Blood flow controls bone vascular function and osteogenesis

Abstract

While blood vessels play important roles in bone homeostasis and repair, fundamental aspects of vascular function in the skeletal system remain poorly understood. Here we show that the long bone vasculature generates a peculiar flow pattern, which is important for proper angiogenesis. Intravital imaging reveals that vessel growth in murine long bone involves the extension and anastomotic fusion of endothelial buds. Impaired blood flow leads to defective angiogenesis and osteogenesis, and downregulation of Notch signalling in endothelial cells. In aged mice, skeletal blood flow and endothelial Notch activity are also reduced leading to decreased angiogenesis and osteogenesis, which is reverted by genetic reactivation of Notch. Blood flow and angiogenesis in aged mice are also enhanced on administration of bisphosphonate, a class of drugs frequently used for the treatment of osteoporosis. We propose that blood flow and endothelial Notch signalling are key factors controlling ageing processes in the skeletal system.

Figure 1. Blood flow dynamics in long bone.

(a) Tile scan image of tibial vasculature immunostained for Emcn (red) and α-smooth muscle actin (green). Arteries covered by α-SMA+ cells connect (arrows) to metaphyseal (mp) type H vessels near growth plate (dashed lines, gp). Sinusoidal type L vessels connect to central large vein (yellow arrowhead). (b,c) Confocal images of 4-week-old metaphysis near growth plate (gp) (b, top panels) or diaphysis (dp) (c, top). CD31+ (green) Emcn- (red) arteries terminate in CD31+ Emcn+ type H vessels in the metaphysis (mp) and endosteum (es) but not in type L vessels in diaphysis (dp). Blue arrows in Emcn-stained (red) overview images (bottom panels) illustrate blood flow direction from metaphyseal vessel columns (b) and endosteum (c) into the adjacent sinusoidal network and veins. (d) Maximum intensity projection of transversal sections of 4-week-old tibia immunostained for α-SMA (green) and Emcn (red). Sinusoidal type L vessels (arrowheads) connect to a large central vein (v). Multiple smooth muscle-covered CD31+ Emcn-arteries cross the diaphysis. Dashed lines mark compact bone. (e) Diagram of arterial (green arrows), type H (red arrows) and sinusoidal/venous flow (blue arrows) in murine long bone. (f) Graph showing blood velocities calculated from line scanning of type H and type L vessels after Dextran (2,000,000 Da) injection. Data represent mean±s.d. (n=5 biological replicates). P values, two-tailed unpaired t test. (g) Overview image and representative line scans showing blood velocities in 2-week-old long bone. Note decreasing velocity of erythrocytes at each branch point (represented as 1°, 2°, 3°, 4°) at the interface between columnar type H (right) and diaphyseal type L vessels (left). Data represent mean±s.d. (n=5 biological replicates). (h) Confocal images of CD31 (green) expression in the metaphysis and diaphysis of 4-week-old tibia. Note differences in CD31 levels. Scale bars are as indicated in the respective images. DAPI (blue) is used for counterstaining of nuclei.

Figure 2. Changes in blood flow impair bone angiogenesis.

(a) Schematic representation of unilateral femoral artery ligation in 3-week-old mice. Results compare operated (ipsilateral) legs with contralateral controls. (b) Representative confocal images showing CD31 (green) and Emcn (red) immunostaining of tibia from ligated and contralateral side at 48 h post surgery (hps). Note decline in vascular front structures, buds and columns on ligation. Insets show vessel buds (blue arrowheads) on the contralateral side and appearance of pointed, sprout-like structures (white arrowheads) after ligation. (c) Analysis of blood flow in Prazosin-treated tibia compared with saline-treated control. Data represent mean±s.d. (n=4 biological replicates). P values, two-tailed unpaired t test. (d,e) Confocal images of 4-week-old Prazosin-treated and control tibial metaphysis immunostained for Emcn (red; d) or CD31 (green; e). Insets in d show vessel buds (blue arrowheads), which appeared collapsed (white arrowheads) after Prazosin treatment. CD31 (green) expression was decreased in Prazosin-treated animals (e). (f) Flow cytometric quantification of tibial type H ECs after treatment with Prazosin or control (saline). (g) Quantitation of EdU+ECs present in the metaphysis of Prazosin-treated and control mice. Data represent mean±s.e.m. (n=4 biological replicates). P values, two-tailed unpaired t test.

Figure 3. Blood vessel growth in bone.

(a) Maximum intensity projection of metaphyseal vessel structures (CD31, red): distal, loop-like arches (A) and bud-shaped protrusions (B) present in proximity of hypertrophic chondrocytes (Ch). (b) Representative confocal images of tibia sections from 4, 12 and 80-week-old mice immunostained for Emcn (red) and CD31 (green). CD31^hi Emcn^hi ECs form columns, distal arches (A) and buds (B) at the vascular growth front, which were abundant at 4 weeks but declined during ageing (12 and 80 weeks). Arrowheads indicate Emcn- arterioles. Ch, chondrocytes. (c) Still images of the indicated time points from a 10 h time lapse movie of 10-day-old Flk1-GFP metatarsal. Arrow indicates bud emerging from arch (A). (d) SEM images of blood vessel (BV) next to chondrocytes (Ch) in 4-week-old tibia with distal arch (A) and bud (B; arrowhead). Right panel shows higher magnification of inset with bud emerging from arch. (e) Merged confocal and differential interference contrast (DIC) image showing distal vessel buds (B) genetically labelled by GFP (green). Arrow indicates a filopodia-bearing endothelial bud displacing an apoptotic chondrocyte, arrowhead points at filopodia-free bud next to intact chondrocyte. (f) 4-week-old Cdh5-CreERT2 R26-Confetti double transgenics at 7 and 21 days post injection (dpi) showing high degree of mosaicism in endothelial columns, arches and buds. (g) Schematic illustration of vessel growth in long bone. Invading vessels buds (B) displace apoptotic growth plate chondrocytes (c) and, through anastomosis, generate new arches (A), from which new buds can emerge. (h) Representative confocal images showing contact formation and anastomosis (arrows) of CD31+ (green) vessel buds in 4-week-old tibia. Note red presence of Ter-119+ RBCs (red) in buds and forming vessel arches. (i) Still images (time indicated) from movie showing anastomosis (arrows) of 2 buds (arrowheads in left image) in 10-day-old Flk1-GFP (green) metatarsal. Arrows show the disappearance of junction between anastomosing buds at later time points.

Figure 4. Flow-mediated angiogenesis is coupled to osteogenesis.

(a) Maximum intensity projections of Emcn (red) and Osterix (Osx, green) immunostaining in contralateral (control) and ipsilateral (ligated) tibia at 48 hps after femoral artery ligation. (b) Quantitation of Osx+ cells in contralateral and ipsilateral tibia after femoral artery ligation. Data represent mean±s.e.m (n=5 mice in three independent experiments). P values, two-tailed unpaired t test. (c) Confocal images of Prazosin and control tibial sections immunostained for Emcn (red) and Osterix (Osx, green). (d) 3D rendering showing micro-CT analysis of mineralized regions in the metaphysis of control and Prazosin-treated tibiae. (e) Histomorphometrical data derived from micro-CT scans. Note reduced bone parameters after Prazosin treatment. (f) RT-qPCR for the expression of known pro-osteogenic factors such as Tgfb1, Tgfb2, Tgfb3, Wnt10b, Pdgfb, Bmp2, Bmp4 and Fgf1 (normalized to Actb expression) in sorted bone ECs from control and Prazosin-treated mice. Data represent mean±s.d. (n=8 biological replicates). P value, two-tailed unpaired t test.

Figure 5. Flow positively regulates endothelial Notch signalling in bone.

(a) Dll4 (green) immunostaining of Prazosin-treated and control tibia sections. (b,c) qPCR analysis of Dll4, Hes1, Hes5, Hey1, Hey2 and Jag1 expression (normalized to Actb) in ECs sorted from Prazosin-treated or control long bone (b) and from ipsilateral (operated) and contralateral sides of femoral artery ligated (16hps) limbs of mice (c). Data represent mean±s.d. (n=5 biological replicates). P values, two-tailed unpaired t test. (d) qPCR analysis of Dll4, Hes1, Hes5, Hey1, Hey2 and Jag1 expression (normalized to Actb) in ECs sorted from 4 (young) and 85-week-old (old) long bone. Data represent mean±s.d. (n=5 biological replicates). P values, two-tailed unpaired t test. (e) Representative confocal images showing the metaphysis region near the growth plate (gp, dashed line) in young (4 week) and aged (85 week) tibia immunostained for Emcn (red) and Dll4 (green). (f) Blood flow measurements in young and old mice show reduced flow upon ageing. Data represent mean±s.d. (n=5 biological replicates). P values, two-tailed unpaired t test. (g) Maximum intensity projections of tibial sections from NICD^iOE-EC (Notch gain-of-function) mice and littermate controls treated with saline (control) and Prazosin, immunostained for CD31 (green) and Emcn (red). Blue arrowheads indicate buds and arches, white arrowheads mark defective buds in the vascular front. (h) Experimental scheme of tamoxifen administration to aged transgenic mice for EC-specific activation of Notch signalling. (i) Confocal images of aged NICD^iOE-EC and littermate control tibiae immunostained for CD31 (green) and Emcn (red). Note increase in CD31+ vessels and buds (arrows) near NICD^iOE-EC growth plate (gp, dashed blue line). (j) Quantification of bud structures in the vascular front of aged NICD^iOE-EC and littermate control tibiae. Data represent mean±s.d. n=4 in three independent experiments. P values, two-tailed unpaired t test. (k) Confocal images of the metaphyseal region of aged NICD^iOE-EC and littermate control tibiae immunostained for Osterix (Osx, white). Nuclei, DAPI (blue).

Figure 6. Alendronate stimulates endothelial Notch signalling and flow in bone.

(a) Graph showing age-dependent changes in EdU+ ECs and bud structures in distal tibia. Blood flow measurements in the same age groups indicate a similar pattern. Data represent mean±s.d. (n=5 biological replicates). P values, two-tailed unpaired t test. (b) Tile scan images of tibia from control and Alendronate-treated 80-week-old mice. Sections were immunostained for CD31 (green) and Emcn (red). Note appearance of CD31+ blood vessels (arrows) in Alendronate-treated metaphysis. (c) Vascular front in sections of control and Alendronate-treated aged tibia immunostained for CD31 (green), Emcn (red) and Osterix (Osx, white). Nuclei, DAPI (blue). Inset shows CD31+ blood vessels in close proximity to osteoblasts. (d) Confocal images showing metaphysis of aged mice after Alendronate administration. Immunostaining shows endothelial Dll4 (green). Nuclei, DAPI (blue). (e) qPCR analysis of Dll4, Hes1, Hes5, Hey1 and Jag1 transcripts expression in ECs sorted from control and Alendronate-treated 80-week-old long bone. Data represent mean±s.d. (n=6 biological replicates). P values, two-tailed unpaired t test. (f) Increased blood flow in tibia of Alendronate-treated aged (80-week-old) mice. Data represent mean±s.d. (n=6 biological replicates). P values, two-tailed unpaired t test. (g) Blood flow measurements in tibia after short-term treatment with Alendronate. Data represent mean±s.d. (n=6 biological replicates). P values, two-tailed unpaired t test.

Figure 7. Schematic representation of key findings.

Postnatal angiogenesis in bone involves the formation of arch and bud structures. Blood flow is higher in metaphyseal type H vessels, which correlates with higher expression and activity of the Notch pathway. Reduction of flow impaired endothelial Notch activity, bone angiogenesis and osteogenesis. In aged mice, type H capillaries, vessel buds, endothelial Notch signalling and osteogenesis were low, all of which were reactivated by induction of EC-specific Notch activity leading to significant increases in osteoprogenitors and mineralized bone.