Conscious rat PET imaging with soft immobilization for quantitation of brain functions: comprehensive assessment of anesthesia effects on cerebral blood flow and metabolism

Background Animal brain functions evaluated by in vivo imaging under anesthesia can be affected by anesthetic agents, resulting in incorrect assessment of physiological brain function. We therefore performed dynamic positron emission tomography (PET) imaging of conscious rats using recently reported soft immobilization to validate the efficacy of the immobilization for brain function assessments. We also determined the effects of six anesthetic agents—a mixed anesthetic agent (MMB), ketamine + xylazine (KX), chloral hydrate (Chloral), pentobarbital (PTB), propofol (PF), and isoflurane (IFL)—on brain function by comparison with conscious rats. Results The immobilization enabled 45-min dynamic [18F]FDG-PET acquisition with arterial blood sampling using conscious rats without the use of special techniques or invasive surgery. The spatial resolution and quantitativity of [18F]FDG-PET were not significantly lower for conscious rats than for anesthetized rats. While MMB, Chloral, PTB, and PF showed ubiquitous reduction in the cerebral metabolic rates of glucose (CMRglu) in brain regions, KX and IFL showed higher reductions in cerebellum and interbrain, and cerebellum, respectively. Cerebral blood flow (CBF) was reduced by MMB, KX, PTB, and PF; increased by IFL; and unaltered by Chloral. The magnitude of decrease in CMRglu and CBF for MMB were not larger than for other five anesthetic agents, although blood glucose levels and body temperature can be easily affected by MMB. Conclusion The six anesthetic agents induced various effects on CMRglu and CBF. The immobilization technique presented here is a promising tool for noninvasive brain functional imaging using conscious rats to avoid the effects of anesthetic agents. Supplementary Information The online version contains supplementary material available at 10.1186/s13550-021-00787-6.

by animal movement during the scan. Anesthetic agents have various significant effects on brain function, including cerebral metabolism [1] and neural activities [2]. Therefore, animal brain functions that are evaluated by in vivo imaging under anesthesia could be affected by anesthetic agents, resulting in incorrect assessment of brain functions under physiological conditions.
To evaluate animal brain functions without the effect of anesthesia, several imaging techniques have been reported. In one approach, PET/SPECT imaging for estimating conscious animal brain functions, conscious animals are administrated radiotracers and kept conscious during tracer distribution [3,4]. At the appropriate time [~ 40-60 min for 2-deoxy-2-[ 18 F]fluoroglucose ([ 18 F]FDG)] post-injection, animals are anesthetized and a static PET/SPECT scan is performed under anesthesia. This approach is applicable only in limited probes that show fast distribution according to the brain functions and stabilize their concentration in the regions of interest (ROIs) that are not affected by anesthetic agents during the scanning. Another approach for imaging conscious animals is using a head holder to prevent head movement [5][6][7]. Because head holders are often directly attached to animal skulls by invasive surgical procedures, the effects on the animals' physiological condition from attachment of the head holder should be considered. Additionally, the gamma-ray scattering and attenuation caused by the head holders should be considered. Reconstruction techniques for free moving animals with motion correction using motion tracking systems have also been reported [8][9][10]. However, these techniques have not been able to be widely used yet, since they require additional hardware and/or complicated software. Recently, soft immobilization for conscious mouse MRI scanning has been reported [11]. Because this immobilization, which uses a simple soft restrainer, suppresses the head movements of conscious mice for at least 30 min without surgery or a metallic apparatus [11], it could be suitable for dynamic PET scanning.
In this study, we assessed the cerebral metabolic rate of glucose (CMR glu ) and cerebral blood flow (CBF) in conscious rats using the soft immobilization technique, and validated its efficacy for brain function imaging of conscious rats. In addition, we also determined the effects of six anesthetic agents on CMR glu and CBF by comparison with conscious rats.

General
Fluorine-18 was produced by 18

Animal preparation
Rats aged 8-9 weeks old were used for the experiments, and their weights ranged from 233.8 to 339.4 g (277.2 ± 24.4 g). Rats (n = 6 or 7) were fasted for 15 h from the evening before and then anesthetized with the anesthetic agents described below. A polyethylene catheter (i.d. 0.5 mm, o.d. 0.8 mm) was inserted into the femoral artery for blood sampling and filled with heparin saline solution (10 IU/mL), and an indwelling needle (Surflo I.V. Catheter 24 G, SR-OT2419C, Terumo Co., Tokyo, Japan) was inserted into the tail vein for the administration of radiotracers and continuous infusion of propofol. These cannulas were in place until euthanasia. Body temperature was maintained at approximately 37 °C by a heating pad during the preparation. Rats were restrained on the bed and a head CT scan was performed using a small animal PET/SPECT/CT system (FLEX; Gamma Medica Ideas, Northridge, CA, USA). Blood gas profiles and glucose levels in arterial blood were measured using an i-STAT analyzer with an i-STAT CG8 + cartridge test (Abbot Point of Care, Illinois, USA) 5 min before [ 18 F]FDG injection.

Handmade restrainer
Two square pieces of denim fabric (200 mm × 200 mm and 160 mm × 160 mm) were sewn together and folded to make a hood. A small cut was made in the hood to create an air vent. Several hook-and-loop fasteners were attached to hold the head, neck, and abdomen of the rats (Fig. 1a).

Anesthetic conditions
Conscious (CONS): To allow the rats to acclimate to the restrainer, the restrainer was applied before being used for the experiments, as described below. Rats were anesthetized with an inhalation of 3% isoflurane (ISOFLURANE Inhalation Solution, Mylan, PA, USA) in 50% oxygen (1 L/min), and wrapped in the handmade restrainer (Fig. 1b) within 5 min. The rats were then restrained on a bed for 75 min without isoflurane inhalation. This acclimation was repeated once daily every weekday at least nine times.
On the experiment day, rats were anesthetized with an inhalation of 3% isoflurane in 50% oxygen (1 L/min) during arterial and venous cannulation (< 30 min), and lidocaine hydrochloride (Xylocaine Jelly 2%, Aspen Japan KK, Tokyo, Japan) was applied to the incision. After cannulation, rats were wrapped in the restrainer, and restrained for 30 min to induce emergence from isoflurane.

Corticosterone assay
To monitor the stress caused by restraining and/or blood sampling during PET acquisition, plasma corticosterone levels were measured. Arterial blood (200 µL) was collected for corticosterone assay from conscious rats and MMB-and IFL-anesthetized rats (before [ 18 F]FDG injection, n = 4). Five-minutes later, these rats were intravenously injected with 0.5 mL of saline instead of [ 18 F] FDG, and arterial blood sampling was performed as described above. After arterial blood sampling at 2700 s post-injection, additional arterial blood (200 µL) was collected for the corticosterone assay (after [ 18 F]FDG-PET). Next, 0.5 mL of saline instead of [ 125 I]IMP was intravenously injected and arterial blood sampling was performed as described above. After arterial blood sampling at 300 s post-injection, additional arterial blood (200 µL) was collected for the corticosterone assay (after [ 125 I] IMP assay). Arterial blood samples for the corticosterone assay were centrifuged at 1500×g at 4 °C for 5 min to prepare plasma, which was stored at − 80 °C until the assay. The plasma corticosterone levels were measured using an ELISA kit (Enzo Life Science, ADI-900-097, PA).

Statistical analysis
Data are expressed as mean ± SD. The statistical significance of the differences between the conscious and anesthetized rats was determined using non-repeated ANOVA with Dunnett's multiple comparison test. The statistical significance of the differences in plasma corticosterone levels between before [ 18 F]FDG injection, after [ 18 F]FDG-PET, and after [ 125 I]IMP assay was determined using repeated ANOVA with Tukey's multiple comparison test. P values of less than 0.05 were considered statistically significant. The association between SUV determined by PET and that determined by γ counter was calculated by Pearson's product moment correlation coefficient. Differences between regression coefficients of the regression lines were determined by testing the t-value. The association between CMR glu and brain SUV or BPG were also calculated by Pearson's product moment correlation coefficient.

Animal condition during experiments and quality of PET data of conscious rats
After acclimation to the restrainer, almost all (six in seven rats) conscious rats remained stable in the restrainer for at least 75 min. During PET/CT acquisition, the rats also remained stable without noticeable head movement, resulting in sufficient spatial resolution (Fig. 2).
The quantitated brain uptake value of the last frame of [ 18 F]FDG-PET (44-45 min post-injection) was well correlated with the [ 18 F]FDG uptake determined using the γ counter after resection (~ 55 min post-injection) (Fig. 3a , y = 0.68 x + 0.97, R 2 = 0.93, P < 0.01). The regression coefficient based on the correlation between conscious rat brain SUV determined by PET and by γ counter was not significantly different from that of anesthetized rats (Fig. 3, t = 0.023, P = 0.98).
The blood glucose levels and plasma corticosterone levels are summarized in Table 1. The arterial blood gas profiles are summarized in Additional file 1: Table S1. MMB-or KX-anesthetized rats showed significantly higher blood glucose levels than conscious rats (P < 0.01). The plasma corticosterone levels of conscious rats just before [ 18 F]FDG injection were 10.4 ± 5.0 ng/mL (n = 4), and were not significantly different from those in MMBand IFL-anesthetized rats (P > 0.05). The plasma corticosterone in conscious and MMB-and IFL-anesthetized rats remained low levels (< 20 ng/mL) during the 45-min [ 18 F] FDG-PET acquisition and [ 125 I]IMP assay, while slight increase in plasma corticosterone levels in conscious rats were observed during the 45-min [ 18 F]FDG-PET acquisition (P < 0.05).

[ 18 F]FDG kinetics in conscious and anesthetized rats
Typical [ 18 F]FDG-PET images and the brain uptake values of conscious and anesthetized rats are summarized in Fig. 2. The whole-brain SUV of the IFL-anesthetized rats was significantly lower than that of the conscious rats (P < 0.05), and PET images showed marked reduction of [ 18 F]FDG uptake in cerebrum and interbrain regions.
There was a trend towards a small ubiquitous reduction in [ 18 F]FDG in MMB-and Chloral-anesthetized rat brains (P > 0.05). KX-anesthetized rats showed low [ 18 F] FDG uptake in cerebellum. The [ 18 F]FDG distribution patterns in PTB-and PF-anesthetized rat brains were similar to those in conscious rat brains.
Additional file 1: Fig. S1 shows typical plasma and whole brain time activity curves (TACs) of [ 18 F]FDG in conscious and anesthetized rats. MMB-and KX-anesthetized rats tended to show slow [ 18 F]FDG clearance from the blood. The plasma SUV at 45 min post-injection in MMB-and KX-anesthetized rats were 1.39 ± 0.20 and 1.34 ± 0.34, respectively, and were significantly higher than those in conscious rats (0.84 ± 0.10, P < 0.01). Table 2 summarizes the [ 18 F]FDG rate constants (K 1 , k 2 , k 3 , and k 4 ) and the CMR glu determined by TAC analysis. Although the K 1 , k 2 , and k 4 values of the anesthetized rats were not significantly different from those of conscious rats (P > 0.05), there was a weak negative correlation between blood glucose levels and [ 18 F]FDG K 1 values (Additional file 1: Fig. S2). [ 18 F]FDG phosphorylation reaction constants (k 3 ) in all anesthetized rats were significantly lower than in conscious rats. The CMR glu in conscious rats was higher than in the anesthetized rats (P < 0.05).
Regional CMR glu (rCMR glu ) is summarized in Fig. 5a. There was a ubiquitous reduction in rCMR glu of MMB-, Chloral-, PTB-, and PF-anesthetized rats. KX-anesthetized rats showed highly reduced rCMR glu in the cerebellum (P < 0.01) and interbrain (thalamus and hypothalamus, P < 0.01) compared with conscious rats. The reduction in cerebellum rCMR glu in IFL-anesthetized rats was comparatively smaller than in other regions (P < 0.05 compared with cortex and thalamus). Table 3 shows the CBF determined using [ 125 I]IMP. Conscious rats and Chloral-anesthetized rats showed similar CBF values. MMB-, KX-, PTB-, and PF-anesthetized rats showed lower CBF rates compared with conscious rats (P < 0.01). The CBF was higher in IF-anesthetized rats than in conscious rats (P < 0.01).

Discussion
In the present study, the efficacy of the PET acquisition procedure using soft immobilization [11] for conscious rats (Fig. 1) was validated. This immobilization technique enabled 45-min dynamic PET acquisition of conscious rats without invasive surgery or the use of special techniques. The majority of rats were successfully habituated to the restrainer with simple immobilization procedures before the experiments. Only a few rats could not keep stable for 75 min in the restrainer. The plasma corticosterone levels, a stress marker for rodents, in conscious rats at just before [ 18 F]FDG injection were not significantly different from those in MMB-and IFL-anesthetized rats, and were within the normal range [19]. Although the plasma corticosterone levels in conscious rats were slightly increased during  Table 1 Blood glucose levels and plasma corticosterone levels of conscious and anesthetized rats Data are expressed as mean ± SD (n = 6-7 for blood glucose levels, and n = 4 for plasma corticosterone levels). *P < 0.05 and **P < 0.01 compared with conscious rats as determined using ANOVA with Dunnett's multiple comparison test. † P < 0.05 and † † P < 0.01 compared with before [ 18  I]IMP assay indicated that physical stress induced in our protocol was negligible. The brain PET images of conscious rats showed sufficient spatial resolution (Fig. 2). The brain SUV calculated from PET images (44−45 min post-injection) of conscious rats correlated well with that determined by γ counter after resection (~55 min post-injection), although the γ counter SUVs showed 1.1-times higher values than the PET images due to the difference in timing between PET and resection and/or attenuation (Fig. 3a). The regression coefficients of the regression lines of the conscious rats were not significantly different from that of the anesthetized rats (Fig. 3b), indicating that the effects of motion artifacts, and the attenuation and scattering by the restrainer were negligible. These results demonstrate that this conscious rat PET acquisition procedure using a soft restrainer could be widely applicable to neuronal PET imaging using various PET probes. The CMR glu in KX-, Chloral-, PTB-, PF-, and IFLanesthetized rats was lower than that in conscious rats (Table 2). While the CMR glu in the cerebellum of KX-anesthetized rats was markedly reduced, there was only moderate reduction of CMR glu in the frontal cortex. Chloral, PTB, and PF ubiquitously reduced the CMR glu in every region in the brain. Although IFL dramatically reduced the CMR glu in the cerebrum and the interbrain regions, the reduction in CMR glu in the cerebellum was slight (Fig. 5a). These findings are in good agreement with the previous reports [1,5,9,23]. MMB is a mixed anesthetized agents that consists of an α 2 adrenergic receptor agonist, a benzodiazepinetype GABA A receptor modulator, and a κ-opiate receptor agonist [12]. MMB, as well as the other anesthetic agents, reduced rat CMR glu (Table 2), and the reduction was distributed ubiquitously (Fig. 5a). Although the mechanism of the region-specific reduction in CMR glu  The effects of anesthetic agents on cerebral blood flow and metabolism. a The change in regional cerebral metabolic rates of glucose (rCMR glu ) by anesthetic agents, calculated from [ 18 F]FDG-dynamic PET. b The change in regional cerebral blood flow (rCBF) by anesthetic agents, determined using [ 125 I]IMP. The values are expressed as ratio to values of conscious rats (mean ± SD, n = 6-7) The CBF in MMB-as well as KX-, PTB-, and PF-anesthetized rats was also lower than that in conscious rats (Table 3), which could be coupled with CMR glu reduction. In contrast, the CBF in Chloral-anesthetized rats was not significantly different from that in conscious rats, and IFL increased the CBF compared with conscious rats, resulting in uncoupling of CMR glu and CBF (Tables 2, 3). These findings are consistent with those of previous reports [23,26]. Uncoupling of CMR glu and CBF in IFL-anesthetized rats might be intricately regulated by the CMR glu decrease caused by an increase in cerebral lactate [27] and CBF increase caused by the vasodilation effect of IFL [28]. MMB could be effective for anesthetized animal neuroimaging because the effects of MMB on CMR glu and CBF were comparatively small compared with those of the other anesthetized agents. However, MMB easily decreased the body temperature, probably due to the effects of medetomidine on central thermoregulation [29]. Because the CMR glu and CBF were significantly lower in hypothermic rats than in normothermic rats (data not shown), the body temperature of MMBanesthetized animals should be carefully controlled.
Anesthetic agents affect not only brain functions but also the whole-body physiological condition. For example, MMB-and KX-anesthetized rats showed higher blood glucose levels than conscious rats (Table 1) even after overnight fasting, which was reported to reduce the effects of anesthesia on blood glucose levels [30,31]. This effect was possibly caused by the inhibition of insulin release on pancreatic islet cells by medetomidine and xylazine (α 2 adrenergic receptor agonists) [32]. PET probe uptake in the brain, indicated by SUV, could be influenced not only by the brain functions but also by the whole-body physiologically condition. In the present study, brain SUV at 45 min post-injection could not accurately assess cerebral glucose metabolism (Fig 4a), which could be caused by delayed [ 18 F]FDG clearance from blood circulation. These findings show that kinetic analysis with dynamic PET scan is generally necessary for the quantification of brain function in both anesthetized and conscious animals. Therefore, the soft immobilization technique presented here, which enables dynamic PET acquisition with/without arterial blood sampling, is a promising tool for noninvasive brain functional imaging studies using conscious rodents. On the other hand, BPG, which can be measured by static PET acquisition, was found to be well correlated with CMR glu (Fig. 4b), suggesting its potential for utilization as an index of the brain glucose metabolic rate. Further studies are necessary to establish the applicability of BPG for the assessment of cerebral glucose metabolism.

Conclusions
We have demonstrated the efficacy of soft immobilization for neuro-imaging of conscious rats. The facile and noninvasive technique enables dynamic PET acquisition without loss of spatial resolution and quantification, indicating this technique's potential for neurofunctional imaging in conscious rodents.