Endothelial cell dynamics in vascular remodelling

1 Introduction to angiogenesis and vascular remodelling

The vertebrate body contains a tree-like tubular network of blood vessels lined by a monolayer of endothelial cells (ECs) devised to allow the transport and distribution of nutrients and oxygen to all tissue [12]. In the early stages of embryo development, new vessels form de novo via the assembly of angioblasts, a set of mesoderm-derived precursors which differentiate into ECs that aggregate in order to form a primitive vascular plexus, through a process called vasculogenesis [11]. However, the expansion of this primitive vessel network in the embryo, but also equally in regenerative processes in adult organisms, is mostly mediated by angiogenesis, a complex multi-step program that allow blood vessels to invade avascular tissues, expand the pre-existing network, and remodel into a hierarchical structure. Angiogenesis is initiated upon activation of pre-existent endothelial cells by numerous growth factors and cytokines, amongst which vascular endothelial growth factor (VEGF)-A plays the most prominent role [27]. The most common form of angiogenesis is sprouting angiogenesis, however intussusceptive angiogenesis also contributes to vascular growth. In sprouting angiogenesis, following activation, endothelial cells in the angiogenic front acquire specific transcriptional and phenotypic states, so called endothelial tip and endothelial stalk cells. The proportion of endothelial cells in these two states is essential for functional angiogenesis and depends on Notch signalling, a topic extensively reviewed [12, 27]. Endothelial tip and stalk cells promote elongation of existent blood vessels through migration and proliferation of endothelial cells into avascular regions. The net outcome of this dynamic endothelial activity is the generation of a highly dense but immature vascular plexus. Yet, in order to evolve into a functional hierarchically branched network, blood vessels must be reorganised, a process that involves extensive vascular remodelling [27]. In the last few years, the number of mouse mutants harbouring defects in vascular remodelling has been growing extensively [4, 27]. Yet surprisingly, very little is known about the molecular and cellular mechanisms responsible for the control of vascular remodelling.

2 Mechanisms of vascular remodelling

In this ESCHM 2016 conference talk report, we summarise two recently published original articles Franco et al. PLoS Biol 2015 and Franco et al. eLIFE 2016. One of the key aspects of remodelling is vessel segment regression, in which previously established connections between two vessel segments are lost. Early reports focusing in vessel regression showed that endothelial cell apoptosis is the main driver for regression of hyaloid and pupillary membrane vascular networks [22]. In this context, WNT7b secreted by macrophages induces endothelial cell apoptosis, which promotes vessel stenosis, and in a cascade event, activation of apoptosis of adjacent endothelial cells due to cessation of blood flow, ultimately driving the regression of the entire hyaloid vasculature [19]. Also, extensive endothelial apoptosis can be detected in obliteration of retinal vessels in the oxygen-induced retinopathy model [34]. Thus, a similar mechanism was thought to drive vessel regression during developmental vascular remodelling, however conflicting evidences existed in the literature [6, 30].

Our recent research focused in this particular question. We investigate with high resolution the cellular mechanisms contributing to vessel regression in mouse and zebrafish. We found that vessel regression in mouse developmental angiogenesis is largely cell-death independent. We demonstrated that, rather, vessel regression involves dynamic rearrangement of endothelial cells, which migrate from regressing vessel segments to integrate in neighbouring vessels. We showed that developmental vessel regression involves four discrete steps: (1) an initial selection step of the regressing vessel, which precedes and triggers the morphological alterations during regression; (2) a stenosis step, in which the lumen is constricted or collapsed; (3) EC retraction, in which endothelial cells migrate and retract, being this process associated with junctional remodelling; (4) resolution of the regressing vessel segment, which comprises the final loss of any endothelial processes in this branch, leaving only basement membrane and pericytes behind (Fig. 1). At the cellular level, we observe junctional arrangements similar to those found during vessel anastomosis, suggesting that vessel regression resembles morphologically anastomosis in reverse. We, therefore, proposed a dynamic model for vessel regression and vascular remodelling challenging our previous vision on the vascular system [9]. The results are consistent with recent reports in zebrafish highlighting the apoptosis-independent nature of developmental vessel regression [5, 18]. More recently, deletion of apoptosis effector proteins BAK and BAX in endothelial cells, which results in the suppression of apoptosis, confirmed that endothelial cell apoptosis is dispensable for pruning [35]. By contrast, apoptosis reduces endothelial numbers during vascular remodelling, and the absence of endothelial cell apoptosis, capillaries exhibit an increase in diameter.

Therefore, it is now established that vessel regression during the process of developmental vascular remodelling is a dynamic process involving extensive cell movements within the vascular network. This generates a new model in which endothelial cells proliferate to provide sufficient numbers to support formation of the primitive plexus and are then rearranged and re-used in the process of making a functional vascular plexus to meet regional demands. Apoptosis in this scenario is a consequence of vessel regression rather than a cause.

3 Blood flow and vessel regression

Blood flow is an important regulator of vascular remodelling [11], but the relevance of and the mechanics of how physical forces and signalling pathways collectively stabilise or disrupt vessel connections remained elusive.

Previous studies showed that endothelial cells in zebrafish brain vessels sense a threshold of low blood flow, below which vessel regression is triggered irreversibly [5]. We therefore hypothesise that blood flow could be a main regulator of vessel remodelling. In order to understand the implications of blood flow and vessel regression, we developed computation modelling approaches of haemodynamic forces in the mouse retinal plexus [2, 10]. Furthermore, we were able to demonstrate that endothelial cell polarisation against the blood flow direction is a conserved feature in response to blood flow stimulation, both in vivo and in vitro [10]. Our findings demonstrate that blood flow acts as a coordinator of cell polarisation and endothelial cells respond to blood flow-induced shear stress (frictional force per unit area), elongating parallel to the flow direction while polarising and migrating against it (Fig. 2A). Both in high flow vessels, such as arteries and arterioles, or in lower flow vessels as in veins or capillaries of the retinal plexus, polarity of endothelial cells was directed against the blood flow (Fig. 2B). However, in low/no-flow vessel segments endothelial cells showed decreased orientation in endothelial cell axial polarities. Based on these observations, we proposed that blood flow is the primary factor dictating the stability or instability of vessel segments, by promoting converging or diverging polarities in endothelial cells. In this model, when flow generates appropriate high shear forces on the endothelial luminal surface, this new directional force breaks the previously established symmetry and drives polarisation against the blood flow direction. This polarisation event directs migration of cells in low-flow or oscillatory flow segments towards the high flow segments, thus disrupting the segment. Consequently, flow-induced endothelial cell polarisation directs migration of endothelial cells that reside in low-flow branches, leading to a regression of this segment and stabilization of the higher flow segments. Increasing flow asymmetry between juxtaposed vessels is the trigger for developmental vessel regression [10]. Interestingly, we noticed an increase in endothelial cell density in remodelled vessels, in retina central areas. Since apoptosis is not the major driver in the transition from a primitive to a remodelled plexus and that proliferation of endothelial cells is minimal in these regions, the increase in density is explained by the incorporation of endothelial cells from regressing low-flow vessels into these higher flow segments.

4 Non-canonical Wnt signalling as a regulator of developmental vessel regression

Wnts are signalling factors known to control several developmental processes, such as proliferation, differentiation, asymmetric division, patterning and cell fate determination [3, 24]. During development, these ligands act as morphogens and regulate the patterning of the embryo by triggering concentration-dependent autocrine and paracrine responses [25, 36]. To date, three major signalling branches have been identified including a canonical or Wnt/β-catenin dependent pathway and the non-canonical or β-catenin-independent pathway which can be further divided into the Planar Cell Polarity (PCP) and the Wnt/Ca2 + pathways. During canonical Wnt signalling, the binding of Wnt ligands to their Frizzled/LRP receptor complexes causes a stabilization of β-catenin, which is normally degraded by the GSK-3/APC complex. Stabilized β-catenin is then able to translocate into the nucleus and modulate the expression of specific target-genes [21]. In contrast, the less-characterized non-canonical Wnt pathways are independent of β-catenin and transduce Wnt signals through either JNK/PCP or PKC/CamKII pathways [16, 23].

While there is some controversy regarding the role of canonical Wnt signalling in vessel regression [19, 26], our recent study provided evidences that non-canonical Wnt ligands – Wnt5a and Wnt11 – act in a paracrine manner to prevent premature blood vessel regression by promoting endothelial cell cohesion and stabilization in the mouse retina. Using a compound Wnt5a endothelial-specific knockout and Wnt11 knockout mice (Wnt5a iEC-KO; Wnt11 KO), we demonstrated that loss of endothelial-derived Wnt ligands resulted in decreased radial expansion and decreased vessel density in the retina, accompanied by a higher number of regression profiles. Interestingly, apoptosis and endothelial cell numbers remained unaffected. However, Korn et al. also used a similar strategy and showed that non-canonical Wnt signalling was involved in the regulation of endothelial cell survival and apoptosis [17]. Thus, further clarification in this question in needed.

Given the recent data demonstrating the specific role of blood flow in regulating developmental vessel regression, we investigated the potential connection between non-canonical Wnt signalling and blood flow response. We use our computational analysis tool [2, 10] together with in vivo and in vitro experiments and we confirmed that depletion of non-canonical Wnt ligands from the endothelium significantly increased the sensitivity of endothelial cells to polarise against the blood flow direction at lower wall shear stress levels. We proposed that this enhanced sensitisation of endothelial cells promotes earlier discrimination of flow asymmetries between neighbouring blood vessels, which acts as the driving force for premature vessel regression and accelerated remodelling in response to the lack of Wnt5a and Wnt11 (Fig. 2C). Importantly, overexpression of Wnt5a in endothelial cells of the retina had no effect on vessel regression or overall vessel morphology, indicating that Wnt signalling is not used as a mechanism to drive flow-dependent vessel regression, but its presence at a basal level seems to be required to lower the sensitivity of endothelial cells to flow-induced regression. However, in vitro studies also demonstrated that Wnt5a/Wnt11-depleted HUVECs retain their capacity to activate the expression of flow sensing components as control cells, suggesting that non-canonical Wnt signalling acts as modulator of the physical reorganization of cell polarity rather than flow sensingitself [9].

Taking into consideration other recent studies showing that: (1) endothelial cells migrate and rearrange dynamically in vascular sprouts [14]; (2) vessel regression events during vascular development are driven by cell migration/rearrangements under the influence of flow [5, 18]; (3) non-canonical Wnt signalling inhibits premature vessel regression [9]; we proposed a novel model to explain the transition from a primitive to a mature network. We propose that the primitive network before flow onset, or at a subthreshold of flow levels, is in a state of dynamic stability where the movement of cells is balanced by counter movements of adjacent cells promoting that the immature vascular network remains open and lumenised in low flow conditions. In this context, non-canonical Wnt signalling is likely to stabilise cell junctions and promote coordinated endothelial cell behaviour. Flow-induced polarity will supersede the dynamic stability proprieties of the network, driving endothelial cells to polarize against the flow direction. Flow-induced polarization introduces a bias in the system that leads to stenosis and regression of low-flow vessel segments (Fig. 2C). What drives the rearrangements of cells in the primitive plexus and how flow in one segment initiates regression in another is poorly understood [9].

5 Perspectives

Recent studies have revealed that besides being an organizing center piece for cortical and junctional actin cytoskeleton [29], VE-Cadherin dynamics act as the main driving force during vascular rearrangements [1]. According to Bentley et al., cells with higher levels of vascular endothelial growth factor (VEGF) and lower Notch activity are highly motile, as they possess a significantly larger mobile fraction of VE-Cadherin present at their junctions. Whether these observations also hold true for events taking place during vascular regression remains unclear. However, considering other studies showing that: (1) Notch signaling remains active during remodeling [8, 20]; (2) VE-Cadherin is a central component of EC shear stress force sensing [7]; (3) VE-Cadherin has been implicated in the coordination of EC polarity during collective cell migration [33]; we envisage the involvement of Notch signaling as a creator of local differences in cell motility during vascular rearrangements, whereas non-canonical Wnt signaling acts as a balancing force element of net movements, by promoting cell coordination, cohesion and symmetry. The breaking of this symmetry in the primitive vascular plexus is then initiated with the onset of flow, which provides an extrinsic directional cue that will ultimately polarize EC from the low-flow segments towards the high-flow segments. However, more studies addressing these issues must be performed in order to fully understand the extent of the molecular and cellular mechanisms cooperating to enable the spatial and temporal control of vascular remodeling.