A labeling strategy for the three-dimensional recognition and analysis of microvascular obstruction in ischemic stroke

Rationale: Large vessel recanalization in ischemic stroke does not always go along with tissue reperfusion, a phenomenon called “no-reflow”. However, knowledge of the mechanism of no-reflow is limited because identifying microvascular obstruction across the cortex and subcortex both in clinical and experimental models is challenging. In this study, we developed a smart three-dimensional recognition pipeline for microvascular obstruction during post-ischemia reperfusion to examine the underlying mechanism of no-reflow. Methods: Transient (60 min) occlusion of the middle cerebral artery (tMCAo) in mice was induced using a filament. Two different fluorophore-conjugated tomato lectins were injected into mice via the tail vein before and after ischemia/reperfusion (I/R), respectively, one to label all blood vessels and the other to label functional blood vessels. Post-I/R microvascular obstruction was visualized using combined iDISCO+-based tissue clearing and optical imaging. Arterioles and capillaries were distinguished using whole-mount immunolabeling with an anti-αSMA antibody. Circulating neutrophils were depleted utilizing an anti-Ly6G antibody. Brain slices were immunostained with the anti-Ly6G antibody to identify co-localized blockage points and neutrophils. MATLAB software was used to quantify the capillary diameters in the ipsilateral brain from the normal and tMCAo mice. Results: Microcirculatory reperfusion deficit worsened over time after I/R. Microvascular obstruction occurred not only in arterioles but also in capillaries, with capillary obstruction associated with local capillary lumen narrowing. In addition, the depletion of circulating neutrophils mitigated reperfusion deficit to a large extent after I/R. The co-localization of blockage points and neutrophils revealed that some neutrophils plugged capillaries with coexisting capillary lumen narrowing and that no neutrophil was trapped in heaps of blockage points. Quantification of the capillary diameter showed that capillary lumen shrunk after I/R but returned to typical measurements when intravascular neutrophils were depleted. Conclusions: According to our findings, both vascular lumen narrowing and neutrophil trapping in cerebral microcirculation are the key causes of microvascular obstruction after I/R. Also, the primary contribution by neutrophils to microvascular obstruction does not occur through microemboli plugging but rather via the exacerbation of capillary lumen narrowing. Our proposed method will help monitor microcirculatory reperfusion deficit, explore the mechanism of no-reflow, and evaluate the curative effect of drugs targeting no-reflow.


Laser Speckle Contrast Imaging (LSCI)
Cortical perfusion was monitored using the LSCI system. First, mice were anesthetized with 3% isoflurane until they were unresponsive to the tail pinch test, and then maintained at 1.5%. Next, the heads of mice were fixed in a stereotaxic apparatus and the eyes were kept wet with ointment.
The scalps were incised longitudinally to fully expose the skull and kept moist with a saline cotton ball. Subsequently, the mice were placed under the LSCI system to acquire a sequence of raw speckle images. The laser speckle temporal contrast analysis method was used to deduce the relative blood flow velocity, and then generate cerebral blood flow (CBF) images. Cortical CBF images on the ipsilateral side were obtained pre stroke (baseline), 30 min after inserting the monofilament, and 10 min after withdrawing the monofilament. Mice without CBF reduction to less than 30% of baseline level after ischaemia or CBF restoration to more than 70% of baseline level after recanalization were excluded from further experimentation.
First, 0.2% Triton X-100 in 0.01 M PBS (PBST) was prepared. Then, 100-μm-thick brain slices were blocked with 10% goat serum in 0.2% PBST for 1 h and incubated with the primary antibody and following the secondary antibody diluted with 5% goat serum in 0.2% PBST for 24 h. Finally, brain slices were washed six times with 0.2% PBST for 30 min. For nuclear staining, the immunostained samples were immersed in 0.01 M PBS containing 1 μg/ml DAPI (Invitrogen, New York, USA) with shaking at room temperature for 2 h.

3D immunostaining
Whole-mount immunostaining was performed as previously described in the iDISCO+ protocol, including methanol pretreatment and immunolabeling protocol. [1] Methanol pretreatment. First, samples were dehydrated in methanol solutions at increasing concentration gradients (20 vol. %, 40 vol. %, 60 vol. %, 80 vol. %, 100 % and 100 % in dH2O) for 1 h/step. The dehydrated samples were incubated overnight with 66% dichloromethane in methanol, followed by two washes in 100% methanol for 1 h/wash. Samples were then bleached with 5% H2O2 in methanol at 4 °C overnight. After bleaching, samples were rehydrated in methanol solutions at decreasing concentration gradients (60 vol. %, 40 vol. %, 20 vol. % in dH2O) for 1 h/step, and washed twice in 0.01 M PBS for 2 h. Except for bleaching, other steps are processed at room temperature.
Immunolabeling protocol. After methanol pretreatment, samples were blocked in PBST with 10% goat serum for 2 h, incubated in rabbit anti-αSMA (ab5694, Abcam, dilution 1:300) diluted in PBST with 5% goat serum for 1 d, and washed with PBST for 12 h., Samples were then incubated in Alexa Flour 647-conjugated goat anti-rabbit IgG (A21244, Thermo Fisher Scientific, dilution 1:500) diluted in PBST with 5% goat serum for 1 d, washed with PBST for 12 h, and stored in 0.01 M PBS at 4 °C before clearing and imaging. All steps were processed at room temperature.

Microscopy
The z-stack fluorescence images of brain blocks were acquired with an inverted confocal Whole-brain vessel images in Figure S4A were acquired with a light-sheet microscope (LiTone XL, Light Innovation technology, Hongkong, China) equipped with a 4×/0.25 dry objective (W.D. 25 mm) and a 10×/0.5 objective (W.D. 5 mm). For imaging whole brains, we chose the 4× objective and an interval of 5 μm. For imaging brain regions, we chose the 10× objective and an interval of 3 μm.

Image processing
ImageJ and Imaris software were used to visualize and analyze 2D and 3D images. The maximum intensity projection (MIP) of the z-stack was performed with sequential images in the ImageJ software. 3D images and movies were reconstructed and captured with the Imaris software.
To extract occlude microvessels, we used the ImageJ software to process the z-stack images ( Figure S5). First, the "Binary" tool was used to calculate the threshold value of each image in the image stack. Next, the binary image of "functional vessels" (the second tomato lectin) was subtracted from the binary image of "existing vessels" (the first tomato lectin) using the "Image calculator" tool to obtain the resulting image, and this resulting image was processed using the "median filter algorithm" in "Filter" tool. Using "And" operation of the "Image calculator" tool, the original image stack of "existing vessels" and the filtered resulting image were calculated to obtain the image stack of "occluded vessels". In addition, the background signals of the image stacks were removed by "And" operation between the original images and the binary images. In the end, the image stacks of "functional vessels", "existing vessels", and "occluded vessels" were merged using the "Color" tool to generate a new three-channel image stack.
To calculate the vascular diameter, we used the ImageJ software to segment the blood vessel images. First, the "Binary" tool was used to calculate the threshold value of each image in the image stack. Then, the "Filter" tool was used to remove the background signal in the binary images. At last, the "Morphology" tool was used to fill the holes in the signal of the blood vessels.

Quantifications
Changes in cortical perfusion. The relative blood flow velocity in cortical branch of the middle cerebral artery was measured. The relative blood flow velocity before tMCAo was used as the baseline, and the ratio of blood flow to baseline was used to determine changes in cortical perfusion during and after tMCAo. Each value was obtained by averaging values for the three regions.
The degree of microvascular obstruction. The ratio of vascular length of "functional vessels" to "existing vessels" and the relative volume of "occluded vessels" were both used to quantify the degree of microvascular obstruction. The Imaris software was used for quantification of vascular length and vascular volume. To quantify vascular length, the "filament" tool was used to trace the blood vessels automatically and the total length of the filament was utilized as the total vascular length. The length of "functional vessels" are divided by the length of "existing vessels" to obtain the ratio of vascular length of "functional vessels". To quantify vascular volume, the "surface" tool was used to segment the blood vessels. Surfaces smaller than 1000 μm 3 were removed as clutter, and then the volumes of the total surface in each channel were calculated. The volume of "occluded vessels" was divided by the volume of "Existing vessels" to determine the relative volume of "occluded vessels".
The number of neutrophils. We immunostained neutrophils with the FITC-conjugated anti-Ly6G antibody, stained the nuclear with DAPI, and labeled blood vessels by intravenously injecting tomato lectin. Images in the cortex and the striatum were observed with 30-μm z-projections. Intravascular markers containing FITC and DAPI signals were recorded as the neutrophils. The number of neutrophils in the ipsilateral and contralateral striatum and cortex was counted. Each value was determined by averaging these numbers in the four regions.
The diameter of the capillary. Segmented blood vessel images were imported into the MATLAB software to calculate the vascular diameter. The "Bwskel" function was used to skeletonized the segmented blood vessel images and obtain the centerline of the blood vessels. The "Bwdist" function was used to calculate the distance between the pixels whose signal is "1" to its nearest pixel whose signal is "0" in the segmented blood vessel images. The calculated distances in the centerline represented the radii of the blood vessels.   (B) Histogram showing fluorescence signal intensity in cleared brain blocks (n=6). (C) Brightfield images of 2-mm-thick brain blocks before and after clearing. Grid size, 1.44 mm × 1.44 mm.
(D) Histogram showing size change of brain blocks after clearing (n=4). (E) Transmittance curves of the brain blocks cleared with iDISCO+, uDISCO, and FDISCO (n=4). Data were presented as the mean ± SD. n.s., p>0.05; ***p<0.001; ****p<0.0001; Brown-Forsythe and Welch ANOVA tests with post hoc Tamhane's T2 multiple comparisons test for each comparison between groups. Figure S4. Tissue optical clearing assisted 3D analysis of microcirculatory changes after I/R (A, B) 3D (A) and 2D (B) fluorescence images of microvessels in whole mouse brain labeled with tomato lectin, cleared with iDISCO+, and imaged with a light-sheet microscope. (C) 3D reconstruction and tracing of microvessels in 1-mm brain slices. Blood vessels were traced using the "filament" tool of the Imaris software to quantify vascular length. The ratio of vascular length of "functional vessels" to "existing vessels" was calculated to obtain the degree of microvascular obstruction. (D) 3D fluorescence images of microvessels in the ipsilateral cortex and striatum. The mice were subjected 1 h of ischaemia and 2 h of reperfusion and injected with two different fluorophore-conjugated tomato lectins before and after I/R to identify microvascular obstruction. 1-mm brain slice was cleared with iDISCO+ method and imaged with a confocal microscope. (E) Bar graph of the ratio of vascular length showing the degree of microvascular obstruction in the ipsilateral cortex and striatum (n = 5). Data were presented as the mean ± SD. *p<0.05, two-tailed t test.