ABSTRACT
The thin endothelial wall of a newly formed vessel is under enormous stress at the onset of blood flow, rapidly acquiring support from mural cells (pericytes and vascular smooth muscle cells; vSMCs) during development. Mural cells then develop vasoactivity (contraction and relaxation) but we have little information as to when this first develops or the extent to which pericytes and vSMCs contribute. For the first time, we determine the dynamic developmental acquisition of vasoactivity in vivo in the cerebral vasculature of zebrafish. We show that pericyte-covered vessels constrict in response to α1-adrenergic receptor agonists and dilate in response to nitric oxide donors at 4 days postfertilization (dpf) but have heterogeneous responses later, at 6 dpf. In contrast, vSMC-covered vessels constrict at 6 dpf, and dilate at both stages. Using genetic ablation, we demonstrate that vascular constriction and dilation is an active response. Our data suggest that both pericyte- and vSMC-covered vessels regulate their diameter in early development, and that their relative contributions change over developmental time.
INTRODUCTION
The cerebral vascular network is stabilized by perivascular cells, which include pericytes, vascular smooth muscle cells (vSMCs), astrocytes, microglia, perivascular macrophages (PVMs) and fluorescent granular perithelial cells (FGPs; originally known as Mato cells) (Gaengel et al., 2009; Galanternik et al., 2017; Mato and Ookawara, 1979; Williams et al., 2001). Two of these cell types, pericytes and vSMCs, play roles in actively regulating vessel diameter through contraction and relaxation. We use zebrafish to study the function of perivascular mural cells in blood flow regulation during development.
Even though pericytes and vSMCs have similar origins, the relative contributions of each cell type to vascular regulation during development is unknown. Moreover, as this process is hidden during mammalian embryogenesis, development of vasoactivity in vivo is unexplored in any species. Although more is known about mural cell function in the zebrafish trunk, pericytes and vSMCs of the head and trunk have different developmental origins. Consequently, there may be differences in mural cell function arising from different anatomical origins. Pericytes of the trunk derive from the sclerotome (Ando et al., 2016; Arciniegas et al., 2000; Santoro et al., 2009; Stratman et al., 2017) whereas those of the head originate from both the neural crest and the mesoderm (Ando et al., 2016; Cavanaugh et al., 2015; Kok et al., 2015; Wang et al., 2014; Whitesell et al., 2014). The trunk dorsal aorta (DA) acquires tone between 48 and 80 h postfertilization (hpf) (Stratman et al., 2017), but the developmental timing of brain vessel vasoactivity is unknown.
The process of mural cell recruitment to the endothelium is still poorly understood, and there are conflicting reports as to the role of blood flow. Blood flow initiates at ∼26 hpf in the zebrafish trunk DA. Flow induces shear stress on the endothelium. In this low-flow state, shear stress causes deflection of endothelial cilia that function as mechanosensors, and induces the recruitment of mural cells to the endothelium (Chen et al., 2017; Goetz et al., 2014). However in a separate study, mural cell recruitment in the head was shown to be flow independent around the basilar artery (BA) and central arteries (CtAs) of the brain, trunk DA and intersegmental vessels (ISVs) (Ando et al., 2016). Thus, our understanding of hemodynamics in driving mural cell recruitment requires further characterization.
Once attached to the endothelium, pericytes are solitary cells that contact other pericytes, astrocytes and endothelial cells. Their morphology varies along the vascular tree but they all have a rounded cell body, long processes and attach to the longitudinal axis of capillaries (Armulik et al., 2005, 2011; Attwell et al., 2016; Gaengel et al., 2009; Hartmann et al., 2015). Pericytes share the basement membrane with endothelial cells, and the two cell types physically contact each other via peg and socket connections (Armulik et al., 2005, 2011). In contrast, vSMCs wrap perpendicularly around vessels. They are connected to endothelial cells via myoendothelial gap junctions, allowing for direct signals from one cell to the other (Borysova et al., 2018). Vessel size determines the degree of coverage by vSMCs, as they form multi-layers around larger vessels and a single layer around smaller vessels (Santoro et al., 2009).
In zebrafish, pdgfrb marks pericytes (and not vSMCs) and is one of the earliest pericyte markers (Ando et al., 2016). pdgfrb transgene expression can be seen as early as the eight-somite stage in the neural crest, and marks cerebral pericytes by 48 hpf. Cerebral vSMCs can be visualized using the transgenic line Tg(acta2:EGF) (Whitesell et al., 2014), which is expressed in vSMCs (and not pericytes) and is first seen at about 3-3.5 days postfertilization (dpf) in the ventral head, much later than pericyte markers. We use pdgfrb and acta2 to identify pericytes and vSMCs because these markers are strongly and almost exclusively expressed in their respective cells on zebrafish brain vessels.
Here, we test the constriction and dilation ability of cerebral vessels covered by different vascular mural cells during development. As there is an increase in expression of contractile markers over the period between 4 dpf and 6 dpf, we hypothesized that vasoactivity also develops during this period. We find that vasoactivity changes dynamically during development. Using live imaging and genetic ablation experiments, we show that pericyte-covered brain capillaries constrict at earlier stages in development, while vSMC-covered arteries develop this ability later.
RESULTS
Pericytes and vSMCs appear on cerebral vessels of different diameters
Pericytes are first visualized in the zebrafish brain at 48 hpf and vSMCs at around 80 hpf. We tested multiple stages of development and chose two stages to study. At 4 dpf, pdgfrb-expressing pericytes have been present for about 2 days, while vSMCs are only starting to express acta2 and are therefore not fully differentiated. At 6 dpf, both cell types have been present on the vessels for several days and appear morphologically differentiated. Fig. 1A shows a schematic of the relative locations of pericytes and vSMCs in the developing zebrafish brain.
Mural cell morphology and coverage of the cerebral vasculature at 6 dpf. (A) A model of vSMC and pericyte locations on 6 dpf zebrafish embryo cerebral vessels. PHS, primordial hindbrain sinus; CaDI, caudal division of the internal carotid artery; BA, basilar artery; CtAs, central arteries. (B) Pericytes are typically found on vessels with diameters of ≤6.5 μm (n=215 pericyte-covered vessel regions), while vSMCs appear on the larger cerebral vessels (≥9.5 μm; n=130 vSMC-covered vessel regions). Cerebral vessels with diameters of 6.5-9.5 μm are covered by a mix of pericytes and vSMCs. (C) Lateral image of cerebral vessel pericyte coverage at 6 dpf. (D-F‴) Enlargements of vessels in C, highlighting contact between processes of different pericytes. Dotted lines outline pericyte cell bodies; arrows mark processes. (G) Lateral image of cerebral vessel vSMC coverage at 6 dpf, focusing on the CaDI and BA. (H-J‴) Enlargements of the vessels in G showing extensive vSMC coverage. Scale bars: 50 µm in C and G; 10 µm in D-F‴,H-J‴. A, P, V and D refer to anterior, posterior, ventral and dorsal.
Mural cell morphology and coverage of the cerebral vasculature at 6 dpf. (A) A model of vSMC and pericyte locations on 6 dpf zebrafish embryo cerebral vessels. PHS, primordial hindbrain sinus; CaDI, caudal division of the internal carotid artery; BA, basilar artery; CtAs, central arteries. (B) Pericytes are typically found on vessels with diameters of ≤6.5 μm (n=215 pericyte-covered vessel regions), while vSMCs appear on the larger cerebral vessels (≥9.5 μm; n=130 vSMC-covered vessel regions). Cerebral vessels with diameters of 6.5-9.5 μm are covered by a mix of pericytes and vSMCs. (C) Lateral image of cerebral vessel pericyte coverage at 6 dpf. (D-F‴) Enlargements of vessels in C, highlighting contact between processes of different pericytes. Dotted lines outline pericyte cell bodies; arrows mark processes. (G) Lateral image of cerebral vessel vSMC coverage at 6 dpf, focusing on the CaDI and BA. (H-J‴) Enlargements of the vessels in G showing extensive vSMC coverage. Scale bars: 50 µm in C and G; 10 µm in D-F‴,H-J‴. A, P, V and D refer to anterior, posterior, ventral and dorsal.
We first identified stereotypical vessel locations for pericytes and vSMCs. At 4 dpf, pericytes are present on smaller cerebral capillaries with diameters ≤6.5 μm (Fig. S1A,B), typically the brain central arteries (CtAs). vSMCs cover the larger vessels that are ≥9.5 μm in diameter (Fig. S1A,F). VSMC-covered vessels at 4 dpf include the caudal division of the internal carotid (CaDI) and the basilar artery (BA). In the zebrafish brain, the CaDI is much smaller than the mammalian equivalent (∼10-20 µm in diameter), and is roughly equivalent to an arteriole as it connects to capillaries. Cerebral vessels with diameters between 6.5 μm and 9.5 μm at the capillary-arteriolar interface have variable coverage by both pericytes and vSMCs. We excluded these vessels from further analysis, as we could not clearly separate mural cells in this size range of vessels due to mixed cell populations on the same vessel.
In the zebrafish brain at 4 dpf, pericytes appear sporadically along vessels with long processes that extend along the surface of the endothelium (Fig. S1C-E‴). vSMCs appear on only a few brain vessels and form a continuous sheet around the vessel (Fig. S1G-I‴). By 6 dpf, the abundance of mural cells increases; there are more pericytes than vSMCs at both stages. At 6 dpf, vessels with diameters ≤6.5 μm continue to be primarily covered by pericytes, while those with diameters of ≥9.5 μm are primarily covered by vSMCs (Fig. 1B). At 6 dpf, pericyte processes are in close proximity and appear to contact one another (Fig. 1C-F‴). vSMC coverage of the cerebral endothelium also extends more dorsally over time (Fig. 1G-J‴).
Pericyte-covered vessels constrict in early development
To determine whether blood vessels constrict in response to stimuli during early development, we used α1-adrenoceptor agonists. We established 10 μM as the most effective dose of phenylephrine (PE) at 6 dpf from published doses in combination with our own dose-response curves. As changes in heart rate might affect vascular dynamics and confound our analysis, we determined that heart rate was unaltered in response to pharmacological agents (Fig. S2). To determine changes in vessel diameter, the cerebral vasculature was imaged before dosing (0 min) and repeatedly every minute after PE treatment (1-5 min; Fig. 2A). At 4 dpf, pericyte-covered vessels constrict 4% within 3 min in response to PE and maintain their constriction to 5 min (P≤0.007, n=55 from 15 embryos; Fig. 2B,C-G; Movie 1; Table S1). Most pericyte-covered vessels constrict, but some dilate, and a subset does not respond to PE (Fig. S3A). Specifically, 35 out of 55 pericyte-covered vessels consistently constrict at both 3 and 5 min. There was no response to vehicle (E3) (P=0.416, n=29 from 8 embryos). We assessed changes in adjacent vessel segments not covered by pericytes as a control, and found they do not constrict in response to PE (P=0.972, n=69 from 18 embryos).
Pericyte-covered vessels constrict in response to phenylephrine at 4 dpf but not at 6 dpf. (A) Schematic of the experimental timeline. Imaging occurs prior to addition of 10 µM PE (0 min) and at 1 min intervals over 5 min after drug addition. (B) At 4 dpf, pericyte-covered vessels constrict in response to PE within 3 min (P=0.007, n=55 from 15 embryos). There was no constriction with vehicle (P=0.416, n=29 from 8 embryos). Vessel regions with no pericytes and diameters ≤6.5 μm did not respond to PE (P=0.972, n=69 from 18 embryos). When pericytes were ablated, ≤6.5 µm diameter vessels constricted in response to PE (P=0.0003, n=46 from 10 embryos) but not to vehicle (P=0.289, n=128 from 22 embryos). (C-G) An example of pericyte-covered vessel constriction in response to PE within 3 min. F is an overlay of D and E. Arrows indicate constriction between the two time points. (H) At 6 dpf, pericyte-covered vessels did not respond to PE (P=0.553, n=68 from 15 embryos). Scale bars: 10 µm in C-F. Values are mean±s.d. with individual data points indicated. Significance was determined using a repeated measure one-way ANOVA (*P≤0.05, **P≤0.005).
Pericyte-covered vessels constrict in response to phenylephrine at 4 dpf but not at 6 dpf. (A) Schematic of the experimental timeline. Imaging occurs prior to addition of 10 µM PE (0 min) and at 1 min intervals over 5 min after drug addition. (B) At 4 dpf, pericyte-covered vessels constrict in response to PE within 3 min (P=0.007, n=55 from 15 embryos). There was no constriction with vehicle (P=0.416, n=29 from 8 embryos). Vessel regions with no pericytes and diameters ≤6.5 μm did not respond to PE (P=0.972, n=69 from 18 embryos). When pericytes were ablated, ≤6.5 µm diameter vessels constricted in response to PE (P=0.0003, n=46 from 10 embryos) but not to vehicle (P=0.289, n=128 from 22 embryos). (C-G) An example of pericyte-covered vessel constriction in response to PE within 3 min. F is an overlay of D and E. Arrows indicate constriction between the two time points. (H) At 6 dpf, pericyte-covered vessels did not respond to PE (P=0.553, n=68 from 15 embryos). Scale bars: 10 µm in C-F. Values are mean±s.d. with individual data points indicated. Significance was determined using a repeated measure one-way ANOVA (*P≤0.05, **P≤0.005).
We next used genetic ablation to determine the role of pericytes in regulation of cerebral vessel diameter. Embryos expressing pdgfrb:Gal4 driver expressed in pericytes (Ando et al., 2016) in combination with a transgenic reporter expressing the nitroreductase (NTR) gene fused to mCherry under the UAS promoter were exposed to metronidazole (Mtz) for 24 h. NTR converts Mtz into a cytotoxic metabolite, resulting in cell death without damage to nearby cells. Ablation efficiency of mCherry-expressing cells was complete and cellular debris was observed. Mtz treatment started 24 h before imaging experiments. Mtz-treated fish had normal morphology and normal vessel structure after mural cell ablation (Fig. S4A-H′).
Pericyte-ablated zebrafish were exposed to vasoconstricting agents to determine residual vasoactivity. At 4 dpf, constriction of cerebral vessels was reduced after pericyte ablation. Cerebral vessels with diameters of ≤6.5 µm (vessels that would normally have pericyte coverage), show reduced constriction in response to PE (P≤0.0003, n=46 s from 10 embryos; Fig. 2B and Fig. S5A-D; Table S1). There was significant constriction at 3 min (P=0.019) and 5 min (P=0.001), the same time points at which constriction occurred when pericytes were unablated. Pericyte-ablated vessels did not constrict in response to vehicle (P=0.289, n=128 from 22 embryos). We hypothesized that incomplete ablation might account for remaining contractile activity, and therefore assessed mosaicism of pericyte transgenes at 4 dpf by crossing two lines using the same driver and two different reporters Tg(pdgfrb:GFP) and Tg(pdgfb:Gal4;UAS:NTRmCherry). We note that the two reporters largely overlap, but not completely, suggesting mosaicism in the transgene expression (Fig. S4I-K″). Thus, reduced contractile ability after pericyte ablation is consistent with pericytes playing an active role in vascular constriction at 4 dpf, with the caveat that there is incomplete ablation and therefore not a complete reduction in contractile activity.
Surprisingly, at 6 dpf, on average pericyte-covered vessels do not constrict in response to PE (Fig. 2H; Table S2, Fig. S3A). There was no significant change in vessel diameter in response to either vehicle (P=0.329, n=52 from 15 embryos) or PE (P=0.553, n=68 from 15 embryos), and vessel regions not covered by pericytes also did not respond to PE at 6 dpf (P=0.642, n=67 from 11 embryos). When pericytes were ablated at 5 dpf and tested at 6 dpf, there was also no response to PE (P=0.123, n=45 from 6 embryos) or to the vehicle control (P=0.141, n=114 from 19 embryos). Heterogeneity in response was noted but, on average, pericyte-covered vessels do not consistently respond to PE at 6 dpf.
We used a second α1-adrenoceptor agonist, noradrenaline (NA), to test whether it would cause contractile behavior in pericyte-covered vessels. Our dose-response curve identified 2 µM as the optimal dose with no effect on heart rate (Fig. S2). To determine changes in vessel diameter, the cerebral vasculature was imaged before dosing (0 min) and repeatedly every 1 min after NA treatment (1-4 min). At 4 dpf, pericyte-covered vessels constrict 6% in response to NA within the first minute of treatment (P≤0.0001, n=60 from 14 embryos; Fig. S6A-F; Movie 2; Table S3). Constriction persists for the duration of NA treatment and, over time, vessels constrict further. There is a heterogeneous response to NA with a subpopulation (31 out of 60) of pericyte-covered vessels constricting consistently (Fig. S3A), whereas other pericyte-covered vessels demonstrate both dilation and constriction during NA treatment. There was no constriction in the control vehicle-treated group (P=0.416, n=29 from 8 embryos). Vessels of a similar diameter but not covered by pericytes constricted 2% in response to NA (P=0.005, n=54 from 13 embryos) at 1 min (P=0.015), 2 min (P=0.008) and 4 min (P=0.019). We then ablated pericytes. After pericyte ablation, cerebral vessels with diameters of ≤6.5 µm (vessels that normally have pericyte coverage) constrict 6% in response to NA (P=0.008, n=18 from 4 embryos; Fig. S5I-L), although they constrict later than in unablated fish, at 3 min (P=0.027) and 4 min (P=0.015). In pericyte-ablated embryos treated with the vehicle, there was no constriction (P=0.289, n=128 from 22 embryos). Residual constriction could be due to incomplete ablation as noted previously.
Similar to the response to PE at 6 dpf, pericyte-covered vessels do not constrict in response to NA at 6 dpf (P≤0.019, n=46 from 15 embryos; Fig. S6G,H; Table S4). There was also no response to the vehicle control at 6 dpf (P=0.329, n=52 from 15 embryos) in vessel regions not covered by pericytes (P=0.059, n=44 from 9 embryos), or in pericyte-ablated vessels treated with NA (P≤0.065, n=50 from 12 embryos) or vehicle (P=0.141, n=114 from 19 embryos).
In summary, pericyte-covered vessels constrict in response to α1-adrenoceptor agonists PE and NA at 4 dpf but not at 6 dpf. Constriction at 4 dpf is active, as it is significantly reduced when pericytes are ablated.
vSMC-covered vessels constrict in later development
The response of vSMC-covered vessels to α1-adrenoceptor receptor agonists was tested next. In response to PE, vSMC-covered vessels constrict at both 4 and 6 dpf (Fig. 3; Tables S5 and S6). vSMC-covered vessels were imaged before dosing (0 min) and repeatedly every minute after treatment (1-5 min; Fig. 3A). At 4 dpf, vSMC-covered vessels constrict within the first minute (P≤0.0001, n=16 from 15 embryos; Fig. 3B), and by 13% over 5 min in response to PE (Fig. 3C-G; 14/16 vessels constrict, Fig. S3B). However, 8% constriction is also observed in the vehicle-treated control group at 4 dpf, which is unexpected (P=0.005, n=17 from 9 embryos).
Vessel constriction in developing vSMCs in response to phenylephrine. (A) Schematic of the experimental timeline. Imaging occurs prior to addition of 10 μM PE (0 min) and at 1 min intervals over 5 min after drug addition. (B) At 4 dpf, vSMCs constrict in response to PE (P≤0.0001, n=16 from 15 embryos) and to the vehicle control (P=0.005, n=17 from 9 embryos). When the vSMCs are ablated, there is constriction in response to both PE (P=0.070, n=14 from 10 embryos) and vehicle (P=0.0001, n=42 from 21 embryos). (C-G) An example of vSMC-covered vessel constriction in response to PE within 3 min at 4 dpf. F is an overlay of D and E. (H) At 6 dpf, vSMC-covered vessels significantly constrict in response to PE (P=0.0001, n=38 from 22 embryos) but not to vehicle (P=0.161, n=22 from 13 embryos). When vSMCs are ablated, there is no response to PE (P≤0.314, n=10 from 9 embryos) but there is constriction in response to the vehicle control (P=0.003, n=24 from 18 embryos). (I-M) An example of vSMC-covered vessel constriction to PE at 6 dpf. L is an overlay of J and K. Scale bars: 10 μm in C-F,I-L. Data are mean±s.d. Significance was determined using a repeated measure one-way ANOVA (*P≤0.05, ***P≤0.0005, ****P≤0.00005). ns, not significant.
Vessel constriction in developing vSMCs in response to phenylephrine. (A) Schematic of the experimental timeline. Imaging occurs prior to addition of 10 μM PE (0 min) and at 1 min intervals over 5 min after drug addition. (B) At 4 dpf, vSMCs constrict in response to PE (P≤0.0001, n=16 from 15 embryos) and to the vehicle control (P=0.005, n=17 from 9 embryos). When the vSMCs are ablated, there is constriction in response to both PE (P=0.070, n=14 from 10 embryos) and vehicle (P=0.0001, n=42 from 21 embryos). (C-G) An example of vSMC-covered vessel constriction in response to PE within 3 min at 4 dpf. F is an overlay of D and E. (H) At 6 dpf, vSMC-covered vessels significantly constrict in response to PE (P=0.0001, n=38 from 22 embryos) but not to vehicle (P=0.161, n=22 from 13 embryos). When vSMCs are ablated, there is no response to PE (P≤0.314, n=10 from 9 embryos) but there is constriction in response to the vehicle control (P=0.003, n=24 from 18 embryos). (I-M) An example of vSMC-covered vessel constriction to PE at 6 dpf. L is an overlay of J and K. Scale bars: 10 μm in C-F,I-L. Data are mean±s.d. Significance was determined using a repeated measure one-way ANOVA (*P≤0.05, ***P≤0.0005, ****P≤0.00005). ns, not significant.
To test whether constriction at 4 dpf is active, we genetically ablated vSMCs using double transgenic acta2:Gal4;UAS:NTR-mCherry animals. Mtz was added 24 h before imaging, and the absence of vSMCs on vessels was determined visually. The expression of acta2:Gal4 is mosaic, as not every cell is labelled in this construct when we compare it with acta2:GFP (Fig. S4L-N″). Therefore, we expect that ablation will reduce, but not eliminate, vSMCs. Cerebral vessels of a diameter and location that would normally be covered by vSMCs constrict 4% in response to PE at 4 dpf after ablation (P=0.070, n=14 from 10 embryos; Fig. 3B, Fig. S5E-H). However, constriction occurred later, was weaker and was not maintained relative to unablated vessels (P=0.016). This reduction in contractile ability is consistent with active contraction by vSMCs at 4 dpf. We found a 5% constriction of the ≥9.5 µm diameter vessels in response to the vehicle when vSMCs are ablated (P=0.0001, n=2 from 21 embryos; Table S5). We conclude that some vSMC-covered vessels can constrict in response to PE at 4 dpf. Embryos with ablated vSMCs or treated with vehicle have delayed and weaker constriction. However, as vSMC-covered vessel constriction is also observed in these controls, we conclude there is both a specific response of vSMCs via the α1-adrenoceptor pathway, as well as a non-specific response that is sensitive to the addition of a vehicle.
At 6 dpf, vSMC-covered vessel constriction is robust and specific. vSMC-covered vessels constrict an average of 9% in response to PE (P=0.0001, n=38 from 22 embryos; Fig. 3H-M; Table S6; 27/38 vessels constrict, Fig. S3B) but not in response to vehicle (P=0.161, n=22 from 13 embryos). When vSMCs are ablated, there is no constriction of the ≥9.5 µm vessels to PE (P≤ 0.314, n=10 from 9 embryos). However, there is constriction in the vehicle-treated group with ablated vSMCs (P=0.003, n=24 from 18 embryos). Our 6 dpf genetic ablation data suggest that vSMCs actively contract at 6 dpf to constrict cerebral vessel diameter.
vSMCs also respond to a second α1-adrenoceptor agonist, NA. At 4 dpf, vSMC-covered vessels constrict 17% in response to NA (P= 0.006, n=10 from 9 embryos, Figs S3B and S7A-F; Movie 3; Table S7) and 8% in response to the vehicle (P=0.005, n=17 from 9 embryos). vSMC-covered vessels constrict to NA typically after 3 min of exposure and remain constricted. When we ablate vSMCs, vessels normally covered by vSMCs and with diameters ≥9.5 µm constrict 18% in response to NA (P=0.001, n=9 from 7 embryos; Fig. S5M-P) at 2 min (P=0.012) and remain constricted. These vSMC-ablated vessels constricted 5% in response to vehicle (P=0.0001, n=42 from 21 embryos). Thus, at 4 dpf, we conclude that the ability of vSMCs to constrict in response to NA is not fully developed. Similar to our conclusions using PE as a stimulus, these data suggest that there is an agonist-specific response of vSMCs at 4 dpf, and a non-specific response to vehicle.
At 6 dpf, vSMC-covered vessels constrict as predicted in response to NA. They constrict faster than they did at 4 dpf, responding within the first minute and constricting 8% for the duration of drug exposure (P≤0.0001, n=32 from 25 embryos; Figs S3B and S7G-M; Table S8). There was no constriction in response to vehicle (P=0.161, n=22 from 13 embryos). When we ablated vSMCs, no constriction in response to NA (P≤0.233, n=12 from 10 embryos) was seen but there was a 7% constriction in the vehicle-treated vSMC-ablated group (P=0.003, n=24 from 18 embryos). Our 6 dpf genetic ablation data suggest that vSMCs actively contract at 6 dpf to constrict vessel diameter.
Our data suggest that vSMCs contract at both 4 and 6 dpf. There is both a specific and non-specific component at 4 dpf, but a more consistent response at 6 dpf.
Pericytes and vSMCs actively maintain resting vessel diameter
A second method for determining whether pericytes and vSMCs actively regulate vascular tone is to compare resting vessel diameter in the presence and absence of each mural cell. Vascular tone is the maintenance of a constant vessel diameter that is less than maximal dilation in response to blood flow. At 4 dpf, unstimulated vessels with ablated pericytes are significantly enlarged by 6% from 5.2±0.9 µm to 5.5±0.7 µm (P≤0.016; Fig. S5Q). At 6 dpf, as expected, the average vessel diameter in the presence and absence of pericytes was not significantly different: 4.9±0.8 µm and 4.9±0.9 µm (Fig. S5R). This data support the observation that pericytes actively regulate vessel diameter at 4 dpf but not 6 dpf.
vSMCs regulate tone in a complementary manner. At 4 dpf, there is no difference in cerebral vessel diameter in the presence and absence of vSMCs (Fig. S5S), consistent with our conclusion that they do not have full capacity to actively regulate vessel diameter at this time. The average vessel diameter with and without vSMCs was 11.8±1.8 µm and 11.7±1.8 µm, respectively. However, at 6 dpf, the average vessel diameter without vSMC coverage was 12% larger (11.6±1.6 µm versus 13.0±2.5 µm in unstimulated ablated vessels; P≤0.001; Fig. S5T), suggesting active regulation of vascular tone by vSMCs at 6 dpf. Thus, using a second method, we demonstrate that pericyte- and vSMC-covered vessels have developed tone by 4 dpf, and 6 dpf, respectively.
Vessel dilation occurs in early development
Nitric oxide is a key molecule mediating vascular mural cell relaxation. To test the ability of developing vessels to dilate, we exposed zebrafish to nitric oxide donors at 4 dpf and 6 dpf. We carried out dose-response curves for each agent and determined there were no significant changes to heart rate (Fig. S2). Doses of 1 mM SNP and 100 µM SNAP were selected from our dose-response curves, supported by doses previously published. At 4 dpf, pericyte-covered vessels dilate an average of 6% in response to SNP (P≤0.005, n=53 from 15 embryos; Fig. 4A-F; Table S9). Specifically, 33/53 vessels dilate (Fig. S3A; Movie 4). There was no dilation in the vehicle control group (P=0.211, n=32 from 6 embryos). We found no dilation of ≤6.5 µm vessels without pericytes (P= 0.528, n=34 from 23 embryos). When we genetically ablate pericytes with Mtz and expose embryos to SNP, vessels with diameters ≤6.5 µm lose the ability to dilate, and instead constrict in response to SNP (P≤0.003, n=33 from 10 embryos; Fig. 4A; 25/33 vessels constrict). Vehicle-treated embryos with ablated pericytes also show constriction (P=0.002, n=115 from 18 embryos). The loss of dilation when pericytes are ablated supports an active role of pericytes in vascular dilation at 4 dpf.
Pericyte-covered vessels dilate at 4 dpf, while vSMCs dilate at 6 dpf but only from a pre-constricted state. (A) Pericyte-covered vessels dilate in response to SNP (P=0.005, n=53 from 15 embryos). When pericytes are ablated, ≤6.5 μm diameter vessels constrict in response to both the vehicle (P=0.002, n=115 from 18 embryos) and SNP (P=0.003, n=33 from 10 embryos). (B-F) An example of pericyte-covered vessel dilation to SNP at 4 dpf. E and F are overlays of C and D. (G) Experimental timeline for the pre-constriction followed by dilation of vSMC-covered vessels. Embryos were imaged prior to drug addition (0 min), followed by 3 min of exposure to PE (10 μM) and then to 4 min of exposure to SNAP (100 μM). Embryos were imaged every minute. (H) At 4 dpf, vSMC-covered vessels do not dilate from a pre-constricted state (P≤0.0001, n=15 from 11 embryos) after SNAP exposure. (I) At 6 dpf, vSMC-covered vessels dilate from a pre-constricted state (P≤0.0002, n=15 from 13 embryos). The shading in H and I highlight drug exposures and the corresponding changes in vSMC-covered vessel diameter according to the experimental timeline (G). Scale bars: 10 μm in B-E. Data are mean±s.d. Significance was determined using a paired two-tailed t-test for A and a repeated measures one-way ANOVA for H-I (*P≤0.05, **P≤0.005, ***P≤0.0005). ns, not significant.
Pericyte-covered vessels dilate at 4 dpf, while vSMCs dilate at 6 dpf but only from a pre-constricted state. (A) Pericyte-covered vessels dilate in response to SNP (P=0.005, n=53 from 15 embryos). When pericytes are ablated, ≤6.5 μm diameter vessels constrict in response to both the vehicle (P=0.002, n=115 from 18 embryos) and SNP (P=0.003, n=33 from 10 embryos). (B-F) An example of pericyte-covered vessel dilation to SNP at 4 dpf. E and F are overlays of C and D. (G) Experimental timeline for the pre-constriction followed by dilation of vSMC-covered vessels. Embryos were imaged prior to drug addition (0 min), followed by 3 min of exposure to PE (10 μM) and then to 4 min of exposure to SNAP (100 μM). Embryos were imaged every minute. (H) At 4 dpf, vSMC-covered vessels do not dilate from a pre-constricted state (P≤0.0001, n=15 from 11 embryos) after SNAP exposure. (I) At 6 dpf, vSMC-covered vessels dilate from a pre-constricted state (P≤0.0002, n=15 from 13 embryos). The shading in H and I highlight drug exposures and the corresponding changes in vSMC-covered vessel diameter according to the experimental timeline (G). Scale bars: 10 μm in B-E. Data are mean±s.d. Significance was determined using a paired two-tailed t-test for A and a repeated measures one-way ANOVA for H-I (*P≤0.05, **P≤0.005, ***P≤0.0005). ns, not significant.
In the mouse, pericytes relax via prostaglandin EP4 receptors in combination with NO to cause pericyte relaxation (Hall et al., 2014). Astrocyte endfeet release prostaglandins (including PGE2, the EP4 ligand) to evoke vasomotor responses in pericytes (Macvicar and Newman, 2015). The spatial relationship between pericytes and astrocyte endfeet in early development is unknown, but close proximity is crucial for these cells to send and receive prostaglandin signals in the neurovascular unit. We observed direct contact between pericytes and astrocytes in the 4 and 6 dpf zebrafish brain (Fig. 5), suggesting that astrocyte-pericyte signaling starts developing at this stage.
Pericytes closely associate with astrocytes in the developing zebrafish brain. (A) Spatial relationship between pericytes and astrocytes (marked with GFAP:GFP) at 4 dpf. (B-E) Enlargements of A. (F) Spatial relationship between pericytes and astrocytes at 6 dpf. (G-J) Enlargements of F. Scale bars: 10 µm.
Pericytes closely associate with astrocytes in the developing zebrafish brain. (A) Spatial relationship between pericytes and astrocytes (marked with GFAP:GFP) at 4 dpf. (B-E) Enlargements of A. (F) Spatial relationship between pericytes and astrocytes at 6 dpf. (G-J) Enlargements of F. Scale bars: 10 µm.
We next assessed vSMC-covered vessel dilation in response to SNP at 4 dpf, but found no activity (P≤0.992, n=42 from 23 embryos; Fig. S8A,B; Table S9). However, there was a binary response from vSMC-covered vessels. Of 42 measurements, there were 16 that dilated, one unchanged and 25 that constricted (Fig. S3B). In the vehicle control, vSMC-covered vessels constrict (P=0.0001, n=14 from 8 embryos). This suggests that, at 4 dpf, the ability of vSMCs to relax has not fully developed.
We next assessed pericyte- and vSMC-covered vessel dilation at 6 dpf and found no dilation of either pericyte (P=0.438, n=79 from 7 embryos; Table S10) or vSMC-covered vessels (P=0.240, n= 25 from 16 embryos; Fig. S8C-D) in response to SNP at this stage. There was no change in 6 dpf pericyte-covered vessels in response to vehicle (P=0.263, n=77 from 19 embryos) while the vSMC-covered vessels constrict in response to vehicle (P=0.046, n=25 from 15 embryos).
We reasoned that we did not observe dilation of larger vSMC-covered vessels at 6 dpf because vessels are already close to maximal dilation. Thus, as is commonly carried out in the field, we pre-constricted vSMC-covered vessels using 10 µM PE and then exposed them to 100 µM of the nitric oxide donor SNAP (Fig. 4G; Table S11). At 4 dpf, vSMC-covered vessels constrict but are unable to dilate (P≤0.0001, n=15 from 11 embryos; Fig. 4H). Some vessels appear to dilate in response to SNAP within the first minute of treatment (4th minute); however, as time progresses, they constrict again and remain significantly constricted for the duration of the experiment. There is no significant dilation in response to the vehicle control (P=0.073, n=6 from 5 embryos). Thus, it appears that, at 4 dpf, the ability of vSMCs to relax and dilate vessels has not yet fully developed.
In contrast, at 6 dpf, vSMC-covered vessels can dilate from a pre-constricted state (P≤0.0002, n=15 from 13 embryos; Fig. 4I; Table S11). In response to PE, vSMC-covered vessels significantly constrict 1 min post-drug application (P=0.001). When SNAP was added, these vessels then actively dilate by 5% within the first minute of exposure (4th minute; P=0.035). The vehicle control showed no significant activity (P=0.155, n=5 from 5 embryos). Thus, by 6 dpf, vSMCs have developed active vasomotor ability to both contract and relax.
SNP was used to test vSMC-covered vessel dilation, from a pre-constricted state, with similar results. At 4 dpf, vSMC-covered pre-constricted vessels dilate in response to SNP (P=0.0009, n=22 from 12 embryos; Fig. S8E; Table S12). There is no dilation in response to the vehicle control at 4 dpf (P=0.363, n=17 from 10 embryos). However, at 6 dpf, SNP is ineffective at dilating pre-constricted vessels (P=0.003, n=11 from 9 embryos; Fig. S8F). However, a trend towards increased vessel diameter is seen after SNP exposure. There is no significant activity in the vehicle control group (P=0.380, n=10 from 8 embryos).
Overall, pericyte-covered vessels can dilate at 4 dpf but not at 6 dpf, whereas vSMC-covered vessels can dilate from a pre-constricted state at both 4 and 6 dpf. We find that SNAP has a stronger ability to cause pre-constricted vSMC-covered vessels to dilate than SNP. Our data also suggest that unmanipulated vSMC-covered vessels are at near maximal diameter in early development.
DISCUSSION
Development of vasoactivity in zebrafish
For the first time, we have harnessed the advantages of optical clarity and availability of transgenic lines for pericyte and vSMC lineages in zebrafish to explore the dynamic role of mural cells in regulating vascular diameter during embryonic development. Development of vasoactivity occurs in utero in mammals and cannot be live imaged. Our intriguing observations show that small vessels respond to agonists to regulate cerebral vessel diameter earlier in development than large vessels. At 4 dpf, pericyte-covered vessels actively constrict and dilate. In contrast, at this stage, vSMCs are still developing the capacity to contract and relax, and their response is slow and heterogeneous. At 6 dpf, only vSMC-covered vessels show vasoactivity in response to pharmacological stimuli (Fig. 6), as pericyte responses become heterogeneous.
Summary figure: developmental changes in mural cell ability to regulate cerebral vascular tone. (A) At 4 dpf, both pericyte and vSMC-covered vessels constrict, whereas only pericyte-covered vessels dilate. (B) At 6 dpf, both pericyte and vSMC populations have increased in number. Pericyte-covered vessels no longer respond to vasoactive agents, while vSMC-covered vessels constrict and dilate. (C) There is no vasomotor response when vSMCs are ablated.
Summary figure: developmental changes in mural cell ability to regulate cerebral vascular tone. (A) At 4 dpf, both pericyte and vSMC-covered vessels constrict, whereas only pericyte-covered vessels dilate. (B) At 6 dpf, both pericyte and vSMC populations have increased in number. Pericyte-covered vessels no longer respond to vasoactive agents, while vSMC-covered vessels constrict and dilate. (C) There is no vasomotor response when vSMCs are ablated.
We used pharmacological agents in our study as a first step to understand the intrinsic ability of pericytes and vSMCs to respond to cues. As of yet, there are no established assays to specifically stimulate local cerebral blood flow in zebrafish. Developing these methods in the future could enhance our understanding of the timing of neurovascular unit differentiation with respect to endogenous cerebral blood flow control.
Pericytes actively contract and relax to regulate cerebral vessel diameters in early development
We provide several lines of evidence to show that pericyte-covered vessels actively regulate their diameter. First, our results show that at 4 dpf, pericyte-covered vessels are responsive to vasoactive agents, including vasoconstrictors (NA and PE) and the vasodilator SNP. Second, genetic ablation of pericytes reduces vasoconstriction and blocks vasodilation. Third, the resting diameter of vessels with ablated pericytes is larger than wild-type pericyte-covered vessels, suggesting that pericytes contribute to the regulation of vessel diameter at this stage.
The roles of pericytes and vSMCs in regulating cerebral vessel diameter is conflicting in the literature for several reasons. All these studies have been undertaken in adult mouse models where identification of pericytes is less clear. Unlike in zebrafish, markers such as Pdgfrb can label both mural cell types in mouse (Armulik et al., 2011; Nakayama et al., 2013). Mural cells have also been identified based on morphology, which can be unreliable (Attwell et al., 2016). Furthermore, pericytes have a variety of morphologies (Hartmann et al., 2015); some pericytes express contractile proteins and some do not, suggesting heterogeneity. These factors make it difficult to clearly separate the contribution of mural cells to the regulation of vascular tone, potentially leading to contradictory conclusions regarding vasomotor activity (Fernandez-Klett et al., 2010; Hill et al., 2015).
We observe heterogeneity in mural cell response to vasoactive agents at our chosen developmental stages. At 4 dpf, the majority of pericyte-covered vessels respond to vasoactive agents in a predicted manner (i.e. constrict in response to vasoconstrictors, dilate to a vasodilator). However, the proportion of pericyte-covered vessels responding consistently to a stimulus decreases by 6 dpf such that a greater number of pericyte-covered vessels have heterogeneous responses than the proportion of pericyte-covered vessels responding in the predicted manner. Heterogeneous pericyte responses have also been observed in the mouse model. Only a small population of pericytes contract in response to the vasocontrictor U46610 in a cranial window model, with half of the capillaries constricting and the other half dilating (Fernandez-Klett et al., 2010). Additionally, whereas some pericyte-covered regions respond to vasoactive agents, other regions may dilate in response to accommodate for the blood flow within the vascular network. Only a subset of pericytes has been observed to express smooth muscle actin (αSMA) in the brain (Bandopadhyay et al., 2001), spinal cord (Roufail et al., 1995), retina (Nehls, 2004) and pancreatic islets (Almaça et al., 2018), suggesting that only a subset of pericytes has vasoactivity.
Several lines of data suggest that pericytes directly regulate vascular diameter rather than it being an indirect consequence of vSMC activity. First, the pericyte-covered vessels we studied here do not have any vSMC coverage; thus, any changes in their diameter can only be attributed to pericyte activity. We observe activity in these pericyte-covered vessels at only one developmental stage, 4 dpf, but not another, 6 dpf. If pericyte-covered vessel diameter changes were due to changes in upstream vSMC-covered vessels, we would expect stronger diameter changes in pericyte-covered vessels at 6 dpf when vSMC-covered vessels strongly constrict. However, at 6 dpf, only vSMC-covered vessels significantly constrict, not pericyte-covered vessels. Second, genetic ablation of pericytes at 4 dpf reverses the ability of small vessels to dilate in response to SNP. Genetic ablation also results in a larger resting diameter of small vessels at 4 dpf, but not at 6 dpf, suggesting active regulation of tone by pericytes at this stage. Third, at 4 dpf, pericyte-covered vessels constrict faster in response to NA than do vSMCs. Taken together, these observations argue that direct pericyte activity modulates cerebral vessel diameter in early development.
One limitation of our study is that ablation resulted in later and smaller magnitude changes relative to when pericytes or vSMCs are present, but did not completely eliminate activity. Incomplete ablation is likely due to our transgene mosaicism; two transgenic lines with the same pericyte or vSMC driver, but different reporters, show incompletely overlapping expression. Thus, the reduced, but not absent, activity could be a result of unlabeled pericytes that are not ablated. Another possibility is that as mural cell numbers are increasing between 4 and 6 dpf, there may be newly differentiating mural cells that have vasomotor activity, but do not express high enough levels of the driver to undergo ablation. Despite these limitations, our evidence strongly suggests that at 4 dpf, pericyte-covered vessels actively regulate their vessel diameter.
Intracellular calcium concentration dictates whether pericytes will contract or relax in mice (Sakagami et al., 1999, 2001; Sugiyama et al., 2004). Pericytes contract in response to endothelial-derived vasoactive agents (Matsugi et al., 1997) but do not express the same contractile machinery as vSMCs. Pericytes lack the calcium-binding protein calponin, which is responsible for regulating contraction in vSMCs (Bandopadhyay et al., 2001). However, pericytes contain microfilaments resembling actin and myosin containing fibers (Ho, 2004; Lebeux and Willemot, 1978). In cultured pericytes, activation of endothelin 1 (ET-1) receptors in pericytes increases calcium, resulting in the alignment of F-actin and intermediate filaments and contraction (Dehouck et al., 1997). We used noradrenaline and phenylephrine to assess the contractile properties of pericytes during development because catecholamines such as noradrenaline modulate tone in cultured pericytes (Markhotina et al., 2007) and in cerebral slices (Hall et al., 2014; Peppiatt et al., 2006).
Our understanding of pericyte relaxation is limited. Similar to vSMCs, pericytes relax by activation of potassium channels (Burnette and White, 2006; Cao et al., 2006; Jackson, 2005; Li and Puro, 2001; Quignard et al., 2003; Von Beckerath et al., 2000; Wiederholt et al., 1995). Activation of potassium channels by vasodilators such as NO reduces the activity of L-type voltage-operated calcium channels (VOCC) and calcium-activated chloride channels (ClCa) in pericytes (Sakagami et al., 2001). The mechanism by which NO causes pericyte relaxation appears to be through prostaglandin I2 activity: a vasodilator (Burnette and White, 2006; Dodge et al., 1991). We show physical contact between astrocytes and pericytes at 4 and 6 dpf of development. However, further work is needed to determine when and whether astrocyte-pericyte interactions lead to pericyte relaxation in the zebrafish during development.
Vascular tone during embryonic development
It is intriguing that at 4 and 6 dpf, vSMC-covered vessels can only dilate from a pre-constricted state, suggesting that vSMC-covered vessels in the zebrafish embryo are close to maximal dilation during normal development. We found evidence of regulation of vascular tone at 4 dpf by pericytes and at 6 dpf by vSMCs by examining resting vessel diameter in ablated mutants. Tone regulation develops at a similar time in the zebrafish trunk aorta. Hemodynamic load increases between 24 and 48 hpf, and the diameter of the DA and posterior cardinal vein (PCV) of the trunk increases (Sugden et al., 2017). However, by 72 hpf, endothelial cells change shape and align themselves to decrease vessel diameter (Sugden et al., 2017). This timing corresponds to the recruitment of mural cells to the trunk aorta (Stratman et al., 2017). Developmental timing, vessel diameter, flow and vessel pattern are very different in the cerebral vasculature, and we observe the development of tone slightly later.
Embryonic vascular mural cells in the neurovascular unit
Understanding when neurovascular coupling develops in an organism is crucial to understanding the role of cerebral vascular mural cells in regulating tone. Ulrich et al. determined that, up until 3 dpf, the development of neuronal structures is not affected by the presence or absence of vessels in the zebrafish hindbrain (Ulrich et al., 2011). However, the importance of blood vessels to the brain changes over the following days. We show that at 4 dpf, at a time when neurovascular coupling has not been experimentally detected, pericytes are present in large numbers on cerebral vessels, and pericyte-covered vessels regulate cerebral vessel diameter in response to a stimulus. At 6 dpf, neuronal activity in the optic tectum in response to visual stimulus is present in fish larvae, but this does not result in increased red blood cell (RBC) speed until 8 dpf, the first stage at which neurovascular coupling is observed in the optic tectum (Chhabria et al., 2018). We show physical contact between astrocytes and pericytes at 4 dpf, but given that neurovascular circuits are operational later in development, at 4 dpf it is likely that these contacts are still developing.
In the adult mouse model, there is no consensus on pericyte vasoactivity and corresponding changes in blood flow regulation. In vitro, in vivo and ex vivo approaches have shown that although there is robust control of vessel diameter by vSMCs, pericytes have more limited and controversial roles. In the adult mouse, the first observations of pericyte contraction was largely based on indirect evidence (Díaz-Flores et al., 2009; Hamilton et al., 2010; Puro, 2007). For example, pericytes contract during ischemia and remain contracted even after reperfusion of the occluded artery (Yemisci et al., 2009). Other studies found that, although pericytes are contractile and capable of modulating cerebral blood flow (CBF), they do not play a major role in the process of neurovascular coupling (Fernandez-Klett et al., 2010). However, in some of these early studies, pericytes were identified by their morphology without the use of markers. Using adult rat brain slices and the mouse brain in vivo, Hall et al. found that pericytes regulate cerebral vessel diameters and the consequent changes in RBC velocity (Hall et al., 2014). They found that pericyte-covered capillaries dilate prior to vSMC-covered arterioles. However, others have shown that smooth muscle actin is present only in mural cells with smooth muscle morphology and not in cells with pericyte morphology (Hill et al., 2015). In the adult mouse cerebral cortex, vSMCs provide baseline contractile tone, and depolarize to regulate vessel diameter and CBF (Hill et al., 2015). Importantly, hemodynamic variability within the microcirculation, as well as non-exclusive molecular markers, may account for some of these differences in findings (Gould et al., 2017).
Our findings in developing zebrafish embryos reveal key similarities to and differences from data generated in the adult mouse brain model. We show that both pericyte-covered and vSMC-covered vessels regulate their diameters in response to agents, but at different stages of development. Pericyte-covered vessels can regulate cerebral vessel diameter at a time when vSMCs are only starting to differentiate at 4 dpf. It is important to remember that, although the vessels covered by pericytes are small, capillaries have the highest surface area of the cerebral vasculature and are the primary site of flow resistance (Gould et al., 2017). Thus, constriction or dilation in one pericyte-covered vessel region might not significantly alter local blood flow, but the sum of constriction or dilation of many pericyte-covered vessels can have a substantial effect on blood flow. Furthermore, zebrafish erythrocytes have a diameter of ∼5 µm, about the same diameter as zebrafish brain capillaries, which forces these cells to deform while squeezing through constricted blood vessel regions (Noguchi and Gompper, 2005; Pawlik et al., 1981; Skalak and Branemark, 1969). Thus, a small change in diameter on the smallest vessels might dramatically decrease resistance and increase blood flow.
Later, at 6 dpf, once greater numbers of vSMCs have differentiated and neuronal activity is increasing, vSMCs become the primary regulators of vascular tone. Although there are gap junctions between pericytes, endothelial cells and vSMCs (Borysova et al., 2013; Cuevas et al., 1984), the mechanism by which neuronal activity causes changes in vessel diameter and blood flow in zebrafish development remains unexplored. We observe that pericyte-covered vessel responses are increasingly heterogeneous throughout development, with subsets of cells that are capable of responding to vasoactive signals, and subsets of pericytes that cannot respond. Does this reflect pericyte heterogeneity? Are there subpopulations of pericytes that are capable of vasoactivity in later development and adulthood, even though the majority of cells are not apparently active? Furthermore, although we show that both pericyte- and vSMCs-covered vessels can respond to exogenous vasoactive agents, the sources and developmental ontogeny of endogenous vasoactive signals remain unknown. Going forward, studies involving local activation of neural or mural cell activity are needed to unravel the interactions between mural cell subtypes and developmental neurovascular coupling.
MATERIALS AND METHODS
Zebrafish husbandry and fish strains
All experimental procedures were approved by the University of Calgary's Animal Care Committee (Protocol AC17-0189). Zebrafish embryos were maintained at 28.5°C and in E3 medium (Westerfield, 1995). Transgenic lines used include: TgBAC(pdgfrβ:GFP)ca41 (Whitesell et al., 2019), TgBAC(pdgfrβ:Gal4)ca42 (Whitesell et al., 2019), Tg(kdrl:mCherry)ci5 (Proulx et al., 2010), Tg(acta2:GFP)ca7 (Whitesell et al., 2014), Tg(acta2:Gal4FF)ca62 (Whitesell et al., 2019), Tg(flk:GFP)la116 (Choi et al., 2007), Tg(UAS:NTR-mCherry)c264 (Davison et al., 2007) and Tg(GFAP:GFP)mi2001 (Bernardos and Raymond, 2006).
Pharmacological agents and nitroreductase ablation
All chemicals were obtained from Sigma. Sodium nitroprusside dihydrate (SNP; 71778) was used at 1 mM. S-Nitroso-N-acetyl-DL-penicillamine (SNAP; N3398) was used at 100 µM. Diethylamine NONOate sodium salt hydrate (D184; NONOate) was used at 5 µM. (R)-(−)-Phenylephrine hydrochloride (P6126) was used at 10 µM. (−)-Norepinephrine (A7257) was used at 2 µM. All vasoactive agents were dissolved in E3 fish medium.
For ablation experiments, Tg(pdgfrβ:Gal4;UAS:NTRmCherry;flk:GFP) and Tg(acta2:Gal4;UAS:NTRmCherry;flk:GFP) zebrafish were used. Metronidazole (Mtz; M3761) was used at 50 µM for 3 dpf mural cell ablation, at 50 µM for 5 dpf vSMC ablation and at 5 mM for pericyte ablation. Embryos were treated for 24 h. Only morphologically normal embryos with no pericyte or vSMC cells associated with the cerebral vessels were selected for experiments. mCherry-positive debris (marking ablated mural cells) was observed around vessels.
Heart rate measurements
A dose-response curve was used to determine optimal drug concentrations. Heart rate was counted manually for a period of 1 min. For each concentration of a vasoactive agent, five or six zebrafish were exposed individually to either E3 fish medium or the vasoactive agent.
Microscopy and image analysis
Zebrafish were mounted on glass-bottomed Petri dishes (MatTek, P50G-0-30-F), using 0.8% low melt agarose (Invitrogen, 16520-050) dissolved in E3 fish medium. Imaging was carried out using a Zeiss LSM 700 or Zeiss LSM 880 in Airyscan super-resolution mode confocal microscopes. All images were obtained with the 488 nm and 555 nm lasers, with a pinhole of 1 airy unit (AU). Images were either 8, 12 or 16 bit with slice intervals of 0.21-1 μm. Objectives were 10× (NA 0.25), 20× (NA 0.8) and 40× (NA 1.1) magnification. Z-stack images were obtained for each time point and processed using Zen and ImageJ/Fiji to obtain maximum intensity projections and measurements.
Vessel diameters were measured at positions where there was an association of a mural cell with the endothelium. All visible pericyte or vSMCs in an image were measured. The vessel diameter was measured from the external diameter of the endothelium. Three measurements were obtained from each region and averaged to obtain a vessel diameter. All vessel diameter measurements were paired before and after treatment. All experiments were replicated three times.
Statistics
All data were plotted as normalized to baseline measurements unless otherwise indicated. GraphPad Prism7 was used to carry out all statistics using either a paired two-tailed t-test, an unpaired t-test or a repeated measures ANOVA with *P≤0.05, **P≤0.005, ***P≤0.0005. From the repeated measures one-way ANOVAs, P values from Dunnett's test are reported. Furthermore, Tukey's multiple comparison tests were carried out and P values are reported for each experimental minute. All data are represented as mean±s.d. The number of vessels measured and the number of embryos used for each experiment is indicated in the figure caption.
Acknowledgements
We thank Dr Thomas Richard Whitesell, Dr Suchit Ahuja, Charlene Watterston, Jasper Greysson-Wong, Dr Grant Gordon, Dr Bill Cole, and Dr Peng Huang and his laboratory for their insightful suggestions and helpful comments on this project. We thank Dr Pina Colarusso and Dr Pia Svendsen for assistance with microscopy.
Footnotes
Author contributions
Conceptualization: N.B.; Methodology: N.B.; Validation: N.B.; Formal analysis: N.B.; Resources: S.J.C.; Writing - original draft: N.B.; Writing - review & editing: N.B., S.J.C.; Supervision: S.J.C.; Project administration: S.J.C.; Funding acquisition: S.J.C.
Funding
This work was funded by National Science and Engineering Council grants (RGPIN 06360-2014 and RGPIN/07176-2019) and a Heart and Stroke Foundation of Canada grant to S.J.C. (G-16-00012741). N.B. is the recipient of a studentship from the Alberta Children's Hospital Research Institute.
References
Competing interests
The authors declare no competing or financial interests.