Endothelial Piezo1 sustains muscle capillary density and contributes to physical activity

Abstract

Piezo1 forms mechanically activated nonselective cation channels that contribute to endothelial response to fluid flow. Here we reveal an important role in the control of capillary density. Conditional endothelial cell-specific deletion of Piezo1 in adult mice depressed physical performance. Muscle microvascular endothelial cell apoptosis and capillary rarefaction were evident and sufficient to account for the effect on performance. There was selective upregulation of thrombospondin-2 (TSP2), an inducer of endothelial cell apoptosis, with no effect on TSP1, a related important player in muscle physiology. TSP2 was poorly expressed in muscle endothelial cells but robustly expressed in muscle pericytes, in which nitric oxide (NO) repressed the Tsp2 gene without an effect on Tsp1. In endothelial cells, Piezo1 was required for normal expression of endothelial NO synthase. The data suggest an endothelial cell-pericyte partnership of muscle in which endothelial Piezo1 senses blood flow to sustain capillary density and thereby maintain physical capability.

Keywords: Ion channels; Microcirculation; Muscle Biology; Skeletal muscle; Vascular Biology.

Figure 1. Endothelial Piezo1 determines physical performance but not desire for activity.

Throughout the figure, data in gray represent control mice (Ctrl) and data in orange are for Piezo1^ΔEC mice. The lighter color is for data sampled during the light cycle (inactive period) and darker color for data during the dark cycle (active period). (A) Pooled and averaged ambulatory activity (XAMB) across 3 light and dark cycles for Ctrl mice and Piezo1^ΔEC mice. (B) Day-by-day averaged ambulatory activity. (C) Similar to A but showing exploratory activity (ZTOT). (D) Day-by-day averaged exploratory activity. (E) Similar to A but showing running wheel rotation counts (voluntary activity). (F) Day-by-day averaged voluntary activity. (G) Cumulative running wheel rotations during 72-hour recording. Gray shaded areas indicate the dark cycles. (H) Number of active bouts of exercise (periods of activity defined as activity seen in 1 or more consecutive 10-minute intervals). (I) Number of interbout pauses (periods of inactivity between 2 bouts of exercise). (J) Percentage of time for which mice were active on the wheel. (K) Percentage of time for which mice were off the wheel (inactive time). (L) Normalization of running wheel rotations per active bouts of exercise. (M) Running-wheel speed. All data are for n = 10 mice per group (mean ± SD). Superimposed dots are the individual underlying data values for each individual mouse. **P < 0.01, ***P < 0.001 vs. light cycle; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. Ctrl mice. Statistical significance was evaluated using 2-way ANOVA followed by Tukey’s HSD post hoc test for multiple comparisons.

Figure 2. Endothelial Piezo1 specifically affects microvascular density.

(A) Immunohistochemistry of gastrocnemius muscle cross sections for myosin heavy chain (MHC) type 2a (green) plus type 2b (red, left) or type 2x (magenta, right). Scale bars: 100 μm. (B) Quantification of the relative frequency of the different fiber types in gastrocnemius muscle. (C) Immunohistochemistry for CD31 (red) to visualize endothelial cells in capillaries of gastrocnemius muscle sections. Scale bars: 50 μm. (D) Mean data for capillary density measured from images of the type shown in C. (E) Similar to D but showing mean data for the ratio of capillaries to muscle fibers. (F) Immunohistochemistry for isolectin B4 (IB4, green) to visualize endothelial cells in capillaries of gastrocnemius muscle sections. Scale bars: 50 μm. (G) Mean data for capillary density measured from images of the type shown in F. (H) Similar to G but showing mean data for the ratio of capillaries to muscle fibers. All data are for n = 7 to 8 mice per group (mean ± SD). Superimposed dots are the underlying data values for each individual mouse. Gray indicates muscles from Ctrl mice and green indicates muscles from Piezo1^ΔEC mice. ^##P < 0.01, ^###P < 0.001 vs. ctrl mice. Statistical significance was evaluated using Student’s t test.

Figure 3. Protection against endothelial microvascular rarefaction.

(A) Immunohistochemistry for CD31 (red) and type IV collagen (Coll IV, green) to visualize capillaries and basement membrane, respectively, in gastrocnemius muscle sections. Merged images are shown on the right. Scale bars: 50 μm. Superimposed circles highlight an example of a regressing vessel (CD31^–Coll IV⁺). (B) Quantification of empty Coll IV sleeves in gastrocnemius muscle sections, based on images of the type shown in A. (C) Pearson’s correlation of capillary density and empty Coll IV sleeves (r = –0.71, P = 0.003). The black line is the correlation fit. All data are for n = 7 to 8 mice per group (mean ± SD). Superimposed dots are the underlying data values for each mouse. Gray indicates muscles from Ctrl mice and green indicates muscles from Piezo1^ΔEC mice. ^###P < 0.001 vs. Ctrl mice. Statistical significance was evaluated using Student’s t test (B) or Pearson’s correlation (r) test (C).

Figure 4. Protection against endothelial cell apoptosis.

(A) Immunohistochemistry for endothelial cells in capillaries (IB4, green), apoptotic cells (TUNEL, red), and nuclei (DAPI, blue) in gastrocnemius muscle sections. Merged images are shown on the right. Scale bars: 30 μm. Superimposed circles highlight an example of an apoptotic endothelial cell (IB4⁺TUNEL⁺DAPI⁺). (B) Quantification of apoptotic endothelial cell percentages in Ctrl and Piezo1^ΔEC gastrocnemius muscle using images of the type shown in A. (C) Pearson’s correlation analysis of capillary density and percentage of apoptotic endothelial cells (r = –0.62, P = 0.014). (D) Pearson’s correlation analysis of apoptotic endothelial cell percentage and capillary regression (r = 0.75, P = 0.001). The black lines are the correlation fits. All data are for n = 7 to 8 mice per group (mean ± SD). (E) Quantitative PCR mRNA expression data for proapoptotic markers (Bax, Bak) and antiapoptotic markers (Bcl2, BclXL) in endothelial cells isolated from skeletal muscle of Ctrl (gray) and Piezo1^ΔEC (blue) mice. (F) Quantitative PCR mRNA expression data for proapoptotic markers (Bax, Bak) and antiapoptotic markers (Bcl2, BclXL) in whole gastrocnemius muscle of Ctrl (gray) and Piezo1^ΔEC (green) mice. RNA abundance was normalized to housekeeping gene expression and is presented as the fold-change relative to that in Ctrl mice. All data are for n = 5 to 6 mice per group for endothelial cells and n = 14 to 16 mice per group for whole muscle (mean ± SD). Superimposed dots are the individual underlying data values for each mouse. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. Ctrl mice. Statistical significance was evaluated using Student’s t test, except in C and D where Pearson’s correlation was used.

Figure 5. Downstream signaling mediated by eNOS and TSP2.

(A) Relative mRNA levels for candidate downstream genes in whole gastrocnemius muscle from control (gray) and Piezo1^ΔEC (green) mice. (B) Representative Western blot for TSP2 protein in gastrocnemius muscle. (C) For data of the type shown in B, quantification of TSP2 protein normalized to GAPDH and expressed as fold-change in Piezo1^ΔEC compared to Ctrl. (D and E) Relative mRNA levels for (D) the Tsp2 gene in isolated endothelial cells from skeletal muscle (SkECs) of Ctrl (gray) and Piezo1^ΔEC (blue) mice and (E) eNOS (Nos3 gene) mRNA in whole gastrocnemius muscle from Ctrl (gray) and Piezo1^ΔEC (green) mice and isolated SkECs of Ctrl (gray) and Piezo1^ΔEC (blue) mice. mRNA abundance was determined by qRT-PCR, normalized to housekeeping gene expression, and is presented as the fold-change relative to Ctrl mice. (F) Representative Western blot for eNOS phosphorylation at serine 1177 (p-eNOS) and total eNOS (t-eNOS) in isolated SkECs. (G) For data of the type shown in F, quantification of p-eNOS relative to t-eNOS in Piezo1^ΔEC compared to Ctrl mice. (H) For data of the type shown in F, quantification of t-eNOS relative to the housekeeper protein GAPDH in Piezo1^ΔEC compared to Ctrl. (I) Immunohistochemistry for CD31 (green) and eNOS (yellow) in gastrocnemius muscle longitudinal sections. Merged images are shown on the right. eNOS fluorescence intensity was measured in CD31⁺ regions (red). Scale bars: 15 μm. (J) Quantification of eNOS fluorescence intensity in CD31⁺ regions corresponding to endothelial cells. Data are for n = 8 to 9 mice per group (A and F–H), n = 10 to 13 (B and C), n = 5 to 6 (D and E), and n = 7 to 8 (I and J) (mean ± SD). Superimposed dots are the underlying data values for each mouse. ^#P < 0.05, ^##P < 0.01 vs. Ctrl mice. Statistical significance was evaluated using Student’s t test.

Figure 6. In situ upregulation of TSP2 in pericytes.

(A and B) Relative mRNA abundance for (A) Tsp2 in isolated SkECs (gray) and pericytes (yellow) from WT mice, and (B) Tsp2 and Tsp1 in isolated pericytes from WT mice: untreated (yellow), treated with NO inhibitor (1 mM L-NMMA, light yellow), or with NO donor (300 μM GSNO, dark yellow/green) for 4 hours. mRNA abundance was determined by qRT-PCR, normalized to housekeeping gene expression, and is presented as fold-change relative to isolated SkECs (A) or untreated pericytes (B). (C) Immunohistochemistry for CD31 (green), NG2 (magenta), and TSP2 (yellow) in gastrocnemius muscle longitudinal sections. Merged images are shown on the right. TSP2 fluorescence intensity was measured in NG2⁺ regions (red). Scale bars: 15 μm. (D) Quantification of TSP2 fluorescence intensity in NG2⁺ regions corresponding to pericytes. (E) Schematic model of the mechanism. Data are for n = 4 mice per group (A and B) and n = 7 to 8 (C and D) (mean ± SD). Superimposed dots are the underlying data values for each mouse. ^#P < 0.05 vs. Ctrl mice; ***P < 0.001 vs. WT isolated SkECs; ^§P < 0.05 vs. untreated WT isolated pericytes. Statistical significance was evaluated using Student’s t test.