Impaired dynamics of precapillary sphincters and pericytes at first-order capillaries predict reduced neurovascular function in the aging mouse brain

Abstract

The microvascular inflow tract, comprising the penetrating arterioles, precapillary sphincters and first-order capillaries, is the bottleneck for brain blood flow and energy supply. Exactly how aging alters the structure and function of the microvascular inflow tract remains unclear. By in vivo four-dimensional two-photon imaging, we reveal an age-dependent decrease in vaso-responsivity accompanied by a decrease in vessel density close to the arterioles and loss of vascular mural cell processes, although the number of mural cell somas and their alpha smooth muscle actin density were preserved. The age-related reduction in vascular reactivity was mostly pronounced at precapillary sphincters, highlighting their crucial role in capillary blood flow regulation. Mathematical modeling revealed impaired pressure and flow control in aged mice during vasoconstriction. Interventions that preserve dynamics of cerebral blood vessels may ameliorate age-related decreases in blood flow and prevent brain frailty.

Extended Data Fig. 1 |. Comparison of absolute diameter changes, cross-sectional area changes and peak diameters by different stimulations in adult and old mice.

(a–c) Absolute diameter changes, cross-sectional area changes and peak diameters and peak diameters of vasodilation by whisker pad stimulation (WP stim). Adult: N = 17 animals, n = 34 vessels. Old: N = 17 animals, n = 25 vessels. (d–f) Absolute diameter changes, cross-sectional area changes and peak diameters and peak diameters of vasodilation by K_ATP channel opener pinacidil puff. Adult: N = 9 animals, n = 15 vessels. Old: N = 11 animals, n = 17 vessels. (g–i) Absolute diameter changes, cross-sectional area changes and peak diameters and peak diameters of vasodilation by papaverine puff. Adult: N = 8 animals, n = 17 vessels. Old: N = 5 animals, n = 8 vessels. (j–l) Absolute diameter changes, cross-sectional area changes and peak diameters and peak diameters of vasoconstriction by l-NAME intravenous infusion (4 min post). Adult: N = 5 animals, n = 5 vessels. Old: N = 8 animals, n = 8 vessels. (m–o) Absolute diameter changes, cross-sectional area changes and peak diameters and peak diameters of vasoconstriction by ET1 puff. Adult: N = 5 animals, n = 10 vessels. Old: N = 7 animals, n = 15 vessels. Linear mixed effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

Extended Data Fig. 2 |. Comparison of control puff and pinacidil puff.

The control puff experiments were performed with the same concentration of fluorescent dye but without an active compound. The initial ‘constricting’ artifact can be perfectly overlaid by both control puff and pinacidil puff, using the time of puffing as a time lock and normalizing the negative peak at the moment of puffing. (a) Vessel diameter change by control puff. (b) Vessel diameter change by pinacidil puff. (c-f) Overlay of the vessel diameter change curves at each capillary order, with (c) as penetrating arteriole, (d) as 1st order capillary, (e) as 2nd order capillary and (f) as 3rd order capillary. Data are given as mean ± SEM.

Extended Data Fig. 3 |. Comparison of vasodilation elicited by papaverine puff, pinacidil puff, and whisker pad (WP) stimulation.

(a–b) Relative diameter change elicited by the three stimulations in (a) adult and (b) old mice. (c–d) Absolute diameter change elicited by the three stimulations in (c) adult and (d) old mice. (e–f) Peak diameter elicited by the three stimulations in (e) adult and (f) old mice. WP stimulation: Adult: N = 31 animals, n = 56 vessels. Old: N = 26 animals, n = 53 vessels. Pinacidil: Adult: N = 9 animals, n = 15 vessels. Old: N = 11 animals, n = 17 vessels. Papaverine: Adult: N = 8 animals, n = 17 vessels. Old: N = 5 animals, n = 8 vessels. Linear mixed effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

Extended Data Fig. 4 |. Representative immunohistochemical images of staining CD31 and Hoechst in the adult and old mouse brains.

This is to identify the vascular endothelial cell somata. Maximum intensity projection of an (left) adult and (right) old brain image stack. Red: NG2; Green: CD31; Gray: Hoechst. Scale bar: 20 μm.

Extended Data Fig. 5 |. Immunohistochemical analysis of mural cell structural change with age.

(a) Image processing procedure for the z-stack was to free-hand draw the background area and then calculate the mean background fluorescence intensity for each image plane. The background intensity was normalized to the peak value of the z-stack. Finally, the z-stack images were projected onto one image by maximal projection. (b) The semi-manual customized code is to identify the whole vessel surface and pericyte-covered area. Upper: Hand-drawn delineation of the vessel region in the green and red imags. Lower: The NG2 image was used to select the mural cell-positive pixels. (c) Non-vascular nucleus volumetric density. The non-vascular nucleus is defined as nuclei with a distance to the nearby vessels. (d) Endothelial cell soma density obtained by dividing the number of endothelial cell somas and the examined vessel length. Adult: N = 3 animals, n = 25 vessels. Old: N = 3 animals, n = 26 vessels. Linear mixed effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as mean ± SEM. * indicates p < 0.05, **indicates p < 0.01, ***indicates p < 0.001.

Extended Data Fig. 6 |. In vivo vascular structural changes with aging.

(a–c) Total vessel number per brain volume in adult and old somatosensory cortex. (a) General total vessel number, (b) capillary order dependent total vessel number, and (c) cortical depth dependent total vessel number. (d–f) Mean vessel length in adult and old somatosensory cortex. (d) General mean vessel length, (e) capillary order dependent mean vessel length, and (f) cortical depth dependent mean vessel length. (g–i) Total vessel volume per brain volume in adult and old somatosensory cortex. (g) General total vessel volume, (h) capillary order dependent total vessel volume, and (i) cortical depth dependent total vessel volume. Adult: N = 6 animals, n = 52 vessels. Old: N = 4 animals, n = 29 vessels. Linear mixed effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, ***indicates p < 0.001. L1–L5 denote cortical sublayers.

Extended Data Fig. 7 |. Math modeling indicates that vascular blood flow and pressure are affected by aging.

(a–b) Mathematical modeling of pressure distribution in 3D reconstructed vasculature from 3 adult (a) and 3 old (b) mouse brains. (c) Summary of the averaged pressures in penetrating arteriole (PAs), precapillary sphincters, and first- to third-order capillaries in the vascular networks under resting-state, with whisker pad stimulation (WPstim), and after ET1 puff. (d) Correlation of mural cell coverage and blood pressure summarized for all three states. The dashed line connects the adult and old measurements at the same vascular segment for easy comparison. (e) Mathematical modeling of flow distribution in 3D reconstructed vasculature from 3 adult and 3 old mouse brains. Only one representative image from each group is shown. (f) Summary of the average flow in PAs, precapillary sphincters, and first- to third-order capillaries in the vascular networks under resting-state, WP stim, and after ET1 puff. Adult: N = 3 animals, n = 3 vessels; Old: N = 3 animals, n = 3 vessels. Linear mixed effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

Extended Data Fig. 8 |. Diagram of vascular mural cell phenotypes at the microvascular inflow tract (MIT) and the mechanisms that contribute to changes in function and morphology with MIT aging.

(1) Vascular mural cell phenotypes at the MIT. Arteriole vascular smooth muscle cells (VSMCs) have spindle-shaped cell bodies. Like contractile bands, they cover the circumferences of the penetrating arterioles in a single layer. We previously reported the existence of precapillary sphincters displaying indentation of the vessel lumen and strong expression of αSMA at 50% of the joints between penetrating arterioles and first-order capillaries. They are important in maintaining and regulating capillary blood flow. Ensheathing pericytes cover capillaries up to the third order. Ensheathing pericytes have short longitudinal processes with dense circumferential offshoots woven together to create a meshed circumferential coverage of the capillary that covers 95% of the capillary endothelium. VSMCs, precapillary sphincters and ensheathing pericytes do express αSMA. A stepwise decrease in αSMA content is observed along the first, second, third, and up to the fourth order of capillary. This αSMA-positive arteriole-proximal capillary segment is termed the microvascular inflow tract (MIT)—here, the pericyte coverage across capillary junctions is ~90%; further downstream, the capillary junctional pericyte coverage is ~45%. Mesh pericytes exhibit mesh-like morphology and are found downstream of ensheathing pericytes and the αSMA terminus; they have lesser endothelial coverage (71.6%) than ensheathing pericytes, and their longitudinal processes are longer and have a less dense network of circumferential offshoots. In higher-order capillaries, thin-strand pericytes display thin and long longitudinal processes, with only a few circumferential offshoots that extend along and cover 51.3% of the capillary endothelium. (2) Changes in morphology and signaling investigated in this study contribute to vascular aging at the MIT. The resting-state diameter of vessels at the MIT increases with age, but the coverage by mural cells decreases. K_ATP channels are downregulated and are less densely expressed by aging. The probability of K_ATP-channel opening is also reduced with age. NO production and bioavailability maybe reduced by aging, and our results suggest that ongoing NO-dependent mechanisms are attenuated in old mice. Although ET1 signaling and synthesis are increased in old relative to adult mice, ETA receptors are downregulated with age. Furthermore, aging is accompanied by fragmented elastin, deposition of collagens, and vessel stiffness. All the above age-related changes at the MIT contribute to reduced responsivity to both vasodilators and vasoconstrictors, indicating the existence of a compensating mechanism to keep cerebral blood flow and oxygen supply at a physiologically healthy level with healthy aging.

Extended Data Fig. 9 |. Max projection and individual planes in Fig. 3a.

The joint of 1^st and 2^nd order capillary in Fig. 3a can be identified by examining individual planes.

Fig. 1 |. Whisker-pad stimulation-induced and pinacidil-induced vasodilations are reduced in aging and are vascular mural cell dependent.

a, Diagram of the in vivo experimental setup. The physiological state of the mouse was monitored throughout the experiment, including arterial blood pressure, exhaled CO₂, body temperature, heart rate and O₂ saturation. Two stimulations were used: WP stimulation and a glass micropipette inserted into the target area for local ejection (puff). b, A two-photon image obtained by maximum intensity projection of a local image stack at MIT, containing the PA, precapillary sphincter (Sphinc), and first-order to third-order capillaries (1stCap, 2ndCap, 3rdCap). Scale bar, 10 μm. c,d, Representative images of MIT in response to WP stimulation in adult (c) and old (d) mice. Scale bar, 20 μm. Small insets show enlarged precapillary sphincters (Scale bar, 10 μm.). e,f, Mean (solid curve) and s.e.m. (shadow) traces of the vessel diameter change at each vessel location upon WP stimulation in adult and old mice. g–j, Comparison of relative dilation amplitude (g), undershoot amplitude (h), half-peak dilation onset (i) and half-peak dilation duration (j) in adult and old brains. Adult, N = 31 animals, n = 56 vessels; old, N = 26 animals, n = 53 vessels. k, Elicited LFPs by WP stimulation and recorded by the same glass micropipette. Mean (solid curve) and s.e.m. (shadow) traces of LFPs in adult and old mice. l–n, Comparison of the amplitudes, latency and area under the curve (AUC) of fEPSPs. Adult, N = 17 animals, n = 34 vessels; old, N = 17 animals, n = 25 vessels. o,p, Representative images of MIT in response to 5 mM pinacidil puffing in adult and old mice. Scale bar, 20 μm. Small insets show enlarged precapillary sphincters (Scale bar, 10 μm). q,r, Mean (solid curve) and s.e.m. (shadow) traces of the vessel diameter change at each vessel location upon pinacidil puffing in adult and old mice. s, Comparison of relative dilation amplitude by pinacidil puffing. Adult, N = 9 animals, n = 15 vessels; old, N = 11 animals, n = 17 vessels. Raw data for pinacidil puffing in adult brains were reused from our previous publication. Linear mixed-effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as the mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. NS, not significant.

Fig. 2 |. Papaverine-induced vasodilation and l-NAME- and ET1-induced vasoconstrictions are attenuated in brain aging and are dependent on vascular mural cells.

a–c, Comparison of vessel dilation with 10 mM papaverine puffing. a,b, Representative images of vascular response to papaverine puffing in adult and old brains. Scale bar, 10 μm. Small insets show enlarged precapillary sphincters (scale bar, 10 μm). c, Comparison of relative dilation amplitude by papaverine puffing. Adult, N = 8 animals, n = 17 vessels; old, N = 5 animals, n = 8 vessels. d,e, Representative images of MIT in response to intravenous bolus infusion of l-NAME at a dose of 30 mg per kg body weight in adult (d) and old (e) mice. Scale bar, 20 μm. Small insets show enlarged precapillary sphincters (scale bar, 10 μm). f,g, Mean (solid curve) and s.e.m. (shadow) traces of the vessel diameter changes at each vascular location upon l-NAME infusion in adult and old mice. h, Comparison of relative diameter changes at 4th minute after l-NAME infusion. i, P values of statistical analysis for comparison of averaged constriction amplitudes at each vascular location at the 1st, 2nd, 3rd and 4th minutes after injection of l-NAME. Red highlighted grids show P values that are smaller than 0.05. Adult, N = 5 animals, n = 5 vessels; old, N = 8 animals, n = 8 vessels. Raw data for the intravenous infusion of l-NAME in adult brains were reused from our previous publication. j,k, Representative images of MIT in response to 0.5 μM ET1 puffing in adult (j) and old (k) mice. Scale bar, 20 μm. Small insets show enlarged precapillary sphincters at the same imaging plane in the green channel. Scale bar, 10 μm. l,m, Mean (solid curve) and s.e.m. (shadow) traces of the vessel diameter change at each vascular location upon ET1 puffing. The initial diameter drop was due to a momentary mechanical pressure change caused by puffing. n, Comparison of relative dilation amplitude upon ET1 puffing. Adult, N = 5 animals, n = 10 vessels; old: N = 7 animals, n = 15 vessels. Raw data for ET1 puffing in adult brains were reused from our previous publication. Linear mixed-effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as the mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. i.v., intravenous.

Fig. 3 |. Change of vascular mural cell structure at the microvascular inflow tract with age.

a,b, Maximum intensity projection of an adult (a) and old (b) brain image stack. Red, NG2; green, αSMA; blue, Hoechst. Enlarged insets of each vessel segment are presented in the middle for comparison. Scale bar, 10 μm. Note the vertical branch sprouting between first-order and second-order capillaries in a. Clearer individual planes are shown in Extended Data Fig. 9. c, Mural cell coverage of vessel surface obtained from dividing mural cell area by the total vessel surface area. d, Mural cell soma density along vessel calculated from the number of co-localized Hoechst and NG2 divided by vessel length. e, Mural cell soma number divided by endothelial cell soma number at each vessel location. f, αSMA density at mural cells obtained from calculating the mean αSMA fluorescence intensity at NG2-positive vessel areas. Adult, N = 6 animals, n = 34 vessels; old, N = 6 animals, n = 40 vessels. Linear mixed-effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as the mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. a.u., arbitrary units.

Fig. 4 |. Aging-induced angioarchitectural remodeling.

a, A representative image stack recorded by two-photon microscopy, containing the PA, the first-order capillary (1stCap), the second-order capillary (2ndCap), the third-order capillary (3rdCap), the fourth-order capillary (4thCap) and the fifth-order capillary (5thCap). Scale bar, 50 μm. b,c, Three-dimensional reconstructed vascular skeleton of the somatosensory cortex in an adult (b) and an old (c) mouse. Raw data were obtained by two-photon microscopy and analyzed by Amira software. Red, PAs; blue, 1stCap; green, 2ndCap; yellow, 3rdCap; orange, 4thCap; brown, 5thCap; purple, ascending venules. d–f, Vessel total length per brain volume in adult and old somatosensory cortex. General vessel total length (d), capillary-order-dependent vessel total length (e) and cortical depth-dependent vessel total length (f). g–i, Vessel mean diameter in adult and old somatosensory cortex. General mean diameter (g), capillary-order-dependent mean diameter (h) and cortical depth-dependent mean diameter (i). j–l, Vessel mean tortuosity in adult and old somatosensory cortex. The tortuosity index denotes division of the curved length by the chord length of each vessel segment. General mean tortuosity (j), capillary-order-dependent mean tortuosity (k) and cortical depth-dependent mean tortuosity (l). Adult, N = 6 animals, n = 52 vessels; old, N = 6 animals, n = 29 vessels. Linear mixed-effect models were used to test for differences among vessel segments, followed by Tukey post hoc tests for pairwise comparisons. Data are given as the mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. L1–L5 denote cortical sublayers.