Acute ischemic stroke (AIS) is a frequent cause of death and adult disability. AIS patient management targets the ischemic penumbra: Hypoperfused, electrically silent brain tissue, which can be salvaged by restoring blood flow during the first, critical hours after symptom onset. Neuroimaging studies in AIS patients suggest that penumbral tissue is characterized not only by hypoperfusion, but also by microvascular flow disturbances that strongly affect tissue outcome. Here, we demonstrate that microvascular flows become increasingly chaotic in the ischemic penumbra in the hours after middle cerebral artery occlusion in a rat model of AIS. Biophysical models suggest that these disturbances are accompanied by increasing hypoxia in the absence of blood flow changes. Unlike findings in severe ischemia, pericyte constrictions do not appear to occlude penumbral capillaries. We propose that microvascular flow disturbances represent a critical feature of penumbral tissue, and a potential target for neuroprotective therapy after AIS.
This manuscript provides a useful characterization of penumbral microvascular flow disturbances over the first hours after ischemia onset in a rat model of middle cerebral artery occlusion. This work suggests that there are microscopic changes (including directionality of capillary blood flow and formation of capillary stalls) during the peri-ischemic timeframe but the data are incomplete and not sufficient to support the claims.
Stroke is the second-most common cause of death, claiming 5.8 million lives worldwide in 20161. Two out of three patients survive their stroke, but about 50% do so with permanent disabilities. As a result, stroke was also the second-most common cause of disability-adjusted life years lost to disease in 20162. The majority of acute strokes are the result of focal brain ischemia caused by arterial occlusion, and the acute management of these patients therefore aims to limit brain infarction by restoring blood supply to the affected tissue3. The pathophysiological rationale for this therapeutic approach is the existence of an ischemic penumbra4: A volume of electrically silent, critically hypoperfused tissue, which can be salvaged by reperfusion within the first few, critical hours after symptom onset5. Penumbral tissue is further defined by its residual blood supply: while tissue injury ensues almost immediately at cerebral blood flow (CBF) below 8-12 mL/100mL/min6, brain tissue can survive for some time at CBF levels below the ischemic threshold of approximately 20mL/100mL/min4,5. Left untreated, the ischemic core lesion therefore grows as penumbral tissue dies, and penumbral contributions to the patient’s neurological symptoms become permanent. Over the past decade, intravenous administration of thrombolytic agents has been shown to improve functional outcome for patients when administered within 4.5 hours after symptom onset2 and in wake-up strokes7 in eligible patients, while endovascular clot-removal is proven safe and efficacious up to 24 hours after symptom onset2.
Several reports suggest that knowledge of CBF alone does not suffice to predict tissue viability after acute ischemic stroke. Perfusion magnetic resonance imaging (MRI) studies in acute stroke patients (< 12 hours after symptom onset) show that the survival of tissue with a certain level of hypoperfusion, as indexed by the prolongation of blood’s mean transit-time (MTT) through each image voxel’s vasculature, also depends on the within-voxel distribution of the blood flow8-10. Accordingly, tissue with an abnormally narrow within-voxel flow distribution on acute MRI shows extreme risk of subsequent infarction in the absence of reperfusion therapy – largely independent of its level of (hypo)perfusion8-10. Quantifying this phenomenon by the coefficient of variation (CoV) of within-voxel vascular transit times (their standard deviation divided by their mean, MTT)11, Engedal et al.12 found that low CoV in acute (< 6 hours after symptom onset) stroke patients is a common property of severely hypoperfused tissue (as measured by MTT), and associated with high risk of infarction if reperfusion cannot be achieved. Notably, in the absence of reperfusion, hypoperfused tissue with normal CoV shows much lower infarct risk than equally hypoperfused tissue with low CoV, confirming that not only the level of hypoperfusion, but also the extent of microvascular flow disturbances, impact the fate of hypoperfused tissue12. Reperfusion, meanwhile, reduces infarct risk of tissue with low initial CoV to the level of similarly hypoperfused tissue with normal CoV12.
The dependency of penumbral fate upon the microscopic distribution of residual blood flow raises an intriguing question: can hypoperfused tissue be protected by restoring or preserving bloods microvascular distribution – irrespective of whether subsequent reperfusion can be achieved? Penumbral microvascular flow disturbances might reflect reversible instances of phenomena observed during more severe ischemia, such as compression by swollen astrocytic end-feet, capillary obstructions by recruited immune cells, or constriction due to capillary pericyte rigor13-18. Indeed, if blood’s microvascular distribution could be preserved by therapies that are suitable for pre-hospital administration (See e.g., Gaudin et al.19), stroke patients might benefit irrespective of their access to or eligibility for subsequent reperfusion therapy.
MRI-based CoV-observations summarize the hemodynamics of thousands of interconnected microvascular segments within 10 mm3 image voxels11. With respect to the nature of underlying microvascular flow disturbances, CoV estimates are therefore ambiguous, as they could reflect either compensatory homogenization of blood flows through open capillaries to optimize oxygenation as perfusion pressure drops8,20, gradual cessation of flow through microvascular pathways with long transit times12, or both. Although ischemia-related microvascular changes have been reported18,21-24, these studies have not, or to a limited degree, distinguished between core and penumbra tissue.
The aim of this study was therefore to characterize microvascular flow disturbances in penumbral tissue in a rat model of acute ischemic stroke and their evolution over the critical, first four hours after ischemia onset. First, we adapted the MRI CoV-method for dynamic two-photon microscopy (TPM) to examine the hemodynamics of the microvasculature that connects individual, diving arterioles and ascending veins in ischemic cortex. Next, we characterized the passage of erythrocytes through individual capillaries in the cortical microvasculature in order to characterize CoV changes in terms of the underlying hemodynamic and their severity. To this end, we applied biophysical models to estimate the impact of microcirculatory flow disturbances on penumbral oxygenation, and examined whether pericyte constrictions18,25 might affect capillary hemodynamics within penumbral tissue.
Dissociated penumbral micro- and macrovascular response to middle cerebral artery occlusion
Following filament occlusion of the middle cerebral artery (MCAo) in Sprague-Dawley rats, we observed immediate reductions in the diameters of cortical pial arterioles (Fig. 1a,b) and in CBF as measured by the Laser Speckle Contrast Imaging (LSCI) signal over the affected cortical area (Fig.1c). We used the extent of LSCI signal reduction to subdivide hypoperfused cortex into penumbral, core, and control tissue, respectively21-22,26-28 (Fig. 1d), and verified the fate of penumbral tissue by TTC staining (Fig. 1e and Supplementary Fig. 1d).
Unlike the cortical LSCI signal (Fig. 1f), RBC velocities and flux (Fig. 1i,h), as measured by TPM line scans across 530 capillaries (270 in controls and 260 in stroke animals in the 100-250μm depth range), remained largely unaltered in penumbral micro vessels compared to those of control tissue. Also, capillary diameter did not change significantly (Fig. 1g), unlike pial artery diameter (Fig 1a,b). See Supplementary Material Fig. 2 for corresponding linear density and velocity variance measurements over time.
Erythrocyte flows in penumbral capillaries deteriorate over time
Although average LSCI flow and capillary RBC fluxes remained largely constant in penumbral tissue for the duration of the experiment, RBC flow dynamics in individual capillaries grew increasingly chaotic over time. Observing RBCs within capillaries by two photon laser scanning microscopy (TPM) line-scanning (Fig. 2a), we observed instances of flow reversal in penumbral capillaries during 30 sec observation periods, as illustrated in figure 2b. Over the 4h observation period, the proportion of capillaries with flow reversal increased dramatically, ultimately involving as many as 25 % of the observed capillaries (Fig. 2d). This phenomenon was accompanied by a high incidence of stalled flow: capillaries with zero flux comprised 15-35% of the observed capillaries throughout the duration of the experiment (Fig. 2c). Capillary red blood cell (RBC) velocities were determined the angle resulting from their movements within FITC filled capillaries during TPM line scans (Fig. 2a). In addition to stalled flow and flow reversal, we observed increasing variability in penumbral RBC velocity when compared to control tissue. RBC variance and linear density are shown in Supplementary Fig. 2. We note that microvascular haematocrit was elevated in penumbral capillaries over the measurement period.
Progressive capillary transit time disturbances across penumbral tissue
To determine the course of hemodynamic disturbances in terms of the CoV of capillary transit times, we performed dynamic TPM of arteries and veins during bolus dye injection (Fig. 3) and conducted indicator dilution analysis equivalent to that of MRI studies to determine the mean and the standard deviation of plasma transit times as blood passes through the cortical microcirculation30 (Fig. 3d,e). Referring to the two as the mean transit time (MTT) and transit time heterogeneity (CTH), respectively, CTH is expected to change in proportion to MTT in normal, passive compliant microvascular networks31 such that their ratio, CoV, remains constant. Meanwhile, MTT is proportional to the dye’s vascular distribution volume (plasma) and inversely proportional to the plasma flow between arteriole and venule.
As expected, penumbral MTT values increased after vascular occlusion, reaching a peak at 120 min after which they decreased to values similar to those of control tissue after 210 minutes (Figure 4a). This biphasic behaviour was paralleled by penumbral CTH changes (figure 4b) such that penumbral CoV remained similar to control values until 180 minutes, after which penumbral CoV was significantly lower that of control tissue at 210 minutes (figure 4c).
We note that the apparent ‘normalization’ of MTT and CTH after 120 minutes happens as an increasing proportion of capillaries becomes affected by stalled flow and flow reversal. As penumbral blood flow, determined by the LSCI signal, changed little in this time period, we ascribe the decrease in MTT values to reduced vascular distribution volume for the injected dye. This volume is expected to decrease if the diameter or total length of passable microvascular paths from arteriole to venule decreases, or if microvascular haematocrit increases. Indeed, more capillary stalls further reduce the number of passable capillaries, just as increased haematocrit further reduces the dye distribution volume. Our measurements thus suggest that the decreases in MTT and CTH after 120 minutes represent a ‘pseudonormalization’. Simulations of microvascular flows suggest that gradual occlusion of capillaries tend to affect long pathways due to their higher resistance, giving rise to a reduction in CoV and a false impression of a ‘beneficial’ capillary flow homogenization12.
Thirty minutes after occlusion, an unexpected increase in blood flow in both ischemic and control animals was observed, resulting in elevated MTT and CTH values in both animal groups (Fig. 4a,b, Supplementary material Fig. 1b,c). These data are shown with transparent symbols in Figure 4, as it occurred in both MCAo animal and controls and seems to be a feature of our animal preparation, rather than of the stroke model.
Metabolic significance of microvascular flow disturbances
To estimate the severity of the microvascular blood flow disturbances, the impact of the measured MTT and CTH changes on tissue oxygenation were calculated post hoc, using biophysical models32,33 to calculate corresponding, steady state cerebral metabolic rate of oxygen (CMRO2), and tissue oxygen tension (PtO2, Fig. 6). Note how CMRO2 and PtO2 in control animals remained constant throughout the experiment and within the normal range (CMRO2 =3.6-3.8 mL/100mL/min; PtO2 = 10-15 mmHg). In penumbral tissue, however, CMRO2 and PtO2, decreases to reach 1.98 mL/100mL/min and 2 mmHg, respectively.
Capillary pericytes do not affect penumbral RBC hemodynamics
To examine whether hemodynamic changes were affected by changes in pericyte morphology, we performed TPM line scanning of individual capillaries, up- and downstream of pericyte soma, respectively, and measured capillary diameter at the location of the pericyte soma (fig. 5). Pericyte staining revealed a uniform distribution of these cells throughout the tissue volume, localized at capillary branch points as well as along individual capillaries. We found the inner diameter of the capillary to be significantly larger at the position of the pericyte soma than further away (Fig. 5b). RBC linear density was modulated in the same way by the soma passage in MCAo and control animals, while RBC flux was unaffected, confirming that no branching occurred at the location of the pericytes. In MCAo animals, the variance of RBCv was elevated upstream of the pericyte, which we ascribe to the increased, general level of flow disturbances across the capillary network, including capillaries with reversed flow (Fig. 5e). Accordingly, penumbral pericytes do not seem to explain trapped erythrocytes or stalled blood flow in penumbral capillaries (Fig. 2).
Our first main finding is that, during the first few hours following filament-induced MCAo in rats, microvascular flow disturbances evolve in penumbral tissue despite constant, residual blood flow. Thus, a growing fraction of capillaries showed reversal or complete cessation of RBC flows, particularly beyond 90 minutes of ischemia. Similar observations of the cortical microvasculature in murine experimental models during stroke-like conditions have revealed reduced capillary perfusion levels and pronounced microvascular flow disturbances24 in concert with irreversible pericyte constrictions15,18,22,34,34. Our study extends these by using established LSCI thresholds to distinguish between ischemic core and penumbral tissue, and by following the progression of hemodynamic changes over time14. With this distinction, our findings extend previous studies by ascribing microvascular flow disturbances in penumbral tissue to mechanisms other than pericyte constrictions.
Our second main finding is that, according to our biophysical models, oxygen availability dwindles over time in the ischemic penumbra, despite constant residual blood flow. Tissue oxygenation is traditionally gleaned from residual CBF, factoring in the higher oxygen extraction fraction that is characteristic of ischemic tissue. Therefore, current stroke management focuses on restoring blood flow as soon as possible. In parallel, neuroprotective strategies have sought to reduce the vulnerability of neurons to the corresponding, low oxygen levels, although translation of successful drug candidates to humans have failed. Due to the poorer oxygen extraction that results as this residual blood flow has to pass through fewer capillaries, however, oxygen availability deteriorates as a result of these microvascular flow disturbances. Indeed, our calculations suggest that normal oxygen metabolism cannot be sustained for more than 60 minutes of ischemia with deteriorating microvascular flows, according to our biophysical models. Together, our findings thus support the notion that penumbral oxygenation availability depends on both CBF and bloods microvascular, and that failing capillary flows contribute to penumbral infarction. Future studies should therefore examine the neuroprotective effects of therapies that support sustained capillary perfusion during ischemia.
The role of pericytes in capillary flow disturbances during ischemia remains unclear. Studies by Hall et al.25, Yemisci et al18. and Zhang et al34. show that pericytes die and constrict during ischemia, closing RBC flow through individual capillaries. It hould be kept in mind, however, that pericyte tone is affected by multiple factors35 that may change during whole animal- and subsequent tissue preparation. We searched for signs of altered pericyte morphology and function by examining whether pericytes located along capillaries in penumbral tissue might affect capillary diameter (Fig. 5b) or RBC passages, as inferred from the relation between upstream and downstream RBC velocities compared to control animals (Fig. 5f) in vivo, but found no clear indications thereof.
Potential causes of capillary flow disturbances include capillary occlusions or more proximal constrictions. Recent studies by Erdener14,36 found that dynamic stalling erythrocytes in capillaries occurred persistently in salvageable tissue (penumbra) after reperfusion and further that leukocytes were traveling slowly through capillary lumen or were stuck. Further, a decreased number of stalls were associated with improvement in penumbral blood flow within 2-24h after reperfusion along with increased capillary oxygenation and hereby improved functional outcome36. In addition, neutrophils adhering to distal capillary segments has shown to obstruct of 20-30% of capillaries in the infarct core and penumbra37. Grubb et al.38 recently showed the presence of a precapillary sphincter at the transition between the penetrating arteriole and the first capillary that links blood flow in capillaries to the arteriolar inflow and further that global ischemia and cortical spreading depolarization constrict sphincters and cause vascular trapping of blood cells. These results reveal precapillary sphincters as bottlenecks for brain capillary blood flow. We further speculate that capillary flow disturbances relate to swelling of astrocytic endfeet39, as we recently found in in a murine model of subarachnoid hemorrhage40.
Limitations to the study
While we used LSCI to assure that our hemodynamic observations were performed in tissue that was initially penumbral tissue, our experimental design did not allow us to establish when in the course of the four-hour observation period tissue succumbed to irreversible tissue injury.
We see an increased degree of disturbed microvascular flow with gradually more irreversible damage already within the first 90 min of occlusion, which in concert with stalling and reversing RBC flows in individual capillaries led to highly elevated CTH. When the ischemic episode is initiated, the oxygen demands from the compromised penumbra tissue is met by increasing flow at 30 min (1st bolus, Fig. 4d) illustrated by the increased MTT. We speculate that beyond 30 min of ischemic stroke, the further increased flow cannot continue to match the oxygen demand from the tissue as the flow pattern becomes more heterogeneous (increased CTH), and further that at 120 min after the stroke onset, the blood flow cannot sustain the oxygen delivery and the capillary network collapses. Future studies should reveal the relation between deteriorating capillary flows and these penumbral no-flow capillaries.
Our indicator dilution technique relies on repeated dye injections with parallel exposure of the microcirculation to laser light (Fig. 4), which might damage individual capillaries. Our repeated measurements in control animals show that these injections, and the vascular exposures to laser light, do not affect the hemodynamic responses and the estimated parameters. Accordingly, the measured parameters (MTT, CTH, Fig. 4a,b) remain stable during the whole experimental period with 7 injected boluses.
Our experimental approach did not allow us to study individual capillaries immediately after ischemia onset, and we would therefore overlook vessels that were closed to plasma flow as we commenced two-photon imaging. Thus, from our light microscope images (Fig. 1a), we found contracting capillaries 10 min after MCA occlusion, and not all of these were filled with the FITC dextran dye. Such capillaries would not appear in the TPM scanning.
Our data provide direct in vivo evidence that the ischemic penumbra is characterized by deteriorating microvascular hemodynamics, as much as a low but constant, residual CBF. According to our biophysical models, these progressive, microvascular flow disturbances may be major contributors to the demise of penumbral tissue (infarction), consistent with earlier MR data. Our data do not support capillary pericytes as culprits in the disturbances. Further studies should directly quantify the effects of altered capillary blood flow patterns on penumbral oxygen availability and cell survival, and demonstrate that pharmacological improvements of capillary blood flow save penumbral tissue.
[Methods 3000 ord - lige nu: 2286]
Thirty-eight male Sprague Dawley rats (Taconic Biosciences, Denmark, mean w=380±1.7g) were studied as part of the experiments. All rats were $housed in standard cages in groups of two and kept on a 12-hour light:dark cycle with free access to food and water. All experiments were carried out in accordance with the regulations of the Danish Ministry of Justice and Animal Protection Committees. The study was approved by the Danish Animal Inspectorate with licenses no 2013-15-2934-00788 and 2018-15-0201-01443, and complies with the ARRIVE guidelines 2009 (Animal Research: Reporting In Vivo Experiments).
The animals were studied according to one out of three protocols to address the following: Group I. Identification of the penumbra (w=334±3.3g, control n=2, stroke n=6); Group II. Capillary blood flow patterns (w=348±3.3g, control n=7, stroke n=8, Fig 1); and Group III Capillary RBC velocity (RBCv) and effects of pericytes (w=434±3.5g, control n=7, stoke n=8). In all groups, aninmals were randomly assigned to either the stroke or the control group by coin flip. None of the animals were used in more than one experimental protocol. The protocol lasted 4h and measurements were performed either continuously or every 30 min throughout this period.
Surgical preparation and monitoring
The rats were intubated and artificially ventilated with 30% O2 using a small animal ventilator (Harvard Apparatus 683, MA, USA) under Hypnorm-Dormicum anaesthesia (1.8 ml/kg for induction and 1/3 dose every 30 min for maintaining). End-tidal CO2 was continuously monitored (MicroCapStar End-tidal CO2 analyzer, Cwe Inc., Ardmore, USA) and ventilation rate was adjusted to ensure end-tidal CO2 levels between 35-40 mmHg. Body temperature was maintained at 37°C by a heating pad controlled by feedback from a rectal thermometer (Homeothermic monitor, Harvard Apparatus, MA, USA). The right femoral artery was cannulated for continuous recording of mean arterial blood pressure (MAP) and heart rate (fH) using a BP-1 system (WPI Inc., Sarasota, FL, USA) and for monitoring of arterial blood pH, PaCO2, and SaO2 (ABL90 Flex, Radiometer Medical ApS, Brønshøj, Denmark). Three blood samples (60μL) was taking from each animal; before the MCAo, just after the occlusion and at the end of the experiment (Fig. 1). The femoral vein was cannulated in animals in experimental groups II and III for intravenous administration of fluorescence dyes. All physiological data were sampled at 50Hz with a Powerlab 35 series 16 bit data acquisition system (ADInstuments Ltd, Oxford, United Kingdom).
Middle cerebral artery occlusion (MCAo)
Permanent focal cerebral ischemia was achieved by occluding the distal portion of the middle cerebral artery (MCA) as previously described by Belayev41 and Koizumi42, using an intraluminal thread and a 4-0 reusable suture (4041PK5Re, Doccol Corporation, Sharion, USA.). In brief, the carotid bifurcation was exposed and the common carotid artery occluded by a suture. The branches of the external carotid artery were dissected and divided, and the filament were then inserted into the common carotid artery and advanced 20 mm. The filament remained in place during the 4h experimental protocol to achieve a permanent MCA occlusion.
Laser Speckle Contrast imaging
For penumbra identification in experimental group I, rats were placed in a stereotactic frame (Small animal stereotaxic, Kopf Instruments, CA, USA) with ear bars. The skull was thinned by a dental drill (Foredom Electrics CO, CT, USA) under continuous cooling with room temperature saline until the pial vessels became visible. The thinning procedure was done immediately prior to the MCA occlusion (MCAo). To increase the signal from the Laser Speckle Contrast Imaging (Moore Instruments, Devon, UK) and to keep the bone moisturised, almond oil was continuously added. Perfusion was measured continuously during the 4h experimental period after MCAo using a full-field laser perfusion imager (MoorFLPI, Moor Instruments Ltd., Devon, UK). A 785 nm class 1 laser diode was employed for illumination of the tissue down to a depth of approximately 1 mm. Laser speckle images was acquired using a 576 x 768 pixel grayscale CCD camera operating at a frame rate of 25 Hz. After acquisition, images were converted to matlab files and analysed using custom written matlab software.
Cranial window preparation and pericyte labelling
For Two-Photon microscopy (TPM) studies, a chronic cranial window was established immediately prior to the MCA occlusion at the calculated expected position of the penumbra as determined from the laser speckle data above. A metal holding bar was glued to the right frontal bone of the rat to immobilize its head during imaging and a cranial window with 5 mm diameter was drilled through the parietal bone. Drilling was performed during continuous cooling by room temperature saline. After the skull was removed, the dura was peeled off carefully to avoid any bleedings. In order to visualise cortical pericytes, a topical application of Nissl bodies (NeuroTrace 500/525) was performed (Damisah et al., Nature Neuroscience, 2017) repeatedly during 10 min. Afterwards the window was cleaned by saline before being filled with 1.5 % agarose in saline (Sigma-Aldrich, Søborg, Denmark) and covered with a glass coverslip secured with dental acrylic (GC Fuji PLUS dental cement, GC Corporation, Tokyo, Japan). The window was allowed to dry for 45 min before subjecting animals to MCAo prior to TPM.
Two-photon microscopy (TPM)
Imaging was performed using a Praire Ultima-IV In Vivo Laser Scanning Microscope (Bruker Corporation, Billerica, MA, USA). The field of view (FOV) was centred in the cranial window over the penumbra. To characterise the microvascular blood flow changes during the infarction of the penumbra, an indicator dilution technique was applied to determine the distribution of transit times through the cortical vasculature in experimental group II. This technique comprises dynamic imaging of the passage of a bolus-injected dye as previously described by Gutierrez et al29. Briefly, a 10X water immersion objective (Olympus 0.30 numerical aperture-NA, 3.3 mm working distance-WD) was used with a pixel resolution of 1.16-2.23μm per pixel (1x-2x) depending on the optical zoom. To adjust the FOV and set the coordinates, shadows originating from the pial vessels vessel were visualized by NADH auto-fluorescence using second harmonics (laser λ=810nm, optical parametric oscillator (OPO, λ=1100). A 50μl bolus of 0.5% Texas-red dextran solution (70,000 MW, 5 mg/mL in 0.9% NaCl, t1/2 ∼ 25min, Termo Fisher Scientific) was injected at a constant rate of 30μl/sec using a syringe infusing pump (GenieTouch, Kent-Scientific, Torrington, CT, USA) while performing a 20 sec spiral-scan at 6.25fps within a single plane (512x512 pixels, dwell time per pixel=1.2μsec). Fluorescent emission was detected by a GaAsP-PMT (Hamamatsu, H7422-40) using a 660/40 nm-emission filter to optimize signal to noise ratio. Laser λ=810nm, OPO λ=1100nm.
Arteries and veins were identified based on the timing of dye arrival (Fig. 4a). After identification of vessels within the FOV, a scan path was defined though the pial vessels of the upper cortically layer by applying the TPMs free-hand drawing tool (PraireView, Bruker Corporation, Billerica, MA, USA). Scanning lines were drawn to cross the largest artery and vein within the FOV as well as arterioles and venules. Subsequently, repeated line-scans were performed (6.7 msec per line-scan path dwell-time per pixel 1.2μsec) for a total scan time of 60 sec, during which the dye was injected after 10 sec. Bolus passage measurements were repeated every 30min for the entire 4h experimental period. To obtain an angiogram that displays the anatomical relation of the vessels, a z-stack was acquired on a FOV of 1.18μm2 (10x objective) from the pial surface down to a depth of 300-400 μm.
In order to characterise capillary RBC velocity (RBCv) and effects of pericytes in experimental group III, single capillaries were scanned to access RBCv and RBC flux (RBCflux) with a 20X water immersion objective (Olympus, 1.0 NA 2.0 mm) with a pixel resolution between 0.19-0.23μm per pixel depending on the optical zoom used (5x-6x). A single bolus of 50μL 0.5% Texas-red dextran solution (70,000 MW, 5 mg/mL in 0.9% NaCl, t1/2 ∼ 25min, Termo Fisher Scientific) was injected, and capillaries were defined as vessels showing single cell passages within the vessel lumen. Two scan paths were performed on each capillary, along the axis for RBCv estimation, and a transversal scan for RBCflux and diameter assessment. This was done on both side of the position of pericyte, and diameter scans were taken just next to the pericyte body and at the far end of the capillary. Laser λ=810nm, OPO=1100nm, Laser for pericyte λ=1000nm. Four sub-FOV at depths of 80-150μm was chosen within the main FOV and all capillaries in each sub-FOV (6-10) were scanned every 30 min during the 4h experimental period. Individual capillaries were scanned for 30 sec in the depth of 80-150μm.
Laser Speckle image analysis and penumbra identification
The spatial extend of the ischemic penumbra was identified from the laser speckle images. First brain pixels were isolated by manually outlining brain pixels and a centre line was defined to separate the control and stroke hemispheres (Fig. 1d, Supplementary materials Fig. 1). Larger vessels had the highest speckle contrast and the corresponding pixels were masked out by applying intensity thresholds. Laser speckle has previously been used to visualise CBF changes throughout the ischemic territory in stroke models in rodents and cats22,26,28,43. Based on this approach, penumbral regions were defined by the group of pixels with intensities within the interval 25%-50% of the maximum intensity of the control hemisphere33,44 (fig. 1d,e). Two ROIs were drawn to represent the penumbra and the core within the stroke hemisphere, and these regions were mirrored along the center line to define corresponding ROIs in the control hemisphere. Finally, pixel intensities were averaged within the ROIs (control ROIs pooled, Fig. 1d).
Bolus tracking image analysis
To identify the primary inputs and outputs for the bolus tracking analysis within each FOV, the vessels were separated according to their diameter and the time-to-peak (TTP) of their concentration curve (CTC). The arterial input function (AIF) was the CTC with a large diameter, which was first to enhance (i.e. had the shortest TTP, Fig. 3b), whereas the venous output function (VOF) was identified as the CTC of a large vein, which was the last to enhance (i.e. longest TTP). Then, diving arterioles and ascending venules were selected (Fig. 3a). From the z-stack scan, vessels were then paired based on their apparent, anatomical connectedness throughout the capillary bed. The dye transport was analysed based on these same pairs every 30 min. The pre-bolus intensity signal was identified for each vessel and the baseline signal was subtracted to create curves proportional to the dye concentration (CTCs). Afterwards, all curves were scaled to the post-bolus level of the AIF to correct for differences in signal intensity due to light traveling from vessels of varying depth (Fig. 3d). Analogous to the approach used when fitting the residue function-based tracer retention observed by DSC-MRI, we determined the transport function by deconvolution of the VOF with the AIF11. The resulting transport function comprises of a gamma cumulative distribution with two parameters α and β, from which the mean transit time (MTT) and capillary transit-time heterogeneity (CTH) were estimated as the mean (αβ) and the standard deviation of the distribution, respectively. The coefficient of variation (CoV) of transit times was determined as the ratio between CTH and MTT . The deconvolution results were discarded if α<1 and β>0.5
Capillary line scan and pericyte imaging analysis
RBCv values were calculated based on the Radon transformation46 as previously described in Gutierrez et al29. Flux was estimated by analysing the intensity variations occurring within the cross-sectional scan of each capillary. As for the velocity estimation, average intensity profiles were derived from 150 msec time intervals within the transversal line scan. Low and high contrast were taken to indicate presence and absence of RBC, respectively. The number of capillaries showing either stalled blood flow (flux=0) and/or reversed flow, where RBCv alternates between positive and negative flow values, was calculated. Linear density (LD) was calculated as flux/velocity and capillaries with a LD higher than 300cells/mm were discarded from further analysis since this represents an unrealistic high cell/plasma ration for an erythrocyte of approximately 2μm. Estimates of capillary RBCv and RBCflux were averaged for the 30 sec line scan time, and their standard deviation during the 30 sec period was calculated as a measure of the variance of flow.
Capillary diameter was estimated based on dye concentration line profiles from the line scan data. The vessel diameters were estimated as full-width at half maximum (FWHM) values of the line profiles as described earlier29,46, assuming the vessel intersects define their full diameters. Vessel cross-sectional areas were estimated from these diameter calculations. Vessels showing inner diameter ζ 10μm were considered to be arterioles or venules and therefore disregarded in further analysis.
Post hoc analysis
To access the effects of capillary flow pattern changes on oxygen availability during stroke, cerebral metabolic rate of oxygen (CMRO2), tissue oxygen tension (PtO2) and oxygen extraction fraction (OEF) were predicted based on the MTT and CTH values in a biophysical model for oxygen extraction33 assuming all vessels stay open (Fig. 6).
Statistical analysis was performed in RStudio (Verision 1.1.456 © 2009-2018, RStudio Inc.) using a statistical significance level of 0.05. Physiological data were analysed by One-way and Two-way ANOVAs and Tukey multiple comparisons test. Differences in the hemodynamic values were tested by linear mixed model (LMM), where the hemodynamic values were treated as fixed effects whereas time and type (i.e. control vs stroke) were random effects. P-values were obtained by likelihood ratio test of the effect in questions. Effects of types at the specific time points was evaluated by t-test. The difference between the hemodynamic upstream and downstream of the pericytes, as well as the fraction of capillaries with stalled or reversed flows, were addressed by a two-sided t-test. All results are presented as their mean value ± SE, if not otherwise indicated.
This study was funded by The Danish Research Council individual postdoctoral grant and Sapere Aude Young Research Talent (NKI).
NKI designed the study, performed the experiment, analyzed the data and wrote the manuscript. EGJ contributed to the TPM scannings and data analysis. PMR and IKM wrote the matlab codes and contributed to data analysis. HA conducted the oxygen calculations, TRH contributed to the experiment and animal models. LØ contributed to the experimental design, data analysis and helped write the manuscript.
- 1.Mortality from ischemic heart diseaseAnalysis of data form the world health organization and coronary artery disease risk factors from NCD risk factor collaboration. Circoutcomes 12
- 2.2018 guidelines for the early management of patients with acute ischemic stroke: A guideline for healthcare professionals from the American heart association/American stroke associationStroke 49:e46–e110
- 3.Pathophysiology and treatment of stroke: Present status and future perspectivesInt J Mol Sci 21
- 4.Thresholds in cerebral ischemia - the ischemic penumbraStroke 12:723–725
- 5.Donnan, Geoffrey A et al. Stroke Lancet (London, England) 371, 1612–1623 (2008).Stroke Lancet (London, England) 371:1612–1623
- 6.Perfusion thresholds in cerebral ischemiaThe Ischemic Penumbra. Informa Healthcare: New York
- 7.MRI-Guided Thrombolysis for Stroke with Unknown Time of OnsetN. Engl. J. Med 379:611–622
- 8.Combined diffusion-weighted and perfusion-weighted flow heterogenity magnetic resonance imaging in acute strokeStroke 31:1097–1103
- 9.Final infarct size after acute stroke: Prediction with flow heterogenityRadiology 225
- 10.Abnormal intravoxel cerebral blood flow heterogenity in human ischemic stroke determined by dynamic susceptibility contrast magnetic resonance imagingStroke 36
- 11.Reliable estimation of capillary transit time distributions using DSC-MRIJ. Cereb. Blood Flow Metab 34:1511–1521
- 12.Transit time homogenization in ischemic stroke – A novel biomarker of penumbral microvascular failure?J. Cereb. Blood Flow Metab 38:2006–2020
- 13.The role of cerebral capillaries in acute ischemic stroke: The extended penumbra modelJ. Cereb. Blood Flow Metab 33
- 14.Dynamic capillary stalls in reperfused ischemic penumbra contribute to injury: A hyperacute role for neutrophils in persistent traffic jamsJ. Cereb. Blood Flow Metab 41
- 15.Pericytes in ischemic strokeAdv Exp Med Biol 1147:189–213
- 16.Cerebral microvascular pericytes and neurogliovascular signalling in health and diseaseBrain Res 1623:3–17
- 17.What is a pericyte?J. Cereb. Blood Flow Metab 36
- 18.Pericyte contraction induced by oxidative-nitrative stress impairs capillary reflow despite successful opening of an occluded cerebral arteryNat. med 15
- 19.Squalenoyl adenosine nanoparticles provide neuroprotection after stroke and spinal cord injuryNat Nanotechnol 9:1054–1062
- 20.Blood flow in the cerebral capillary network: a review emphasizing observations with intravital microscopyMicrocirculation 4:233–252
- 21.Impact of stroke co-morbidities on cortical collateral flow following ischaemic strokeJ Cereb Blood Flow Metab 40:978–990
- 22.Vasoconstrictive neurovascular coupling during focal ischemic depolarizationsJ Cereb Blood Flow Metab 26:1018–30
- 23.Dynamic imaging of cerebral blood flow using laser speckleJ Cereb Blood Flow Metab 21:195–201
- 24.Dynamic capillary stalls in reperfused ischemic penumbra contribute to injury: A hyperacute role for neutrophils in persistent traffic jamsJ Cereb Blood Flow Metab 41:236–252
- 25.Capillary pericytes regulate cerebral blood flow in health and diseaseNature 508:55–60
- 26.Laser speckle contrast imaging in biomedical opticsJ. Biomed. Opt 15
- 27.Evaluation of laser speckle flowmetry for imaging cortical perfusion in experimental stroke studies: quantification of perfusion and detection of peri-infarct depolarizationsJ Cereb Blood Flow Metab 26:645–653
- 28.Peri-infarct depolarizations lead to loss of perfusion in ischemic gyrencephalic cerebral cortexBrain 130:995–1008
- 29.Effect of electrical forepaw stimulation on capillary transit-time heterogenity (CTH)J Cereb Blood Flow Metab 36:2072–2086
- 30.The effect of transit time heterogenity on brain oxygenation during rest and functional activationJ Cereb Blood Flow Metab 35:432–442
- 31.The roles of cerebral blood flow, capillary transit time heterogenity, and oxygen tension in brain oxygenation and metabolismJ Cereb Blood Flow Metab 32:264–277
- 32.The effect of capillary transit time heterogenity (CTH) on brain oxygenationJ Cereb Blood Flow Metab 35:806–817
- 34.Mild induced hypertension improves blood flow and oxygen metabolism in transient focal cerebral ischemiaStroke 39
- 35.Temporal alterations in pericytes at the acute phase of ischemia/reperfusion in the mouse brainNeural Regen. Res 17:2247–2252
- 36.The Ca2+-gated channel TMEM16A amplifies capillary pericyte contraction and reduces cerebral blood flow after ischemiaJ. Clin. Investing 132
- 37.Dynamic capillary stalls in reperfused ischemic penumbra contribute to injury: A hyperacute role for neutrophils in persistent traffic jamsJ Cereb Blood Flow Metab 41:236–252
- 38.Neutrophils obstructing brain capillaries are a major cause of no-reflow in ischemic strokeCell Rep 33
- 39.Precapillary sphincters maintain perfusion in the cerebral cortexNat. Comm 11
- 40.Diverse functions of pericytes in cerebral blood flow regulation and ischemiaJ Cereb Blood Flow Metab 35
- 41.Capillary flow disturbances after experimental subarachnoid hemorrhage: A contributor to delayed cerebral ischemia?Microcirculation 26
- 42.Middle cerebral artery occlusion in the rat by intraluminal suture. Neurological and pathological evaluation of an improved modelStroke 27:1616–1622
- 43.Experimental studies of ischemic brain edemaJPN Stroke 8:1–8
- 44.Dynamic imaging of cerebral blood flow using laser speckleJ Cereb Blood Flow Metab 21:195–201
- 45.Ischemic cell death in brain neuronsPhysiol Rev 79:1431–1568
- 46.Chronic optical access through a polished and reinforced thinned skullNat. met 7:981–984
- 47.Imaging pericytes and capillary diameter in brain slices and isolated retinaeNat. prot 9:323–336