|Home | About | Journals | Submit | Contact Us | Français|
Blood brain barrier opening during ischemia follows a biphasic time course, may be partially reversible, and allows plasma constituents to enter brain and possibly damage cells. In contrast, severe vascular disruption after ischemia is unlikely to be reversible and allows even further extravasation of potentially harmful plasma constituents. We sought to use simple fluorescent tracers to allow wide-scale visualization of severely damaged vessels and determine whether such vascular disruption co-localized with regions of severe parenchymal injury.
Severe vascular disruption and ischemic injury was produced in adult Sprague Dawley rats by transient occlusion of the middle cerebral artery (tMCAo) for 1, 2, 4, or 8 hours, followed by 30min reperfusion. Fluorescein isothiocyanate-dextran (2 MDa) was injected intravenously prior to occlusion. After perfusion-fixation, brain sections were processed for ultra-structure or fluorescence imaging. We identified early evidence of tissue damage with Fluoro-Jade staining of dying cells.
With increasing ischemia duration, greater quantities of high molecular weight dextran-FITC invaded and marked ischemic regions in a characteristic pattern, appearing first in the medial striatum, spreading to the lateral striatum, and finally involving cortex; maximal injury was seen in the mid-parietal areas, consistent with the known ischemic zone in this model. The regional distribution of the severe vascular disruption correlated with the distribution of 24-hour TTC pallor (r=0.75, p<0.05) and the cell death marker Fluoro-Jade (r = 0.86, p<0.05). Ultrastructural examination showed significantly increased areas of swollen astrocytic foot process and swollen mitochondria in regions of high compared to low leakage, and compared to contralateral homologous regions (ANOVA p<0.01). Dextran extravasation into the basement membrane and surrounding tissue increased significantly from 2 to 8 hours of occlusion duration (Independent samples t-test, p<0.05).
Severe vascular disruption, as labeled with high molecular weight dextran-FITC leakage, is associated with severe tissue injury. This marker of severe vascular disruption may be useful in further studies of the patho-anatomic mechanisms of vascular-disruption mediated tissue injury.
Pathological responses to ischemia in the microvasculature play a central role in the evolution of infarction; a critical event following ischemia is blood brain barrier (BBB) breakdown1, an antecedent event to cerebral infarction and hemorrhagic transformation2. Increasing awareness of the interplay between vessels, glia and neurons has led to improved understanding of the mechanisms of infarction3 and has partially begun to explain the failures of previous neuroprotective therapies. In parallel with new understanding of the neurovascular and glial-vascular unit, preliminary data suggests direct cytotoxicity of serum constituents, such as thrombin and plasminogen4, in addition to the toxic effects of water entry due to oncotic pressure shifts. The sequence of events is complex, however, because these same compounds also could arise de novo in injured parenchyma, or in the endothelium. Traditional studies of BBB leakage relied upon either simple measures of water flux (edema), leakage of small molecular weight markers (IgG or albumin labeled with Evan’s Blue), or very complicated and expensive ultra-structural imaging of endothelial cells, so it has been difficult to fully characterize the patho-anatomic mechanisms of injury following severe vascular disruption and to separate the effects of edema (water shift) from other putative toxic molecules. Further progress in these investigations is limited by 1) a paucity of data regarding the time course of BBB leakage to differentially sized markers; 2) the absence of a simple, reliable marker of severe vascular disruption; and 3) quantitative measurements of leakage over time after severe vascular disruption. We sought to characterize severe vascular disruption with high molecular weight dextran-FITC using fluorescent, immunohistochemical, and ultrastructural confirmation, and then compared such vascular disruption to evidence of tissue injury.
All protocols were approved by the Animal Research Committee of the Veteran's Affairs Medical Center, San Diego and by the IACUC of UCSD, following all national guidelines for the care of experimental animals. The (n=71) subjects were adult male Sprague-Dawley rats (Harlan, San Diego, CA), average weight 300 g. All animals received tail-vein injections of fluorescein isothiocyanate (FITC) conjugated to a high molecular weight dextran (2 MDa; Sigma); 0.3 ml of 5 % (w/v) solution in sterile phosphate buffered saline (PBS) at the start of surgery, e.g. about 20 minutes prior to occlusion of the middle cerebral artery. The subjects were allowed to awaken from anesthesia during the occlusion and reperfusion periods to allow neurological exams with a dichotomized version of the published rodent neurological grading system 5, 6. To assure sufficient ischemia with the MCAo, only animals that registered abnormal on 3 behavioral signs were used, otherwise the subject was excluded from further analysis. Animals were also excluded for subarachnoid hemorrhage found at post-mortem dissection.
We used our version of the standard model of filament occlusion of the middle cerebral artery5 7. Briefly, animals were induced with isoflurane anesthesia and maintained with a mixture of 4% isoflurane in oxygen:nitrous oxide 30:70 by face mask. After adequate anesthesia and aseptic preparation, an incision was made in the neck, exposing the left common carotid artery. The external carotid and pterygopalatine arteries were ligated with 4-0 silk. An incision was made in the wall of the common carotid artery, which was then threaded with a 4-0 nylon suture (Ethicon, Piscataway, NJ) that was blunted in a microforge (Narishige MF83, NY); filament diameters were measured using microscopy and image analysis and only filaments between 290 and 310µm were selected for further use. The suture was advanced 17.5mm from the bifurcation point of the external and internal carotid arteries, thereby blocking the ostium of the MCA. At the end of the reperfusion period, the rat was euthanized with an overdose of pentobarbital and then intracardially perfused with 200 to 300ml saline followed by 300ml of 4 % (w/v) paraformaldehyde in PBS.
After rapid removal from the skull, each brain was post-fixed in 4% (w/v) paraformaldehyde in PBS and then cryoprotected in 30% sucrose to obtain 50µm thick sections with a freezing sliding microtome. To characterize the distribution of high molecular weight dextran-FITC, sequential sections through the anterior-posterior axis of the MCA territory subsuming about 4.5 mm of brain were sampled from about −0.3 bregma as an anchoring level and mounted onto glass slides. Sections were cover-slipped with Pro-long Gold Antifade mountant (Molecular Probes, Eugene, Oregon). An additional series of sections was slide-mounted for determination of regional co-localization in the ischemic striatum of retained fluorescein with neuronal degeneration marked by Fluoro-Jade C staining (Chemicon, Temecula CA). Sections processed for immunocytochemistry for light microscopy incubated free-floating in antibody solutions and endogenous peroxidase was quenched with a 10 minute incubation in 3 % (v/v) hydrogen peroxide in PBS. Primary antibody (anti-universal IgG, Vector, Burlingame, Ca) was diluted in a blocking diluent (phosphate buffered saline (PBS) with 10 % (v/v) blocking serum and 0.2 % (v/v) Triton X-100) was applied for 2 days and was followed by incubation for 4 hours in biotinylated anti-rabbit secondary antibody diluted in blocking diluent. Biotinylated secondary antibody was visualized by overnight incubation of sections, in Cy5 conjugated streptavidin (Jackson Immunoresearch, West Grove, PA). Fluorescent immunostained sections were mounted on slides and cover-slipped with Pro-Long Antifade mountant (Molecular Probes, Eugene, OR). Background staining was assessed in sections processed without primary antibody.
For ultrastructural localization of high molecular weight dextran-FITC after stroke, animals were prepared for transcardial perfusion-fixation and perfused with Ringer’s solution, followed by 4% paraformaldehyde and 0.1% glutaraldehyde in PBS solution. Brains were removed from the skull, fixed in 4% buffered paraformaldehyde overnight, and cut into 100um thick slices on a Leica VT1000S microtome. Bound fluorescein was visualized by incubation of sections with biotinylated anti-fluorescein antibody (BA-0601, Vector, Burlingame Ca; 1:1000 dilution) for 1 day followed by peroxidase catalysis of diaminobenzidine reporter (ABC kit PK6100 Vector and DAB kit, SK4100 Vector). Immunostained brain slices were post-fixed in 2.5% glutaraldehyde for 15 minutes on ice and then 1% osmium tetroxide for 1 hour, dehydrated, embedded in Durcupan, sectioned at 50nM on a Reichert-Jung Ultracut E system, and mounted on coated copper grids.
Fluorescence from retained high molecular weight dextran-FITC was quantified within the hemisphere by semi-automated image analysis. Digitized images of the ischemic half of the brain were taken with a Zeiss microscope outfitted with CCD camera (KAF32MB; Apogee, Auburn, CA). Images were obtained at 500µm intervals across the anterior to posterior axis of the MCA territory. Fluorescence was quantified using Image Pro Plus (Cybermedia, Bethesda, MD). An operator without knowledge of the subject’s group or occlusion duration examined each section after first setting the magnification and performing a linear calibration using a scale bar. The operator then examined each section and set the brightness and contrast levels to optimize the appearance of the fluorescence. Using semi-automated thresholding, segmentation, and size filtering the operator measured the total area of fluorescence. Total fluorescence typically consisted of multiple discrete ‘islands’ on each section, and within each island there were pale areas of extravasated label intermixed with areas of very bright vascular labeling. Using an image of the islands as an overlay, the operator then re-thresholded to emphasize the bright objects contained within the islands of total fluorescence, i.e., labeled vessels, and again using segmentation and size filtering, the area of all bright vessels was measured. The area of extravasation was obtained by subtracting the area of bright fluorescence in the vessels from the total area fluorescence.
To quantify the presence of multiple fluorescent markers on single sections, we adapted a laser-scanning technique. Image acquisition was performed on an Olympus BX50 Microscope retrofitted with a CompuCyte™ laser scanning cytometry (LSC) acquisition system (Cambridge, MA). Tissue was illuminated with a focused argon laser (488nm) and fluorescein fluorescence was collected through an emission filter of 530±30nm. Cy5 fluorescence was illuminated with a focused helium-neon laser (633nm) and fluorescence was collected through an emission filter bandwidth of 675±50nm. The fluorescence was averaged over 20µm diameter bins (scanned areas) that encompassed the entire tissue section. Histograms were constructed to plot the ROIs or “counts” as a function of integrated fluorescence intensity in those areas. Background was determined by scanning sub-areas on the non-occluded side of the section to obtain a histogram of the distribution of background signals. We conservatively determined a threshold for “signal” according to the maximum level of tissue background. Data was expressed as the number of scanned areas detected above background fluorescence intensities, divided by the total number of scanned areas/section. In cases with fluorescent immuno-staining, similar routines were imposed to quantify immunostained regions of interest in register with fluorescein-Dextran retention sites. The percent of counts with double fluorescent signals (fluorescein and Cy5) was determined from scattergram plots that were divided into four quadrants using Wincyte™ software.
Two photon laser scanning microscopy was used to scan and reconstruct labeled vessels and extravasated fluorescent markers. Stacks of optically sectioned images were acquired with a two photon laser scanning microscope of custom design8 using the MPScope software9. We used a 40×, 0.8 NA water dipping objective (IR Achroplan, Carl Zeiss Inc, Oberkochen, Germany), 0.4 µm per pixel lateral sampling, and 0.5 µm per plane axial sampling. Fluorescein-isothiocyanate-dextran was excited at 800 nm and the fluorescence was detected by low-pass filtering of the emission light below 700 nm. Image rotation and projection operations were performed with the ImageJ software program. (NIH, Bethesda, MD).
Blocks for plastic embedding were selected, based on the immuno-staining for high molecular weight dextran-FITC above, and categorized as originating from a region of high or low leakage. Blocks were also selected from homologous regions in the contralateral hemisphere. After processing as above, images were taken at 5000× on a JEOL 1200EX microscope. An examiner blind to the region of origin for each image then placed a 6×6 grid (with an interval of 2µm between grids) on the images centered on a vessel. Using standard stereological technique10 the grid crossings that hit the swollen astrocytic food processes or structureless space were counted as points of “swollen cells”. For the points aligning on the boundary of grids, only those on the right and bottom side were counted. To normalize the results, the points that corresponded to an area of endothelial cells (including lumen space) were counted as well. Enlarged abnormal mitochondria were counted separately in each field of view. The point counts were summarized and normalized to the containing space, standardized to the calibrated grid.
Transient MCAo caused uptake of circulating high molecular weight dextran-FITC into vessels and extravasation into parenchyma as shown in Figure 1. The rest of the tissue section was not fluorescent because the saline perfusion removed intravascular dextran-FITC at sacrifice11. At longer occlusion times the sub-regional distribution of fluorescence expanded to include the more lateral aspect of the striatum (Fig. 1B). Islands of leakage appeared in the cortex after 8 hours MCAo (Fig. 1C). Image analysis identified regions showing both labeling of the vessel walls as well as parenchymal extravasation (Fig. 1D). With increasing occlusion duration, vascular damage increased in the regions supplied by the middle cerebral artery, as indicated by a significant increase in the accumulated dextran-FITC (Fig. 1E).
To demonstrate that the extravasated fluorescence seen in Figure 1 was truly extravascular, we used 2-photon imaging to optically section and reconstruct a labeled vessel and adjacent leakage (Fig. 2). Epifluoresence microscopy demonstrates a swath of intensely fluorescent vessel segments (Fig.2A) on the ischemic side of the brain. In a subset of labeled vessels, a hazy fluorescence also appeared to extend into the neighboring parenchyma (Fig. 2B). In the 2-photon maximal projections (Fig. 2C and D) the intense vascular fluorescence was associated with the vessel wall. The vessel lumen was clear (Fig. 2D), consistent with the effective washout of labeled plasma with transcardial perfusion fixation). This suggests that macromolecular dextran-FITC can lodge at high concentration in ischemic endothelial cells and also escape from these vessels into surrounding parenchyma. Animals (n=6) were infused with high molecular weight dextran-FITC intravenously and then subjected to transient MCA occlusion of 2 or 8 hours, followed by 30 minutes reperfusion. Sections from the mid parietal cortex were processed for optimal ultrastructural visualization as in Methods, and three blocks of tissue were selected based on the pattern of high molecular weight dextran-FITC labeling to include regions with high dextran distribution, low dextran distribution, and contralateral site (Fig. 3). To image the ultrastructural deposition of bound fluorescein-dextran, the tissue was reacted with anti-fluorescein antibody and converted to diaminobenzidine (DAB), an electron dense reaction product. Ischemic vessels that were labeled with DAB were distinguished by electron-dense cytoplasmic labeling in constituent endothelial cells. These labeled endothelial cells were typically swollen (Fig. 3C). Endothelial cells of vessels from low leakage ischemic areas or from the contralateral side were not obviously enlarged. (Fig. 3D and E). We used stereological analysis to quantify ultrastructural changes in tissues surrounding vascular disruption as labeled with high molecular weight dextran-FITC (Fig. 4). Ipsilateral to the occlusion, the total areas of endothelial cells (including lumena) were slightly decreased compared to the opposite control side, consistent with edematous compression (data not shown), but the ratio of endothelial cell area to lumen was markedly increased in areas of high leakage, compared to areas of low leakage or contralateral side (Fig. 4C, p<0.001, one-way ANOVA, Tukey’s post hoc test.) We demonstrated significant increases in the density of swollen mitochondria and astrocytes, and empty voids indicative of severe edema (Fig. 4C). In regions of high leakage the average area of astrocytic end-feet and structureless spaces were significantly larger—consistent with swelling as seen in cytotoxic edema—compared to areas with lower amounts of high molecular weight dextran-FITC signal, or corresponding sub-regions on the contralateral side (p<0.001, one-way ANOVA, Tukey’s post hoc test). Greater numbers of swollen mitochondria were found in association with vascular labeling with high molecular weight dextran-FITC (Fig. 4C, p<0.001, one-way ANOVA, Tukey’s post hoc test).
We demonstrated tissue damage associated with areas of severe vascular disruption using multiple approaches. In 11 animals with variable durations of occlusion followed by reperfusion until 24 hours after onset of occlusion, the area of high molecular weight dextran-FITC leakage correlated well with the volume of tissue damage as labeled by 2,3,5-Triphenyltetrazolium chloride (TTC) staining (Fig. 5, correlation coefficient r=0.75, r2=0.56, p<0.05). Neuronal degeneration, labeled with Fluoro Jade, was also observed in regions of high molecular weight dextran-FITC leakage in 10 animals (Fig. 6, r=0.86, r2=0.75, p<0.05). These findings together establish that high molecular weight dextran-FITC leakage serves to identify areas of severe vascular disruption and ischemic tissue injury.
To determine whether vascular disruption labeled with high molecular weight dextran-FITC merely served to identify areas of BBB leakage, which could be reversible, we compared the distribution of IgG leakage to vascular disruption in 20 animals (Fig. 7). We found a significant dissociation between BBB leakage as labeled with IgG, compared to areas of severe vascular disruption, at multiple durations of MCA occlusion (Fig. 7, one-way ANOVA p<0.001, Tukey’s post hoc test). At all occlusion durations, the area of BBB leakage greatly exceeded the area of vascular disruption. The extent of BBB leakage reached a maximum by 4 hours of occlusion, but in contrast, the time course of severe vascular disruption was slower, and showed greatest leakage at 8 hours, the longest duration we studied.
Our data demonstrate that severe vascular disruption allows passage of high molecular weight dextran-FITC that is time-dependent and maximal in the brain regions made ischemic after occlusion of the MCA (Fig. 1, Fig 2, and Fig 3)12. With longer durations of ischemia, the high molecular weight dextran-FITC label accumulates with greater concentration in areas of basal lamina disruption, endothelial and astrocytic swelling, and eventually with total vascular poration (Fig. 3 and Fig 4). Further, as the degree of vascular disruption increases, there is a corresponding increase in the extent of associated tissue damage (Fig. 4, Fig 5, and Fig 6). The label can be used to identify regions of brain undergoing severe vascular disruption as early as one hour after onset of ischemia (Fig. 7). This suggests that brain regions suffering the most severe degree of ischemic injury after vascular occlusion can be labeled easily with a simple intravenous infusion of an inexpensive fluorescent maker. As shown in Figure 3, the presence of the marker can be used to select tissue that is undergoing significant vascular disruption and tissue damage for further detailed study. To our knowledge, this is the first simple marker of severe tissue injury that can be used easily and reproducibly as early as 1 hour after ischemia onset. We used high-molecular weight dextran-FITC leakage to demonstrate a significant down-regulation of the Aquaporin 4 receptor in regions of severe vascular disruption, further supporting the relationship between severe vascular disruption and tissue injury13.
The loss of BBB function in ischemic vasculature has been extensively studied and well established with a variety of quantifiable tracers including isotopically-labeled proteins and amino acids14–18, isotopically labeled sucrose 19, fluorescent tracers20–22 and with MRI contrast agents23. Typically, quantification is made in terms of average vessel leakiness. The time course of increased leakage observed in the present study is consistent with that observed by previous spectrophotometric methods21. Additionally, the gross regional specificity of our results are consistent with the low-resolution spatial patterns of BBB observed with MRI, which demonstrate early contrast enhancement (2.5 hours of occlusion) in the ventral striatum of the adult rat23.
The mechanism of high molecular weight dextran leakage is not established unequivocally by our data, although inspection of the ultrastructure (Fig. 3 and Fig 4) suggest that both increased transcytosis and gross cellular poration are involved. Using similar approaches, the leakage of small molecular weight albumin was shown to precede and show more extensive staining than the leakage of large molecular weight dextran12. Considerable literature addresses the leakage of smaller molecular weight molecules via loosening of the tight junctions between endothelial cells but there is debate over the roles of transcytosis and transmembrane poration24–26. Our data do not address the role of tight junction loosening in the mechanism of higher molecular weight dextran leakage, but the time course and spatial distribution of vessel ischemia, as demonstrated by high molecular weight dextran-FITC uptake, is consonant with the molecular events that underlie BBB breakdown after experimental large vessel occlusion27. Proteolytic breakdown of BBB structural molecules19 is an early event that occurs within 1 to 2 hours after an ischemic insult19, 27–31. Immunocytochemical imaging studies of molecular changes in other neural compartments illustrate that they also occur in a spatially heterogeneous fashion32, 33, forming “islands” of altered protein expression that have been purported to represent small infarctions in both the ischemic striatum and cortex. The temporal and spatial development of severe vascular disruption (Fig. 1) appears to coincide with these well-documented molecular events and while our data does not allow us to confirm a mechanistic link, the technique we present here will greatly facilitate such mechanistic investigations in the future.
Our study comes with some limitations. High molecular weight dextran-FITC is detected directly, without amplification; other markers, including the low molecular weight marker IgG, are detected after antibody labeling which could amplify the signal. The degree of difference between IgG leakage vs high molecular weight dextran leakage (Fig. 7) likely exceeds the possible amplification step, but such an effect cannot be excluded by our data. Tissue injury markers such as TTC exclusion and Fluoro-Jade uptake, while standard in the field, are not synonymous with cell death, nor do they differentiate between necrotic and apoptotic cell death mechanisms. Indeed, we cannot assert that the severe vascular disruption identified with high-molecular weight dextran-FITC marks regions of brain undergoing irreversible ischemic damage that will inevitably become infarct, although it is difficult to believe that areas showing such severe injury (Fig. 3 and Fig 4) could recover. Further studies are needed to 1) show that such marked areas do not possess the ability to recover; 2) identify the mechanisms of severe vascular disruption; and 3) determine if protective agents that inhibit severe vascular disruption also block the associated tissue injury. Another limitation is the relatively biased sampling strategy we used for selecting sections for ultrastructure. On the one hand, a purely random sampling strategy would be unbiased, and might allow one to determine whether evidence of severe vascular disruption is found only in areas of greatest high molecular weight dextran leakage. On the other hand, such a sampling strategy is difficult for cost and time reasons. To overcome this selection bias, all sections were reviewed in semi-blinded fashion. In other words, in double labeling experiments, images were taken from all sections using one label, then scrambled, and all sections were re-imaged using the second label. Similarly, in the planimetry and point counting assessments, the investigator examined all sections of one label, then reviewed all the sections using the other label after scrambling de-identified sections. With these limitations in mind, the argument that high molecular weight dextran leakage occurs in areas of vascular disruption and tissue injury is supported by the fact that we used multiple, complimentary imaging techniques, all of which clearly show the same, highly statistically significant relationship. We specifically hypothesized that vascular disruption and tissue injury would occur in areas of greater dextran leakage, and our data support this hypothesis directly (Fig. 5 and Fig 6).
We have demonstrated that a simple, inexpensive, intravenous marker—high-molecular weight dextran-FITC—reproducibly labels brain regions undergoing severe vascular disruption and associated tissue injury prior to the development of obvious infarction. Labeling is greatest in the regions of brain known to suffer the greatest degree of ischemia after MCAo, and increasing durations of ischemia cause greater amounts of labeling. This marker will allow for further studies of the mechanisms of vascular injury, and the relationships between vascular, glial, and neuronal cell injury mechanisms, by providing a simple way to identify severely damaged tissue as early as one hour after ischemia onset.
We thank Rodolfo Figueroa for fabricating occluding filaments, Kai Yang for assisting in fluorescence quantification, Judy Norberg for performing the laser scanning cytometry, Dr. Maryann Martone for helpful suggestions in Ultrastructural imaging, and Drs. Donald Pizzo and the late Leon Thal for use of their photomicroscope. BC thanks the HHMI-NIBIB Interfaces Training Program at UCSD. This work was funded by the Veteran’s Affairs Medical Research Department (P.D.L), by the National Institute of Health grants NS/043300 and NS/052565 (P.D.L.), NS/041096 and EB\003832 (D.K.).