Increased capillary stalling is associated with endothelial glycocalyx loss in subcortical vascular dementia

Animals

All animal preparations and experiments were conducted with permission from the Institutional Animal Care and Use Committee of Korea Advanced Institute of Science and Technology (KA2018-04) in accordance with the Animal Protection Act of Korea (Act No.15278). All experiments were reported in accordance with the ARRIVE (Animal Research: Reporting in vivo Experiments) guidelines (https://www.nc3rs.org.uk/arrive-guidelines). ³⁰ A total of 108 male C57BL/6J mice aged 10–12 weeks (25–29 g) were used in this study.

Injection of the cEG degrading enzyme

Hyaluronidase was used to degrade the hyaluronan of the cEG and heparinase I and III were used to effectively degrade heparan sulfate regardless of its sulfation. Heparinase I (H2519, Merck, 0.1 unit/μL, 140 unit/kg), heparinase III (H8891, Merck, 0.05 unit/μL, 1.5 unit per mouse), and hyaluronidase (H3884, Merck, 50 mg dissolved in 750 μL saline, 5 mg per mouse) were dissolved in saline and mixed before injection.^17,31,32 Sterile saline was used for experiment control. To deliver the mixed enzymes effectively to the brain, the enzymes were infused intra-arterially under urethane anesthesia (1.5 mg/kg, intraperitoneally (i.p.)). ³³ Details of the intra-arterial injection are in the Supplementary Information. Enzyme or saline was circulated for an hour. Body temperature, SO₂, heart rate, and respiratory rate were measured (75-1501, Harvard Apparatus).

Ex vivo and in vivo capillary stalling observation

To quantify capillary stalling ex vivo, 4 μm diameter polystyrene microspheres (S37225, ThermoFisher) coated with DiI dye were injected through the tail vein under urethane anesthesia (1.5 mg/kg, i.p.) as previously described. ⁴ The microspheres were circulated for an hour after which 20 mL of phosphate-buffered saline was perfused followed by 20 mL of 4% paraformaldehyde. The harvested brain was sliced at 30 μm thickness. The remaining DiI-coated microspheres were counted and defined as the number of capillary stalls.

To quantify capillary stalling in vivo, DiI-coated microsphere or rhodamine 6G (R6G) (83697, Merck) dye was utilized. DiI-coated microspheres were circulated for an hour before image acquisition by TPM with an 800 nm laser. Images were acquired from pial surface to 300 μm depth with 10 μm step size with a 420 × 420 μm²-field of view (FOV). To identify the cause of capillary stalling in vivo, R6G, which labels leukocytes and platelets, was utilized. ³⁴ Texas red conjugated dextran 70 kDa (D1830, ThermoFisher) was used to label blood plasma. The emission bandpass filter for Texas red conjugated dextran was 605–625 nm. R6G was injected 20 minutes before imaging by TPM with an 850 nm excitation laser. The emission bandpass filter for R6G was 506–534 nm. Repeated volume imaging was performed from pial surface to 200 μm depth with 5 μm step size with 280.56 × 280.56 μm ² FOV for an hour. R6G dye accumulates in the mitochondria of cells. ³⁵ Therefore, leukocytes and platelets were labeled with R6G dye but not red blood cells (RBCs). To identify cell types of R6G-positive cells, 50 µl of Hoechst 33342 (62249, ThermoFisher) was infused intravenously with R6G dye. The emission bandpass filter for Hoeschst 33342 was 455–500 nm. Only the leukocytes were co-labeled with Hoechst 33342 and R6G. Therefore, it was possible to classify the cause of capillary stalling. The time for one volume of angiogram was 351.4 seconds. Stalled capillaries were classified by their cause (stalled capillary by leukocyte plugging or RBCs plugging).

Cranial window surgery

To acquire cEG images 8 weeks after SVaD induction, mice were anesthetized by intraperitoneal injection of urethane (1.5 mg/kg) and their left somatosensory cortex was exposed and sealed with 1.5% agarose solution as previously described. ³⁶ For capillary stalling observation in SVaD, a chronic cranial window was implanted for which the mice were anesthetized with a mixture of Zoletil (30 mg/kg) and Xylazine (10 mg/kg) via intraperitoneal injection. After left somatosensory cortex exposure, sandwich cover glass (stacked 5 mm and 3 mm diameter glass) and head-plate were implanted at 2 weeks before imaging as previously described. ³⁷

SVaD induction

To induce SVaD, mice were anesthetized with a mixture of Zoletil (30 mg/kg) and xylazine (10 mg/kg) via intraperitoneal injection and placed on a heat blanket to maintain body temperature at 37 °C. A spiral microcoil with an internal diameter of 0.18 mm (Sawane Spring Co.) was placed at the right common carotid artery (CCA), and an ameroid constrictor with an inner diameter of 0.5 mm (Research Instrument SW) was placed at the left CCA (Supplementary Figure 1A). ³⁸ Sham surgery was performed for the control mice.

Capillary isolation and mRNA sequencing

Capillaries in the cerebral cortex were isolated using dextran solution. ³⁹ After anesthesia with urethane (1.5 mg/kg. i.p.), mice were decapitated without blood washout, and the brains were stored in ice-cold MCDB131 media (10372019, ThermoFisher). The cortexes from two mice were pooled and were dissected for capillary isolation; further details for capillary isolation can be found in the Supplementary Information. RNA was isolated from the isolated capillary pellet and high-throughput sequencing was performed using NovaSeq 6000 (Illumina); further details for mRNA sequencing can be found in Supplementary Information.

OCTA and D-OCT imaging

A swept-source OCT system with a center wavelength of 1.3  μ m and an A-scan rate of 240 kHz was used. ⁴⁰ OCTA was performed with a FOV of 1 × 1 mm. Serial OCT angiograms were continuously acquired every 7 seconds for 91 seconds to observe temporal changes in microcirculation. D-OCT imaging was performed to measure arterial flow. A single pial arteriole directly feeding the region of FOV. The vessel was repeatedly scanned across its width along a 250- μ m line for 30 seconds at a temporal resolution of 0.06 seconds. The corresponding D-OCT imaging provides a quantitative measurement of the two-dimensional flow profile of the vessel. ⁴¹ Further details for OCTA and DOCT can be found in Supplementary Information. Throughout imaging, animals were anesthetized with isoflurane (5% for induction, and 1.5–1.7% for maintenance with a gas mixture of 40% oxygen and 60% air), and the physiological signals were monitored (Physio-Suite, Kent Scientific). After longitudinal imaging, the brains were collected for immunohistochemistry (IHC) analysis as described in the Supplementary Information.

OCTA and TPM imaging

A temporal series of OCT angiograms of the mouse cortex was obtained before TPM imaging. Stalled segments located 150–200  μ m from the cortical surface were identified using 3D OCT angiograms and used for cEG imaging. Capillaries closer to the surface were selected for ease of identification later during TPM imaging. Each animal was moved to the TPM device (LSM 510, Zeiss) after the OCTA session. A femtosecond laser with a center wavelength of 800 nm and a Plan-Apochromat 20x/1.0 DIC VIS-IR water immersion objective lens were used for imaging. WGA-lectin conjugated with fluorescein isothiocyanate (WGA-lectin FITC) was dissolved in saline (1 mg/mL) and injected intravenously via the tail vein at 45 minutes before imaging at a dose of 6.25 mg/kg, to label the cEG. ⁴² Dextran (2 MDa; 0.25% (w/v) in saline, 100 μL) conjugated with tetramethylrhodamine (D7139; Thermo Fisher) was intravenously injected just before the imaging to label blood plasma. Animals were anesthetized with isoflurane (5% for induction, and 1.5–1.7% for maintenance with a gas mixture of 40% oxygen and 60% air). Stalled segments found in OCT angiograms were two-dimensionally scanned one by one by TPM (Supplementary Figure 1B). The FOV was 60 × 60 μm², with a pixel resolution of 1024 × 1024. The pixel dwell time was 0.64 μs, and each line was generated by averaging eight lines. TPM was performed within 1 hour to avoid signal reduction.

Data analysis

Changes in capillary stalling and vessel density were measured using en face OCT angiograms. A time-averaged OCT angiogram was generated by averaging all angiograms acquired. The overall contrast of the averaged OCT angiogram was adjusted by contrast limited histogram equalization. After equalization, vessels were segmented using the Weka trainable segmentation plugin in Fiji software (Supplementary Figure 1 C).^43–46 Shadows from large pial vessels were manually removed from the image. The vascular areas of the large pial vessels were measured. Subsequently, a binary mask of the capillary network was obtained, which was then skeletonized. Skeletonized segments smaller than 4 pixels were removed. Vessel density was defined as the total number of skeleton segments divided by the area excluding the area occupied by the large pial vessels. Capillary stalling was evaluated using a temporal series of OCT angiograms. Each angiogram was multiplied by the binary mask to clearly identify the vessels. Capillaries that showed a sudden drop in the signal in at least one frame of the OCT angiograms (<7 seconds) during the entire acquisition were defined as stalled segments and were identified automatically using MATLAB (Mathworks) and double-checked by manual inspection. The density of the stalled segments was defined as the number of stalled segments divided by the vessel density to correct for the possibly high stalled segment numbers due to the high vessel density. A time course of changes in the vessel density and the density of stalled segments during the 9 weeks of imaging were obtained for each animal. Each time course was normalized to the baseline course (W0), and the area under the curve of the time course of the density of the stalled segments was defined as the severity of capillary stalling. Some portions of the stalled segments at baseline (W0) were stalled again during the 9 weeks of imaging; these capillaries were defined as re-stalled segments. Alterations in vessel density and the density of stalled segments were investigated using six sham and nine SVaD mice. Two of the eight sham mice were excluded from analysis due to regrowth of the thick dura mater, which affected the quality of the OCT angiograms.

cEG images by TPM were analyzed using Fiji software. The intensity profiles of WGA-lectin FITC and dextran-TMR in the direction perpendicular to the vessel wall were measured, as previously described. ⁴² We measured the extent of cEG covering rather than the thickness because cEG thickness does not reflect the cEG density. The boundary between the cEG layer and blood plasma was defined as the inflection point of the minimum-maximum normalized intensity profile of dextran-TMR, while the boundary between the endothelial cells and cEG layer was defined as the location where the intensity of WGA-lectin exceeded the mean plus five times the standard deviation of background noise (Supplementary Figure 1D). The area under the curve of the relative intensity profile of WGA-lectin FITC between the boundaries was defined as the extent of the cEG.

Capillary stalling quantification and classification of the capillary stalling causes were analyzed using TPM images after R6G injection. Capillaries with blood flow blocked by leukocytes and RBCs were measured. To measure long capillary stalling (>700 seconds), we measured the stalling period for each mouse, which was calculated by multiplying the number of long capillary stalling (>700 seconds) by the duration of each stalling. Leukocytes were identified by the Hoechst 33342-positive and R6G-positive pixels. Platelets were identified by the Hoechst 33342-negative and R6G-positive pixels. RBCs were identified by unstained blood plasma.

Arterial hemodynamic parameters, such as vessel diameter and flow speed were measured using D-OCT imaging. ⁴⁰ Among the 9 SVaD mice, one was excluded from analysis as the flow speed of the pial artery decreased below the minimum detectable limit of the D-OCT measurements before W8.

For IHC data analysis, all confocal images were analyzed using Fiji software. To evaluate gliosis, the percentage of GFAP-positive pixels among all pixels within the region of interest was measured. Blood-brain barrier (BBB) disruption was evaluated as the mean fluorescence intensity of the anti-IgG antibody signal in the image. IHC analysis was performed using the five sham and nine SVaD mice that were used for the OCTA analysis. One of the six sham mice was excluded from the IHC analysis because of severe brain damage that occurred during brain harvesting.

For RNA sequencing data, quality was controlled with FastQC. Adaptor and low-quality reads (<20) were removed using FASTX-Trimmer and BBMap. After trimming, reads were mapped to the reference genome using TopHat. ⁴⁷ Fragments per kb per million reads (FPKM) values were used to approximate gene expression level. ⁴⁸ The FPKM values were normalized based on a quantile normalization method using EdgeR.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 8 (GraphPad Software) and SPSS 25 (IBM). All data are presented as mean ± standard deviation, median and interquartile range, or 95% confidence interval of the mean. Data were tested for normality using Kolmogorov−Smirnov and Shapiro−Wilk tests. All statistical analysis results and number of animals are summarized in Supplementary Tables 1–4. Differences between groups were tested using parametric or non-parametric statistics depending on the distribution of data. For comparing two groups, Student’s t-test, Welch’s t-test, or Mann−Whitney test were used. All longitudinal data from OCT imaging were tested using two-way repeated measure analysis of variance, followed by Bonferroni’s correction. A log-transformation (natural logarithm) was performed to satisfy normality. Pearson correlation coefficient was calculated to measure correlation. Details regarding the number of samples, p values, and other notes for each group are included in the figure legends.

Less extent of cEG observed in the stalled capillary segment

A capillary bed at 150–300 μm depth was imaged with OCTA. Capillary stalling was observed with serial OCTA angiograms. Because of sudden blockage of blood flow, a steep decrease of decorrelation signal occurred in some of capillary segments (Figure 1(a) and Supplementary Video 1). These segments were classified as stalled capillary segments and demarcated by blue color. There were re-stalled capillary segments among the stalled capillary segments. Approximately 38.4% of segments were stalled more than twice during 9 weeks of OCTA imaging (six mice used, mean = 38.41 ± 11.17; Figure 1(b)). Stalled capillary segments found with OCTA were searched using TPM. The cEG of the stalled capillary segments and adjacent non-stalled capillary segments was imaged (Figure 1(c)). Visible WGA-lectin signal along the capillary wall was detected in non-stalled capillary segments but not in stalled capillary segments. The quantified extent of the cEG was significantly smaller in the stalled capillary segments than in the non-stalled capillary segments (Figure 1(c); nine mice used, n = 38 for non-stalled capillary segments, 0.54 ± 0.42, n = 24 for stalled capillary segments, 0.21 ± 0.13, p < 0.0001). There was no difference in the diameter of the stalled and non-stalled capillary segments (4.262 ± 1.301 μm, n = 38 and 4.376 ± 1.532 μm, n = 24 for non-stalled and stalled capillary segments, respectively; p = 0.7543).

OPEN IN VIEWER

Figure 1.

Stalled capillary segment has less extent of cerebral endothelial glycocalyx. (a) (left) Representative high-quality angiogram acquired by optical coherence tomography angiography (OCTA). OCTA provides temporal dynamics of blood flow at a single capillary level. Capillary stalling occurs in certain capillary segments as demarcated by blue color. (right) Capillary stalling occurs in magnified images of capillaries from yellow boxes in the left angiogram. At a certain time, a sharp drop of signal inside the capillary occurs because of blocking of capillary blood flow. The letter F indicates flowing frames and the letter S indicates stalled frames. (b) In sham mice, 38.4% of the stalled capillary segments are re-stalled capillary segments (n = 6). (c) (left) The schematic figure demonstrates cEG in capillaries. The intensity profiles of WGA-lectin-FITC and dextran-TMR are overlaid along the scan direction. Two dotted lines indicate the cEG boundaries. The extent of cEG is defined as the region marked by the diagonal line. (middle) TPM images of the blue-colored stalled capillary segment and neighboring non-stalled capillary segment demonstrates the different distribution of cEG. Averaged WGA-lectin-FITC and dextran-TMR intensity profile across the dotted square in TPM images reflects the different distribution of cEG. WGA-lectin-FITC and dextran-TMR profiles are minimum-maximum normalized for better presentation. (right) A smaller extent of cEG is measured in stalled capillary segments (9 mice were used, n = 38 for non-stalled capillary segments, n = 24 for stalled capillary segments). Data are presented as mean ± standard deviation or median and interquartile range. The Mann–Whitney test is used (****: p < 0.0001). Scale bar: 100 μm for Figure 1(a) (left), 30 μm for Figure 1(a) (right), 10 μm for Figure 1(c) (middle).

Increased capillary stalling after cEG degradation

A mixture of heparinase and hyaluronidase was infused through the left CCA an hour before microsphere injection. The enzyme treatment degraded the WGA signal in the cerebral vessels (Supplementary Figure 2). We measured the number of microspheres after 1 hour. We confirmed that the microspheres were located inside the capillaries (Figure 2(b)). Increased number of microspheres, i.e. increased capillary stalling, in the cortex, hippocampus, and subcortex after 1 hour of enzymatic degradation of cEG was observed (Figure 2(c); n = 8 for each group; 2.0 ± 0.4 and 3.4 ± 1.1 for cortex treated with saline and enzyme, respectively, p = 0.01; 1.9 ± 0.3 and 3.5 ± 1.3 for hippocampus treated with saline and enzyme, respectively, p = 0.13; and 1.6 ± 0.4 and 2.4 ± 0.9 for subcortex treated with saline and enzyme, respectively, p = 0.04). Additionally, the number of microspheres increased after the separate injection of heparinase or hyaluronidase (Supplementary Figure 3). Furthermore, the number of microspheres increased both in the ipsilateral and contralateral hemispheres (Supplementary Figure 4). After enzyme administration, the animal physiology was altered over time and a difference in respiratory rate was found with a post-hoc test, 1 hour after enzyme treatment (Supplementary Figure 5). Additionally, there was no significant correlation between the number of microspheres and altered physiology parameters.

OPEN IN VIEWER

Figure 2.

Capillary stalling increases after cerebral endothelial glycocalyx (cEG) degradation. (a) The experimental timeline to explore the effect of cEG on capillary stalling is shown. (b) Immunohistological staining of cerebral vessels shows DiI dye-coated microspheres located inside capillaries and marked capillary stalling. (c) The number of DiI dye-coated microspheres increases after cEG degradation in the cortex, hippocampus, and subcortex (n = 8 for saline and enzyme-treated group). (d) In vivo two-photon microscopy imaging confirms an increased number of DiI-coated beads after enzymatic degradation of cEG (n = 4 for saline and enzyme-treated group). Data are presented as mean ± standard deviation. The unpaired t-test with Welch correction (Figure 2(c), *: p < 0.05) and unpaired t-test (Figure 2(d), *: p < 0.05) are used. Scale bar: 100 μm (Figure 2(b)), 1 mm (Figure 2(c)), 100 μm (Figure 2(d)).

In vivo imaging was also performed as the number of microspheres may be affected during ex vivo sample preparation. The number of microspheres increased after cEG degradation (Figure 2(d); n = 4 for each group, 2.5 ± 1.3 for saline-treated, 6 ± 1.6 for enzyme-treated, p = 0.015). Additionally, there was no significant difference in pial artery blood flow after saline or enzyme treatment (Supplementary Figure 6).

Increased leukocyte plugging after cEG degradation

To reveal the cause of increased capillary stalling, in vivo TPM imaging was performed after leukocyte labeling. Stalled capillary segments were identified by R6G-positive cells or by RBCs. As expected, the total number of capillary stalls over an hour increased (Figure 3(a); n = 5, 25.8 ± 14.8 and n = 4, 72.3 ± 32.1 for the saline- and enzyme-treated groups, respectively, p = 0.023) and also the total number of capillary stalls over an hour by plugged R6G-positive cells increased (Figure 3(b); n = 5, 19.8 ± 13.6 and n = 4, 57 ± 24.7 for the saline- and enzyme-treated groups, respectively, p = 0.023) after enzymatic cEG degradation. However, the total number of capillary stalls over an hour by cells other than R6G-positive cells did not significantly increase after cEG degradation (n = 5, 6 [0.5–11.5] and n = 4, 16.0 [7.8–22.0] for the saline- and enzyme-treated group, respectively, p = 0.127). Additionally, stalling period >700 seconds was quantified through serial TPM imaging. The stalling period was increased after enzyme treatment (n = 5, 2460 [351.4–9488] and n = 4, 13705 [11596–31099] seconds for saline- and enzyme-treated groups, respectively, p = 0.016). Stalling period by plugged R6G-positive cells was increased after cEG degradation (n = 5, 2460 [351.4–6501] seconds and n = 4, 9312 [8434–20996] seconds for the saline- and enzyme-treated groups, respectively, p = 0.016). In contrast, the period of stalling by cells other than plugged R6G-positive cells had not increased (n = 5, 0 [0–2987] seconds and n = 4, 4393 [2723–10542] seconds for the saline- and enzyme-treated group, respectively, p = 0.056). In particular, in the three mice treated with saline, capillary stalling by the RBCs was not observed for >700 seconds. Although it did not reach statistical significance, maybe due to statistical power, the period of stalling by cells other than plugged R6G-positive cells showed an increasing trend. To identify increased plugged R6G-positive cells after enzyme treatment, we labeled leukocytes with Hoechst 33342 and R6G (Figure 3(d)). ³⁴ Most of the capillary stalling was caused by plugged leukocytes (Figure 3(e); p = 0.0079, 11.20 ± 5.119 for the number of plugged leukocytes and 1 ± 1.414 for the number of plugged platelets; five mice were analyzed). There was no difference in capillary diameter after saline or enzyme treatment (median = 3.288 μm, n = 75 for saline and median = 3.562 μm, n = 60 for enzyme-treated capillaries, p = 0.097).

OPEN IN VIEWER

Figure 3.

Leukocyte plugging increases after enzymatic degradation of cerebral endothelial glycocalyx (cEG). (a) Representative angiogram of saline- or enzyme-treated mouse. Arrowheads indicate capillary stalling by R6G-positive cells or by RBCs (R6G negative). (b) The total number of capillary stalls during 1 hour increases after cEG degradation. Capillary stalling by R6G-positive cells during 1 hour increases after cEG degradation. Capillary stalling by cells other than R6G-positive cells is not increased (n = 5 for the saline-treated group, n = 4 for the enzyme-treated group). (c) Total period of capillary stalling more than 700 seconds increases after cEG removal. The total period of capillary stalling by R6G-positive cells is increased. The total period of capillary stalling by cells other than R6G-positive cells has an increasing trend (n = 5 for the saline-treated group, n = 4 for the enzyme-treated group). (d) Representative angiogram of enzyme-treated capillaries with leukocyte plugging. Leukocytes are identified with colocalized Hoecht 33342 and R6G signal. (e) The quantified number of capillary stalls indicates the main cause of capillary stalling as leukocyte plugging (p = 0.0079, 11.20 ± 5.119 for the number of plugged leukocytes, 1 ± 1.414 for the number of plugged platelets, five mice are analyzed, Mann−Whiney test). Data are presented as mean ± standard deviation or median and interquartile range. The unpaired t-test and Mann−Whitney test are used (*: p < 0.05, **: p < 0.01). Scale bar: 50 μm (Figure 3(a)), 20 μm (Figure 3(d)).

Capillary stalling increases with SVaD progression

Before microcirculation imaging in SVaD, we validated decreased arterial blood flow in the SVaD model with D-OCT. After inducing SVaD, rapid decrease in arterial flow was measured (Supplementary Figure 7). Alterations in vessel density and the density of stalled segments were measured. Charge-coupled device (CCD) images in Figure 4(a) show the regions where OCTA was performed in sham and SVaD mice. Alterations in microcirculation were hardly detectable in the CCD images because of insufficient spatial resolution. Time-averaged OCT angiograms acquired at the same regions during the 9 weeks of imaging showed an apparent capillary loss in SVaD mice, while no apparent changes were observed in sham mice (Figure 4(b)). Vessel density (normalized to the baseline value acquired at W0) decreased in SVaD mice (Figure 4(c)), but not to a significant level (n = 6 for sham and n = 9 for SVaD mice, F_(8,104)=1.12, p = 0.3274 for interaction; F_(1,13)=2.06, p = 0.1751 for group factor). The density of the stalled segments were significantly higher in the SVaD than in the sham mice (Figure 4(d); n = 6 for sham and n = 9 for SVaD mice; F _(8,104)=4.23, p = 0.0002 for interaction; F_(1,13)=7.54, p = 0.0166 for group factor). The number of stalled segments were increased in the SVaD mice (Supplementary Figure 8). At W8, the density of stalled segments in SVaD mice was 2.61 times higher than that in the sham mice (n = 6 for sham and n = 9 for SVaD mice, sham: 1.24-fold change, SVaD: 3.24-fold change, p = 0.024). The number of stalled segments for only one frame (<7 seconds) was significantly increased in the SVaD mice (F_(8,104)=2.289, p = 0.027, n = 6 for sham, n = 9 for SVaD) (Supplementary Figure 9). The number of stalled segments for more than two frames (>14 seconds) or more than three frames (>21 seconds) was not different in both groups. The averaged stall duration showed a decreasing trend in the SVaD mice (F_(8,104)=2.007, p = 0.0526, n = 6 for sham, n = 9 for SVaD). The number of stalled segments at W0 (n = 6 for sham and n = 9 for SVaD mice, sham: 23.6 ± 6.4, SVaD: 20.9 ± 7.8, p = 0.4939) was much smaller than the total number of capillaries in both the sham and SVaD mice (n = 6 for sham and n = 9 for SVaD mice, sham: 4367 ± 929.3, SVaD: 4972 ± 1033.0, p = 0.2688). Approximately 40% of the stalled segments identified at W0 stalled again at least once during the 9 weeks of imaging in both the sham and SVaD mice (n = 6 for sham and n = 9 for SVaD mice, Figure 4(e); sham: 38.41 ± 11.2, SVaD: 40.77 ± 13.6, p = 0.7298). This result indicates that stalling did not occur randomly but that certain segments have a higher susceptibility to stalling. It is noteworthy that more than 40% of the re-stalled segments were lost at W8 in SVaD mice, which is twice the rate of capillary segments loss compared to that in sham mice (n = 6 for sham and n = 9 for SVaD mice, Figure 4(f); sham: 19.96 (15.42–30.00), SVaD: 36.84 (27.88–53.26), p = 0.0042). Additionally, astrogliosis and BBB leakage were correlated with capillary stalling (Supplementary Figures 10). During the 9 weeks of imaging, there were no differences in physiological signals between the sham and SVaD mice (Supplementary Figure 11).

OPEN IN VIEWER

Figure 4.

Cerebral microcirculation is altered in subcortical vascular dementia (SVaD). a. No detectable changes with CCD images both in sham and SVaD mice. b. Longitudinal imaging with OCTA demonstrates cerebral microcirculation changes during 9 weeks of imaging. Reduced vessel density is detected in this representative SVaD mouse. c. Even though some SVaD mice show apparent vessel density reduction, it does not significantly reduce in SVaD mice (F _(8,104)=1.165 for interaction, p = 0.3274, F_(1,13)=2.058 for group factor, p = 0.1751). d. Density of stalled capillary segments increases in SVaD mice (F _(8,104)=4.232 for interaction, p = 0.0002, F_(1,13)=7.544 for group factor, p = 0.0166). e. Proportion of re-stalled capillary segments is not different between the two groups (38.41 ± 11.2 for sham, 40.77 ± 13.6 for SVaD mice, p = 0.7298). F. Percentage of re-stalled segments disappearing at 8 weeks after surgery significantly increases (19.96 (15.42–30.00) for sham, 36.84 (27.88–53.26) for SVaD mice, p = 0.0042). Data are present as mean ± standard deviation. (c, e), median and interquartile range (F), or 95% confidence of interval of the mean (d). n = 6 for sham and n = 9 for SVaD mice. Log transformation is performed for the density of stalled segments. A two-way analysis of variance followed by Bonferroni’s post-hoc test, unpaired t-test, and Mann−Whitney test is used (**: p < 0.01, ***: p < 0.001). Scale bar: 150 μm (Figure 4(a)), 200 μm (Figure 4(b)).

Degraded cEG in SVaD

Capillary segments from the pial surface to 200 μm in depth were randomly imaged using TPM. There was a clear distribution of WGA-lectin signal on the luminal side of the capillaries in the sham mice (Figure 5(a)). However, there was no clear distribution of WGA-lectin signal at capillaries of the SVaD mice even when leukocytes were labeled with WGA-lectin. The quantified extent of cEG was significantly reduced in SVaD (Figure 5(b); four mice for each group, n = 79, 0.51 [0.37–0.78] for sham mice, n = 74, 0.2 [0.16–0.40] for SVaD mice, p < 0.0001). The extent of cEG was reduced after one week of SVaD induction (Supplementary Figure 12).

OPEN IN VIEWER

Figure 5.

Extent of cerebral endothelial glycocalyx (cEG) is reduced in subcortical vascular dementia (SVaD). (a) A representative two-photon microscopy image of the capillary bed from sham and SVaD mice shows cEG and blood plasma. In the sham mice, cEG is clearly distributed on the luminal side of capillaries but not in SVaD capillaries. (b) The extent of cEG is significantly reduced in the SVaD model mice (n = 79, 0.51 (0.37–0.78) for sham mice, n = 74, 0.2 (0.16–0.40) for SVaD mice). Four mice are used for each group. Data are presented as median and interquartile range. The Mann−Whitney test was used (****: p < 0.0001). Scale bar: 50 μm.

Glycocalyx-related gene expression changes in SVaD

Isolated cortical capillaries were collected for RNA sequencing. We purified mRNA at 1 week and 8 weeks after SVaD induction (Figure 6(a)). Among the total mRNA, genes with >1.3-fold up/down change with statistical significance (p < 0.05) were classified as upregulated and downregulated genes. A week after SVaD induction, 427 genes were upregulated and 208 genes were downregulated (Figure 6(b); N = 3 for each group). Eight weeks after SVaD induction, 756 genes were upregulated and 386 genes were downregulated (N = 4 for the sham group, N = 3 for the SVaD group). Gene set enrichment analysis revealed that genes involved in extracellular matrix (ECM)-related biological pathways had changed expression after SVaD induction. More changes in ECM-related pathways were detected at 8 weeks after SVaD induction (Figure 6(c)). Endothelial cell-specific genes and glycocalyx-related genes were identified based on a previous study. ⁴⁹ More modified gene expression was measured at 8 weeks than 1 week after SVaD induction (Figure 6(d) and Supplementary Table 5). It should be emphasized that angiopoietin2 (Angpt2) expression was upregulated at only 8 weeks after SVaD induction. Glycocalyx-related gene expression was also modified after SVaD induction. At 1 week after SVaD induction, dermatan sulfate epimerase (Dse), heparan sulfate 6-O-sulfotransferase 1, matrix metalloproteinase 9 (MMP9), and hyaluronan mediated motility receptor had modified gene expression. There were more modified glycocalyx-related gene expressions 8 weeks after SVaD induction. Decorin, galectin 1, galectin 9, a disintegrin and metalloproteinase with thrombospondin motifs type 1 motif 9, Dse, heparanase (Hpse), hyaluronidase 3, MMP14, and hyaluronan and proteoglycan link protein 3 (Hapln3) were upregulated. Beta-1,4-galactosyltransferase 6 (B4galt6), hyaluronan and proteoglycan link protein 2 (Hapln2), hyaluronan and proteoglycan link protein 4 (Hapln4), and exostosin-like glycosyltransferase1 (Extl1) were downregulated. Furthermore, since protein kinase A (PKA) and protein kinase C (PKC) signaling pathways are involved in the secretion of heparanase ⁵⁰ we performed gene ontology analysis for PKA, PKC, and lysosome secretion involved gene set and found that seven genes were significantly changed (Supplementary Figure 13). These results suggest that the PKA, PKC, and lysosome secretion signaling pathways are involved in heparanase secretion and increased when Angpt2 increased. mRNA expression for leukocyte adhesion molecules such as Icam-2 and Pecam-1 was changed after 8 weeks of SVaD induction (Figure 6(e)).

OPEN IN VIEWER

Figure 6.

Glycocalyx-related gene expression is modified in subcortical vascular dementia (SVaD). (a) Isolated capillaries from the cerebral cortex are collected from both groups. RNA is isolated and sequenced. (b) At different time points, different mRNA expression patterns between sham and SVaD mice are observed (N = 3 for each group at 1 week after SVaD induction, N = 4 for sham, N = 3 for SVaD model at 8 weeks after SVaD induction). (c) Changes in gene-enriched profile are detected with gene set enrichment analysis. The ECM-related biological processes are changed. (d) The endothelial cell and glycocalyx-related gene expression changes are shown. Relative expressions are shown. Upregulated and downregulated genes are marked red and blue, respectively. (e) Leukocyte adhesion molecule gene expression is shown. Two mice are pooled for one sample of cortical capillaries. The unpaired t-test is used (*: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001). Scale bar: 50 μm.