The tumor microenvironment is now recognized as critical to a complete understanding of the pathophysiology of cancer. Tumor stroma, an important element of the tumor microenvironment, is composed of extracellular matrix (ECM), fibroblasts, endothelial cells (ECs), and mural cells. Mural cells are associated with blood vessels and can be subdivided into a continuum of phenotypes ranging from pericytes to vascular smooth muscle cells (vSMCs) . Normal pericytes are embedded within the basement membrane of capillaries, either as solitary cells or a single-cell layer, where they coordinate intercellular signaling with ECs and other components of the blood vessel wall to prevent leakage. In contrast, vSMCs form single or multiple layers around arteries and veins to mediate vascular tone and contraction. Mural cells with phenotypes between those of classical pericytes and vSMC are associated with arterioles and venules . Pericyte markers vary depending on the tissue of origin, with the most common markers being desmin, α-smooth muscle actin (SMA), regulator of G protein signaling 5 (RGS-5), platelet-derived growth factor receptor (PDGFR) [3, 4], and chondroitin sulfate proteoglycan 4 .
Pericytes and pericyte-like cells have been implicated as mediators of several processes associated with cancer pathophysiology including tumor angiogenesis and metastasis. While not discussed in this review, pericyte-derived tumors may represent a broad continuum of mesenchymal neoplasms from soft tissue to primary bone tumors given the inherent plasticity of pericytes and the ubiquitous distribution of these cells within different tissues . Although the role of pericytes in tumor pathophysiology is still unclear, evidence indicates that aberrations in EC–pericyte signaling networks could contribute to tumor angiogenesis and metastasis. The purpose of this review is to provide an integrated view of recent findings on the role of pericytes in tumor biology and evaluate potential therapeutic targets for anticancer intervention.
Pericytes in Tumor Angiogenesis
The initial stage of angiogenesis begins with pericyte–EC dissociation followed at later stages by EC invasion and proliferation and subsequent endothelial tubulogenesis and vessel stabilization. During the initial stage, pericytes adopt an angiogenic phenotype morphologically evident as bulging cells with shortened cytoplasmic processes. Compared to quiescent pericytes, activated pericytes can change their expression profiles , leading to phenotypes that are highly proliferative with the pluripotent ability to differentiate into other pericytes, matrix-forming cells, smooth muscle cells, or adipocytes. Activated pericytes are loosely attached to microvessels and develop cytoplasmic extensions into the tumor parenchyma . Detachment of pericytes from the vessel wall enables ECs to migrate into the surrounding matrix to form new blood vessels, and increased vessel permeability leads to leakage of plasma proteins that serve as a provisional matrix for EC and pericyte migration through interactions between integrins and plasma proteins. Although the role of pericytes in angiogenic sprouting is still unclear, a better understanding of the variable expression of different pericyte markers has revealed the presence of pericytes in the immature vascular plexus, leading to the view that pericyte-mediated signaling may be essential for the growth phase of angiogenesis . Pericytes are also necessary for vessel maturation and quiescence . Indeed, it has been shown that pericyte recruitment is required for vessel stabilization, in which new blood vessels are stabilized through a range of heterotypic contacts and soluble factors that inhibit EC proliferation while providing survival signaling to ECs. A variety of signaling factors mediate perictye–EC interactions, including VEGF, PDGF-beta (PDGFB), and Ang/Tie 2 .
Pericyte-mediated vessel destabilization
During the initial stage of angiogenesis, the stable association between pericytes and ECs in preexisting vasculature is disrupted, leading to pericyte activation and EC proliferation and migration into surrounding tissue. Tumor hypoxia has dramatic effects on pericytes. The hypoxic state of tumors initiates a cascade of cellular signaling that leads to vasodilation via released nitric oxide, increased vascular permeability via vascular endothelial growth factor (VEGF) and angiopoietin 2 (Ang-2), and degradation of the vascular basement membrane and ECM via proteases (Fig. 1). As a result of this signaling cascade, the association between quiescent ECs and pericytes in the perivascular mesenchymal niche is destabilized. VEGF has emerged as a critical mediator of pericyte–EC dissociation. VEGF is secreted by pericytes especially under hypoxic conditions , where pericyte production of VEGF establishes a paracrine loop resulting in EC proliferation . In addition, VEGF with other proangiogenic factors generates vasodilation and increased permeability of the endothelial barrier . Under conditions of PDGF-mediated angiogenesis, VEGF reduces pericyte coverage on nascent vascular sprouts through inhibition of PDGFRB signaling in mural cells. This in turn results in vessel destabilization . Following vessel destabilization, pericytes participate in angiogenesis by directing expression of membrane type 1 metalloproteinase (MT1-MMP) at the migrating tip of newly formed endothelial vessels  and increasing EC motility and cord angiogenesis through expression of NG2 proteoglycan . However, at the site of developing sprouts pericyte investment is scarce .
Pericyte-mediated vessel stabilization
Pericyte coverage is required for the stabilization of immature endothelial tubes . The recruitment of pericytes to newly formed blood vessels was demonstrated in mice coinjected with T241 tumor cells and embryonic stem cells engineered to express the marker protein lacZ in pericytes . Several factors have been implicated in pericyte recruitment to the tumor vessel wall, including sphingosine-1-phosphate, Ang-1, PDGFB, and metalloproteinases . VEGF and other proangiogenic factors induce expression of matrix metalloproteinase 9 (MMP9) in pericytes to promote migration to the tumor vessel wall in mice . Similarly, pericytes are recruited to immature sprouting capillaries and proliferating arteries by PDGFB-expressing ECs , where they are integrated into the vessel wall [21, 22]. At the molecular level, the C-terminal retention motif of PDGFB anchors secreted proteins to the cell surface and ECM, creating a concentration gradient that guides pericytes to EC tubes [23–25]. In vitro evidence indicates that pericyte–EC crosstalk leads to sprout stabilization as a result of mutual expression of tissue inhibitors of metalloproteinase (TIMPs)—TIMP-2 by ECs and TIMP-3 by pericytes—resulting in inhibition of matrix proteolysis by MMPs and vessel stabilization . Inhibition of TIMP expression results in vessel regression in a process dependent on MMPs. Stratman et al.  recently demonstrated that pericyte recruitment to EC-generated vascular guidance tunnels leads to formation of vascular basement membrane matrix and restriction of blood vessel width (through fibronectin). Both cell types contribute to release of fibronectin, nidogen-1, perlecan, laminin isoforms 10 and 11, and increased expression of α5β1, α3β1, α6β1, and α1β1 integrins. This leads to construction of ECM bridging between laminins, nidogens, and collagen type IV, as well as fibronectin, collagen type IV, and perlecan.
Although tumor cell expression of PDGFB leads to pericyte recruitment to the tumor, these pericytes fail to firmly attach to the vessel wall, indicating that PDGFB expression by the endothelium is necessary for proper pericyte attachment to the vessel wall rather than simply needing to be present in the tumor stroma . Genetic deletion of PDGFB or PDGFRB in mice leads to a marked reduction in pericyte coverage of blood vessels, resulting in defective endothelial junctions, endothelial hyperplasia, microvascular leakage, vessel dilation, poor capillary blood flow, and hemorrhage [29, 30]. In a normal physiological setting, a significant reduction in pericyte coverage can be lethal in animal models . Although mutant mice lacking the C-terminal retention motif in PDGFB are viable, they have fewer pericytes, which are poorly integrated into the vessel wall, especially in the retina and kidney . It has been recently demonstrated that PDGF-BB increases tumor pericyte coverage by activation of the stromal-derived factor-1A (SDF-1A)/CXCR4 axis. The ensuing SDF-1A chemotaxis gradient corresponds to PDGF-BB-induced pericyte recruitment during angiogenesis, partially because of increased pericyte motility . Inhibition of PDGFR signaling with the receptor tyrosine kinase inhibitor SU6668 causes pericyte detachment and vessel regression in several tumor models, leading to diminished tumor growth [32–34]. Consistent with this finding, glioma vasculature growing in a dorsal skinfold chamber regresses when it is treated with SU5416, a tyrosine kinase inhibitor of both VEGFR-2 and PDGFRB, in the presence of pericytes . Notably, the ability of pericytes to inhibit the proliferation of adjacent ECs can be disrupted through increased Rho GTPase signaling in pericytes .
Both pericytes and vSMCs express Ang-1, which binds to the Tie-2 receptor on ECs to establish pericyte–EC interactions that inhibit proliferation and stabilize neovessels. In addition, Tie-2 signaling induces ECs to express and secrete endothelial-derived heparin binding epidermal-like growth factor (HB-EGF), which binds to epidermal growth factor receptors to promote further migration of pericytes . In contrast, overexpression of Ang-2, which binds to Tie2 in competition with Ang-1, reduces pericyte coverage and destabilizes vessels within the tumor, even in the presence of continued VEGF stimulation . Inhibition of Tie2 blocks mural cell recruitment to developing vessels and leads to increased expression of MT1-MMP by ECs in the vessel trunk . In addition, transgenic mice overexpressing Ang-2 in the retina develop dense vascular networks with reduced pericyte coverage . These findings indicate that the balance of Ang-1 and Ang-2 signaling functions to regulate pericyte recruitment.
Recent evidence has also implicated pericytes in vessel branching. Several in vitro studies demonstrate that stromal cells promote sprouting and EC tubulogenesis [40–42]. Notably, vascular guidance, branching and development of capillaries networks require pericyte–endothelial cell contact, which could not be saved by exogenous VEGF . In addition, neuropillin-1 is essential for pericyte–EC association during the formation of branching vessels in subcutaneous matrigel plugs and sprouting of intersegmental vessels in developing zebrafish . Expression of Class 3A semaphorin leads to vessel pruning and increased pericyte coverage while also reducing tumor growth and possibly normalizing blood vessels . Although overexpression of delta like-4 ligand results in deceased vessel branching, it also increases fibronectin expression and pericyte coverage . Similarly, the ephrin B4 receptor activates Ang-1/Tie2 signaling between pericytes and ECs, and inhibition of ephrin B4 decreases vascular density and pericyte recruitment [47, 48]. The role of pericytes in tumor vessel stabilization and δ-like 4/Notch 1 signaling was recently explored in mathematical modeling , where it was shown that outward tumor expansion was dependent on pericyte coverage of tumor blood vessels.
Pericyte coverage and dysfunction in different tumor types
Pericyte coverage of blood vessels varies depending on the tumor tissue, with islet carcinomas exhibiting higher pericyte density and glioblastomas and renal cell carcinomas showing lower pericyte density than their respective normal tissues. In addition, tumors with fewer pericytes are usually characterized by dense vascular networks with active EC proliferation . Intriguingly, it has been shown that pancreatic and colorectal cancer cells overexpressing PDGFB produce smaller tumors with increased pericyte recruitment to the tumor vasculature and growth arrest of ECs . Both glioblastomas and fibrosarcomas overexpressing PDGFB exhibit greater pericyte coverage around blood vessels , though it has yet to be studied whether this observation correlates with a less aggressive tumor phenotype. Absence of the PDGFB retention sequence in mice with transplanted murine fibrosarcoma leads to a deficit of pericytes, defective attachment to the vascular wall, hyperdilated tumor microvessels, and hemorrhage . Although pericytes associate with greater than 97% of tumor vessels, the association between pericytes, ECs, and matrix is abnormal [8, 53]. The importance of pericyte coverage may be in part due to survival factors produced by pericytes. Studies in human tumor xenograft models demonstrate that EC survival is dependent on pericyte production of VEGF , with EC death occurring in vessels lacking pericytes and VEGF stimulation . Although glioblastomas have a substantial fraction of blood vessels without pericyte coverage, these tumor vessels are nevertheless dependent on other sources of VEGF and recede when VEGF production by cancer cells is reduced . Besides tumor cells and pericytes, other significant sources of VEGF production in the tumor microenvironment are SMA-expressing myofibroblasts  and carcinoma-associated fibroblasts (CAFs) , which comprise up to 80% of stromal fibroblasts in some tumors . In addition to tumor pericyte coverage, the degree of pericyte investments in the organ of origin or metastatic site may influence the dynamics of tumor growth. Several studies have shown that hepatocellular carcinoma and liver metastases are both aided by a tumor microenvironment in the liver that consists of a large population of specialized pericytes called hepatic stellate cells . Tumor-associated pericytes in the liver are responsible for matrix deposition that limits tumor cell anoikis and promotes invasion [60, 61], which may in part contribute to the liver being the second most common site of metastasis (discussed below).
Pericytes in Tumor Metastasis
Recent studies have shown that blood metastasis is limited by pericytes. Abnormal pericyte integration into the capillary wall along with deficient coverage could be partly responsible for the vessel abnormalities that contribute to metastasis. Immature and leaky vessels in tumors could elevate tumor interstitial fluid pressure and promote the flow of tumor cells into the vessel lumen. Indeed, deficient pericyte coverage of the tumor vasculature in human colorectal carcinoma was associated with an increase in the number of metastases . This finding was confirmed in a mouse model of pancreatic cancer, in which metastatic dissemination of tumor cells was increased in pericyte-deficient animals . Similarly, in a model of prostate cancer, increased pericyte coverage of blood vessels corresponded with decreased frequency of tumor invasion into those vessels . At the molecular level, metastatic dissemination was limited by neural cell adhesion molecule (NCAN), which facilitated the recruitment of pericytes and subsequent assembly of ECM components . In contrast, metastatic colonization may be dependent on pericyte support, in which altered pericytes may assist circulating tumor cells in the process of extravasation . In human liver metastasis, the number of hepatic pericytes (stellate cells) was increased , with evidence indicating that activation of these cells was tumor-dependent  and that these cells, once activated, were involved in the initial establishment of micrometastases and subsequent EC recruitment for generation of a tumor vascular network [67–69]. Thus, these findings indicate that pericytes may function to limit intravasation at the primary tumor site, whereas tumor-coopted pericytes may serve to facilitate extravasation at the metastatic site.
Pericyte-Targeted Therapy for Cancer
Tumor pericytes appear to play a critical role in regulating vessel maturation and function, even in settings where they are less abundant and more loosely attached to vessels than in healthy tissues. Preclinical and clinical studies have largely focused on the role of tumor pericytes in promoting EC survival and stabilizing the tumor vasculature through a variety of signaling networks. As noted earlier, pericyte recruitment to tumor neovessels is dependent on signaling through the PDGFB/PDGFRB and Ang-1/Tie2 networks. PDGFB/PDGFRB signaling also appears to be critical for maintaining the pericyte–EC contacts needed for vessel stabilization. Once recruited to the tumor vessel endothelium, pericytes produce VEGF in the local tumor microenvironment. Therefore, tumor vessels with disrupted pericyte–EC interactions or completely lacking pericyte coverage may be more responsive to drugs targeting ECs. Indeed, most therapeutic strategies have sought to disrupt pericyte–EC interactions or inhibit pericyte recruitment to achieve intratumoral vascular regression.
Antipericyte therapies for inhibiting tumor growth
Antiangiogenic strategies targeting VEGF or its receptor VEGFR2 have been shown to potently prevent neovascularization and the growth of many tumor types. However, targeting VEGF signaling alone is often ineffective at inducing vascular regression or preventing the rapid regrowth of tumor vessels . Pericytes in lung tumor have been shown to directly form a network of tubes in matrigel . In addition, pericytes may protect ECs from VEGF withdrawal, leading to pericyte-mediated resistance to antiangiogenic therapies. Studies have shown that vessels without pericyte coverage are more dependent on VEGF signaling for survival , and VEGF inhibition leads to increased pericyte coverage of the tumor vasculature . Pericytes appear to stabilize blood vessels and provide EC survival signals through the Ang-1/Tie2 pathway . Therefore, by targeting tumor pericytes it may be possible to overcome pericyte-mediated resistance to VEGF pathway inhibition and achieve more effective tumor vessel destabilization through disruption of pericyte–EC association or directly through pericyte loss. Vascular regression could also lead to the normalization of tumor microvessels and the opening of previously collapsed vessels  via decreased interstitial fluid pressure . This possibility would account for how PDGF pathway inhibition leads to improved drug delivery [74–76].
Several studies have tested the effects of combining anticancer agents with antipericyte agents that target PDGF or other pericyte markers, such as NG2 proteoglycan . As predicted, inhibition of only PDGFR resulted in pericyte loss, as well as an increase in VEGF/VEGFR expression and the promotion of angiogenesis, in an animal model of retinopathy . Involvement of the SDF-1A/CXCR4 axis in pericyte recruitment within PDGF-BB–overexpressing tumors suggests that blockade of this axis may provide an additional target in antiangiogenic tumor therapy . Combining PDGFRB tyrosine kinase inhibition with VEGF inhibition more efficiently blocked tumor angiogenesis than VEGF inhibition alone in several experimental models [32–35, 79]. Similarly, PDGF inhibition disrupted pericyte support and sensitized ECs to antiangiogenic chemotherapy, resulting in regression of preexisting tumor vasculature in a mouse model . However, a human clinical trial for renal carcinoma showed that inhibition of both the VEGF and PDGF pathways resulted in no therapeutic benefit when compared to inhibition of the VEGF pathway only; in fact, the combined regimen exhibited toxicity . Given these results, further preclinical studies are needed to clarify the mechanism(s) by which PDGF-targeted agents affect pericyte–EC interactions, and additional clinical studies are needed to clarify the potential benefits and risks associated with antipericyte cancer therapy.
Propericyte therapies for inhibiting tumor growth
At least two alternative therapeutic approaches appear plausible given the role of pericytes in promoting tumor angiogenesis, although both are highly speculative and no proof-of-principle studies have been conducted in animal models. The first approach is to promote excessive pericyte recruitment, thereby causing vessel stabilization and restricted vessel sprouting. This approach may limit tumor angiogenesis in blood vessels with normal pericyte investment of the endothelium and may prevent the dissemination of tumor cells into the circulation by reducing the leakiness of intratumoral blood vessels and, perhaps, by also blocking extravasation of circulating tumor cells. The second approach involves the use of pericyte progenitor cells as a cellular vehicle for gene delivery. This idea is supported by previous work using progenitor ECs [82–84], and more recently pericytes , to delivery antiangiogenic gene therapy. Bone marrow–derived hematopoietic cells expressing the pericyte marker NG2 were identified in close contact with tumor blood vessels in animal models of melanoma , pancreatic islet carcinomas , and brain tumors [19, 88]. Thus, pericyte progenitor cells appear to be recruited to sites of angiogenesis from the bone marrow niche; however, intravenous injection of pericyte progenitor cells may fail to migrate and integrate into the tumor vasculature . Neither of these approaches promoting pericyte recruitment to the tumor vasculature has been tested in preclinical models or clinical trials.
Pericyte–EC interactions involve soluble factors, matrix components, and direct cell–cell contact that are responsible for regulating the normal physiology of the developing and adult vasculature, in addition to the pathophysiology associated with tumor angiogenesis and metastasis. Future research is needed to clarify the role of tumor pericytes in stabilizing the tumor vasculature and the effect of pericyte coverage on limiting tumor angiogenesis and metastasis. More research is also need to better understand in which settings to employ antipericyte or propericyte therapies or which combinations are most effective. In some settings, the goal may be to promote pericyte recruitment and proper pericyte integration into the tumor endothelium to stabilize and normalize tumor blood vessels, with the goal of limiting tumor angiogenesis and metastasis, as well as allowing for more effective delivery of tumor cell cytotoxics. In other settings, however, optimal therapy may be to limit pericyte coverage or disrupt pericyte–EC interactions to destabilize the tumor vasculature and thereby induce vascular regression, especially in combination with antiendothelial agents.