Pericyte Dynamics during Angiogenesis: New Insights from New Identities

INTRODUCTION

Vascular pericytes were first discovered in 1873 by the French physiologist, Charles Marie–Benjamin Rouget, and defined as “Rouget” cells. Zimmerman introduced the term pericyte in 1923 and described them as a periendothelial support cell that wrapped around the length of microvessels [1]. While their presence in the microcirculation has been long documented, their functional roles and importance in microvascular physiology have been largely under-investigated. In more recent years, clarification of their versatile functionality has thrust them into the spotlight. Because of their putative influence on endothelial cells, pericytes have become a research area of growing interest and are increasingly being evaluated as potential targets for pro- or anti-angiogenic therapies.

Vascular pericytes are still defined morphologically as periendothelial support cells that elongate and wrap along endothelial cells [2]. Through trans-basement membrane interactions, pericytes are thought to regulate capillary diameter and physically influence endothelial cell behavior [3]. During capillary sprouting, vascular pericytes are recruited by endothelial cells via platelet-derived growth factor (PDGF)-B/PDGF receptor-β (PDGFR-β) paracrine communication [4, 5]. We do know that the presence of pericytes is critical for normal angiogenesis [6]. Yet, the inability to distinguish specific pericyte subpopulations has limited 1) our understanding of how pericytes directly influence endothelial cell dynamics and 2) the development of pericyte-targeted angiogenic therapies. For example, pericytes have been shown to both stabilize and promote capillary sprouting. This apparent dual role motivates the questions: Do specific subtypes of pericytes exist? Do pericytes change their phenotype in order to perform their specific functions? In endothelial cell biology, specialized cell types are exemplified by the discovery of tip cells versus stalk cells during angiogenesis [7]. The tip cell paradigm highlights the potential for two neighboring cell types to be delineated by their phenotype and function. In the context of pericyte biology, determining the spatial and temporal phenotypic differences along angiogenic vessels versus mature vessels represents an equal opportunity to add new insight and perspective.

Common pericyte markers include smooth muscle α-actin (SMA), desmin, vimentin, PDGFR-β, and neuron-glial antigen 2 (NG2). However, these markers identify multiple cell populations along the entire hierarchy of microvascular networks and are not angiogenesis specific [3, 8, 9]. What is equally challenging is that pericyte marker expression varies across a network and per tissue. In this review we propose a new paradigm for pericyte biology wherein transient expression patterns by pericyte subtypes during angiogenesis can potentially be used to distinguish pericyte functionality. In the first half of this review, we briefly summarize the current understanding of pericyte physiology and function. Then in the second half we introduce the potential for defining functionally distinct pericyte subpopulations, by discussing the temporal expression dynamics of two neural molecules, NG2 and class III β-tubulin, by pericytes in the adult microcirculation.

PERICYTE DYNAMICS INVOLVED IN ANGIOGENESIS

Vascular pericytes are critical players in the microcirculation and their various functions are comprehensively covered by multiple reviews [3, 10–13]. Yet, the mapping of specific pericyte functions and downstream mechanisms to specific pericyte dynamics remains elusive. Pericytes are defined morphologically as cells within the basement membrane that elongate along the long axis of capillaries, and wrap around endothelial cells with shorter processes [3, 14]. Direct interdigitations protrude from pericytes into endothelial cells at their interface, and endothelial processes reciprocate [15]. Pericytes are functionally associated with the regulation of blood vessel diameter, vessel permeability, endothelial cell proliferation, angiogenesis, and even leukocyte recruitment [11, 14, 16–18]. Recent evidence shows that pericytes mediate capillary diameter by contraction in response to electrical or neurotransmitter stimulation [19], implicating pericytes as highly localized modulators of blood flow. Lack of pericyte recruitment leads to endothelial hyperplasia and increased vascular permeability [6, 20], associating pericytes with vessel stability and maturation.

Work in the pericyte biology field has established several mechanisms that regulate pericyte-endothelial communication [3, 11, 13]. The prominent signaling pathways are PDGF-B/PDGFR-β, angiopoietin 1 (Ang1)/Tie2, and transforming growth factor-β (TGF-β) [21], which regulate pericyte recruitment, endothelial cell viability and mural cell differentiation, respectively [22]. Pericytes can also influence the local extracellular matrix (ECM) to guide endothelial migration by modulating deposition of basement membrane proteins [23]. As pericytes have been linked to microvascular function, it makes sense that their coverage is heterogeneous among organs depending on the needs of the local tissue environment. In the brain, for example, a high density of pericytes tightly regulates the blood-brain barrier as a selectively permeable fortification which prevents fluctuations in the sensitive environment of the central nervous system (CNS). Pericyte coverage in skeletal muscle, however, may be as low as one pericyte for every 100 endothelial cells [2]. Liver (stellate cells) and kidney pericytes are specialized to aid parenchymal functions. Recently, pericytes of both organs have garnered attention for their central role in liver and kidney fibrosis [24–26].

Alterations in pericyte coverage are not surprisingly linked to multiple pathologies. In the case of diabetic retinopathy, the disassociation of pericytes from endothelial cells has been associated with uncontrolled angiogenesis [12]. Decreased pericyte coverage in the brain causes neurodegeneration [27]. This is similar to a tumor growth scenario, in which loose pericyte coverage leads to vascular dysfunction [28, 29]. Pericytes contribute to angiogenesis during tumor vascularization. However, loss of pericytes or abnormal pericyte function in tumors increases vessel permeability and decreases vessel integrity, which enhances tumor metastatic potential [28, 30]. This illustrates the dual role of pericytes in regulating angiogenesis and vessel integrity.

During angiogenesis, pericyte dynamics include recruitment, ECM modulation, paracrine signaling and direct interactions with endothelial cells (Figure 1) [3]. They have also been shown to lead sprouting endothelial cells and bridge the gaps between two sprouting segments in some cases [31, 32]. Endothelial cells of newly forming vessels recruit pericytes by secreting PDGF-B [4, 5]. Manipulating pericyte investment by altering recruitment to the endothelium can cause lethal microvascular dysfunction, deviate vascular patterning during development, and affect the formation of new vessels during physiological and pathological angiogenesis [3, 33]. As previously mentioned, this can be attributed to pericyte regulation of endothelial cell proliferation. Recent work has also identified pericytes as important players in regulating basement membrane organization. During developmental vascular assembly, pericytes catalyze deposition of basement membrane proteins including collagen IV, laminin, and fibronectin [11]. Pericyte actomyosin-mediated contraction can modulate the local mechanical environment for adjacent endothelial cells [34]. Endothelial cell behavior can also be mediated by direct connections between the two cell types via N-cadherin adhesions [13, 35], pericyte NG2 binding to β1 integrins on endothelial cells [29, 36] and potentially connexin 43 [37, 38]. Furthermore, pericyte-endothelial cell interactions induce changes in expression of ECM-binding integrins by both cell types, which are necessary for investment in newly formed matrix [39]. Once vascular tubes are assembled, pericyte-derived tissue inhibitor of metalloproteinase-3 (TIMP-3) prevents proteolysis of matrix proteins [40].

Because of their necessity in microvascular remodeling and proximity to endothelial cells, pericytes have emerged as a cellular target to manipulate angiogenesis [41–43]. For example, inhibition of PDGFR-β+ tumors depletes pericytes and disrupts tumor vasculature, which curbs tumor growth [44]. Conversely, increasing pericyte coverage via Ang2 inhibition can represent an anti-angiogenic tumor therapy [45]. Pericytes have been implicated in preventing tumor metastasis [28, 30], and depletion of pericytes in vascular retinopathy leads to hypervascularization and blindness [33]. The therapeutic potential for pericytes can also be exemplified by considering the similarities between pericytes and mesenchymal stem cells (MSCs) [46–51] (a concept that will be re-visited later in the next section) and appreciating that the delivery of MSCs represents a viable approach for pericyte-targeted angiogenic therapies [43]. Successes of this approach has been well documented [43] and therapeutically delivered MSCs have been convincingly shown to acquire pericyte characteristics and stabilize aberrant angiogenesis in various models of normal and pathological models of vessel remodeling. Still, the exact mechanisms that regulate pericyte and MSC physiology in these scenarios remain unclear, and understanding the pericyte (and MSC) subtypes that contribute to angiogenesis and vessel stabilization will open new avenues for therapeutic manipulation of the microvasculature.

While it is clear that pericyte-endothelial cell interactions play a crucial role in establishing and maintaining a normal microcirculation, questions about pericyte mechanisms, especially during angiogenesis in the adult remain. Determining phenotypic differences between pericytes on angiogenic vessels and mature vessels may provide insights to the varying functions of pericytes across the microvasculature. However, pericyte-specific phenotypes remain elusive, and recent studies document the overlap between pericytes and multiple other cell types. Interestingly, pericytes have even been suggested to display multipotency similar to stem cells. Many questions also certainly remain regarding pericyte lineage. Yet in the midst of any debate, the recognition of their phenotypic dynamics offers a valuable opportunity to discover new pericyte spatial and temporal identities that define specific cell subpopulations.

PERICYTE IDENTITY

DO PERICYTE SUBPOPULATIONS PLAY FUNCTIONAL ROLES?

The concept of specialized subpopulations of vascular pericytes forces the integration of studies using pericyte markers to identify pericytes and studies focused on elucidating the functional roles of those markers. NG2 is sometimes referred to as a marker of angiogenic pericytes along capillaries [31, 62] because it typically colocalizes with pericytes along all capillaries and capillary sprouts. In support of the angiogenic classification, we have demonstrated that during angiogenesis NG2 is dramatically upregulated along venules during angiogenesis implicating its involvement in capillary sprouting [64]. Importantly, NG2 is more than just a marker, and its identification lead to its discovery as a regulator of capillary sprouting. NG2 binding of PDGF-AA and basic fibroblast growth factor (bFGF) [84, 85], as well as the kringle domains of plasminogen, which is important for activation of latent TGF-β [86], implicates its role as a regulator of local angiogenic growth factor presentation. NG2 expression has also been shown to orient with actin stress fibers in a linear array and has been linked to downstream Rho-Rac signaling [54, 87]. Another possible role for NG2 is manipulation of the local ECM environment as a receptor for type VI collagen [88]. The soluble form of the proteoglycan promotes endothelial cell motility and angiogenesis via engagement of galectin-3 and α3β1 integrin [89] suggesting a direct interaction between NG2-expressing perivascular cells and neighboring endothelial cells. Transmembrane signaling between the two cell types is further supported by NG2 knockdown pericyte-endothelial cell co-culture models. Pericyte specific NG2 inhibition can reduce β1 integrin activation by endothelial cells and endothelial cell tube formation, junction formation, and permeability [29].

The functional importance of this proteoglycan has been demonstrated in vivo with the use of ng2^−/− mice and functional blocking antibodies [90, 91]. Genetic deletion or blockade of NG2 in hypoxia-induced microvascular growth in the retina and bFGF-induced growth in the cornea resulted in fewer vessels [91]. NG2 antibody inhibition in rat mesenteric microvascular networks has been shown to inhibit bFGF and vascular endothelial growth factor (VEGF) induced sprout formation [92]. During tumor angiogenesis, NG2 has been shown to sequester angiostatin, increasing tumor growth rate [93]. Tumors in mice with pericyte-specific NG2 deletion display highly dysfunctional and destabilized vasculature, which leads to increased vessel permeability and tumor hypoxia [29], further advocating the functional role of NG2.

The influences of NG2 on pericyte and endothelial cell dynamics highlight the association between a pericyte phenotype and function. However, spatially correlating NG2 expression by pericytes to dynamics involved in angiogenesis is not so straightforward. For instance, SMCs along quiescent arterioles also express NG2, and one generally assumes that the functions of SMCs along arterioles differ from the functions of pericytes along capillary sprouts. Nonetheless, considering the functional roles of NG2 and its coordinated angiogenesis-specific expression patterns supports that pericytes change their phenotype to play specific functions during angiogenesis. To further explore this possibility, let us reconsider our recent discovery of class III β-tubulin as an angiogenic pericyte marker. Maybe not surprisingly, experimental results do support a functional role for class III β-tubulin.

Class III β-tubulin appears to play a role in pericyte migration, which is necessary for pericyte recruitment during the sprouting process. Cultured human brain-derived pericytes express class III β-tubulin in vitro (Figure 5). Given that MSCs in culture also express class III β-tubulin [83] and that pericytes and MSCs share phenotypic similarities, this result might have been expected and we have also observed class III β-tubulin expression by human-derived placental pericytes in vitro (data not shown). Considering pericyte class III β-tubulin expression in vitro together with our observations of class III β-tubulin up regulation by pericytes in adult rat mesenteric microvascular networks during angiogenesis (Figure 4; [80]) raises questions regarding 1) cell state in vitro and 2) the potential differences between pericytes in vivo versus in vitro. Both issues will prove critical for applying our gained knowledge from in vitro studies.

To investigate the functional role of class III β-tubulin, we inhibited its expression by specific siRNA treatments and measured its effect on pericyte invasion in an in vivo scratch wound assay. Commercially obtained human brain-derived pericytes (ScienCell) were grown to confluence and treated with no siRNA, control (non-targeting) siRNA or class III β-tubulin siGENOME SMARTpool siRNA from Dharmacon/Thermo Scientific for 24 hours. siRNA inhibition of class III β-tubulin decreased the percentage of cells with observable class III β-tubulin immunolabeling (Figure 6C) and relative quantities of RNA compared to control groups based on qRT-PCR (data not shown). Inhibition of class III β-tubulin reduced cell migration into the area void of cells in a monolayer scratch test (Figure 6). The scratch-wound area closure fraction for the class III β–tubulin siRNA group was 35% smaller compared to the non-targeting Control siRNA group (0.45 +/− 0.04 vs. 0.69 +/− 0.06; p = 0.002) and 27% smaller than the Sham Control group (0.45 +/− 0.04 vs. 0.62 +/− 0.02; p = 0.009; Figure 6D). In the class III β-tubulin siRNA treatment groups, cells were both class III β-tubulin positive and negative indicating that the siRNA knockdown effect was not homogeneous. Class III β-tubulin+ cells were frequently found leading migration into the scratch area in the Target siRNA groups (Figure 6E, F). The effect of class III β-tubulin inhibition on the scratch wound closure is consistent with reports by others [94] that have shown that class III β-tubulin in vitro can influence cell motility and proliferation. Our results in the context of pericytes suggest that expression of class III β-tubulin by pericytes during angiogenesis would influence their ability to be recruited to newly forming capillary sprouts. Undoubtedly, further follow-up studies including various pericyte sources and additional assays are required, but the preliminary results do motivate the need to understand the different roles of specific pericyte subpopulations. If we can define specific pericyte subpopulations, as in the endothelial tip versus stalk cell scenario, then we might be able to differentiate the importance of one pericyte type versus another.

SUMMARY

Pericytes are essential for angiogenesis and vessel maturation. Their specific mechanisms during these processes remain unclear due to the heterogeneity of pericyte identities across individual microvascular networks and different tissues. Although this has been a barrier in the past, transient pericyte phenotypes during capillary sprouting suggest that subpopulations perform specific angiogenic functions. Transient expression of NG2 and class III β-tubulin, typically associated with the neural system, implicate two neural phenotypes in pericyte function during angiogenesis. Correlating known angiogenic dynamics with such subpopulations represents an opportunity to pinpoint specialized pericyte functions based on phenotype, or target pericyte mechanisms for therapeutic benefits. Accordingly, defining specific pericyte subpopulations involved in angiogenesis versus vessel maturation will be advantageous, and maybe compulsory, for future studies that aim to elucidate pericyte mechanisms. As we attempt to focus on specific spatial and temporal pericyte subpopulations, the need for tunable model systems will emerge. The development of such in vivo and ex vivo systems in which pericytes can be dynamically probed within intact microvascular networks will help advance our understanding of pericyte identity [29, 92]. In the meantime, the discoveries of different pericyte subpopulations offer new foundations for the future.