A systematic observation of vasodynamics from different segments along the cerebral vasculature in the penumbra zone of awake mice following cerebral ischemia and recanalization

Abstract

Different segments of the cerebral vascular network may react distinctly to brain ischemia and recanalization. However, there are limited systematic observations of these vascular responses in mice under a physiological state following ischemic stroke. Herein, we aimed to investigate the vasodynamics among several segments along the cerebral vessels in awake mice following cerebral ischemia/recanalization via two-photon imaging. Plasma in the blood vessels were labelled with fluorescein isothiocyanate dextran. Smooth muscle cells and pericytes were labelled via a genetic mouse line (PDGFRβ-tdTomato). We observed a no-reflow phenomenon in downstream microcirculation, and the vasodynamics of different segments of larger cerebral vessels varied in the penumbra area following cerebral ischemia-reperfusion. Despite obtaining reperfusion from the middle cerebral artery, there were significant constrictions of the downstream blood vessels in the ischemic penumbra zone. Interestingly, we observed an extensive constriction of the capillaries 3 hours following recanalization, both at the site covered by pericyte soma and by the pericyte process alone. In addition, we did not observe a significant positive correlation between the changed capillary diameter and pericyte coverage along the capillary. Taken together, abnormal constrictions and vasodynamics of cerebral large and small vessels may directly contribute to microcirculation failure following recanalization in ischemic stroke.

Introduction

Ischemic stroke is one of the leading causes of mortality and disability worldwide. ¹ Reperfusion therapy, including intravenous thrombolysis with recombinant tissue plasminogen activator and mechanical thrombectomy, is the primary treatment for patients with acute ischemic stroke during time window.^2,3 There were growing evidences indicated that the success rate of recanalization was made above 80% of standard senior stroke centre. ⁴ However, approximately 50% of the patients with successful recanalization experienced poor clinical prognosis.^5,6 Several studies reported on the clinical risk factors correlated with worse neurological scores, such as aging, hypertension, diabetes, the time to puncture, and brain collateral vessels.^7,8 Long lasting brain tissue hypoperfusion (no-reflow) of the penumbra zone following recanalization is one of the most important potential mechanisms. ⁹ The no-reflow of microcirculation compromises the benefit of the recanalization of large vessels, thus leading to a growth of the ischemic core and a deterioration of neurological functions. ¹⁰ However, despite upstream recanalization, the mechanism of no-reflow of microcirculation is unclear, including large vessel vasospasm, the contraction of pericyte, microclots formation, and tissue oedema associated with the collapse of veins.^11,12

Microcirculation dysfunction is associated with abnormal alterations of any other part of the upstream and downstream cerebral vessels.^13,14 The cerebral vascular system is not only an elaborate pile system for nutrition supply and metabolic waste elimination but also a dynamic network with a fantastic autoregulation ability and interaction with neurons, astrocytes, microglia, and circulating immune cells. ¹⁵ In the brain cortex, the vasculature is divided into several segments by the morphology and function, consisting of an intact vascular network. Along the blood flow direction, the pial artery dives into the brain parenchyma and diverges into the penetrating arterioles through the pre-capillary arteriole, capillary, post-capillary venule, and ascending venule; eventually, the blood flow drains into the pial vein and venous sinus. ¹⁶ The blood flow of any other segment could influence the flow in the capillary because the cerebral vessels are highly interconnected. Previous studies reported on the substantial dynamics of different segments following cerebral ischemia and recanalization.^15,17–19 However, most of these studies focused on a particular segment rather than all vascular segments simultaneously. Another concern is that the in vivo imaging experiments were performed under an acute cranial window and anaesthesia state, which could affect the neural activity and neurovascular coupling. ²⁰ This necessitates investigating the vasodynamics of all segments and their roles on the no-reflow of ischemic stroke in a more physiological condition. Herein, we aimed to use a mural cell-labelled mouse line to explore the systematic vasodynamics of different cerebral vessel segments by two-photon in vivo imaging through a chronic cranial window under an awake state.

Materials and methods

Animals

All animal experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee at the Chinese Institute for Brain Research (CIBR), Beijing. Animal data are reported in accordance with ARRIVE guidelines (Animal Research: Reporting in Vivo Experiments). All animals were housed in a specific pathogen-free facility with a temperature and humidity-controlled 12-h light/dark cycle. PDGFRβ-Cre mice, a generous present from the Volkhard Lindner’s Lab of the Maine Medical Centre Research Institute, were bred with the Ai14 reporter line (#007914, Jackson Laboratory) to produce PDGFRβ-tdTomato offspring. Male PDGFRβ-tdTomato mice were used at 10 weeks to 12 weeks of age. All surgical operations were conducted under isoflurane anaesthesia, and we made every attempt to minimise suffering and the number of experimental animals. We used only male mice for our experiments for the recognized evidence that male and female respond differently to ischemic stroke.

Cranial window surgery

A long-term and clear cranial window was built before two-photon in vivo imaging. Briefly, the male mice (10–12 weeks old) were anaesthetised by isoflurane (4–5% for induction, 1–2% for maintenance), and fixed in a home-made stereotaxic apparatus. We performed strict disinfection before the cranial procedures. Craniotomy (3 mm in diameter) was performed and covered with a sterile coverslip over the somatosensory cortex (centred at AP = −2.0 mm, L = 1.5 mm). Eventually, a home-made head plate was attached to the skull with dental cement. The mice were kept warm in a heating pad (stable 37°C) during the operation and transferred to a warm cage until completely awake. They were allowed to recover for 2 weeks following this procedure before two-photon in vivo imaging and the transient middle cerebral artery occlusion (tMCAO) model.

Mouse model of transient focal cerebral ischemia

The ischemic stroke was induced by tMCAO. ²¹ Briefly, male PDGFRb-TdTomato mice (10–12 weeks, 25–30 g) were anaesthetised by isoflurane (4–5% for induction, 1–2% for maintenance). Following disinfection and the incision of midline skin of the neck, we exposed the common carotid artery and external carotid artery under the cervical triangle muscles. Subsequently, a heparinised filament (602256PK5Re, Doccol) was gently inserted to the origin of the middle cerebral artery via an incision of the external carotid artery. Following 60 min of occlusion, the filament was gently removed, and blood flow was restored. The mice were allowed to recover in a warm chamber and transferred to cages with a gel containing sterilised water post operation.

Laser speckle contrast imaging

The regional cerebral blood flow (CBF) was measured at baseline, during ischemia, and following recanalization (3 h and 24 h post recanalization) via a laser speckle contrast imaging system (Simopto, Wuhan XunWei Optoelectronic Technology Co., Ltd, China) through the chronic cranial window or in mice without a cranial window. The ischemic penumbra zone was defined as the area in which CBF declined to 30–50% of the baseline value. ²² The ischemic core was defined as the area in which the CBF declined to <30% of the baseline value. ²²

Two-photon in vivo imaging

We used the same imaging protocols for all four mice before and after tMCAO. The mice were trained to get accustomed to the cylinder platform with a head fixing design for 3 days before imaging to avoid struggling during image acquisition. To maintain a physiological state, they were in awake state without anaesthesia during image acquisition. We recorded the observations before cerebral ischemia (baseline), 3 h following ischemia and recanalization (re-3 h), and 24 h following ischemia and recanalization (re-24 h). For vasodynamics comparison, all image acquisitions in re-3 h and re-24 h were acquired in a same region with the baseline values. We injected 2,000 KDa fluorescein isothiocyanate (FITC) dextran dye (100 μl, 1%-w/v, saline. FD2000S, Sigma) through the retro-orbital plexus. Cerebral vasculature images were acquired by an upright two-photon laser scanning microscope (TPLSM, FVMPE-RS, Olympus) with 10 × 0.6 NA and 25 × 1.05 NA water immersion objective lens (Olympus), controlled by a software (F31S-SW, Olympus). Scanning laser was generated from a Mai Tai HP Ti:Sapphire laser device (InSight X3TM, Spectra physics), and the scanning speed was 2.0 μs/pixel at a 1024 × 1024 scan size by a Galvano scanner. We performed line scanning to measure the velocity of red blood cells (RBCs) through a centreline of the measured blood vessel at a speed of approximately 1.2 ms/line with a 2,000-cycle repetition. The suitable excitation wavelength was set as 990 nm for both FITC and TdTomato. The fluorescence emitted from FITC and TdTomato was detected with a non-descanned GaAsp detector (Olympus); BA495-540 (green) and BA575-645 (red) were the emission bandpass filters. Z-stack images were acquired at a 2-μm step size and a depth of 300 μm to 400 μm from the brain pia. We have optimized a strategy for two-photon imaging acquisition (Supplementary methods).

Imaging processing and analysis

All images were processed by ImageJ/Fiji and MATLAB R2020b software. ²³ The surface vascular pattern images were rendered by the Imaris software (Bitplane, version 9.7). We quantified the diameter of blood vessels from maximum projection image. To measure the diameter of a vessel, a perpendicular line was plotted along the centre line of the vessel in Fiji, which calculated the full width at half maximum of the luminal dye fluorescence. We analysed the line scanning data for velocity by a modified MATLAB algorithm. ²⁴ Line scanning data for the flux of capillary, pre-capillary arteriole, and post-capillary venule were analysed by a customised MATLAB algorithm. The flux was described as the number of RBCs per seconds in small vessel. The flux of remaining larger vessels was calculated according to a previously reported formula. ²⁵ The pericyte coverage of the capillary is defined by four grades based on the projection image as follows: (1) only one side of the capillary is covered by a pericyte process; (2) two sides are covered by two pericyte processes; (3) two sides are covered by two pericyte processes and the lumen is partly covered; (4) the capillary is completely covered. VasoMetrics was performed to measure multiple consecutive diameters along a single blood vessel or in a time-lapse imaging of the pial artery and penetrating arteriole. ²⁶ Imaging analyses were performed by investigators blinded to the imaging timepoints.

Statistical analyses

Statistical analysis was performed by the GraphPad Prism software (version 9.0 for windows, GraphPad Software, San Diego, CA, USA, www. graphpad.com), and the Shapiro-Wilk test (S-W test) was performed to assess for normality. Data accorded with Gaussian distribution are presented as mean ± standard deviation (SD). We examined significant group differences for multiple and two group comparisons using the one-way analysis of variance, followed by the Holm-Sidak’s multiple comparisons test, and the Student’s t-test, respectively. The alteration of the capillary diameter at a single line position was dotted, and we plotted the best-fit linear line between the distance from the soma of pericyte and the alteration of the capillary diameter. We presented a linear fit to examine the relationship between the coverage of pericyte and the alteration capillary diameter. Moreover, the correlation coefficients were calculated by Spearman’s correlation coefficients. For non-normally distributed data, we performed a non-parametric test (Friedman test, Dunn’s multiple comparisons test) to assess the group differences. P-values <0.05 were considered statistically significant.

Results

A total of 10 male PDGFRβ-tdTomato mice were used in our study, including three mice for tMCAO without a cranial window (for assessing the CBF distribution in tMCAO) and three mice for tMCAO with a cranial window (for validating the position of the penumbra area) and four mice for two-photon imaging following tMCAO. Animals were randomly assigned to each group with a random number generator (function Rand () in Excel software).

Cerebral vascular network hierarchy along seven different segments was visualised by in vivo two-photon imaging of the awake mice

The cerebral vasculature of the cortex can be generally divided into seven segments by the morphology and function (Figure 1(a), Supplementary Figure 1). To label the mural cells of vessels, we used PDGFRβ-tdTomato mice to distinguish different vascular segments in two-photon imaging through a chronic cranial window (Figure 1(b)). We acquired a distinct pial vascular pattern (Figure 1(c, d)) with the combination of FITC-positive lumen and tdTomato-positive mural cells (smooth muscle cells and pericytes). The pial artery and penetrating arteriole are wrapped in loops of smooth muscle cells (Figure 1(e, f)). The pre-capillary arteriole is surrounded by several mural cells (smooth muscle cells, sphincters, and pericytes) (Figure 1(g)) along the blood flow direction. ¹³ However, the definition of hybrid cells and their functions in this segment are debatable. The capillary is largely covered by thin-strand pericytes with protruding ovoid soma, which extend long and thin processes that traverse longitudinally (Figure 1(h)). The capillary is followed by a post-capillary venule and ascending venule, which are surrounded by an intermittent pericyte (Figure 1(i)). Moreover, the pial vein (Figure 1(j)), collecting the blood flow and draining to the venous sinus, is covered by venous smooth muscle cells.

Figure 1.

In vivo two-photon imaging of cerebral vascular hierarchy in awake mice. (a) Schematic illustration of cerebral vascular hierarchy in cortex. (b) Schematic illustration of in vivo two-photon imaging in an awake mouse through a chronic cranial window. (c) Representative images of cerebral pial vessels from a PDGFRβ-tdTomato mouse. FITC, a dye injected intravenously. (d) Representative rendered image from (c) by Imaris software. Green, vessel lumen; red, SMCs or pericytes and (e–j) Representative images of pial artery, penetrating arteriole, pre-capillary arteriole, capillary, post-capillary venule, ascending venule and pial vein. Arrow in (g) indicates the penetrating trunk (0th order). SMC, smooth muscle cells; vSMC, venous smooth muscle cell; PC, pericyte.

An evaluation of the ischemic penumbra area in tMCAO mouse model

The ischemic penumbra zone is the salvageable brain tissue that endures ischemia, which is commonly evaluated by computed tomography perfusion in clinical practice. In our study, the penumbra zone was evaluated by the CBF via laser speckle contrast imaging, according to clinical definitions and a previous animal study. ²² For exploring the position of the penumbra area of mouse skull, we measured the regional CBF from three mice that underwent right side tMCAO operation without a cranial window. The CBF of the penumbra zone declined to 30% to 50% of the baseline value, whereas that of the ischemic core declined to <30% of the baseline value (Supplementary Figure 2 (a, b). The CBF of penumbra area was significantly higher than that of the ischemic core (region of interest, ROI 1 vs. ROI 2, p = 0.02). Based on the distribution of penumbra in this ischemic model, we built a cranial window on the somatosensory cortex (centred at AP = −2.0 mm, L = 1.5 mm). Moreover, the regional CBF of the cranial window at ischemia was significantly lower than that of the baseline value (ischemia vs. baseline, p = 0.003), which accorded with the definition of penumbra (Supplementary Figure 2 (c–g)).

Aberrant vasodynamics of the pial artery and penetrating arteriole following cerebral ischemia and recanalization

To track the change of different segments along the cerebrovascular tree, we performed in vivo two-photon imaging in a more physiological condition. First, we performed cranial window operation, and the mice were allowed to recover for 2 weeks. They were subsequently trained to get accustomed to the cylinder imaging platform for 3 days before imaging to avoid any struggling during image acquisition. Following the acquisition of baseline imaging, the tMCAO model was followed by two-photon imaging at Re-3 h and Re-24 h (Figure 2(a)). We observed a significant constriction of the pial artery at Re-3 h than that at baseline (p = 0.025), followed by a dilation at Re-24 h (p < 0.0001) (Figure 2(b) to (d)). Similarly, the velocity and flux decreased at Re-3 h (Re3 h vs. baseline, velocity, p = 0.037; flux, p = 0.0023) (Figure 2(e) and (f)), and increased to the baseline level or even exceed the level at Re-24 h (Figure 2(e) and (f)). Interestingly, we observed abnormal vasoconstriction/vasodilation frequency and amplitude at Re-3 h, compared with baseline. Higher frequency and amplitude of vasoconstriction/vasodilation from the pial arteries were observed at Re-3 h than those at baseline (Re-3 h vs. baseline, p = 0.0035) (Figure 2(g), Supplementary Figure 3 (a-c), Supplementary video 1-3).

Figure 2.

Persistent constriction and aberrant vasodynamics of pial artery and penetrating arteriole of PDGFRβ-tdTomato mice after cerebral ischemia and recanalization. (a) Illustration of experimental timeline. Cranial window surgery was performed and mice were allowed to recover for at least two weeks. Before in vivo imaging, mice were trained to accustomed to staying in the fixture platform used for two-photon imaging. 2 P, two-photon imaging; Baseline, before ischemia; Re-3 h, 3 hours after recanalization; Re-24 h,Continued. 24 hours after recanalization. (b and c) Representative images of the same pial artery in the region of interest (dashed rectangle) at three time points. Dashed lines represent borders of vessel lumen. Distance between two dashed lines is the luminal diameter of the pial artery at baseline. (d to f) Quantification of diameters (d), velocity of luminal RBCs (e), and flux (f) of same pial arteries at three tMCAO time points. n = 67 from 4 mice. (g) Quantification of standard deviation of diameters of same pial arteries during 50 frames of three tMCAO time points. SD, standard deviation. n = 8 from 3 mice. (h to i) Representative images of the same penetrating (pn) arteriole in the region of interest at three time points. Dotted circle represents the border of vessel lumen. Diameter of dotted circle is the luminal diameter of the penetrating arteriole at baseline. Dashed cycles at baseline were copied to images of Re-3 h and Re-24 h for showing difference. (j to l) Quantification of diameters (j), velocity of luminal RBCs (k), and flux (l) of same penetrating arterioles at three tMCAO time points. n = 41 from 4 mice and (m) Quantification of standard deviation of diameters of same penetrating arterioles during 50 frames (3.219 sec/frame) of three tMCAO time points. SD, standard deviation. n = 8 from 3 mice. Statistical analyses: Friedman test with Dunn’s multiple comparison test for d, f, g, j to m. One-way repeated measures ANOVA with Holm-sidak’s multiple comparison test for e. Shown are before-after matched plots. *p < 0.05, **p < 0.01, ***p < 0.001 as indicated, ns: no significance. Baseline, before ischemia; Re-3 h, 3 hours after recanalization; Re-24 h, 24 hours after recanalization.

The pial artery extends branches, termed the penetrating arteriole, which perpendicularly enters the brain parenchyma ¹³ . The diameter of the penetrating arteriole was significantly smaller at Re-3 h than that at the baseline and Re-24 h (Re-3 h vs. baseline, p = 0.0028; Re-3 h vs. Re-24 h, p < 0.001). Consequently, the diameter of the penetrating arteriole at Re-24 h was recovered to the baseline level (Figure 2(h) to (j)). Interestingly, the velocity and flux of blood flow in the penetrating arteriole did not significantly increase or decrease at Re-3 h than that at the baseline (Figure 2 (k) and (l)). Consistent with the pial artery, we also identified abnormal vasoconstriction/vasodilation of the penetrating arteriole, revealing higher frequency and amplitude of vasoconstriction/vasodilation at Re-3 h than those at baseline (Re-3 h vs. baseline, p = 0.081) (Figure 2(m), Supplementary Figure 3 (d–f), Supplementary video 4–6).

Dramatical dysfunction of the pre-capillary arteriole and post-capillary venule following recanalization

The pre-capillary arteriole is the first to second order branch from the penetrating arteriole, immediately controlling the downstream blood flow of the capillary and venule. ¹⁶ It constricted substantially at Re-3 h and recovered to the baseline level at Re-24 h (Re-3 h vs. baseline, p < 0.001) (Figure 3(a) to (c), Supplementary Figure 4 (a, b)). Simultaneously, the velocity and flux of the RBCs decreased at Re-3 h (Re-3 h vs. baseline, p < 0.001); however, it partly recovered at Re-24 h (velocity, Re-24 h vs. baseline, p = 0.02; flux, Re-24 h vs. baseline, p = 0.003), whereas the flux of three baseline time points were stable (Figure 3(d) to (f)).

Figure 3.

Striking constriction of pre-capillary arteriole of PDGFRβ-tdTomato mice after recanalization. (a and b) Representative images of the same pre-capillary arteriole in the region of interest at three time points. Dashed lines represent borders of vessel lumen. Distance between two dashed lines is the luminal diameter of the pre-capillary arteriole at baseline. Dashed lines at baseline were copied to images of Re-3h and Re-24h. (c to f) Quantification of diameters (c), velocity of RBCs (d), and flux (e, recanalization; f, baseline) of same pre-capillary arteriole at three tMCAO time points. c-e, n = 39 from 4 mice; f, n = 19 from 4 mice. (g and h)Continued.Representative images of the same post-capillary venule in the region of interest at three time points. Dashed lines represent borders of vessel lumen. Distance between two dashed lines is the luminal diameter of the post-capillary venule at baseline. Dashed lines at baseline were copied to images of Re-3h and Re-24 h. (i to l) Quantification of diameters (i), velocity of RBCs (j), and flux (k, recanalization; l, baseline) of same post-capillary venule at three tMCAO time points. i to k, n = 40 from 4 mice; l, n = 20 from 4 mice. Statistical analyses: Friedman test with Dunn’s multiple comparison test for d, e and i to l. One-way repeated measures ANOVA with Holm-sidak’s multiple comparison test for c and f. Shown are before-after pairing plots. *p < 0.05, **p < 0.01, ***p < 0.001 as indicated, ns: no significance. Baseline, before ischemia; Re-3 h, 3 hours after recanalization; Re-24 h, 24 hours after recanalization.

At the opposite side of the capillary network, the post-capillary venule immediately connects with the capillary and controls the efflux of blood flow ¹⁶ . Consistent with the pre-capillary arteriole, we observed a distinct decline in the diameter, velocity, and flux of the RBCs of the post-capillary venule at Re-3 h, followed by almost complete restoration at Re-24 h (Re-3 h vs. baseline, p < 0.001) (Figure 3(g) to (k)). Likewise, the flux of the post-capillary venule was maintained at a stable range before ischemic stroke (Figure 3(l).

Collapsed lumen and reduced blood flow in the ascending venule and pial vein following recanalization

All blood flow from microcirculation converges to the ascending venules and surface pial veins, and eventually to the venous sinus ¹⁶ . The lack of contractile smooth muscle cells or pericytes in the venous vessels are chiefly different from the artery and capillary with dilation and constriction features. An alteration of the artery is the focus of investigating CBF vasodynamics, whereas limited studies have assessed changes in the venules and veins. Herein, we tracked alterations of the ascending venule and pial veins before and following ischemic stroke. We observed a significant decline in the diameter, velocity, and flux of the ascending venule at Re-3 h than that at the baseline level (Re-3 h vs. baseline, p < 0.001) (Figure 4(a) to (e)). The diameter of the ascending venule restored to the baseline level at Re-24 h; nonetheless, the blood flow did not completely recover (Re-24 h vs. Re-3 h, p = 0.02) (Figure 4(c) to (e)). Furthermore, we validated the stability of vasodynamics of the ascending venule at baseline (Figure 4(f)).

Figure 4.

Collapse of ascending venule and pial vein of PDGFRβ-tdTomato mice after recanalization. (a and b) Representative images of the same ascending venule in the region of interest at three time points. Dashed circles represent borders of vessel lumen. The diameter of the dashed circle is the luminal diameter of the ascending vein at baseline. Dashed circle at baseline were copied to images of Re-3 h and Re-24 h. (d and e) Quantification of diameters (c), velocity of RBCs (d), and flux (e, recanalization; f, baseline) of same ascending venule at three tMCAO time points. c and e, n = 23 from 4 mice; f, n = 20 from 4 mice. (g and h) Representative images of the same pial vein in the region of interest at three time points. Dashed lines represent borders of vessel lumen. Distance between two dashed lines is the luminal diameter of the pial vein at baseline. Dashed lines at baseline were copied to images of Re-3 h and Re-24 h. (i to k) Quantification of diameters (i), velocity of RBCs (j), and flux (k, recanalization; l, baseline) of same pial veins at three tMCAO time points. i to k, n = 23 from 4 mice; l, n = 20 from 4 mice. Statistical analyses: Friedman test with Dunn’s multiple comparisons test for d to f, i to l. One-way repeated measures ANOVA with Holm-sidak’s multiple comparisons test for c. Shown are before-after pairing plots. *p < 0.05, **p < 0.01, ***p < 0.001 as indicated, ns: no significance. Baseline, before ischemia; Re-3 h, 3 hours after recanalization; Re-24h, 24 hours after recanalization.

Pial veins do not comprise myogenic contractile components within the vessel walls and drain the blood from the cerebral cortex to the venous sinus and peripheral venous system ¹⁶ . Similarly, we identified an evident collapse of the pial vein at Re-3 h, accompanied by decelerated RBC velocity and slashed blood flow flux, followed by a restoration at Re-24 h (Re-3 h vs. baseline, p < 0.001) (Figure 4(g) to (k)). However, the vasodynamics were stable at baseline (Figure 4(l)).

An extensive constriction along the capillaries at the site covered by pericyte soma or only by the pericyte process following recanalization

Capillary network, a site for the direct exchange of nutrient and oxygen, is the most abundant and complicated vascular segment of brain vascular hierarchy ¹⁶ . However, it endures the highest vascular resistance. In our study, the capillaries displayed an extensive and significant constriction at Re-3 h (Re-3 h vs. baseline, p < 0.001) and recovered to the baseline level at Re-24 h, despite the position covered by pericyte soma or process (>10 μm from the pericyte soma) (Figure 5(a) to (h)). However, the diameters and changed diameters displayed insignificant differences along the capillaries covered by the pericyte soma and process at Re-3 h (Figure 5(i) and (j)). To investigate the basic dynamics of the capillaries at baseline, we compared the vasodynamics (the diameter, velocity, and flux of RBCs) of the capillaries from three baseline time points, and found no significant differences among the baseline phases (Supplementary Figure 5 (a-l)). Paralleled with the diameter, the velocity and flux of RBCs in the capillaries decreased acutely at Re-3 h and were partly restored at Re-24 h, compared with the baseline level (Re-3 h vs. baseline, p < 0.001) (Figure 5(k) and (l)).

Figure 5.

There was an evident constriction both at the sites of capillaries covered by pericyte soma and pericyte process along capillaries of PDGFRβ-tdTomato mice after recanalization. (a and b) Representative images of the same capillary in the region of interest at three tMCAO time points. arrowheads, soma of pericyte. (c, f) Quantification of diameters of same positions covered by pericyte soma (c) and pericyte process (f) at three tMCAO time points. c, n = 135 from 4 mice; f, n = 132 from 4 mice. (d, e, g, h) Pie charts represent the percentage of dilated (pink, dilate>0.2 μm), constricted (blue, constrict>0.2 μm) and stable (green, change < 0.2 μm) capillaries covered by pericyte soma (d, e) and pericyte process (g, h) from two adjacent time points. (i, j) Quantification of diameter (i) and changed diameter (j) of capillaries covered by pericyte soma and process at Re-3 h. n = 135 for soma, n = 132 for process. (k, l) Quantification of velocity of RBCs (k) and flux (l) of same capillaries at three tMCAO time points. k, i, n = 80 from 4 mice. Statistical analyses: One-way repeated measures ANOVA with Holm-sidak’s multiple comparison test for c and f. two-tail unpaired t test for i and j. Friedman test with Dunn’s multiple comparison test for k and l. Shown are before-after pairing plots (c, f, k, l) and mean ± SD (i, j). *p < 0.05, **p < 0.01, ***p < 0.001 as indicated, ns: no significance. Baseline, before ischemia; Re-3 h, 3 hours after recanalization; Re-24 h, 24 hours after recanalization.

No significant correlation was found between the site of pericyte soma and the altered diameter of the site along capillaries

Based on the absence of significant differences in diameter change between the site covered by the pericyte soma and only by the pericyte process along the capillaries (Figure 5), we investigated detailed changes along single capillaries before and following recanalization. First, we divided the capillaries into two types at Re-3 h, namely the patent capillaries with partly recovered blood flow and the stalled capillaries without reflow. In patent capillaries, the site covered by the pericyte soma was defined as an origin location (L = 0). From the soma, we measured the diameter along the capillary at a 1-μm step size (Figure 6(a) and (d)). We did not identify the site covered by pericyte soma constricted or dilated most significantly at Re-3 h (Figure 6(b) and (e)). Moreover, our data did not support that capillaries constricted or dilated most severely at a position with the highest pericyte coverage. We observed a negative correlation between the pericyte coverage and changed diameter (capillary 1, r = −0.36, and p = 0.02; capillary 2, r = −0.08, and p = 0.47) (Figure 6(c) and (f)).

Figure 6.

No evident correlation between the distance from pericyte soma and alteration of capillary of PDGFRβ-tdTomato mice after recanalization. (a, d) Representative images of patent capillaries covered by pericytes at baseline and re-3 h, the location of pericyte soma was defined as original location (L = 0), the step size of every paralleled line is 1 μm. (b, e) A smooth curve between the distance from the soma of pericyte and the alteration of diameter of capillary. (c, f) Linear correlation diagram between the coverage of pericyte and the alteration of diameter of capillary, the correlation coefficients were calculated by spearman correlation coefficients. n = 42 (b, c), n = 76 (e, f). (g, j) Representative images of stalled capillaries covered by pericytes at baseline and re-3 h, the location of the branch point was defined as original location (L = 1), the distance of every measured point is 1 μm as shown in (a). (h, k) A smooth curve between the line position along the capillary and the alteration of diameter of the same capillary. (i, l) Linear correlation diagram between the coverage of pericyte and the alteration of diameter of capillary, the correlation coefficients were calculated by spearman correlation coefficients. n = 60 (h, i), n = 63 (k, l) Baseline, before ischemia; Re-3 h, 3 hours after recanalization.

By contrast, we measured the stalled capillaries, despite the constriction of capillaries that may not directly interrupt the blood flow. RBCs were captured as black stripes in the plasma dye (FITC) in normal condition, whereas a big black stack in the vessel lumen denoted an interruption (Figure 6(g) and (j)). Likewise, the single capillary was marked by several locations for before and after the paired measurement (right side of Figure 6(g) and (j)). Consistent with the patent capillaries, we neither observed a direct correlation between the distance from the pericyte soma and changed diameter nor identified a direct correlation between the position of the halt and the site of pericyte soma (Figure 6 (h) and (k)). Capillaries at Re-3h did not display a positive correlation between the pericyte coverage and changed diameter (capillary 1, r = 0.12, and p = 0.36; capillary 2, r = −0.12, and p = 0.34) (Figure 6(i) and (l)). Collectively, our data indicated a significant constriction of the capillary at the sites covered by both pericyte soma and pericyte process alone following ischemia/recanalization. Moreover, there was no significant difference between the alterations of these sites under a more physiological condition.

Discussion

We aimed to systematically investigate the vasodynamics of various segments along the brain cortex vasculature following cerebral ischemia/reperfusion under a more physiological state. Furthermore, we compared the capillary changes covered by the pericyte soma and that by pericyte process alone following ischemic stroke and recanalization. We observed an extensively compromised blood flow of different segments along the cerebral vessels in the penumbra area following the recanalization of tMCAO, which may contribute to long-lasting microcirculation failure (no-reflow) in ischemic stroke. Interestingly, we identified a significant constriction of the capillaries at the sites covered by both pericyte soma and pericyte process alone. However, we did not observe an evident difference between the altered capillary diameter covered by the pericyte soma and pericyte process alone following brain ischemia/recanalization.

No-reflow is a common phenomenon in patients with cerebral ischemic stroke with recanalization via endovascular therapy, and is associated with adverse outcomes. ¹⁰ The potential mechanisms of no-reflow are unclear, including abnormal vasodynamics of upstream and downstream vessels, the embolization of micro-clot in small vessels, the contraction of pericytes, neutrophils stalls, and tissue oedema-related vessel collapses.^11,12,27 Abnormal vasodynamics of larger vessels principally refer to the high resistance of arteries following ischemia, thus leading to a directly decreased influx into downstream microcirculation. ²⁸ Abnormal vasodynamics of the middle cerebral artery is associated with adverse events and long-term prognosis in patients with ischemic stroke. ²⁹ Animal experiments also demonstrated an incomplete recovery of flux in large surface arteries following reperfusion, whereas the flux of penetrating arterioles returned to baseline. ¹⁷ Consistent with a previous study, we observed similar results in the pial artery and penetrating arteriole. Furthermore, we identified an abnormal vasomotion of the pial artery and penetrating arteriole following recanalization, which exhibited a higher frequency and amplitude of vasoconstriction and vasodilation. This in turn may indicate a compensatory and autoregulatory mechanism following an episode of ischemic stroke. Interestingly, we observed a consistent and dramatic decline of the diameter and flux of pre-capillary arterioles following recanalization. The pre-capillary arteriole, covered by pre-capillary sphincter, is a bottleneck that controls the influx of capillary network. ³⁰ Decreased flux in the arteriole may cause a failure of downstream microcirculation. Therefore, the dilation of pre-capillary arterioles may improve blood flow of microcirculation following recanalization.

Cerebral veins are an important component of the vascular system, not only for the drainage of blood flow but also for the removal of metabolic waste and brain immune surveillance. ³¹ In addition to the description of the artery (arteriole) system, we identified a significant decline of the diameter and blood flow in the vein (venule) system following stroke/recanalization. However, no direct evidence indicated that the failure of veins (venules) was a cause or an outcome of capillary no-reflow. Generally, cerebral veins and venules do not actively contract and lack a venous valve. High intracranial pressure and brain tissue oedema may selectively compress and collapse the veins or venules, which lack a strong smooth muscle cell layer. Animal experiments have indicated an increased influx of the cerebrospinal fluid, which contributed to brain oedema and the compression of veins shortly following brain ischemia. ³² A recent clinical study demonstrated that good venous flow was associated with better outcomes in patients undergoing recanalization. ³³ By contrast, another study indicated that abnormal vein drainage may lead to the acute phase of brain oedema and adverse outcomes in patients with ischemia. ³⁴ Collectively, poor blood flow of veins and venules may predict the outcome in patients with ischemic stroke. Future studies should adopt greater strategies targeting vein dysfunction.

Pericyte refers to a type of mural cells along the cerebral pre-capillary arterioles, capillaries, and post-capillary venules.^35,36 However, their role in contracting and regulating local CBF is controversial.^18,19,37,38 This controversy is partially attributed to an obscure definition of pericytes; however, the role of capillary pericyte in blood flow regulation is unclear. Researchers have reported on two unequivocally opposite results of pericyte contraction along the capillary following ischemic stroke.^18,19 We observed an extensive and consistent constriction of the capillary in the sites covered by both pericyte soma and pericyte process alone following ischemia/recanalization through a paired comparison method of the identical capillary at three different timepoints. There was no significant difference in the alterations of diameter between these sites. The capillary pericyte is like a thin strand with long thin processes surrounding the capillary, whereas sites covered by the pericyte soma are enriched with circumferential processes. ³⁹ According to this definition, the site covered by the pericyte soma displays higher coverage than any other position of the capillary. However, we did not observe a significant correlation between the coverage of the pericyte and changed capillary diameter in ischemic stroke. These novel findings can be partially attributed to different methods and conditions of experiment and approaches of data processing. First, we compared a particular position of the identical capillary at different timepoints, whereas previous studies pooled all individual capillaries for an analysis, thereby neglecting the variation of each individual capillary. Second, we performed in vivo imaging through a chronic cranial window, and the animals were under an awake state during imaging acquisition. By contrast, previous studies have commonly used in vitro brain slices or in vivo imaging through an acute window and anaesthetised animals. The cranial window operation is a craniotomy surgery, which can induce a dramatic inflammatory response in acute phase, and will affect neural activity and neurovascular coupling. ⁴⁰ Whereas anaesthetics eliminate pain, reduce movement artifacts during imaging, and provide a controlled experimental platform, they exert dramatic effects on the neural activity. ²⁰ Thus, our present findings were based on a more physiological condition.

There are also some limitations in our study. First, only young mice were used in our study, old mice not included. However, accumulating evidences have shown that pericyte remodelling is slower in the aged brain and resulted in capillary hemodynamic dysfunction. ⁴¹ Aging related blood vessel regression is also associated with pericyte degeneration, contributing to the dysfunction of regional blood flow and impaired neuronal activity. ⁴² Therefore, more experiments are needed to investigate the pericyte biology and functions in aging animals with or without ischemic stroke in the future. Second, we cannot draw firm conclusions since this is a descriptive study. For example, pericyte-deficient genetic mice or pharmacological interventions should be used to evaluate the role of pericyte in vascular diameter changes.

In summary, we performed a systematic observation of the vasodynamics from different segments along the brain vasculature of the penumbra area following ischemia/recanalization. Besides the decline in the blood flow of the brain arteries after recanalization, we observed an extent of blood flow failure in the veins, venules and capillaries. Eventually, we observed a significant constriction of the capillary at the sites covered by both pericyte soma and pericyte process alone following ischemia/recanalization. These abnormal vasodynamics may contribute to the no-reflow phenomenon in ischemic stroke, and vasodynamics improvement may be a promising direction for the treatment of no-reflow in ischemic stroke.

Authors’ contributions

YW, WG and WS contributed to conception of the study; BQ, ZZ, NW, and ZF performed the animal breeding management, surgery operation, and in vivo imaging; BQ, ZZ, WC, and XC performed the data collection, imaging processing, and analysis; BQ wrote the manuscript draft; All authors revised and approved the final submission.

ORCID iD

Baoshan Qiu https://orcid.org/0000-0002-9194-2532

Supplemental material

Supplemental material for this article is available online.