Animals
Experiments were performed with 95 female Fisher rats (F344) aged 3 to 5 months (122–203 g, 171 ± 17.5 g body weight). Animals were obtained from Charles River Laboratories (Erkrat, Germany) or bred in the central animal facility of the university of Muenster. Animals received from Charles River were housed at least one week before experiments were performed, animals from the central animal facility were housed at least for one day before experiments. Rats were kept in groups of 2–3 animals under 12-hour light/12-hour dark cycle with ad libitum access to food and water. All surgeries were performed during the light phase to rule out possible effects of the circadian cycle. Experiments were conducted according to the German Tierschutzgesetz and were approved by local authorities (Landesamt für Natur, Umwelt- und Verbraucherschutz Nordrhein-Westfalen, Germany, 84-02.04.2016.A135).
Study Design
Two sets of experiments with different animal preparations were performed. One subset of animals (n = 49) received a contrast agent (CA) administration via cisterna magna and underwent repeated 3D T1 weighted (T1w) whole brain MRI over six hours (Table 1) under six different anesthetic regimens. Data from this group was used to study contrast agent distribution in the brain and to model solute clearance. In some of these animals simultaneous optical Ca2+-recordings were performed. The other group (n = 37) was intubated and a series of MRI examinations were first performed under isoflurane, and then repeatedly after switching to one of the six anesthetic regimens. In this group, size of arteries (Time Of Flight angiography (TOF)), subarachnoidal space volume (T2w MRI), aqueductal flow (DWI), and apparent diffusion coefficient (ADC, diffusion tensor imaging (DTI)) in neocortex were measured (Table 1).
Table 1
anesthetic protocol
|
3D T1w
|
bench
|
T2w MRI/DWI/DTI/TOF
|
|
only
|
with simultaneous Ca2+-recordings
|
Ca2+-recordings
|
|
ISO
|
5
|
3
|
|
6
|
MED
|
4
|
3
|
|
6
|
ISO + MED
|
3
|
5
|
|
6
|
MED + AZE
|
8
|
|
3
|
6
|
ISO + AZE
|
10
|
|
3
|
6
|
ISO + MED + AZE
|
8
|
|
3
|
7
|
Animal preparation was generally performed under isoflurane anesthesia (ISO, see dosing details below) (Forene, Abbott, Wiesbaden, Germany). For MRI, animals were placed in a warmed animal cradle and fixed with bite bar and ear plugs. Animals were supplied with a gas mixture of 75% air and 25% oxygen during MRI. Small animal MRI was performed at 9.4 T in a 94/20 Biospec interfaced to a Bruker Avance console controlled by Paravision 5.1 software (Bruker BioSpin, Ettlingen, Germany). For all experiments, a surface receive three-element array coil (Rapid Biomedical, Rimpar, Germany) was positioned above the head of the animal and inserted in a quadrature volume transmit coil (Bruker). Respiratory rate and rectal temperature were continuously monitored, and kept in the physiological range by adjusting the cradle temperature and respiration volume (for intubated rats only).
Six different anesthetic regimens were applied. For the ISO condition, animals were anesthetized with 1.5–1.7% ISO (1.5% contrast agent injection experiments and 1.7% for intubated animals). The MED condition consisted of continuous subcutaneous (s.c.) Medetomidine (Orion Corporation, Espoo, Finnland) application (40 µg/kg bolus, followed by infusion of 50 µg/kg/h, s.c.). The ISO + MED condition supplemented continuous s.c. MED application with additional ISO (0.8%). All three conditions were performed both with and without intraperitoneal (i.p.) injection of Acetazolamide (AZE, Diamox 500 mg, Vifor SA, Villars-sur-Glane, France, i.p., bolus 55.5 mg/kg of Acetazolamide diluted in 0.9% saline).
Catheter Implantation And Dce
For catheter implantation animals were anesthetized with ISO (initiation 5% / continuous dose 2.5% ISO in 100% oxygen) and received buprenorphine (bolus s.c. 0.05 mg/kg, Temgesic, Reckitt Benckiser Healthcare UK Ltd., Hull, UK) 30 min before surgery. Animals were positioned in a stereotactic frame (Stoelting, Dublin, Ireland) on a feedback-controlled heating pad to keep a temperature of 37°C. Absence of pain was verified by pedal and corneal reflex testing.
The atlanto-occipital membrane was exposed via midline neck incision and blunt removal of neck musculature. Puncture of the atlanto-occipital membrane was performed with a modified standard venous catheter (Terumo, Surflow-W, 26G, Eschborn, Germany) which limited catheter penetration to a depth of 1 mm. The catheter was fixed via tissue glue (Histoacryl, B Braun, Melsungen, Germany) and remaining dead volume in the catheter was cautiously filled with NaCl 0.9%. A PE20 line was attached to the catheter and to a 5 ml syringe in a syringe pump (World Precision Instruments, Sarasota, USA, AL-300). Both contained the MR Contrast agent (CA) Gadobutrol (Gadovist, Bayer, Schering, Leverkusen, Germany, 83 mM) diluted in 0.9% NaCl. Skin incision was sewed to minimize movement of catheter during animal positioning.
After surgery the animal was fixed in prone position in the MRI cradle via custom-made ear plugs and bite bar, and anesthesia was switched to ISO (1.5%). A baseline 3D T1w Fast Low Angle Shot (FLASH) sequence was performed (TE = 3.83 s, TR = 15 s, scan time 3 min 4 s, flip angle = 15°, FOV = 30 x 30 x 32 mm3, spatial resolution = 0.117 x 0.234 x 0.25 mm3, matrix = 256 x 128 x 128). Then, anesthesia was switched to one of the six anesthetic regimens and maintained at least for 40 minutes to ensure habituation to the respective anesthetic condition. Either 80 µl 21 mM Gd-BT-DO3A, or 20 µl 83 mM Gd-BT-DO3A were delivered intrathecally via syringe pump at an infusion rate of 1.6 µl or 0.4 µl per minute, respectively (total infusion time of 50 minutes). 3D T1w FLASH scans of the whole brain were continuously acquired during infusion and during the following 5 hours, resulting in a total of 40 scans.
Calcium recordings
Twenty animals (Table 1) additionally received an optic fiber implantation to record calcium release as a surrogate parameter for the brain state, which was measured during the MRI experiments in a subgroup (n = 11). For conditions containing AZE, Ca2+-recordings were performed on the bench (n = 9). These animals had received an intracranial injection of the viral construct encoding for the calcium indicator GCaMP6f (pAAV.Syn.GCaMP6f.WPRE.SV40 was a gift from Douglas Kim & GENIE Project (Addgene plasmid # 100837) at least 4 weeks earlier, as described previously (33). Briefly, rats were anesthetized with Isoflurane (2.5%) and received the analgesic buprenorphine (bolus s.c. 0.05 mg/kg). After fixation in a stereotactic frame via ear and bite bars the skull was exposed and a small craniotomy was accomplished via a dental drill (Ultimate XL-F, NSK, Trier Germany, and VS1/4HP/005, Meisinger, Neuss, Germany). 1 µl AAV1.Syn.GCaMP6f.WPRE.SV40 was injected via a glass capillary into S1FL (anterio-posterior (AP) 0.0 mm, medio-lateral (ML) + 3.0 mm, dorso-ventral (DV) -1.2 mm) at an 35° angle from medial. At the day of DCE a craniotomy (AP + 0.2 mm, ML + 3.3 mm) was performed as described before and a 200 µm optic fiber (Thorlabs, Newton, NJ, USA) was implanted perpendicular to the dura at DV -300 µm, above the GCaMP6f expressing region (33). After verifying detection of calcium signal upon spontaneous brain activity in the stereotactic frame, the fiber was fixed to the skull with UV glue (Polytec, PT GmbH, Waldbrunn, Germany). The amount of the UV glue was reduced to a minimum to avoid MR image distortions.
Anatomical, Tof, And Diffusion Mri
Following anesthetic induction (see above), animals shortly (minutes) received 3.5% ISO in 100% oxygen to ensure full relaxation of the animal during intubation. Subsequently, the animal was placed in the MRI cradle with the head fixed as described above and the tracheal tubus (Introcan Safety-W, 14G, B Braun Melsungen, Germany) was connected to a ventilator (MRI-1 Ventilator, CWE Inc., Ardmore, PA, USA, 1.8–2.2 mL tidal volume, 53 breaths per minute) delivering 1.7% ISO in a mixture of 75% air and 25% oxygen. CO2 concentration in the expired air was analyzed continuously (CapStar-100 CO2 Analyzer, CWE Inc., Ardmore, PA, USA) and kept stable by adjusting tidal volume. After positioning the animal in the scanner and performing localizer scans, shimming was performed using MAPSHIM (Bruker). Four different scan protocols were run under ISO anesthesia:
1. TOF to assess vascular volume (3D T1wFLASH, TE = 2.58 s, TR = 1.5 s, scan time = 12 min 17 s, flip angle = 30°, FOV = 26 x 26 x 28 mm3, spatial resolution = 102 x 102 x 109 µm3, matrix = 256 x 256 x 256),
2. T2w Rapid Acquisition with Relaxation Enhancement (RARE) to measure the size of the subarachnoidal space (TE = 9 ms, RARE factor = 16, TR = 1500 ms, scan time = 6 min 24 s, flip angle = 180°, FOV = 30 x 30 x 8 mm3, spatial resolution = 117 x 234 x 250 µm3, matrix = 256 x 128 x 32),
3. DWI to estimate aqueductal flow (TE = 30.045 ms, TR = 5000.001 s, scan time = 23 min 20 s, b-value = 1000 s/mm2, FOV = 34 x 28 mm2, spatial resolution = 133 x 109 µm2, matrix = 256 x 256 x 20)
4. Axial DTI, to measure the apparent diffusion coefficient in neocortex (DTI-EPI, TE = 42.8 ms, TR = 4250 ms, scan time = 7 min 56 s, b-values = [10 200 400 600 800 1000 1200 1400 1500] s/mm2, segments = 4, FOV = 30 x 30 mm2, spatial resolution = 0.156 x 0.156 mm2, slice thickness = 1 mm, matrix = 192 x 192 x 17).
Subsequently, anesthesia was switched to one of the six anesthetic conditions. After a waiting period of 40 minutes, all scan protocols were acquired again.
Postprocessing
For each experiment, brains were segmented from 3D T1w FLASH, and all images were registered to the initially acquired scan of its measurement series, to correct for animal movement. Next, data were registered on the Waxholm Space rat atlas (34) with 79 brain regions (from the 76 brain regions of the publication the central canal is omitted and the hippocampal formation is divided into cornu ammonis regions 1–3, the fasciola cinereum, and the dentate gyrus) via ANTS (Advanced Normalization ToolS (RRID:SCR_004757)) (35). Mean signal intensities were retrieved for each brain region at each time point. Time signal curves (TSC), showing relative signal changes from baseline (BL) for each brain region were calculated by region-wise normalization to the mean intensity in the pre-contrast scan. In order to quantitatively characterize contrast agent distribution, the following parameters were calculated region-wise for statistical analysis and voxel-wise for 3D visualization using MATLAB (MATLAB 2021a, The MathWorks, Inc., Natick, Massachusetts, United States). Data are presented both region-wise and averaged over all brain regions: 1. arrival time ta, in accordance with previous work (36), was estimated by finding the time point where signal increased for at least three successive measurements, 2. maximum signal and time to maximum tmax, 3. area under the curve (AUC) was calculated as the integral of the TSC from a scan before CA injection (t = 0) to 6 h after injection, as
\(AUC={\int }_{0}^{6h}TSC\left(t\right)dt\) (1),
4. Signal decay rate b was calculated by fitting a mono-exponential decay (exp) to the TSC
\(exp=\text{a}\cdot {e}^{-\text{b}\cdot t}\) (2),
where a is an amplitude factor. For region-wise analysis, MR signal was averaged over each of the 79 brain regions and fitting was performed over the time period between 90 minutes after maximum of the specific TSC, tmax, and end of data acquisition. Data were discarded if tmax was reached less than 90 minutes before the end of data acquisition, or if b was not positive.
ADC maps from scans before and after switch of anesthesia condition were calculated from DTI data using Paravision 5.1 by applying a mono-exponential fit. For ROI-analysis neocortex was manually segmented in one axial slice, matching the region selected for Ca2+-recordings. Similarly, for aqueductal flow, ADC maps were calculated from DWI data and aqueduct was manually segmented in the slice depicting the aqueduct. In order to determine the size of the subarachnoidal space, the basal cistern was manually segmented in axial T2w scans before and after anesthetic switch (Fig. 1). Resulting values were divided by the initial value (ISO condition), to reveal relative changes from BL.
Amplitude Power Spectra Of Calcium Traces
High resolution calcium traces (sampling rate 2 kHz) were analyzed using a customized MATLAB script. First data were smoothed using a lowpass filter (sampling rate 60 Hz; upper bound 200 Hz; lower bound 20 Hz; damping 40 Hz). Every other six minutes, a two-minutes trace was extracted from the original data. For each of these cropped traces a power spectrum analysis was performed. The power spectra of the frequency ranges 0.1–1 Hz, 1–4 Hz and 1–20 Hz were summed up.
Mechanism-independent Two-compartment Model
We used a two-compartment model to quantify efficiency of solute clearance. We choose neocortex (as defined by the Waxholm Rat atlas) because of the decent size of this brain region and the fact, that Ca2+-recordings, ADC data and TSCs were available from this brain region. Neocortex was subdivided into two compartments, one coherent outer compartment that completely surrounded one inner compartment.
The following differential equation describes the model, where Iin is the signal of the inner compartment and Iout is the signal of the outer compartment. k1 and k2 denote the exchange rates between these compartments.
$$\frac{d}{dt}{I}_{in}\left(t\right)={k}_{1}\cdot {I}_{out}\left(t\right)-{k}_{2}\cdot {I}_{in}\left(t\right)$$
3
The differential equation was solved via Laplace transformation.
$${I}_{in}\left(t\right)={k}_{1}\cdot \left[{e}^{-{k}_{2}\cdot t}*{I}_{out}\right(t\left)\right]={k}_{1}\cdot {\int }_{-\infty }^{\infty }{e}^{-{k}_{2}\cdot t}\cdot {I}_{out}(t-l)dl$$
4
where \(*\) represents a convolution. Since Iin and Iout are given by the measured TSCs from the inner and outer compartment, two unknown parameters \({k}_{1},{k}_{2}\) remain. Optimal parameters \({k}_{1},{k}_{2}\) in \(\mathbb{R}\) were found by minimizing the difference between the true Iin from the TSC and the Laplace transform of Eq. (3).
$$optimal({k}_{1},{k}_{2})=\underset{{k}_{1},{k}_{2}\in \mathbb{R}}{min}\parallel {I}_{in}\left(t\right)-{k}_{1}\cdot \left[{e}^{-{k}_{2}\cdot t}*{I}_{out}\right(t\left)\right]{\parallel }_{{L}^{2}}$$
5
To find the optimal size of the inner and outer compartment we performed this procedure for each animal by stepwise changing the thickness from 1 to 17 voxels of the outer compartment and calculated the difference between the true Iin from the TSC and the optimal one from Eq. (4), denoted as \({I}_{in}^{\text{s}\text{o}\text{l}}\left(t\right)\), for each thickness.
$$optimal s=\underset{s\in 1,\dots ,18}{min}\parallel {I}_{in}\left(t\right)-{I}_{in}^{\text{s}\text{o}\text{l}}\left(t\right){\parallel }_{{L}^{2}}$$
6
Via multiplication of the TSC with the exchange rates, the exchanged volumes at every time point were calculated.
$${k}_{1}\cdot {I}_{out}\left(t\right), {k}_{2}\cdot {I}_{in}\left(t\right)$$
7
Since it was assumed that \({k}_{1},{k}_{2}\) do not change over time, these functions are multiples of \({I}_{out}\left(t\right) and {I}_{in}\left(t\right)\). The total volumes are defined by their area under the curves:
$${V}_{out}={\int }_{0}^{end of experiment}{I}_{out}\left(t\right)dt$$
8
$${V}_{in}={\int }_{0}^{end of experiment}{I}_{in}\left(t\right)dt$$
9
Consequently, the exchanged volumes during the total experiment over time are:
$${V}_{out\to in}={\int }_{0}^{end of experiment}{k}_{1}{I}_{out}\left(t\right)dt={k}_{1}\cdot {V}_{out}$$
10
$${V}_{in\to out}={\int }_{0}^{end of experiment}{k}_{1}{I}_{in}\left(t\right)dt={k}_{1}\cdot {V}_{in}$$
11
With \({V}_{out\to in}\) being the volume that is exchanged from the outer to the inner compartment and \({V}_{in\to out}\) the volume that is exchanged from the inner to the outer compartment.
The net exchange ratio (NER) is defined by the ratio of \({V}_{out\to in}\) and \({V}_{in\to out}\),.
$$NER=\frac{{V}_{in\to out}}{{V}_{out\to in}}$$
12
Statistical analysis
Continuous variables were reported as mean ± standard deviation (SD) when normally distributed and as medians (interquartile range) otherwise. Normal distribution was assessed using a Kolmogorov-Smirnov test. Comparisons between groups were performed for continuous variables using a 2-tailed unpaired Student’s t-test or a rank-sum test depending on normality. Hypothesis testing was two-tailed. All p-values < 0.05 were considered statistically significant. Data were presented as boxplots: the central line indicates median, the central mark indicates mean, bottom and top edges of the box indicate the 25th and 75th percentiles, respectively, whiskers indicate the 5th and 95th percentiles. Significant differences are marked with one asterisk if p < 0.05, two asterisks if p < 0.01 and three if p < 0.001. All statistical analyses were performed using MATLAB 2020b.