Keywords

Introduction

The role of the tumour microenvironment (TME) in promoting tumour growth and metastasis is widely recognised and consists of a variety of cells including cancer-associated fibroblasts (CAFs), bone marrow-derived mesenchymal stem cells (BM-MSC), endothelial cells, pericytes and immune cells and the growth factors and proteins they produce. Studies of various types of cancer including ovarian, colorectal, pancreatic and breast demonstrate that stromal signatures predict relapse and recurrence in patients lending strong support to the notion that the TME is a strong contributor to malignant progression (Tothill et al. 2008; Finak et al. 2008; Tsujino et al. 2007; Fujita et al. 2010; Calon et al. 2015). It is possible that not all elements of the TME are pro-tumourigenic—indeed, many tumours are encapsulated by fibroblasts perhaps limiting metastatic spread. It has been suggested that the TME has a role in contributing to resistance against anti-cancer therapeutics (Frame and Serrels 2015). It is therefore important that we gain a better understanding of the contributions of specific subsets of cells found in the TME to cancer progression. This discussion focuses on the role of pericytes given the relatively recent discovery that they have angiogenesis-independent roles in promoting malignant cancer.

Difficulties in Distinguishing Pericytes from CAFs in the TME

CAFs are defined as fibroblasts that universally promote tumour growth, invasiveness and metastasis compared to normal “fibroblasts” (Olumi et al. 1999; Cunha et al. 2003; Kalluri and Zeisberg 2006; Pietras and Ostman 2010). CAFs can originate from diverse sources including tissue-resident myofibroblasts, activated adipocytes and distal bone marrow-derived MSCs/BM-MSCs (Kalluri and Zeisberg 2006; Cirri and Chiarugi 2011), and are mostly identified retrospectively using equivocal “CAF markers”, such as α-SMA (smooth muscle actin) which is activated in all mesenchymal cell types in conditions of stress. Thus, α-SMA is expressed in cultured or activated fibroblasts, myofibroblasts, pericytes and most BM-MSCs (Kalluri and Zeisberg 2006; Cirri and Chiarugi 2011). Lineage-marking studies in animals show that GFP-tagged BM-MSCs home to developing tumours inducing increased metastases (Karnoub et al. 2007; Hung et al. 2005; Mishra et al. 2008; Studeny et al. 2002; Quante et al. 2011), making up about ~20–50% of CAFs. Thus, 50–80% of CAFs are not BM derived and may originate from local fibroblasts or other MSC-like populations such as pericytes. Indeed, our lab has shown that pericytes accelerate tumour growth rates and promote metastatic spread in a xenograft model of ovarian cancer (Sinha et al. 2016). Moreover, we showed that pericytes recruited BM-MSCs to developing tumours, suggesting that they may act upstream of BM-MSCs.

Classic Functions of Pericytes in Cancer: Stabilising Tumour Blood Supply and Limiting Hypoxia

In cancer, pericytes have been widely studied in the context of their well-known capacity to stabilise blood vessel structure and permeability. Dual targeting of vascular endothelial cells and pericytes using kinase inhibitors or anti-VEGF and anti-PDGFβ antibodies, has a synergistic anti-angiogenic/anti-tumour effect, resulting in increased tumour cell killing in animal models attributed to destabilisation of the tumour microvasculature (Bergers et al. 2003; Druker 2002; Erber et al. 2004; Kuhnert et al. 2008; Maciag et al. 2008). Other studies claiming that tumour growth was unaffected after pericyte removal are equivocal given that a maximal 50% pericyte knockdown (KD) was achieved using AX102, an inhibitor of PDGFB signalling (Sennino et al. 2007), or in PDGFret/ret mice that harbour a mutation in the PDGFB retention motif (Nisancioglu et al. 2010; Lindblom et al. 2003). On the other hand, complete KD of pericytes with NG2 promoter driven thymidine kinase, caused tumour hypoxia which led to epithelial-mesenchymal transition and metastases to the lungs in mouse models of breast/renal cell carcinoma and melanoma (Cooke et al. 2012). These data support the idea that pericytes limit metastatic spread through otherwise leaky tumour blood vessels (Xian et al. 2006).

Tumour Blood Vessel Remodelling Leads to Displacement of Pericytes from Their Vascular Niche

Blood vessel remodelling during tissue repair is a dynamic process requiring the initial detachment of pericytes from endothelial cells from pre-existing vessels resulting in the removal of paracrine signalling between the two cell types that keep vessels in a homeostatic state. Pericyte detachment permits endothelial cell sprouting and proliferation, and subsequent re-association of the two cell types in the newly extended blood vessels – processes driven by angiopoietin-1/2 and Tie2 (Ang/Tie2), transforming growth factor-β (TGF-β), and platelet-derived growth factor-B (PDGFB) and its receptor PDGFR-β (Lindblom et al. 2003; Stapor et al. 2014). Similar mechanisms underlie tumour vessel remodelling although tumour vasculature is typically disorganised with torturous vessels, excessive branching and altered gene expression resulting in impaired vascular structure and increased vessel leakiness (Ruoslahti 2002). Notably, pericytes are more loosely attached to endothelial cells with cytoplasmic projections invading the tumour stroma (Morikawa et al. 2002). It has been shown that detachment of perivascular cells (and subsequent endothelial cell sprouting) is mediated by angiopoietin-2 secreted by activated endothelial cells (Scharpfenecker et al. 2005). Similarly, it has been shown that tumour cells overexpressing PDGFBB xenografted onto mice, induced dissociation of pericytes from tumour blood vessels in a dose-dependent manner increasing vascular permeability leading to vascular impairment (Hosaka et al. 2013). Notably, continued exposure of pericytes to PDGFBB led to down-regulation of PDGFβR that in turn decreased the expression of the α1β1 integrin receptor from the cell surface of pericytes, abrogating their adhesion to extracellular matrix proteins in the blood vessel walls resulting in their detachment from them (Hosaka et al. 2013). Thus, paracrine signalling between pericytes and cancer cells provides an important mechanism by which pericytes can be persuaded to leave their normal microenvironment within microvessels and associate more closely with tumour cells. Given the reports that the precise location of pericytes can alter their functionality , it is clear that pericytes may be able to act directly on tumour cells as part of their mesenchymal microenvironment.

Mesenchymal Stem Cell Properties of Pericytes: Similarities and Distinctions

Given that many tissues are well vascularised, it has been speculated that perivascular cells throughout the body serve as a reservoir of multipotent mesenchymal stem cells that can be recruited upon tissue injury. Certainly, the phenotypic and functional similarities between pericytes and mesenchymal stem cells have been widely reported, serving to underpin the idea that the two cell types are in fact one and the same. Thus, both pericytes and BM-MSCs are CD45−CD31−αSMA+CD146+NG2+PDGFRB+CD73+CD90+, located in a perivascular niche and can differentiate into fat, bone, cartilage, muscle, and neuronal cells (Crisan et al. 2008; Caplan 2008; Paquet-Fifield et al. 2009). However, important distinctions exist between pericytes and MSCs in that an immunosuppressive role has been described for MSCs (Shi et al. 2018) whereas the indications are that in animal studies pericytes or perivascular cells are pro-inflammatory (Mills et al. 2015; Dulauroy et al. 2012) and likely to contribute to delayed healing and fibrosis.

Pericytes and Fibrosis

Pericytes are involved in various fibrosis -related pathologies in the kidneys, liver, and skin, acting as progenitors to myofibroblasts, which are the main mediators for extracellular matrix deposition, leading to fibrogenesis and ultimately fibrosis during the healing process (Greenhalgh et al. 2015; Kramann and Humphreys 2014). In a transgenic reporter mouse model, coll1α1-GFP-expressing pericytes were shown to be the main source of myofibroblasts leading to kidney fibrosis (Lin et al. 2008). Similar studies in the liver, where pericytes are known as hepatic stellate cells, also showed that they were the main source of myofibroblasts, and a major player in liver fibrosis (Greenhalgh et al. 2015). Consistent with this, studies with a Cre-transgenic mouse model which labelled hepatic stellate cells in various models of liver injury, demonstrated that they accounted for 82–96% of the myofibroblast pool, which contributes to liver fibrosis (Mederacke et al. 2013). Pericytes have a similar role in skin fibrosis and scar formation as illustrated by genetic fate mapping, revealing that the majority of collagen producing myofibroblasts originate from ADAM12 expressing cells, derived from PDGFRB+ NG2+ perivascular cells or pericytes (Dulauroy et al. 2012). Notably, ablation or knockdown of ADAM12+ cells, was sufficient to limit collagen production in the healing site of injury and reduce fibrosis. These studies demonstrate the ability for pericytes to differentiate into myofibroblasts thereby contributing to fibrosis. They also illustrate the ability of pericytes to contribute to the remodelling of tissue stroma to achieve wound repair thus pointing to ways in which these cells can affect biological processes in an angiogenesis-independent manner.

Pericytes in Cancer and Metastasis

It has been variously postulated that pericytes affect tumour growth and metastasis both positively and negatively. Many of the tumour growth promoting effects are related to establishing a stable vascular network thus ensuring delivery of nutrients to rapidly growing tumour cells and preventing tumour cell dissemination through blood vessels by maintaining vascularity permeability. Both these aspects require adequate pericyte investment on the abluminal surface of tumour blood vessels—experimental depletion of pericytes does indeed result in tumour regression (Bergers et al. 2003) but also leads to hypoxia-induced epithelial-mesenchymal transition increasing metastasis (Cooke et al. 2012). Consistent with this, normalising tumour vasculature by abrogating RGS5 expression (a cell surface protein that is abnormally expressed in tumour vessels), makes tumours more susceptible to chemotherapeutic agents (Hamzah et al. 2008). Moreover, the context in which pericyte dissociation from tumour vessels occurs also affects the response to chemotherapeutic agents as shown for variable levels of PDGFBB expression by various tumours (Hosaka et al. 2013).

Angiogenesis-Independent Mechanisms by Which Pericytes Promote Metastasis

A more direct role for pericytes in promoting cancer growth and metastasis without impact on angiogenesis has recently emerged from several laboratories including our own. It has become increasingly evident that pericytes are potent mesenchymal stem cell-like cells with an ability to promote organ repair and regeneration and multiple mesenchymal lineage differentiation capacity (Crisan et al. 2008; Sa da Bandeira et al. 2017). In the haemopoietic system, pericytes are an integral part of the haemopoietic stem cell niche regulating their maintenance and quiescence through paracrine effects (Sacchetti et al. 2007) supporting haemopoiesis both in vitro and in vivo (Birbrair and Frenette 2016; Morrison and Scadden 2014), as reviewed in Sa da Bandeira et al. (2017). In the process of studying the cellular microenvironment of epithelial renewal in human skin, we discovered a novel, paracrine role for pericytes in influencing skin tissue regeneration in 3D organotypic cultures completely lacking any blood vessels (Paquet-Fifield et al. 2009).

In view of the fact that pericytes are MSC-like and that MSCs had been reported to promote breast cancer metastasis (Karnoub et al. 2007) and ovarian cancer growth (McLean et al. 2011), we sought to establish whether pericytes may be a critical element of the TME with a more direct role in cancer progression. Recognising that the stromal signature of serous ovarian cancer patients reported by the Australian Ovarian Cancer Study Group (AOCS) (Tothill et al. 2008) had markers of both fibroblasts and pericytes, we used the molecular signature previously generated by us for both individual cell types (Paquet-Fifield et al. 2009) to interrogate the AOCS patient dataset annotated for patient outcomes (Sinha et al. 2016). Remarkably, the pericyte signature outperformed the ovarian cancer stromal signature at predicting early relapse revealing that those serous ovarian cancer patients carrying a high pericyte score (evidenced by a set of 146 genes co-expressed by both pericytes and ovarian cancer stromal cells), relapsed significantly earlier with a mean progression-free survival/PFS time of 9 months (vs. 29 months in those with a low pericyte score; n = 215), despite similar treatment (p = 0.00067 vs. p = 0.0011 from Tothill et al. 2008). Notably, the fibroblast signature was relatively poorer at predicting relapse (p = 0.01). Subsequently, we used a xenograft model to demonstrate that pericytes could act as CAFs when co-injected with ovarian cancer cell lines and that critically the tumour vasculature derived entirely from host murine cells remained unaffected with respect to the number of blood vessels or pericyte investment . Thus, co-injection of human pericytes with OVCAR-5 or OVCAR-8 cells, accelerated tumour growth rates and caused rapid dissemination to local tissues increasing metastases in a dose-dependent manner, typical of ovarian cancer spread clinically (Sinha et al. 2016). Notably, the human pericytes remained in the tumour stroma not associating with the tumour vasculature presumably due to species-specific incompatibility of signals that might otherwise result in their incorporation into tumour microvessels. This study provided the first clear evidence uncoupling the pro-angiogenic versus pro-metastatic function of pericytes in cancer. This suggests to us that when pericytes are dissociated from blood vessels they promote metastasis—a novel site of pericyte action, whereas their normal location in blood vessels restricts metastasis (Fig. 5.1). Consistent with a paracrine role for pericytes in promoting tumour cell metastasis, transwell co-culture experiments showed that pericytes increased ovarian cancer cell migration and invasion through matrigel; in other experiments we were able to demonstrate increased ovarian cancer cell proliferation with pericyte co-culture (Sinha et al. 2016). Interestingly, recent work with a variety of epithelial cancers has shown that pericytes contribute to cancer progression by giving rise to CAFs when dissociated from tumour blood vessels (Hosaka et al. 2013).

Fig. 5.1
figure 1

Schematic model of the role of pericytes in promoting malignant cancer progression. Localisation of pericytes in intimate association with blood vessels results in a tightly encapsulated ovarian cancer tumour (left) whilst placing pericytes directly within the tumour microenvironment leads to increased tumour cell proliferation, increased recruitment of αSMA+ stromal cells including cells expressing markers of BM-MSCs, induction of EMT and invasion, and metastatic spread to distant organs. Based on Sinha et al. (2016)

Full size image

It is also likely that pericytes promote aggressive tumour growth by affecting the cancer stem cell compartment within tumours preferentially by secreting paracrine regulators. In support of this, it has been reported that human cancer-associated MSCs found in the TME of ovarian cancer increased the incidence of ALDH+CD133+ cancer stem cells via BMP-2 (McLean et al. 2011). In recent work from our laboratory, we have shown that pericytes can dictate the orientation of cell divisions within the skin’s proliferative compartment, i.e. the basal layer, increasing planar presumed symmetric cell divisions delaying differentiation thereby maintaining epidermal cells in a more primitive state via BMP-2 (Zhuang et al. 2018). These data and work from lower organisms such as Drosophila demonstrating a role for BMP signalling in maintaining “stem-ness ” of neighbouring cells (Kawase et al. 2004; Song et al. 2004) points to a conserved mechanism for the stem cell niche in regulating the fate of adjacent cells.

Future Trends and Directions

The full spectrum of functional capabilities of pericytes are only just starting to emerge and the closer investigators look beyond their classic role in vascular stability and permeability, the more seems to be uncovered (Ribeiro and Okamoto 2015). For instance, their ability to influence the inflammatory response by acting as a physical barrier to extravasation of immune cells and secreting a vast array of cytokines and extracellular matrix molecules are only recently being appreciated (reviewed in Navarro et al. 2016). Pericytes have been implicated in metastasis by increasing tumour cell intravasation at distal sites through endosialin (Viski et al. 2016), by regulating the metastatic niche via KLF4 (Paiva et al. 2018) and suppressing the immune response to brain tumours (Sena et al. 2018). The exact nature of molecular crosstalk between pericytes and cancer cells needs to be studied—one possibility is that they may contribute to tumour growth by differentiating into fat cells, which act as a source of energy driving cancer cell growth and metastasis (Huang et al. 2018). Another an exciting prospect is that exosomes secreted by cancer cells signal pericytes to become CAFs (Ning et al. 2018). Thus, a clear driver of future work has to be the identification of specific subsets of pericytes with cutting-edge technologies such as single cell RNA seq as reported recently for murine brain vascular cells including pericytes (Vanlandewijck et al. 2018). Moreover, the commonalities and distinctions in functional gene expression related to specific anatomical sites and organs needs to be addressed urgently to broaden our understanding of how these cells contribute to tissue renewal, wound repair, cancer, and ageing. A thorough understanding of pericyte cellular and molecular biology and their immense impact on neighbouring cells is essential to devise improved stem cell and regenerative medicine and interventions in cancer progression. An underlying concept is that just as we acknowledge that there are cancer stem cells within tumours that can drive cancer progression, a similar recognition of subsets of “cancer stromal stem cells” is much overdue.