The immunolocalization of cluster of differentiation 31, phalloidin and alpha smooth muscle actin on vascular network of normal and ischemic rat brain

Cluster of differentiation 31 (CD31), phalloidin and alpha smooth muscle actin (α-SMA) have been widely applied to label the cerebral blood vessels in the past years. Although CD31 is mainly used as endothelial marker in determining the cerebral capillaries, it seems likely that its labeling efficiency is closely correlated with the antibodies from the polyclonal or monoclonal one, as well as the conditions of blood vessels. In order to test this phenomenon, we compared the labeling characteristics of goat polyclonal anti-CD31 (gP-CD31) and mouse monoclonal anti-CD31 (mM-CD31) with those of phalloidin and α-SMA on the rat brain in health and ischemia/reperfusion (I/R) with the middle cerebral artery occlusion. By multiple immunofluorescence staining, it was found that gP-CD31 labeling expressed extensively on the cerebral capillaries forming the vascular networks on the normal and ischemic regions, but mM-CD31 labeling mainly presented on the capillaries in the ischemic region. In contrast to the vascular labeling with gP-CD31, phalloidin and α-SMA were mainly expressed on the wall of cortical penetrating arteries, and less on that of capillaries. By three-dimensional reconstruction analysis, it was clearly shown that gP-CD31 labeling was mainly located on the lumen side of vascular wall and was surrounded by phalloidin labeling and α-SMA labeling. These results indicate that gP-CD31 is more sensitive than mM-CD31 for labeling the cerebral vasculature, and is highly compatible with phalloidin and α-SMA for evaluating the cerebral vascular networks under the physiological and pathological conditions.


Results
Distributional characteristics of the vascular labeling with gP-CD31 in the healthy and I/R brain. In order to show the regional characteristics of gP-CD31 labeled blood vessels in the healthy and I/R rat brains, the brain sections were counterstained with fluorescent Nissl staining for demonstrating the neuronal alternation on the cerebral tissue under the physiological and pathological conditions. On the background of Nissl staining, we focused our observation on the cortical region from 80-μm-thick slice, and traced the vascular labeling from the cortical penetrating arteries to the intracerebral capillaries to compare the vascular status on the normal, ischemic, penumbra, and matching contralateral regions ( Figs. 1 and 2).
In the healthy rat brain, the cerebral structures, including cortex, thalamus, striatum, hippocampus and other regions, were symmetrically showed on the coronal section (Fig. 1A), in which gP-CD31 labeled blood vessels ran through the neurons forming cerebral vascular networks (Fig. 1B,C-C4).
In the I/R rat brain, the area of ischemic hemisphere was obviously larger than that of contralateral hemisphere, showing with the sign of edema (P < 0.05, Fig. 2A,B). In contrast to the contralateral and penumbra areas, the gP-CD31 labeling was more strongly expressed on the capillaries of ischemic area in parallel with neuronal loss (P < 0.05, Figs. 2, 3), which was more easily identified in the three-dimensional views ( Fig. 3B4-D4,B5-D5).
Labeling difference between gP-CD31 and mM-CD31 in the healthy and I/R rat brain. The labeling characteristics of gP-CD31 and mM-CD31 for cerebral vascular network were critically assessed in both healthy and I/R brains including the normal, ischemic, penumbra, and matching contralateral regions (Fig. 4). In contrast to vascular labeling with gP-CD31, mM-CD31 was preferentially expressed on the cerebral capillaries in the ischemic region where coexisted with that of gP-CD31 ( Fig. 4B-B4), but there was less vascular labeling with mM-CD31 observed on the normal, penumbra, and matching contralateral regions (P < 0.05, Fig. 4A-A4,C-C4,D-D4,E). The spatial correlation of vascular labeling with gP-CD31 and mM-CD31 in ischemic region was clearly demonstrated in the three-dimensional views (Fig. 4A4,B4,C4,D4).
Comparing the labeling characteristics of gP-CD31 and phalloidin. For the vascular labeling, phalloidin was mainly expressed on the wall of cortical penetrating arteries, rarely on that of capillaries (Fig. 5). Comparatively, gP-CD31 labeling was located on the lumen side of vascular wall and wrapped with phalloidin labeling (Fig. 5A4-D4). Although the morphology of phalloidin labeled blood vessels was similar in both healthy and I/R brains, its expressing area and labeling intensity were larger and stronger in the ischemic region respectively (P < 0.05, Fig. 5E,F).
Comparing the labeling characteristics of gP-CD31 and α-SMA. In this study, α-SMA was also used for the vascular labeling. Similar to the phalloidin, α-SMA was also expressed on the wall of rat cortical penetrating arteries while less on capillaries (Fig. 6). The α-SMA labeling formed a disconnected layer surrounding the gP-CD31 labeling, which was more obvious under the three-dimensional views (Fig. 6A4-D4). Like gP-CD31, α-SMA was also expressed more and stronger in the ischemic area (P < 0.05, Fig. 6E,F). Spatial correlation of vascular labeling with gP-CD31, phalloidin, and α-SMA. Based on the above observations, the spatial correlations of vascular labeling among gP-CD31, phalloidin and α-SMA were further examined in the healthy rat brain and analyzed with three-dimensional reconstruction along the longitudinal axis of penetrating artery and intracerebral capillaries ( Fig. 7A-A4,B-B4), as well as with the transverse view of penetrating artery (Fig. 7C-C4). It was further obvious that gP-CD31 labeling was surrounded  2. Regional alteration of the rat brain with ischemia/reperfusion. (A) Representative montage views from every 12th coronal sections of brain immunofluorescently stained with goat polyclonal anti-CD31 (gP-CD31) and Nissl showing the regional alteration induced by ischemia/reperfusion at vascular and cellular level. (B) Quantitative analysis of the areas of ischemic hemisphere and its contralateral hemisphere (n = 4 rats, **P < 0.05). www.nature.com/scientificreports/ www.nature.com/scientificreports/  www.nature.com/scientificreports/ by phalloidin and α-SMA labellings in the cortical penetrating arteries, but mostly independent in capillaries ( Fig. 7A-A4,B-B4), while some of α-SMA labeling occasionally passed through the wall of the cortical penetrating arteries (Fig. 7C-C4). Although phalloidin and α-SMA were both expressed on the wall of cortical penetrating arteries, there was few colocalization to be observed under the three-dimensional views ( Fig. 7C-C4).

Discussion
In this study, we compared the cerebral vascular labeling properties of gP-CD31 with that of mM-CD31, as well as with those of phalloidin and α-SMA on the rat brains in health and I/R. Taking the advantage of multiple immunofluorescence staining on the thick brain sections, we provided histochemical evidence to show that gP-CD31 is superior to mM-CD31 on the vascular labeling in the normal and ischemic regions, and is highly compatible with phalloidin and α-SMA for insight into the cerebral vascular network in rat brain. These results indicate that the present technological approach should be beneficial in investigating the detailed morphology of cerebral vascular network under the physiological and pathological conditions, and has the potential to assess novel therapeutic interventions of cerebral vascular disorders.
Technological consideration. With the development of microscopic techniques, multiple immunofluorescence staining has been the standard procedure for investigating morphological and molecular aspects in the sliced tissues for several decades [24][25][26][27][28] . In line with this field, we adopted the techniques of panoramic tissue slice scanner and laser scanning confocal microscope in this study. The former can catch the overall view of tissue section and the latter can take the clear image from the thick tissue 29,30 . Taking these advantages, we can simultaneously obtain the labeling information from two dimensions to three dimensions 29,30 . Based on this consideration, the regional characteristics of gP-CD31 labeling were firstly assessed together with fluorescent Nissl in the healthy and I/R rat brains. Beyond the 2,3,5-triphenyltetrazolium chloride (TTC) staining for evaluating the changes of cerebral infarction volume, Nissl staining can not only clearly show the ischemic area, but also demonstrate its morphological alteration at cellular level 13,14 . On the background of fluorescent Nissl staining, it was clear that gP-CD31 labeling formed the cerebral vascular network among the neurons in whole brain specimens under the physiological and pathological conditions. In contrast to the gP-CD31 labeling, the mM-CD31 labeling was mainly detected on the capillaries in the ischemic region. Therefore, it limits the application of mM-CD31 to combine with the other bio-markers for the vascular labeling. Comparatively, our present results indicate that gP-CD31 is a better candidate together with phalloidin and α-SMA for labeling the cerebral vascular networks in the three-dimensional views in rat. www.nature.com/scientificreports/ Monoclonal versus polyclonal antibodies of CD31. As it is known, antibodies are the key players in the immunohistochemistry to bind their corresponding target antigens. No matter what kinds of antibodies, monoclonal or polyclonal, their ability determined by the degree of affinity and specificity to bind an antigen [31][32][33][34][35][36][37] . In general, monoclonal antibody is regarded as an antibody of single and unique specificity for their intended targets, serving as a powerful tool for labeling macromolecules and cells 38,39 , and comparatively polyclonal antibody can detect a multiplicity of epitopes and recognize antigen from different orientations 31 , which may be important in certain assays where the detection of an analyte would be compromised by the use of a single epitope 39 . In addition, polyclonal antibodies are also more stable over a broad pH and salt concentration, and have better specificity than monoclonal antibodies 31 .
In the case of CD31, it has been widely used as an endothelial marker in determining microvascular density 40,41 . Although gP-CD31 labeling and mM-CD31 labeling were co-localized on the cerebral capillaries in the ischemic region (Fig. 4B, Supplementary Fig. 1), in contrast to the gP-CD31 labeling, the mM-CD31 labeling was less detected in the normal, ischemic penumbra, and matching contralateral regions. For the cerebral vasculature, gP-CD31 was expressed more extensively than that of mM-CD31, which was in consistent with the previous studies (see Supplementary Table 1). According to the specification of antibodies, gP-CD31 is a 130 kDa type I transmembrane glycoprotein adhesion molecule in the immunoglobulin superfamily 42,43 , its immunogen is mouse myeloma cell line NS0-derived recombinant mouse CD31/PECAM-1 (Glu18-Lys590, Accession # Q08481). And its expression is restricted to cells involved in circulation, platelets, monocytes, neutrophils, lymphocyte subsets, especially endothelial cells 42,43 . In contrast, mM-CD31 is an antibody approximately 130 kDa, which is expressed on platelets and leukocytes, and primarily concentrates at the borders between endothelial cells 44,45 . However, the underlying mechanism of the difference between both antibodies still remains elusive.

Difference of vascular labeling between gP-CD31, phalloidin, and α-SMA.
On the basis of gP-CD31 labeling, cerebral blood vessels were also successfully labeled with phalloidin and α-SMA. Along the cerebrovascular tree, three kinds of vascular labeling were located orderly on the vascular wall closely without coexisting between one another. This result supports the idea that the cerebrovascular tree is constituted by different kinds of vascular cells, and these cellular components were varied from the cortical penetrating arteries to the intracerebral capillaries 46 . Firstly, gP-CD31 labeling is located on the lumen side of the wall of cortical penetrating arteries and their main branches are surrounded with phalloidin and α-SMA labeling, which implies that gP-CD31 labeled endothelial cells stand on the first line to respond the harmful effects of cerebrovascular insufficiency 46 . Moreover, gP-CD31 labeling takes predominant advantage in the intracerebral capillaries, where phalloidin and α-SMA labeling is less detected, which suggests that gP-CD31 labeled endothelial cells are distributed more extensively than the vascular cells labeled with phalloidin and α-SMA. Therefore, gP-CD31 is not only suitable for labeling the cerebral vascular network, but also highly compatible with phalloidin and α-SMA for demonstrating the sub-types of vascular cells from the cortical penetrating arteries to the intracerebral capillaries in the health and disease. Potential application. According to the quantitative analysis of vascular labeling, gP-CD31 was expressed more strongly on the capillaries in the ischemic region associated with the neuronal degeneration. Under the physiological condition, endothelial, neuronal, and glial cells are interconnected for contributing to brain homeostasis [46][47][48][49] . However, our understanding of multicellular alteration associated with the I/R brain is still incomplete. Under the pathological condition, the precise role of higher expressed gP-CD31 in the ischemic region, whether detrimental or beneficial, remain poorly understood. It might be associated with the promotion of endothelial cell proliferation, migration, capillaries formation, vascular permeability or angiogenesis 48 . Higher permeability can lead to the edema of I/R hemisphere with neuronal and glial alteration, directly causing brain parenchyma lesions 13,50 , which is supported by our present histological evidence. Conversely, angiogenesis is a crucial process for neural repairing during I/R or other vascular injury in the brain, although the cerebral vasculature is quiescent in the adult animals under the normal condition 51 . Previous studies have shown that endothelial cells are mitotic and can generate new vessels in response to brain injury, and its proliferation occurs from 12 h after stroke and continues for a number of weeks 3,52 . However, our present observation is limited on the spatial alteration of vascular labeling at one time-point in the I/R brain, and the temporal changes of proliferation with endothelial cells remain to be determined in the same experimental model. Whether enhanced gP-CD31 expression could be the histochemical evidence for evaluating the angiogenesis, it might be an important issue for the further exploration in the future study. In addition, our results suggest that the combination of the multiple immunofluorescence staining with three-dimensional reconstruction techniques on the thick tissue sample is a beneficial approach for demonstrating the vascular alteration within I/R brain in a three-dimensional view, and it is reasonably used to show the vascular disorder with other diseases in histochemical detail [5][6][7] .
In summary, we have compared the vascular labeling characteristics of gP-CD31, mM-CD31, phalloidin and α-SMA on the cerebral blood vessels in this study. The results indicate that gP-CD31 is more sensitive than mM-CD31 for vascular labeling in both healthy and injured rat brain. Furthermore, our results suggest that gP-CD31 can be an ideal candidate, separately or together with other vascular, neural or glial bio-markers, for revealing the vascular network or neurovascular unit in the rat brain under the physiological and pathological conditions.

Materials and methods
Animals. Eight healthy male Sprague-Dawley rats (weighting 200 ± 20 g) were obtained from the National Institutes for Food and Drug Control (license number SCXK (JING) 2017-0005). Animals were housed in temperature-and humidity-controlled rooms (25 ± 1 °C; 45 ± 5%). All experimental procedures were approved by

Animal model of ischemia/reperfusion with MCAO.
Rats were randomly assigned to control and model groups (n = 4). The rats in I/R model group were conducted with MCAO. Four rats were anaesthetized with tribromoethanol solution (150 mg/kg) via intraperitoneal injection. Briefly, the right common carotid artery and external carotid artery were exposed, after ligating the external carotid artery, the internal carotid artery was carefully separated and a small incision was then cut on the right common carotid artery. After that, a 0.28 mm nylon filament was introduced into the distal right internal carotid artery for blocking the right middle cerebral artery until a slight resistance was felt at approximately 18 mm from the carotid artery bifurcation which ensured the occlusion at the origin of the middle cerebral artery. The filament was fastened around the distal of the right common carotid artery. After 1.5 h of occlusion, the filament was gently pulled out about 10 mm to allow complete blood reperfusion of the ischemic area 15,20 . No operations were performed on the rats in control group.
Perfusion. After 24 h survival, the rats in control and I/R groups were anesthetized with an overdose of tribromoethanol solution (250 mg/kg) via intraperitoneal injection, and then transcardially perfused with 0.9% sodium chloride solution (100 mL) followed by 4% paraformaldehyde (PFA) in 0.1 mol/L (M) phosphate buffer (PB, pH 7.4, 300 mL). After perfusion, the brain was carefully dissected out and post-fixed for 2 h, then cryopreserved in 0.1 M PB containing 25% sucrose at 4 °C overnight.
Section. Serial coronal sections of brain were cut at the thickness of 80 μm on a sliding microtome system (Yamato, REM-710) and collected in order in a six-hole Petridish with 0.1 M PB (pH 7.4).
Multiple immunofluorescence staining. The five kinds of immunofluorescence staining were performed in this study, including ① gP-CD31 + Nissl; ② gP-CD31 + mM-CD31 + 4′,6-diamidino-2-phenylindole (DAPI); ③ gP-CD31 + phalloidin + DAPI; ④ gP-CD31 + mouse anti-α-SMA + DAPI; ⑤ gP-CD31 + phalloidin + mouse anti-α-SMA. The sections were incubated in blocking solution containing 3% normal donkey serum and 0.5% Triton X-100 in 0. Observation. The samples were firstly scanned with the panoramic tissue slice scanner (VS120, Olympus, Japan). Accordingly, the selected cortical regions were further observed and recorded by using a laser confocal imaging system (FV1200, Olympus, Japan). Single image (10 ×, 1260 μm × 1260 μm) was scanned at a resolution of 1024 × 1024 pixels (pinhole = 152 μm), and a series of images (about 40 images were obtained with Z-stacks in the interval of 2 μm/step) in magnificent view (40 ×, 310 μm × 310 μm) were captured at a resolution of 640 × 640 pixels (pinhole = 105 μm). Three-dimensional reconstruction of the cerebral blood vessels was performed using www.nature.com/scientificreports/ Imaris 7.7.1 software (http:// www. bitpl ane. com) as previous described 16,53 . Briefly, the .oib file was dragged and dropped into the main window of Imaris, and the blend mode was used to display the data. Then Edit menu was accessed and Show Display Adjustment was selected. Subsequently, the arrows were moved as needed to adjust each channel's brightness and contrast of the images to optimize the visualization. Care was taken so as not to remove any data from the images. Images were processed with Adobe Photoshop CS6 (Adobe Systems, San Jose, CA, USA) and marked with Adobe Illustration CC 2019 (Adobe Systems, San Jose, CA, USA).

Statistical analysis.
The areas of the ischemic hemisphere and contralateral hemisphere were quantitatively analyzed using the analysis system of the panoramic tissue slice scanner. Three lower power images per rat (10 ×, 1260 μm × 1260 μm) from every 12th coronal brain sections were used to calculate the mean intensity of gP-CD31, mM-CD31, phalloidin and α-SMA labeling by using Image J (National Institutes of Health developed imaging processing program, Bethesda, Maryland, USA), respectively. Besides, the percentage of gP-CD31, Nissl, phalloidin and α-SMA labeling was also analyzed by Image J respectively. All data were expressed as mean ± standard error (SEM) and processed with GraphPad Prism version 8.2.1 (La Jolla, CA, USA). Two-tailed t-test was applied for the comparison between two groups, and a one-way analysis of variance (ANOVA) was carried out followed by Tukey test for multiple group comparisons. P < 0.05 was considered statistically significant.
Ethical approval. This study was approved by the ethics committee of Institute of Acupuncture and Moxibustion, China Academy of Chinese Medical Sciences, Beijing, China (No. D2018-09-29-1). All experimental procedures were approved by the institutional and local committee on the care and use of animals. The study is reported in accordance with ARRIVE guidelines.

Data availability
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request. www.nature.com/scientificreports/