Introduction

Vascular smooth muscle cells (VSMCs) are contractile mural cells wrapping around endothelial cells (ECs) of large vessels, especially arteries (Ando et al., 2022; Donadon & Santoro, 2021; Stratman et al., 2020). The expression of contractile proteins, such as alpha-smooth muscle actin (encoded by acta2), distinguishes VSMC from pericytes, the other type of mural cells primarily associated with small vessels including capillaries (Bahrami & Childs, 2020; Donadon & Santoro, 2021; Stratman et al., 2020). VSMCs are important for arterial homeostasis (Basatemur, Jorgensen, Clarke, Bennett, & Mallat, 2019). In the brain, VSMC constriction and relaxation is essential for functional hyperemia in neurovascular coupling (Hill et al., 2015; Kaplan, Chow, & Gu, 2020). Phenotype switching or dedifferentiation of VSMCs, marked by lower expression of contractile proteins, is prominent in pathological progression of atherosclerosis (Basatemur et al., 2019). Recently, similar VSMC phenotype switching or dedifferentiation have also been described in cell, animal, and human studies of children and adult neurological and cerebrovascular conditions (Aguilar-Pineda et al., 2021; Chou et al., 2022; Milewicz et al., 2010; Oka et al., 2020). These studies suggested that understanding the development of VSMC differentiation on brain arteries might help restore normal VSMC contractility to alleviate various cerebrovascular diseases; such a regenerative strategy requires a deep understanding of how VSMCs acquire a contractile phenotype while differentiating from their progenitors.

VSMCs spatially differentiate from different progenitors from mesoderm and neural crest origin (Ando et al., 2019; Donadon & Santoro, 2021; Whitesell et al., 2019). Previous research described various mechanisms regulating the differentiation of VSMC positive for acta2 in the trunk vasculature. Arterial Notch signaling activated by blood flow is necessary for acta2+ VSMC appearance on the zebrafish dorsal aorta (X. Chen, Gays, Milia, & Santoro, 2017). Autonomous Notch activation is required for specification of pdgfrb+ mural cell progenitors from mesenchyme around arteries, and these progenitors later differentiate into acta2+ VSMCs (Ando et al., 2019). Chemokine signaling promotes VSMCs association with the zebrafish dorsal aorta in the trunk, whereas blood flow modulated transcription factor krüppel-like factor 2 (encoded by klf2a in zebrafish) prevents their association with the adjacent cardinal vein (Stratman et al., 2020). Differentiation into acta2+ VSMCs from pdgfrb+ progenitors in the brain requires autonomous expression of an ATP-sensitive potassium channel (Ando et al., 2022). In contrast, acta2+ VSMCs in the ventral head of zebrafish are not derived from pdgfrb+ progenitors, and their differentiation is regulated by endothelial BMP signaling (Watterston, Zeng, Onabadejo, & Childs, 2019; Whitesell et al., 2019). These studies suggested that VSMC differentiation is highly organotypic and may even be vessel specific.

The circle of Willis (CW) consists of major arteries that supply blood to the vertebrate brain, including the internal carotid arteries and posterior communicating arteries (Campbell et al., 2019; Schröder, Moser, & Huggenberger, 2020). CW arteries are wrapped by VSMCs, and VSMC dedifferentiation and hyperplasia are described in carotid atherosclerosis and the pediatric Moyamoya disease (Chou et al., 2022; Fox, Dorschel, Lawton, & Wanebo, 2021). The internal carotid arteries in the CW are among the earliest circulated arteries in the brain due to their connection to the common carotid arteries (lateral dorsal aorta in zebrafish) (Campbell et al., 2019; Isogai, Horiguchi, & Weinstein, 2001; Schröder et al., 2020). VSMC differentiation on CW arteries, particularly internal carotid arteries, and whether this differentiation is associated with early arterial blood flow in the brain, have not been investigated. The mouse CW resembles that of humans, but its VSMC differentiation is difficult to observe in vivo, as the embryos develop in utero and depend on maternal circulation (Isogai et al., 2001; Schröder et al., 2020). The zebrafish CW also resembles that of humans, and the embryos external growth and optical clarity allows for confocal live imaging of developing blood vessels (Isogai et al., 2001). In addition, a few fluorescent transgenic (Tg) lines have been generated to label VSMCs and their progenitors (Ando et al., 2016; Whitesell et al., 2014). Taking advantage of the zebrafish model, we found a spatiotemporal pattern of VSMC differentiation on CW arteries, and we associated this pattern with hemodynamic signaling pathways in arterial ECs.

Results

Artery muscularization is spatiotemporally regulated in the CW arteries

Arterial ECs in zebrafish brain emerge from primitive veins (Fujita et al., 2011). Recent single-cell transcriptome profiling of vascular cells in prenatal human brain also supported the fate progression from venous to arterial ECs (Crouch et al., 2022). To establish the stage of CW artery specification in the zebrafish brain, we performed confocal live imaging of Tg(flt4:yfp)^hu4881, which labels venous ECs, and Tg(kdrl:hras- mcherry)^s896, which labels all ECs, at four stages starting at 32 hour post-fertilization (hpf), when CW arteries are assembling (Chi et al., 2008; Hogan et al., 2009; Isogai et al., 2001). As arterial ECs express higher kdrl, we used the intensity of mcherry as an indicator of artery specification (Figure 1A-1E) (Chi et al., 2008). We found that ECs in CW arteries are primarily venous at 32 hpf (Figure 1A and 1E), and gradually gain kdrl expression from 54 hpf to 3-day post fertilization (dpf) (Figure 1B, 1C, and 1E). The mcherry intensity at 4 dpf is similar to 3 dpf, suggesting that artery specification of the CW is completed by 3 dpf (Figure 1C-1E).

Figure 1.

Then we characterized CW artery muscularization. Previous research on the CW arteries described that VSMCs derive from pdgfrb+ mural cell progenitors (Ando et al., 2022; Ando et al., 2019). Thus, to determine VSMC differentiation in real time we performed confocal live imaging of Tg(acta2:mcherry)^ca8, TgBAC(pdgfrb:egfp)^ncv22, and Tg(kdrl:cerulean)^sd24, to visualize respectively VSMCs (red), pdgfrb+ progenitors (green) and the developing arteries (white) at four stages starting from 32 hpf (Ando et al., 2016; Page et al., 2013; Whitesell et al., 2014). We found that overall pdgfrb+ progenitors appear on CW arteries after their lumen form around 54 hpf, and most of them do not express acta2 (Figure 2A-2B, 2E-2F, and S1A-S1B). The number of pdgfrb+ progenitors increase temporally from 54 hpf to 3 dpf and spatially on CW arteries from anterior to posterior - from the internal carotid arteries (CaDI), the basal communicating artery (BCA) to the posterior communicating segments (PCS) (Figure 2B-2C). Notably, while many pdgfrb+ progenitors on CaDI differentiate into acta2+ VSMCs by 3 dpf nearly no acta2+ VSMCs were observed in the nearby artery BCA or PCS (Figure 2B-2C, 2E-2F, and S1A-S1B). By 4 dpf, the number of pdgfrb+ cells is similar to 3 dpf, but acta2+ VSMCs on CaDI continue to differentiate in a greater number over the BCA, and PCS arteries (Figure 2C-2F and S1A-S1B). These results show an anterior to posterior differentiation of pdgfrb+ progenitors into acta2+ VSMCs with the CaDI being the first CW artery muscularized and having a higher number of acta2+ VSMCs.

Figure 2.

CW arteries have spatiotemporal difference in hemodynamics

Our data suggest that VSMCs in the anterior CaDI differentiate significantly earlier than the anterior BCA and the posterior PCS, even if their progenitors have the same origin and these CW blood vessel formation and arterialization occurs at the same time. We speculate that different hemodynamic distribution may contribute to the spatiotemporal difference in VSMC differentiation on CW arteries. In fact, CaDI is the only vessels in the CW receiving pulsatile arterial blood flow directly from the lateral dorsal aorta and cardiac outflow tract. Furthermore, the CaDI differs from the BCA or PCS in terms of overall vascular geometry, which has been shown to affect local intravascular forces such as the wall shear stress (WSS) (Katritsis et al., 2007). To explore this possibility, we performed microangiography with a high molecular weight fluorescent dextran and used computational fluid dynamics (CFD) to simulate the effect of pulsatile flow on the CW intravascular forces (ANSYS, 2014; Z. Chen et al., 2019; Fernandes et al., 2022). First, we determined each CW artery diameter, the principal determinant of hemodynamics, and found that there was no significant difference between structures (Figure 3A-3D). Notably, however, the flow velocity across CW arteries was differentially distributed, with the CaDI having the highest blood flow speed starting from 54 hpf to 4 dpf (Figure 3A-3C, and 3E). This is consistent with the CaDI being the inlet vessel for the arterial blood flow within the entire CW circulation. Next, we calculated the WSS across all CW arteries and found an overall increase in average and total WSS from 54 hpf to 4 dpf (Figure 3A-3C, and 3F). Importantly, however, when analyzed individual vessels, we found that the WSS in the CaDI is significantly higher from 54 hpf to 4 dpf, while this increase was not noted in the BCA or PCS (Figure 3A-C, and 3G). Further, WSS in the CaDI is significantly higher than in the BCA and PCS at 3 dpf (Figure 3G), when considerable VSMC differentiation starts on the CaDI but not the BCA or PCS (Figure 2B-2C, 2F, and S1A-S1B). Altogether, the pulsatile blood flow simulation suggests that CW arteries experience different hemodynamic loads that might confer an heterogenous muscularization of CW arteries.

Figure 3.

Blood flow is required for differential CW artery muscularization

Our data suggest that ECs under pulsatile arterial flow might favor the differentiation of acta2+ VSMCs from pdgfrb+ progenitors. To test this hypothesis, we set up an in vitro cell co-culture experiment where GFP-PDGFRB+ human brain pericytes (GFP-HBVPs) were cultured in flow amenable outlet slides, then covered by a thin layer of collagen type 1. Confluent human endothelial cells (HUVECs) were seeded on top of that and then exposed to steady-state laminar flow or pulsatile flow conditions (Figure 4A) (Abello, Raghavan, Yien, & Stratman, 2022). Maturation of PDGFRB+ human brain pericytes into ACTA2+ VSMCs was then quantified after 24 hours from the introduction of flow addition. We found that brain pericytes under pulsatile flow were larger and showed higher expression of VSMC differentiation markers including PDGFRB, ACTA2, and TRANSGELIN (Robin et al., 2013) (Figure 4B-4D), suggesting that pulsatile blood flow favored PDGFRB+ progenitor differentiation.

Figure 4.

Next, we tested the effect of pulsatile flow in the CW on VSMC differentiation in vivo. Previous research suggested that blood flow hemodynamics is not necessary for pdgfrb+ progenitor recruitment at CW arteries, as shown by zebrafish embryos lacking blood flow such as the tnnt2a morphants (MO) (Ando et al., 2016), however their differentiation to acta2+ cells has not been documented. Thus, to determine whether blood flow affects acta2+ VSMC differentiation on CW arteries, we injected 0.35 ng of tnnt2a MO into Tg(acta2:mcherry, kdrl:gfp)^ca8/zn1 at the one to two-cell stage, to abrogate cardiac contractility and thus pulsatile flow (Sehnert et al., 2002; Whitesell et al., 2014; Zhong et al., 2006). Notably, we found no acta2+ VSMCs on CW arteries of tnnt2a MO at 3 to 4 dpf (Figure 4F, 4H-4I, and 4K), suggesting that blood flow is dispensable for pdgfrb+ progenitor recruitment but is required for acta2+ VSMC differentiation on the CW arteries.

To further explore the temporal requirement of blood flow, we treated embryos with 25 μM nifedipine, a drug shown to reduce heart rate in zebrafish embryos (Gierten et al., 2020), from 54 hpf, right before the onset of pdgfrb+ progenitors’ expression of acta2 (Figure. 2A and B). We found that after 18 hours at 3 dpf the number of acta2+ VSMCs is greatly reduced on the CaDI of treated embryos compared to untreated controls (Figure 4G and 4M); acta2+ VSMCs on the BCA and the PCS at 4 dpf are also reduced significantly (Figure 4L and 4N), suggesting that blood flow reduction after 54 hpf delays VSMC differentiation on CW arteries with the earliest and strongest effect on the CaDI muscularization (Figure 4G and 4M).

To determine whether blood flow is differentially required for the maintenance of VSMCs on CW arteries, we treated embryos with nifedipine from 4 dpf, after VSMC differentiation on CW arteries, and imaged at 5 dpf. The number of acta2+ VSMCs on the CaDI and PCS in treated embryos is similar to untreated controls (Figure S2A-S2C), suggesting that blood flow is required for differentiation but not for short-term maintenance of VSMCs.

Blood flow-regulated transcription factor klf2a is required for spatiotemporal CW artery muscularization

Our results suggest that ECs might express a flow pattern-dependent transcriptional program that favors spatiotemporal VSMC muscularization in the CW arteries. For example, previous studies showed that the site and level of the transcription factor Klf2 (Parmar et al., 2006; Sweet, Fan, Hsieh, & Jain, 2018; Warboys, Amini, de Luca, & Evans, 2011) in artery ECs closely follow the predicted pattern of elevated intravascular forces (Lee et al., 2006). Consistently, higher Klf2 expression can be induced by unidirectional pulsatile flow mimicking arterial flow (Dekker et al., 2002). Furthermore, klf2 is implicated in VSMC migration on arterial blood vessels during mouse development: loss of klf2 leads to aorta VSMC deficiency (Wu, Bohanan, Neumann, & Lingrel, 2008) while endothelial klf2a expression prevents VSMC association with primitive veins in zebrafish trunk (Stratman et al., 2020). Hence, we tested the hypothesis that klf2a might be the signaling spatially regulated in the CW artery to control VSMC maturation. We first imaged Tg(klf2a:h2b-egfp, kdrl:hras- mcherry)^ig11/s896, which labels nuclei with active klf2a expression, and quantified the number of klf2a+ ECs in the CW arteries (Chi et al., 2008; Steed et al., 2016). To account for the overall changes in EC number during development (Ulrich, Ma, Baker, & Torres-Vazquez, 2011), as well as the length of each vessels, we also imaged Tg(fli1:nls-gfp, kdrl:hras-mcherry)^y7/s896, which labels all EC nuclei, at four stages from 32 hpf (Roman et al., 2002). Notably, we found a slight increase in ECs number per 100 μm vessel in the CaDI from 32 to 54 hpf (Figure S3A-S3C), but a much more significant increase in klf2a+ EC nuclei in the CaDI (Figure 5A and 5B), before acta2+ VSMCs appear (Figure 2B-2C, 2F, and Figure S1A-S1C). The number of ECs in the CaDI stayed the same from 54 hpf through 4 dpf (Figure S3A-S3C), while klf2a+ ECs in the CaDI increased until 3 dpf (Figure 5A and 5B). In tandem, acta2+ VSMCs on the CaDI increase through 4 dpf (Figure 2C-2D, 2F, and Figure S1A-S1C). EC number per 100 μm vessel in the BCA and PCS are similar from 54 hpf to 4 dpf (Figure S3A-S3C), whereas the number of klf2a+ ECs increase significantly over time (Figure 5A and 5B). This increase predates the presence of acta2+ VSMCs on the BCA and PCS (Figure 2B-2D, 2F, and Figure S1A-S1C). These results suggest that the increase of klf2a+ ECs in the CW arteries is not due to an increase in EC number and precedes the spatiotemporal VSMC differentiation observed on the CW arteries.

Figure 5.

To further define the role of klf2a in CW artery muscularization, we knocked down klf2a in Tg(acta2:mcherry, kdrl:gfp)^ca8/zn1 with 11 ng MO injection at the one to two-cell stage and imaged at 3 to 4 dpf (Nicoli et al., 2010; Whitesell et al., 2014; Zhong et al., 2006). We validated the effect of klf2a MO knockdown using the Tg(klf2a:h2b- egfp, kdrl:hras-mcherry)^ig11/s896 transgenic line (Figure S3D-S3F) (Chi et al., 2008; Steed et al., 2016). Compared to uninjected control, klf2a morphants have significantly less acta2+ VSMCs on the CaDI at 3 dpf but a normal number by 4 dpf (Figure 6A-6D), suggesting that klf2a promotes timely initiation of CaDI muscularization. Together these data suggest that CW muscularization is associated with endothelial klf2a activation and necessary for spatiotemporal VSMC differentiation under pulsatile blood flow conditions.

Figure 6.

Discussion

Here, we used confocal live imaging of fluorescence transgenic zebrafish embryos to characterize the spatiotemporal dynamics of VSMC differentiation on the CW, which consists of major arteries that supply blood to the vertebrate brain. We found that CW morphogenesis preceded arterial specification. pdgfrb+ mural cell progenitors start to express the VSMC marker acta2 after these mural cell progenitors were recruited to CW arteries. VSMCs differentiated earlier on anterior CW arteries, which are under higher WSS than their posterior counterparts, due to the high velocity of incoming pulsatile blood flow. We used an in vitro co-culture assay, genetic manipulation, and drug treatments to provide evidence that pulsatile blood flow can contribute to spatiotemporal VSMC differentiation. We found that the flow responsive transcription factor klf2a is activated from anterior to posterior in the CW arteries, preceding VSMC differentiation. klf2a knockdown delayed VSMC differentiation on anterior CW arteries. Together, these data support the conclusion that pulsatile flow activation of endothelial klf2a promotes spatiotemporal VSMC differentiation on CW arteries in the brain (Figure 7).

Figure 7.

Our work suggests klf2a-mediated blood flow regulation of VSMC differentiation on zebrafish brain arteries, and thus raises the question of how Klf2 transduces endothelial signals to mural cell progenitors and VSMCs. In vitro, endothelial Klf2 upregulates miR- 143/145, which are transported into co-cultured VSMCs within extracellular vesicles (EV) to promote a contractile phenotype in VSMCs (Hergenreider et al., 2012). How EV transport of miR-143/145 works in vivo during development remains unknown.

Notch signaling appears a plausible downstream effector of Klf2 activation. Like Klf2, Notch also responds to flow in heart valve development (Fontana et al., 2020), and Klf2 appears to upregulate Notch (Duchemin, Vignes, & Vermot, 2019). Non-canonical Notch signaling is induced by shear stress to mediate endothelial barrier formation (Polacheck et al., 2017). Notch1 is a proposed mechanosensor in adult arteries (J. J. Mack et al., 2017). In addition, Notch regulation of VSMC development is well established. Additionally, flow activation of Notch in the dorsal aorta (DA) is required for VSMC recruitment (X. Chen et al., 2017). Notch signaling is also activated in DA VSMC progenitors during development, when DA ECs express Jag1 (jagged canonical Notch ligand 1) (Chang et al., 2012; High et al., 2008). Notch2 and Notch3 are compensatory for VSMC development on the DA (Q. Wang, Zhao, Kennard, & Lilly, 2012). In VSMC differentiation on coronary arteries, ECs express Jag1 and pericyte progenitors express Notch3 (Volz et al., 2015). Thus, reciprocal Notch signaling between ECs and VSMCs may regulate VSMC differentiation on brain arteries, although how Notch signaling activates Myocd (myocardin) (Huang et al., 2008; Li, Wang, Wang, Richardson, & Olson, 2003), Srf (serum response factor) (C. P. Mack & Owens, 1999), and miR- 143/145 (Boettger et al., 2009) to enable expression of Acta2 and other contractile proteins remain incompletely understood.

It is important to recognize the versatile roles of Notch signaling in vascular development for process both prior to and in tandem with VSMC differentiation. Notch signaling represses proliferation of arterial ECs during angiogenesis (Hasan et al., 2017; Pitulescu et al., 2017), and maintains arterial identity after morphogenesis (Lawson et al., 2001; Shutter et al., 2000; Villa et al., 2001). Notch signaling is required for recruitment of mural cells to brain arteries, in which Notch2 and Notch3 are redundant for upregulation of Pdgfrb in mesenchymal progenitors (Ando et al., 2019). Notch3, in which mutations cause cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), is broadly required for mural cell development in the brain (Domenga et al., 2004; Joutel et al., 1996; Y. Wang, Pan, Moens, & Appel, 2014). Thus, thorough dissection of how Notch signaling regulates VSMC differentiation on brain arteries would require development of novel methods that enable spatiotemporal resolution of Notch ligands and receptors in different vascular cell types.

Wnt signaling is another possible downstream effector of Klf2. In heart valve development, endocardial Klf2 upregulates Wnt9 (wingless-type MMTV integration type family member 9), which activates Wnt signaling in neighboring mesenchymal cells and thereby regulates their proliferation and condensation (Goddard et al., 2017). The pathway interacts with Notch signaling in heart valve morphogenesis (Paolini, Fontana, Pham, Rodel, & Abdelilah-Seyfried, 2021). Non-canonical Wnt signaling attenuates EC sensitivity to shear stress and stabilizes blood vessels (Franco et al., 2016). In brain vascular development, Wnt signaling controls brain specific developmental angiogenesis and endothelial barrier formation (Daneman et al., 2009; Liebner et al., 2008; Zhou et al., 2014). Wnt signaling in mural cells activates expression of Lama2 (laminin subunit alpha 2), a major component of the brain vascular basement membrane, to promote the neurovascular unit and blood-brain barrier maturation (Biswas et al., 2022). Wnt signaling regulates VSMC proliferation, migration, and survival in cardiovascular diseases (Mill & George, 2012), but relatively little is known about the role of Wnt signaling in VSMC differentiation, especially organotypic VSMC differentiation on brain arteries.

The biochemical pathway by which Klf2 activates Notch or Wnt signaling remains incompletely understood. Recent chromatin occupancy and transcription studies showed that Klf2 has context specific binding patterns and transcriptional targets in the heart and lung endothelium (Sweet et al., 2023). Thus, Klf2 may have multiple direct or indirect paths to activate Notch or Wnt signaling. The context specificity may also explain different effects of klf2a on VSMC development in the zebrafish brain and trunk: previous research suggested that klf2a expression in the cardinal vein prevents VSMC association (Stratman et al., 2020), whereas this study suggests that klf2a activation in brain arteries promotes VSMC differentiation. It is plausible that klf2a has different expression levels, binding patterns, transcriptional targets, and downstream effects in brain arterial ECs compared with trunk venous ECs.

Our findings raise interesting new questions on whether stable klf2a expression in CW arterial ECs supports further maturation of VSMCs. Previous research found that acta2+ VSMCs on the BCA and PCS express pericyte enriched abcc9 (ATP binding cassette subfamily C member 9) at 4 dpf (Ando et al., 2022), which is gradually lost from 5 dpf to 6 dpf (Ando et al., 2022) as these cells become acta2+. Interestingly, acta2+ VSMCs on CaDI, BCA, and PCS still retain expression of pdgfrb at 6 dpf (Ando et al., 2021). Thus, it is possible that stable expression of klf2a in CW arterial ECs will support later on complete VSMC maturation, for example in Transgelin positive cells (Colijn, Nambara, & Stratman, 2023). Alternatively, acta2+ VSMCs with pdgfrb expression are maintained as such in the CW arteries (Ando et al., 2021).

It is important to note the differences between the trunk and our data focused on the CW brain arterial VSMC differentiation. On the dorsal aorta of the trunk, as soon as VSMCs are recruited, they begin expressing Acta2 and Tagln, suggesting simultaneous recruitment and differentiation from sclerotome progenitors (Ando et al., 2016; Stratman et al., 2017), compared with the spatiotemporal dynamics we see on brain arteries. It is also important to note the difference between arterial and venous VSMCs. Brain arterial VSMCs strongly express Acta2, whose expression is weak in venous VSMCs (Crouch, Joseph, Marsan, & Huang, 2023; Hill et al., 2015). Venous VSMCs share more common gene expression signature with pericytes, as they both express Abcc9 and Kcnj8 (Ando et al., 2019; Bondjers et al., 2006; Chasseigneaux et al., 2018; He et al., 2016; Vanlandewijck et al., 2018). Based on our data, it is possible to speculate that pulsatile flow reinforces brain arterial specification before VSMC differentiation. Pulsatile flow contributes to both arterial specification and VSMC differentiation, although the two processes are difficult to dissect in vivo, as both require Notch signaling and overlap temporally. In addition, pulsations dampen from dorsal aorta to internal carotid arteries, and flow eventually becomes stable in capillaries and veins (Humphrey & Schwartz, 2021). Current in vivo methods involving genetic manipulations and drug treatments are only capable of qualitative reduction in all types of flow, including pulsatile. In vitro three-dimensional (3D) vascular culture models, which combine ECs and mural cells (Mirabella et al., 2017; Vila Cuenca et al., 2021), could be further optimized to simulate complex geometry of brain arteries. Combining these models with microfluidics, which allows precise calibration of flow velocity and pulse, would enable a more thorough analysis of endothelial mechanotransduction and its contribution to VSMC differentiation on different arterial beds (Abello, Raghavan, Yien, & Stratman, 2022; Gray & Stroka, 2017; Griffith et al., 2020).

In conclusion, our work highlights organotypic differences in VSMC differentiation in the CW brain arteries and points to hemodynamics as a main driver of spatiotemporal dynamics of VSMC maturation on the brain arteries. Impaired VSMC differentiation in development could be implicated in cerebrovascular diseases, such as Moyamoya disease, which is linked to mutations in the human ACTA2 gene that results in VSMC hyperplasia specifically on the internal carotid arteries (Fox et al., 2021; Guo et al., 2007; Guo et al., 2009; Lin et al., 2012). Our study points to endothelial cell mechanotransduction of pulsatile flow as a key signaling network that could render specific VSMC populations on brain arteries more susceptible compared to VSMCs associated with other vessels in the same tissue. Our data will therefore inform the screening of new genes that in combination to known genetic variants, such as ACTA2, could contribute to our understanding of the susceptibilities in cerebrovascular diseases.

Material and Methods

Zebrafish husbandry and transgenic lines Zebrafish were raised and maintained at 28.5 °C using standard methods. Protocols are approved by the Yale Institutional Animal Care and Use Committee (2020-11473). Transgenic lines in Table 1 were established previously.

Table 1.

Confocal fluorescence microscopy

Zebrafish embryos were raised in 0.003% 1-phenyl-2-thiourea (PTU, phenylthiocarbamide, or n-phenylthiourea, Sigma P7629) from gastrulation stage to prevent pigmentation. Embryos imaged live by confocal fluorescence microscopy were anesthetized in 0.1% tricaine methanesulfonate (TMS, MS-222, or Syncaine, Western Chemical, NC0872873) and mounted in 1% low melt agarose within glass bottom microwell dishes. Fluorescence images were captured with an upright Zeiss LSM 980 confocal microscope using a 20X objective.

Image analysis

Confocal fluorescence images were analyzed with Imaris microscopy image analysis software (Bitplane, Oxford Instruments). Average fluorescence intensity of mcherry in Tg(flt4:yfp, kdrl:hras-mcherry)^hu4881/s896 was estimated by generating volume objects covering the artery of interest with Surfaces module. Vessel lengths of the arteries were manually traced with Filament module. Numbers of pdgfrb+ mural cell progenitors, acta2+ vascular smooth muscle cells, and klf2a+ and fli1+ endothelial nucleus were counted with Spots module.

Morpholino injections

Morpholino antisense oligonucleotides were synthesized by Gene Tools. Morpholinos in Table 2 were validated previously. Optimized dose injected into each embryo are listed. Uninjected siblings were used as controls.

Table 2.

Nifedipine treatment

Nifedipine was dissolved into 20 mM in dimethyl sulfoxide (DMSO). The stock solution was diluted into 25 μM in egg water with 0.003% PTU to treat zebrafish embryos from 54 hpf to 3 dpf or 4 dpf. The stock solution was diluted into 20 μM in egg water with 0.003% PTU to treat zebrafish embryos from 4 dpf to 5 dpf. The same volume of DMSO was added into egg water with 0.003% PTU for sibling control embryos.

Microangiography

Embryos were anesthetized in 0.1% tricaine methanesulfonate, placed on an agarose mold, and injected pericardially with 4 nL Tetramethylrhodamine Dextran (2,000,000 molecular weight, Thermo Fisher) at a concentration of 10 mg/mL. Subsequently, the vasculature of embryos was checked for Dextran fluorescence signal under a stereomicroscope (Olympus, MVX10). Embryos with Dextran fluorescence were mounted in 1% low melt agarose within glass bottom microwell dishes and imaged with confocal microscope.

Computational fluid dynamic simulation

Computational fluid dynamic simulation was performed as previously described with modification (ANSYS, 2014; Barak et al., 2021). Three-dimensional (3D) geometry models of the circle of Willis (CW) arteries were reconstructed from confocal microangiography images of Dextran injected Tg(kdrl:gfp) embryos with the Filament module in Imaris (Bitplane, Oxford instruments). The CW 3D geometry models were pre-processed in ANSYS SpaceClaim 2022 R1 software. Pre-processed models were then meshed in ANSYS Fluent 2022 R1 software using computational fluid dynamics application settings. Inlets and outlets were specified. Meshing orthogonal quality was calculated, and objects less than 0.01 were excluded. Blood density and gauge pressure were considered as 1,060 kg/m³ and 13,332 pascals, respectively. Each analysis consisted of 200 iterations.

Cell Culture

Pooled primary Human Umbilical Vein Endothelial Cells (HUVEC) (PCS-100-010^TM, ATCC) were seeded at an initial concertation of 5,000 cells per cm², in 1x M119 media (110343-023, Gibco) supplemented with 16% FBS (Gibco), 84 µg/mL of heparin sodium salt (H3393, Sigma-Aldrich), 25 µg/mL of endothelial cell growth supplement (02-102, EMD Millipore Corp.) and 1x Antibiotic-antimycotic solution (15240-062, Gibco). Cell cultures were maintained at 37°C, 5% CO₂ and 95% humidity, until the cell reached 80% confluence.

To facilitate cell visualization in cell co-cultures, Human Brain Vascular Pericytes (HBVP) (1200, ScienCell) were transfected using lentiviral particles (LVP310, GenTarget, Inc.) to induce GFP expression under EF1a promotor. Cells cultures were initiated by seeding 5,000 cells per cm² in 175 cm² plastic flasks pre-coated with gelatin and 1x DMEM (11995-065, Gibco) supplemented with 10% FBS (Gibco), and 1x Antibiotic-antimycotic solution under the same cell culture conditions described above. When cells cultures were at 80% confluence, 200 µL containing 2 x 10⁶ GFP-lentiviral particles were added to each 175 flasks. After 72 hours, fresh cell culture media supplemented with 10µg/mL of Blasticidin (15205, Sigma-Aldrich) were added to each flask to select the positive transfected cells.

Flow assays and immunostaining

HBVP cells were harvested and seeded at a concentration of 1.3 x 10⁵ cells in 0.4 optical plastic flow microslides (80176, Ibidi) precoated with 1 mg/mL gelatin and incubated for 24 hours under standard culture conditions. After the initial incubation, 100 µg/mL of collagen I (354249, Corning) diluted in DMEM cell culture media was added to the slides to create a thin layer on top of the HBVP cells. After 2 hours, the media was removed and 2.5 x 10⁵ HUVECs were seeded on top of the collagen I layer and incubated for additional 24 hours. After cell co-cultures were established, the slides were exposed to laminar (15 dyn/cm²) or pulsatile (12-15 dyn/cm²) flow for 24 hours, implementing a peristaltic pump adapted to produce different types of flow (Abello, Raghavan, Yien, & Stratman, 2022). After 24 hours, cultures were rinsed with 1x PBS and fixed for 30 minutes in 4% paraformaldehyde at room temperature. Cell cultures were immunostained with α-Smooth muscle actin D4K9N XP^® rabbit monoclonal antibody (19245, Cell Signaling), followed by Alexa Fluor 633 goat anti-rabbit IgG (A21071, Invitrogen). Confocal images were obtained using a 40x objective with a W1 Spinning Disk confocal microscope, a Fusion camera, and the Nikon Eclipse Ti2-E base. Fiji image processing software was used for image analysis and fluorescence intensity quantification.

Statistics

All statistical analyses were performed with GraphPad Prism (version 10.0.3). Mann–Whitney test was used to compare two groups to test mean differences (protein level, morpholino and nifedipine treatment). Two-way analyses of variance followed by Tukey’s multiple comparisons was used to compare more than two groups to test mean differences (average fluorescence intensity, cell number, vascular diameter, flow velocity, and WSS).

Acknowledgements

We thank the labs of S. Schulte-Merker, D.Y.R. Stainier, S.J. Childs, N. Mochizuki, D. Traver, J. Vermot, and B.M Weinstein for sharing zebrafish transgenic lines. We thank N. Semanchik for all the assistance with zebrafish adult colonies and husbandry. Experiments in the manuscript were supported by R01NS109160 and R01DK118728 awarded to S.N. and AHA23POST1025829 post-doctoral fellowship awarded to I.F.X. The paper is based on a dissertation submitted by S.C. to fulfill in part the requirements for the degree of Doctor of Philosophy, Yale University. S.C. was supported by a Gruber Science Fellowship from Yale Graduate School of Arts and Sciences.

Author Contributions

S.C. and I.F.X contributed to the experiments, the data analysis, and composed the figures. R.W. performed dextran microangiography experiments. J.A. and A.N.S. performed the in vitro experiments. S.C., I.F.X and S.N. wrote the manuscript and prepared figures. J.A. and A.N.S. edited the paper.

Supplementary Figures and Supplementary Figure Legends

Figure S1.

Figure S2.

Figure S3.