2.1.

Dynamic Imaging of CSF IR Tracer Influx with the LICOR Pearl Imager

The primary goal was to determine if an IR CSF tracer could be dynamically imaged with the LICOR Pearl IR imaging system, allowing for assessment of tracer movement over the surface of the cerebral convexity as a surrogate measure of glymphatic exchange. The experimental pipeline [shown in Fig. 1(a)] describes the overall experimental approach including intracisternal injection of the tracers and dynamic imaging [Figs. 1(a.i) and 1(a.ii)], followed by ex vivo whole-brain surface imaging [Fig. 1(a.iii)] and tracer quantification in coronal brain slices [Fig. 1(a.iv)]. We also performed immunofluorescent co-staining and imaged with the fluorescent tracer, demonstrating the ability to follow the dynamic imaging of tracer influx with concomitant histological studies [Fig. 1(a.v)].

Fig. 1

Dynamic IR imaging of CSF tracer influx. (a) Schematic of experimental time course with representative images showing injection of dual tracers into the cisterna magna at (i) $t=0$ and dynamic in vivo imaging of IR tracer movement with (ii) a LICOR Pearl Trilogy imaging system through an intact skull, followed by surface imaging of both (iii) IR and fluorescent tracer, and (iv) subsequent imaging of fluorescent tracer and (v) immunofluorescence-probed slices. (b) Representative dynamic time series of IR LICOR images from mice anesthetized with 2% Iso or K/X acquired every 2 min from 20 to 40-min post-tracer infusion. (c) Representative image of the left and right dorsal cortical surface ROIs used for analysis of dynamic transcranial images and (d) quantification of the average cortical fluorescent intensity of dynamic images over time (20 to 40 min). Data are mean ± SEM from $n=10$ mice/group and were analyzed by repeated measures two-way ANOVA followed by Šidák’s post hoc test ($p<0.01$ for time points 1–5, $p<0.001$ for time points 6–12). (e) Representative dorsal and ventral LICOR whole-brain surface images taken following perfusion and removal of the brain from the skull. (f) Comparative analysis of average fluorescence intensity of IR tracer of LICOR dorsal and ventral ex vivo surface images of mice anesthetized with K/X or 2% Iso (**$p<0.0019$; ****$p$ $<0.0001$), data are mean ± SEM from $n=10$ mice/group and were analyzed by two-way ANOVA followed by Šidák’s post hoc test.

To determine whether this dynamic imaging technique could detect differences in perivascular glymphatic exchange, we evaluated well established effects of different anesthesia regimes on glymphatic function. Ketamine/xylazine (K/X) anesthesia modulates noradrenergic signaling and enhances glymphatic function,^5,6,12 while isofluorane (Iso) impairs glymphatic exchange.^5,13 Figure 1(b) and Fig. S1 in the Supplemental Material show representative imaging time series with dynamic tracer movement from $t=20$ to 40 min post-injection. Tracer accumulation was reflected by increased IR tracer intensity imaged over the dorsal brain surface. Cortical regions of interest (ROI) [Fig. 1(c)] were selected, and IR fluorescence intensity was quantified in mice under either Iso or K/X anesthesia [Fig. 1(d)]. Consistent with previous studies measuring tracer movement under Iso or K/X anesthesia,^2,3,5–7,14 significantly higher IR tracer intensity was observed in K/X-treated animals across all timepoints compared to 2% Iso-treated animals [$p<0.01$, repeated measures two-way analysis of variance (ANOVA)]. Ex vivo brain surface images showed higher IR tracer intensity on the ventral surface compared to the dorsal surface in animals anesthetized with both Iso and K/X [Figs. 1(e) and 1(f)], which is consistent with anatomical movement of tracer after cisternal injection.^1,9 Finally, both dorsal and ventral surface images show a significant increase in IR tracer distribution in K/X treated animals compared to 2% Iso treatment ($p<0.01$ and $p<0.0001$, repeated-measures two-way ANOVA, respectively).

2.2.

Increased CSF Tracer Distribution in K/X-Treated Animals Assessed by Conventional Whole-Slice Fluorescence Imaging Following Dynamic IR Imaging

Next, we tested how these dynamic IR and surface imaging results compared to the method of slice-based quantification of fluorescence intensity, which has been a standard technique used to assess glymphatic function. A cocktail of IR carboxylate dye and fluorescent tracer was injected into the cisterna magna, allowing for direct comparison of fluorescence with IR imaging. Stereoscopic fluorescence imaging of the dorsal and ventral brain surfaces showed higher CSF tracer fluorescence on the ventral compared to the dorsal surface ($p<0.0001$), and a significant increase in fluorescence of brains from K/X-treated animals compared to 2% Iso-treated animals [Figs. 2(a) and 2(b), $p<0.0001$, repeated measures two-way ANOVA]. Five coronal brain slices (0.75-mm intervals anterior-posterior) were selected for analysis, and segmented into five ROIs [Figs. 2(c) and 2(d)]. Analysis of the mean fluorescence intensity across the entire brain slice showed that CSF tracer fluorescence was higher in K/X-treated animals compared to those treated with Iso [Figs. 2(e) and 2(f), $p<0.0001$ repeated-measures two-way ANOVA]. When regional analysis was conducted, CSF tracer fluorescence intensity was greater in the cortex ($p<0.0001$), corpus callosum ($p=0.0378$), and subcortical structures ($p=0.0002$) in K/X-treated animals compared to Iso-treated animals. Post hoc analysis revealed that the effect of K/X anesthesia on CSF tracer influx was distributed throughout the slices within the cortex, but differences in subcortical structures were driven by changes in the posterior brain. This post hoc analysis found no significant effect of anesthesia in the corpus callosum across slices. No significant effect of anesthesia was observed on CSF tracer influx into the hippocampus. As an alternative approach, we measured regional changes in threshold-based area coverage of CSF tracer fluorescence (Fig. S2 in the Supplemental Material). While this approach resulted in similar anesthesia-based differences in glymphatic influx, we observed that measurement of mean regional fluorescence resulted in less variable data.

Fig. 2

Fluorescence-based ex vivo imaging following dynamic IR imaging. (a) Stereoscopic fluorescence imaging of CSF tracer (Texas Red-conjugated 3 kD dextran, TR-d3) distribution over the dorsal and ventral brain surface. (b) Comparison of dorsal and ventral surface fluorescence in 2% Iso- or K/X-anesthetized animals. Data are mean ± SEM from $n=8$ mice/group and were analyzed by two-way ANOVA followed by Šidák’s post hoc test. (c) Schematic representation of cortical slice regions #1–5 moving anterior to posterior through the brain. Slice locations are shown as distance from bregma in mm. (d) Coronal ROI across five brain slices: cortex, corpus callosum, hippocampus, and subcortical regions. (e) Representative slice images #1–5 from a 2% Iso-anesthetized (left) and K/X-anesthetized animals (right). (f) Regional average fluorescence intensity of the full slice, cortex, corpus callosum, hippocampus, and subcortical regions between 2% Iso- and K/X-treated animals. Data are mean ± SEM from $n=10$ mice/group and were analyzed by two-way ANOVA followed by Šidák’s post hoc test (*$p<0.05$, **$p<0.01$, ***$p<0.001$, ****$p<0.0001$).

2.3.

Results from Dynamic IR and Whole-Slice Fluorescence Imaging Are Highly Correlated

The effect of anesthesia on glymphatic function assessed by dynamic IR imaging was consistent with prior studies reporting increased glymphatic function in K/X-versus Iso-treated animals.^5,15 Yet it is unclear whether the results of dynamic IR imaging through the skull and over the dorsal brain surface are consistent with the results gained through more conventional slice-based fluorescence imaging. We compared IR and fluorescence intensity values from the final dynamic image taken at 40-min post-injection to the fluorescent intensity of the surface and slice images. Dynamic image fluorescence significantly correlates with IR dorsal and ventral surface images, fluorescent dorsal and ventral surface images, and full slice and region-specific slice fluorescence [Fig. 3(a)]. Plotting individual values of dynamic IR intensity against slice intensity from slices 1, 3, and 5 reveals a significant correlation between IR intensity and full slice fluorescence and cortical regions in slices 1, 3, and 5 [Figs. 3(b)–3(d)]. No significant correlations are seen between the dynamic IR image fluorescence and hippocampal or white matter ROIs, and among the subcortical ROIs, a significant association was observed only posteriorly.

Fig. 3

Association between in vivo dynamic IR imaging, ex vivo surface imaging, and fluorescent slice-based imaging. (a) Correlation table comparing in vivo dynamic IR tracer intensity ($t=40\min$) and ex vivo IR surface imaging with IR and fluorescence (F) surface imaging, and slice-based fluorescence imaging across the full slice or specific ROI. (*$p<0.05$, **$p<0.01$, ***$p<0.001$, ****$p<0.0001$; Pearson’s correlation coefficient reflected in heatmap). (b)–(d) Pearson correlation analysis between the in vivo dynamic LICOR IR intensity ($t=40\min$) and coronal slice fluorescence intensity of the full slice as well as the cortex, corpus callosum, hippocampus, and subcortical regions within slices 1, 3, and 5, respectively (Pearson’s correlation, $r$, and $p$ values as noted).

2.4.

Concurrent Assessment of Histological Immunofluorescence with Dynamic IR Imaging

We next demonstrated the compatibility of dynamic IR imaging of CSF tracer influx using the LICOR Pearl system, slice-based fluorescence imaging of CSF tracer influx, and concurrent histological examination. As shown in Fig. 4, a representative $20\text{-}\mu \mathrm{m}$ coronal brain slice from a K/X-anesthetized animal probed for the astroglial marker glial fibrillary acidic protein (GFAP) by immunofluorescence. Both longer- and shorter-exposure time $20\times$ stitched images are provided. Higher magnification ($40\times$) zoomed images are also provided showing CSF tracer distribution, GFAP immunofluorescence, and 4′,6-diamidino-2-phenylindole (DAPI) counter-labeling across distinct brain regions.

Fig. 4

Assessment of histological outcomes after dynamic IR imaging. At left, representative stitched $20\times$ coronal slice images from a mouse anesthetized with K/X and injected intracisternally with both IR and fluorescence CSF tracer, perfusion-fixed 45-min post-injection, and immunostained for the astroglial marker GFAP. Whole-slice images were acquired at (a) higher and (b) lower exposure times to capture the wide range in fluorescence intensity observed throughout the slice. At right, representative $40\times$ Z-stack images from the (c) dorsal cortex, (d) hippocampus, (e) ventral cortex, and (f) third ventricle.

4.1.

Mice

Wild type C57BL/6 male mice aged 11 weeks were maintained on a 12-h light/12-h dark cycle with food and water ad libitum. All experiments were carried out following the National Institutes of Health (NIH) Guide for Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the University of Washington, Seattle.

4.2.

Intracisternal Tracer Injection

Dual IR (non-fixable) and paraformaldehyde (PFA) fixable fluorescent CSF tracer was prepared with Texas red-dextran (3 kD, TR-d3; Invitrogen) at a concentration of 0.5% and $100\mu \mathrm{M}$ IR dye (LI-COR 800CW carboxylate 100 nM; Fisher Scientific) in artificial CSF. Mice were anesthetized with either 2% Iso or a K/X cocktail (intraperitoneal; 130-mg/kg ketamine, 8.8-mg/kg xylazine, and 0.02-ml/g body weight). Once anesthetized, mice were fixed in a stereotaxic frame, the dura mater was exposed through blunt dissection of neck muscles below the occipital crest, and a 30G steel needle was inserted at 45 deg (relative to the mouse head) 1 to 2 mm into the cisterna magna.³⁶ Once the cannula was secured to the dural membrane with cyanoacrylate glue, $10\mu \mathrm{l}$ of CSF tracer was injected over a 10-min period with a syringe pump (Harvard Apparatus). Following the 10-min injection period, CSF tracer distribution was permitted for an additional 5 min.

4.3.

LICOR Dynamic Imaging

To visualize tracer movement from the subarachnoid space of the cisterna magna into the brain parenchyma, mice were placed on the heated LICOR Pearl imager platform. Animals in the 2% Iso anesthesia treatment group were maintained under 2% Iso, while K/X treated animals received an additional 1/3 dose of K/X by weight 45 min after the first K/X dose to maintain the plane of anesthesia. Transcranial images were taken of the exposed skull surface every 2 min from 20 min post tracer injection to 40 min post tracer injection. Imaging from 20 to 40 min captures tracer influx into the mouse cortex, however, longer imaging timepoints could be used depending on the outcome of interest. In the present study, we sought to conduct perfusion fixation at 45-min post-injection, thus dynamic imaging ended at $t=40\min$ post-injection. Imaging was performed at $85\mu \mathrm{m}/\text{pixel}$, though images can also be acquired at 170 or $255\mu \mathrm{m}/\text{pixel}$.

4.4.

Perfusion and Sectioning

After 40 min of tracer influx and imaging, mice were transcardially perfused with ice-cold phosphate-buffered saline (PBS), followed by 4% PFA. Brains were removed, and images of the dorsal and ventral surfaces of the extracted brain were taken on the LICOR. Brains were then post-fixed overnight in 4% PFA at 4°C, protected from light. Brains were rinsed in PBS and cryoprotected in 30% sucrose until saturated.

4.5.

Ex Vivo Fluorescence Imaging

Once brains are appropriately cryoprotected in 30% sucrose, dorsal and ventral surface fluorescent and bright-field images were acquired using fluorescent stereoscope images. Brains were then fixed in optical coherence tomography and maintained at $-80°\mathrm{C}$ and then cryosectioned in $20\text{-}\mu \mathrm{m}$ coronal sections and mounted with ProLong™ Diamond Antifade Mountant with DAPI. Fluorescent penetration from the cisterna magna was imaged on five ex vivo sections rostral to caudal through the brain with a Keyence BZ-X800 fluorescence scope. Tile scans were taken at $10\times$ magnification, and images were stitched and saved as TIFF files for analysis.

4.6.

Immunofluorescence

Additional representative immunofluorescence was performed on frozen K/X treated $20\text{-}\mu \mathrm{m}$ slices. Slices were PBS washed and blocked for 1 h at room temperature with 2% BSA in PBS-Triton (0.3%) with 5% donkey serum, incubated with primary antibody overnight at 4°C, and incubated with secondary antibody for 2 h at room temperature. Primary and secondary antibodies used for representative staining were rabbit anti-GFAP (1:750, Abcam) and donkey anti-rabbit Alexa Fluor 647 (1:500; Life Technologies).

Once antibody staining was complete, the section was mounted as described above and imaged at $20\times$ magnification with the Keyence microscope. Representative 40× images were also taken within the left hippocampus.

4.7.

Image Processing and Analysis

Stitched fluorescent images were processed in ImageJ. Overlay images were RGB separated, and the DAPI/blue channel was used to manually draw ROI for the cortex, corpus callosum, hippocampus, and sub-cortical regions as shown in Fig. 2(d). These regions were used to calculate the mean raw fluorescence intensity for each region within that section. Additional analysis of slice 4 was performed with manual thresholding to remove background fluorescence to confirm that approaches utilizing raw fluorescence intensity and threshold-based assessment are comparable. This is visually represented in Fig. S2 in the Supplemental Material.

4.8.

Statistical Analysis

All statistical testing was performed on GraphPad Prism 9.2 and was assembled in Adobe Illustrator. Tests were chosen based on the data set being analyzed and are reported in the figure legends. They include repeated-measures two-way ANOVA, Sidak’s multiple comparisons test, and Pearson’s correlation. All statistical testing was two tailed, and exact $p$ values were calculated at a 0.05 level of significance. All values are expressed as the mean ± standard error of the mean (SEM).