differences in metabolite concentrations and increase reproducibility and therefore improve the quality of the acquired spectra.
Rodents are used extensively in neuroscience research to study disease processes that involve cognition and emotion such as pain, addiction, depression and psychiatric disorders. These conditions cannot be studied reliably in the presence of agents that may alter consciousness, cognition, and neuronal transmission. Similarly, MRS has proven very useful in studying the pathophysiologic mechanisms of a large range of neurological and psychiatric disorders (2) and many of these studies are not possible if the animals are not restrained to prevent movement. An awake rodent MRS preparation as demonstrated here allows one to study these conditions and correlate changes in brain metabolic profiles during resting and functional states with changes in neuronal physiology, chemistry, and behavior without the confounds of anesthetic and paralytic agents.
We conclude that high-resolution and high-quality 1H MRS can be obtained from unanesthetized rat brains at 7 T. With proper training and restraint apparatus, several proton metabolites can be reliably measured even without post processing correction schemes. The current development offers a novel approach to study major brain metabolites in awake rodents. This method circumvents the effects of anesthesia, and allows for longitudinal experiments for prolonged periods of time to study progression of disease especially in the field of chronic pain research. It will also open the door for translational research that bridges the gap between animal and human studies.
This research project was supported by a research grant from the National Institute of Neurological Disorders and Stroke (R01-NS069568 to R.M.) and research grant from the Department of Defense (SC090126 to R.M.).
The authors of this paper have no financial or other conflicts of interest to declare.