The chemokine system – a major regulator of angiogenesis in health and disease


  • Invited Review.

Mette M. Rosenkilde, Laboratory for Molecular Pharmacology, Department of Pharmacology, The Panum Institute, Copenhagen, Denmark. e-mail


The chemokine system controls leukocyte trafficking during homeostasis as well as during inflammation and is necessary for the linkage between innate and adaptive immunity. Tissue regulation outside the hematopoietic compartment, for instance, angiogenesis, organogenesis and tumor development, growth and metastasis, is another important function of the chemokine system. The chemokine-mediated regulation of angiogenesis is highly sophisticated and fine tuned, and involves pro-angiogenic chemokines, for instance, CXCL8/IL8 interacting with the CXCR2 receptor, and anti-angiogenic (i.e. angiostatic) chemokines, for instance, CXCL10/IP10 interacting with the CXCR3 receptor. Chemokines also regulate angiogenesis in a receptor-independent manner by means of a perturbation of bFGF and VEGF function. The current review focuses on the influence of the chemokine system in angiogenesis. Examples of the delicate angiogenesis regulation by the chemokine system in, for instance, wound healing and of the dysregulation in, for instance, tumor development are provided along with the interesting phenomenon of molecular piracy of host-encoded genes within the chemokine system. This phenomenon is a general strategy to circumvent and exploit the immune system – and thereby improve survival – for many viruses. Yet, a certain group of herpesviruses – the γ2-herpesviruses – encode a functional CXCR2 receptor homolog that is activated by angiogenic chemokines and antagonized by angiostatic chemokines, and this particular gene seems to cause the development of a vascular tumor – Kaposi's sarcoma – in the host.

Chemokines constitute a large family of chemotactic cytokines that exert their action via an interaction with receptors belonging to the superfamily of rhodopsin-like G-protein-coupled 7TM receptors. The chemokine system in humans comprises approximately 50 chemokine ligands ( and approximately 20 chemokine receptors (Fig. 1). The principal target cells are bone-marrow derived and the chemokine system is crucial for the regulation and control of the basal homeostatic and inflammatory leukocyte movement. The functional consequences of chemokine receptor activation are not limited to leukocyte locomotion, but also include degranulation, gene transcription, mitogenic and apoptotic effects (1). Many cell types outside the hematopoietic compartment express chemokine receptors. These include endothelia, smooth muscle cells, stromal cells, neurons and epithelial cells (2), and activation of these cell types extends the functional implications of chemokine receptor activation to other aspects of tissue regulation and homeostasis, such as, for instance, angiogenesis and the morphogenetic movement during organogenesis in addition to tumor development and metastasis (3). The control of leukocyte movement, activation and differentiation provides the chemokine system with a pivotal role in the host immune response against invading pathogens. This is supported by the fact that several viruses induce or encode chemokines, chemokine receptors or chemokine-binding proteins, which in different ways manipulate the immune system and exert local control of, for instance, angiogenesis and cell growth through chemokine mimicry (4, 5).

Figure 1.

The endogenous chemokine system. The inner circle consists of the endogenous chemokine receptors arranged according to their properties as “selective” receptors with one chemokine ligand (yellow), or “redundant” or “shared” with two or more chemokine ligands (red). The “orphan” receptors with structural homology to known chemokine receptors are also included (purple) together with the “non-signaling” chemokine receptors (blue). These are more correctly named chemokine-binding proteins due to the lack of signaling. The outer circle consists of the endogenous chemokines (black) together with the three HHV8-encoded chemokines, vMIP1–3 (italic). vMIP-1 and -3 are selective agonists (green) and vMIP-2 is a broad-spectrum antagonist for many chemokine receptors (red) and an agonist for a few receptors (green). For chemokine nomenclature see Table 1. The figure was kindly provided by Morten Lindow.

Table 1. Chemokine nomenclature – past and future
ClassNew nameOld nameAbbreviation
  1. In 2000 a new chemokine nomenclature was introduced (9). At present, scientific papers use both the old and the new names, which often leads to more confusion than clarity. Some chemokines have several complicated names in addition to the CC or CXC number, and only the most commonly used name for each chemokine is given in the current table.

CXCCXCL1GROαGrowth-related oncogene α
 CXCL2GROβGrowth-related oncogene β
 CXCL3GROγGrowth-related oncogene γ
 CXCL4PF-4Platelet factor-4
 CXCL5ENA-78Epithelial cell-derived neuthrophil-activating factor 78
 CXCL6GCP-2Granulocyte chemoattractant protein-2
 CXCL7NAP-2Neutrophil-activating protein-2
 CXCL9MigMonokine induced by γ-interferon
 CXCL10IP10γ-Interferon -inducible protein-10
 CXCL11ITACInterferon-inducible T-cell α-chemoattractant
 CXCL12SDF-1Stromal cell-derived factor-1
 CXCL13BCAB-cell-activating chemokine 1
 CXCL14BRAKBreast and kidney chemokine
 CCL2MCP-1Monocyte chemoattractant protein-1
 CCL3MIP-1aMacrophage inflammatory protein-1α
 CCL4MIP-1bMacrophage inflammatory protein-1β
 CCL5RANTESRegulated on activation normal T-cell-expressed and secreted
 CCL7MCP-3Monocyte chemoattractant protein-3
 CCL8MCP-2Monocyte chemoattractant protein-2
 CCL13MCP-4Monocyte chemoattractant protein-4
 CCL14HCC-1Hemofiltrate CC-chemokine-1
 CCL15HCC-2Hemofiltrate CC-chemokine-2
 CCL16HCC-4Hemofiltrate CC-chemokine-4
 CCL17TARCThymus- and activation-related chemokine
 CCL18PARCPulmonary- and activation-regulated chemokine
 CCL19ELCEpstein-Barr virus-induced receptor ligand chemokine
 CCL20LARCLiver- and activation-related chemokine
 CCL21SLCSecondary lymphoid tissue chemokine
 CCL22MDCMacrophage-derived chemokine
 CCL23MPIF-1Myeloid progenitor inhibitory factor-1
 CCL24MPIF-2Myeloid progenitor inhibitory factor-2
 CCL25TECKThymus-expressed chemokine
 CCL28MEGMucosa-associated epithelial chemokine
CXCL1 Lymphotactin α
 XCL2 Lymphotactin-β
CX3CCX3CL1 Fractalkine


Functionally, the chemokines can be divided into two groups – the inducible chemokines that recruit leukocytes in response to (patho)-physiological events (also called inflammatory chemokines) and the constitutively expressed chemokines responsible for the basal leukocyte trafficking and the architecture of secondary lymphoid organs (also called homeostatic chemokines). Structurally, the chemokines – in their monomeric forms – are small proteins with a molecular weight of 8–12 kD; however, many chemokines form dimers or higher order oligomers (6). The majority of the chemokines are basic proteins and the corresponding binding pockets/areas in the receptors are acidic (7). Besides the interaction with the other chemokines (dimerization/oligomerization) and with the receptors, the chemokines also interact with glycoaminoglycans (GAGs) on the surface of the cells, and the haptotactic gradient, formed by the immobilization of the chemokines on GAGs is essential for the directional movement (chemotaxis) of chemokine receptor-expressing cells (8). Dimerization/oligomerization and GAG interactions are essential events for chemokine action in vivo, whereas obligative monomeric or GAG-deficient chemokine mutants retain their activity (by means of receptor activation) in vitro– but not in vivo (6). The chemokines are divided into two main families based on the presence or absence of an amino acid residue between the two first of four conserved cysteines: CXC chemokines (e.g. CXCL8/IL-81 and CXCL1/GROα) and CC chemokines (e.g. CCL5/RANTES and CCL11/eotaxin). The CXC chemokines are subdivided based on the presence of a Glu-Leu-Arg (ELR) motif close to the first conserved cysteine. ELR-containing chemokines (CXCL1–8) are induced under acute and chronic inflammatory conditions and preferentially attract neutrophils, whereas non-ELR chemokines (CXCL9–12) have a more constitutive expression pattern and function on lymphocytes and cells outside the hematopoietic compartment. The CC chemokines (CCL1–28) influence most of the cells within the lymphoid and myeloid compartments (monocytes, eosinophils, basophils and macrophages) during inflammation as well as during homeostasis. Two additional minor chemokine families are represented by CX3CL1/fraktalkine with three residues between the two cysteines, and by XCL1 that lacks one of the two first cysteines. A special refinement is found in CX3CL1/fractalkine and in CXCL16, where the chemokine domain is anchored to a transmembrane α-helix in the C-terminus by a large mucin stalk that comprises 40–75% of the molecule. This enables the chemokine to remain attached to the membrane or to be shed from the cell by enzymatic cleavage (10).

The receptors are divided into CXCR1–6, CCR1–11, CX3CR1 and XCR1. In addition, a few non-signaling chemokine receptors2 exist (11) as well as some orphan receptors with structural homology to existing chemokine receptors. The ligand-receptor interaction in the chemokine system is characterized by great promiscuity and redundancy for some receptors and ligands, and by high specificity for other receptors and ligands (Fig. 1). An example of a highly specific monogamous receptor-ligand pair is the SDF-1/CXCL12 interaction solely with the CXCR4 receptor as the only ligand for this receptor, whereas great promiscuity is found among the majority of the receptor-ligand interactions. For instance, there is the interaction of CXCL9–11 (Mig, IP10 and ITAC) with CXCR3 and the interaction of CXCL1–3/GROα, β,γ and CXCL5–8/ENA78, GCP-2, NAP-2 and IL8 with CXCR2, where CXCL6 and -8/GCP-2 and IL8, in addition, bind to the CXCR1 receptor. The promiscuity and redundancy in the receptor-ligand interaction are also reflected in the (lack of) phenotypes observed in knock-out (KO) animals with the majority of chemokines and their receptors. Thus, for the promiscuous chemokines and chemokine receptors the KO animals only reveal specific phenotypes related to the absent gene(s) in situations where they are challenged with a certain pathogen (12). Yet for the monogamous (e.g. CXCR4-CXCL12/SDF-1) or “narrower” receptor-ligand pairs (e.g. CCR7 and CXCR5) (see Fig. 1) highly specific phenotypes occur in the KO animals (13–17). Thus, CCR7 and CXCR5 are both essential for the organization and maturation of the secondary lymphoid organs (15, 16), yet their absence is not lethal in utero. In contrast, the absence of either the ligand CXCL12/SDF-1 or its receptor CXCR4 is lethal in utero, and the KO animals reveal pivotal roles for carcinogenesis, vascular development, myelopoiesis, and neuronal development for this monogamous pair (13, 14, 17).

The 7TM chemokine receptors couple to G-proteins upon activation (7TM receptors are alternatively known as GPCRs (G-protein-coupled receptors)) and the main signal transduction pathway (through Gi) is a rapid mobilization of intracellular calcium, which is associated with induction of chemotaxis. It has been shown that the βγ subunits released from activated Gαi, mediate the cell migration through an activation of PI3Kγ (18–20). The downstream events include activation of the three different members of the MAP-kinase superfamily as being important for proliferation and adhesion molecule expression (ERK1/ERK2 and JNK/SAPK) and for chemotaxis (p38) (21–25). The signaling also includes the tyrosine kinases, e.g. activation of related adhesion focal tyrosine kinase for regulation of JNK/SAPK activity, and of Src-kinases linking receptor signaling to activation of small GTP-ases (26). Chemokine receptor dimerization seems to be important for at least some of the reported activities. For recent reviews of chemokine receptor signaling and dimerization see (1, 27).


Angiogenesis, characterized by the neoformation of blood vessels, is essential for a number of physiological and pathophysiological events, such as embryonic development, wound healing (the formation of inflammatory granulation tissue), chronic inflammation and growth of malignant tumors (28–30). All of these aspects of angiogenesis are influenced by the chemokine system through several different mechanisms which will be highlighted below.

CXC-chemokine interaction with chemokine receptors expressed on endothelial cells

The division of the CXC chemokines based on the presence, respectively absence, of the ELR motif has tremendous physiological implications in terms of the effect on angiogenesis, since ELR CXC chemokines have been shown to be angiogenic, whereas most of the non-ELR chemokines are anti-angiogenic (angiostatic) (Fig. 2). The first angiogenic chemokine to be described was CXCL8/IL8 back in 1992 (31). This chemokine is a strong inducer of angiogenesis, equivalent on a molar basis to, for instance, basic fibroblast growth factor (bFGF) (Fig. 3). The first – and to date the most powerful – angiostatic chemokine, CXCL4/PF-4, was described as the first member of the chemokine family already in 1990 (32). Since then most of the other non-ELR chemokines besides CXCL4/PF-4 – CXCL9–11/Mig, IP10 and ITAC – have been described as angiostatic, both with respect to ELR-chemokine-induced angiogenesis and chemokine-receptor-independent angiogenesis by, for instance, bFGF (33–35) (Fig. 3). The effect of the chemokines on angiogenesis does not require preceding inflammation and has been characterized in vitro by, for instance, the endothelial cell chemotaxis assay (33, 34, 36, 37) or the chick chorioallantoic membrane assay (38, 39), and in vivo by means of the rat corneal micropocket vascularization assay (Fig. 3) (33, 37). From a pathophysiological point of view, the chemokine regulation of angiogenesis seems to be very important in, for instance, tumor formation and growth, as described below. The molecular mechanism behind the non-ELR CXC-chemokine-mediated anti-angiogenic (or angiostatic) effects is based upon chemokine interaction with the chemokine receptor(s), and upon chemokine perturbation of growth factor function (40) (see next section). The receptor-dependent angiostatic mechanisms are mainly driven by the interaction of the non-ELR chemokines CXCL9–11/Mig, IP10 and ITAC with the CXCR3 receptor (33, 35, 41). The strong angiostatic action of CXCL4/PF4 and the targeting in vivo to endothelial cells that undergo active angiogenesis has been known for more than a decade (42, 43). Yet the target receptor was unknown until very recently, when Romagnani et al. described a splice variant of CXCR3 (CXCR3B) as a putative target for CXCL4/PF4, in addition to the three known CXCR3 ligands (44). On endothelial cells, both splice variants of the CXCR3 receptors are mainly expressed in human microvascular endothelial cells (HMVECs) (44–46), but are also found in small vessels, where the expression is enhanced during inflammation and neoplasm – presumably reflecting a self-regulatory mechanism for avoiding excessive angiogenesis (40, 46). Another recently described angiostatic (non-ELR) chemokine CXCL13 (47) mainly interacts with the CXCR5 receptor (48); however, it is not known whether the angiostatic effect is mediated through an interaction with CXCR5 or with CXCR3 (49). CXCL12/SDF-1 belongs to the non-ELR CXC-chemokines and has in a single case been reported to be angiostatic with respect to VEGF-induced angiogenesis (50), consistent with the rest of the non-ELR CXC chemokines (33–35). Yet very recent evidence suggests that the overall effect of CXCL12/SDF and its receptor CXCR4 is angiogenic, since CXCL12/SDF-1 stimulates chemotaxis and proliferation of endothelial cells in addition to endothelial tube formation in a MAP-kinase (Erk1/Erk2) and PI3K-dependent manner (51, 52). The angiogenic effect is partly mediated through an induction of VEGF secretion by CXCL12/SDF-1. VEGF can, in turn, upregulate the CXCR4 receptor expression on endothelial cell surfaces, indicating that CXCL12/SDF and VEGF might act in an additive or synergistic fashion to amplify angiogenesis. The activation of angiogenesis by the CXCL12/SDF-CXCR4 axis is regulated by the proinflammatory cytokines tumor necrosis factor α (TNFα) and interferon γ (IFNγ), and neutralizing antibodies against either CXCL12/SDF or CXCR4 inhibit angiogenesis (52, 53). Thus, variable roles for the non-ELR CXC-chemokine CXCL12/SDF-1 have been reported and these differences could very well reflect tissue-specific endothelial microvascular heterogeneity (54–56).

Figure 2.

Role of CXC chemokines in the angiogenic balance. Angiogenesis is orchestrated by a variety of different effectors – hereunder the chemokine system with focus on the endogenous CXC chemokines. Angiogenic chemokines are listed in the green weight, whereas the angiostatic chemokines are listed in the red weight. Non-ELR CXC chemokines are highlighted in bold italics. An endothelial cell is shown above the weights with receptors for the angiogenic and angiostatic chemokines indicated in green and red, respectively.

Figure 3.

Angiogenic effect of CXCL8/IL8 and angiostatic effect of CXCL10. Rat cornea neovascularization in response to ELR and non-ELR CXC-chemokines as adapted from (33, 35). Angiogenic effect of 10 nM bFGF in the absence (A) or presence of 10 nM CXCL10/IP10 (B). Anti-CXCR3 antibody reverses the inhibitory effect of CXCL10/IP10 on the bFGF-induced angiogenesis (C). Angiogenic effect of CXCL8/IL8 (10 nM) in the absence (D) or presence of 10 nM CXCL10/IP10 (E). Anti-CXCR3 antibody reverses the inhibitory effect of CXCL10/IP10 on the CXCL8/IL8-induced angiogenesis (F). 25×magnification. The pictures were kindly provided by Robert Strieter.

Chemokine receptor-independent effects on angiogenesis

Some chemokines have been shown to interfere with the angiogenic effect of bFGF or VEGF in a chemokine receptor-independent manner. Several mechanisms have been suggested for this inhibition, yet the most important seems to be perturbation of the growth factor interaction with glycosaminoglycans (GAGs), especially heparin sulfate proteoglycans. The GAG binding contributes to the high-affinity binding to the receptors by stabilizing the growth factor/receptor complex, protecting the growth factors from degradation, and facilitating growth factor oligomerization/dimerization (57). Two non-ELR CXC-chemokines, CXCL4/PF-4 and CXCL10/IP10, have been shown to inhibit the interaction of VEGF and bFGF with their appropriate receptors by interference with the GAG binding (57, 58) and thereby also the dimerization (59, 60). Although several studies (57–60) have demonstrated the chemokine receptor-independent angiostatic effects of the chemokines to be based upon perturbation of the growth factor function, evidence points towards other additional mechanisms. For instance, it has been described how CXCL4/PF-4 inhibits the function of VEGF-121 (an isoform with deficient heparin-binding ability) (58). Another study has described that CXCL8/IL8 prevents bFGF function in a protein kinase C (PKC)-dependent noncompetitive manner – with no influence on the binding of bFGF to either the receptor or to the GAGs, but instead with a PKC-dependent downregulation of the bFGF receptors (61). In fact, not only CXCL8/IL8, but also other ELR CXC chemokines – in addition to non-ELR CXC chemokines and even a few CC chemokines – have been shown to inhibit endothelial cell proliferation via this mechanism (61). Thus, even though the ELR CXC chemokines (e.g. CXCL8/IL8) have been shown to be angiogenic upon interaction with the appropriate chemokine receptors expressed on endothelial cell surface as described above (33, 34, 37, 62), the receptor-independent effect of these chemokines on angiogenesis seems to be independent of the ELR motif, and also to include the CC chemokines. Endothelial cells are characterized by a large degree of heterogeneity dependent upon tissue types, species and vessel caliber (54–56), and it is therefore difficult to predict which effect of CXCL8/IL8 (and the other ELR CXC chemokines) will be dominant.


The fine-tuned regulation of angiogenesis by the chemokine system is exemplified in the wound healing or tissue repair process. The initial repair process is initiated immediately after injury of a blood vessel through the local release of a large number of growth factors (VEGF, PDGF), cytokines and ELR CXC chemokines (CXCL1/GROα, CXCL5/ENA78 and CXCL7/NAP-2) from activated platelets. These chemokines mediate neutrophil recruitment, i.e. diapedesis and accumulation of neutrophils in the tissue. Also the non-ELR CXC-chemokine CXCL4/PF-4 is secreted in large amounts, yet this chemokine acts very differently in the tissue repair process, with a low chemotactic potential but a powerful procoagulant function in the initial phase (63–65). Angiogenesis is stimulated concomitantly with the neutrophil recruitment (between day 1 and 4), resulting in granulation tissue formation (40). The importance of CXCR2 (and its interaction with ELR CXC chemokines) in wound healing is supported by the notion that CXCR2-deficient mice – or wild-type mice treated with the CXCR2-specific non-peptide antagonist SB225002 (66, 67) – have delayed wound healing (46, 66, 68). Concomitantly with granulation tissue formation – and after day 4 – the expression of non-ELR CXC-chemokines, especially CXCL9–11/Mig, IP10 and ITAC, increases (in part stimulated by IFNγ and TNFα). Also, the CXCR3 receptor expression is increased during inflammation and the CXCL9–11/CXCR3 biological axis hinders uncontrolled angiogenesis by inhibiting the growth and migration of proliferating endothelial cells (40, 46). In addition, the non ELR CXC chemokines mediate migration and proliferation of CXCR3-expressing pericytes around nascent vessels and promote chemotaxis of a large number of T-lymphocytes (40).


Tumor growth and angiogenesis are closely connected. The impact of angiogenesis on tumor development and growth, and the reciprocal situation, i.e. the impact of the tumor on the microvasculature, has been studied in great detail, as shown in recent reviews (69). The central part played by the CXC chemokines in angiogenesis provides these cytokines with essential roles in the dysregulation of angiogenesis necessary for tumor development. Several studies support this notion. For instance it has been shown that the expression of angiogenic CXC chemokines (e.g. CXCL8/IL8) in NSCLC (human non-small-cell lung cancer) cells in SCID mice enhanced tumor growth (70) and that CXCL8/IL8 expression in humans was correlated with an increase in tumorigenesis of bronchogenic carcinomas (36). In contrast, neutralizing antibodies against CXCL8/IL8 attenuated tumor growth and suggested that CXCL8/IL8 accounted for 40–80% of angiogenesis of the bronchogenic carcinomas (36, 70). In fact, increased expression of ELR CXC chemokines (e.g. CXCL8/IL8, CXCL6/GCP2, CXCL1/GROα and the murine CXCL1/GROα analog KC) has been seen to correlate with increases in angiogenesis in several tumors, e.g. glioblastomas and different lung cancers (Lewis Lung Cancer (LLC), squamous cell carcinoma (SCC) and other non-small-cell lung cancers (NSCLC)) (71–76). The chemokine expression is initially driven by inflammatory stimuli and later by the hypoxia in the tumor microenvironment (71). The CXCR2 receptor is the main target receptor for the angiogenic chemokines. This is supported by the notion that CXCR2 −/− mice had significantly reduced tumor growth in a murine LLC tumor model system and CXCR2 +/+ mice treated with a neutralizing CXCR2 antibody showed the same phenotype (73).

The non-ELR CXC chemokines have also been studied extensively with respect to expression and influence on tumor growth, and as expected from their angiostatic effects, these chemokines reduce tumor growth. For instance, the expression (or intratumor reconstitution) of CXCL10/IP10 has been shown to inhibit NSCLC-derived angiogenesis, tumor growth, and spontaneous metastasis in a model of human NSCLC in SCID mice, whereas neutralizing antibodies against CXCL10/IP10 resulted in enhanced tumor-derived angiogenic activity (35). Similarly, expression – or local administration – of CXCL4/PF-4 reduced the tumorigenesis of melanoma or colon carcinoma cells (43, 77, 78).

The influence of the chemokines on angiogenesis is therefore extremely important during oncogenesis. Yet the chemokine system also influences oncogenesis by means of at least three other mechanisms. First of all, by controlling leukocyte infiltration. In 1862, Virchow described the presence of leukocytes in cancer (79). Leukocytes may contribute to tumor survival as sources of growth- and angiogenic factors, as shown for certain CC chemokines (80, 81). Meanwhile, they could also represent attempts to reject the tumor. Thus, many chemokines have been used in animal models of experimental immunotherapy to elicit tumor-specific immune responses for tumor rejection (82, 83). For recent review see (84). Secondly, several chemokines act as growth factors for tumor cells, such as, for instance, CXCL1/GROα and CXCL8/IL8, that stimulate melanomas and bronchogenic carcimonas, respectively (85–87). Thirdly, the chemokine system influences tumor cell migration and metastasis. Certain tumors have been shown to express chemokine receptors, e.g. CXCR4 and CCR7 on the surface of breast cancer cells and CCR10 on malignant melanoma cells (88). The corresponding ligands are found in high concentrations at the site of the metastasis and the chemokine gradient towards the site of the metastasis has been shown to “drive” the tumor-cell migration (88). The importance of the CXCL12-CXCR4 biological axis for metastasis is supported by the notion that also NSCLC cell migration and metastasis is driven by this receptor-ligand pair (89, 90).


The molecular mimicry of viruses and certain parasites deserves further attention in the context of angiogenesis, since a large group of the virus-encoded chemokines and chemokine receptors are functional homologs to the (angiogenic) ELR CXC chemokines and the CXCR2 receptor. In general, both chemokines and chemokine receptors are found in the genome of large DNA viruses belonging to the herpes- and poxviruses (4, 5, 25). These chemokine mimics corrupt and manipulate the immune system and also contribute to control of the local environment by influencing angiogenesis, cellular growth and differentiation. To date, more than 30 different virus-encoded chemokines, chemokine-binding proteins and chemokine receptor have been identified. They are believed to have been incorporated in the viral genomes by an ancient act of molecular piracy, and the high conservation through the different virus strains strengthens the importance of these for the virus life cycle. Some ligands and receptors are still orphans, yet most have been characterized functionally and structurally, as recently reviewed in (4, 5, 25). In connection with the issue of chemokines and angiogenesis, a particular virus – the human herpesvirus 8 (HHV8) or Kaposis sarcoma- associated herpesvirus (KSHV) – is particularly interesting. This virus belongs to the family of γ2-herpesviruses, and homologs of HHV8 are found in other species, e.g. murine γ-herpesvirus 68 (MHV68) and Herpesvirus saimiri (HVS). HHV8 gives rise to at least three different diseases in humans: Kaposis sarcoma, from which the virus was originally cloned (91), multicentric Castleman's disease, and primary effusion lymphoma. Kaposi's sarcoma results from multicentric hyperplasia, and not metastatic neoplasia (92). It is a highly vascularized tumor characterized by spindle cells, vessels and infiltrating leukocytes. All γ2-herpesviruses encode a 7TM chemokine receptor located in the open reading frame 74 (ORF74). The ORF74 receptors are functional homologs to CXCR2, although the structural similarities are surprisingly few. Thus, they all bind ELR CXC chemokines as agonists, whereas some of them, in addition, bind non-ELR CXC chemokines as antagonists (93–95). Constitutive activity is a trait of most of the virus-encoded chemokine receptors – also among the majority of the ORF74 receptors, as reviewed in (25). Thus, ORF74 from HHV8 is highly constitutively active and the non-ELR CXC-chemokines function in this receptor as inverse agonists, in addition to antagonizing the agonist-stimulated receptor (93, 96). In vitro and in vivo experiments have strongly suggested ORF74-HHV8 as the causative agent of Kaposi's sarcoma (Fig. 4). Thus, in vitro studies of receptor signaling pattern have revealed that the constitutive as well as ligand-mediated activation of ORF74-HHV8 induces various growth factors (e.g. VEGF), angiogenic and pro-inflammatory cytokines (97–99). The increase in VEGF expression results in proliferation of transfected cells and induction of angiogenesis of human endothelial umbilical vein endothelial cells (HUVECs) (97, 100). Injection of ORF74-HHV8-expressing fibroblasts into the flank of nude mice results in vascularized tumors at the site of injection (100), and, most significantly, transgenic mice expressing ORF74-HHV8, either ubiquitously in endothelial cells or within the haematopoietic compartment, develop Kaposi's sarcoma-like lesions with macroscopic and histologic resemblance to Kapsosi's sarcoma in humans (101–103). Importantly, the lesions in the mice were shown to be dependent upon the ligand binding and constitutive activity of the ORF74-HHV8 receptor (103). A further refinement with respect to angiogenesis modulation by HHV8 exists, since this virus – as the only γ2-herpesvirus – encodes three angiogenic CC chemokines, vMIP1–3 (38, 39). vMIP-2 is a broad-spectrum chemokine antagonstist (104), whereas the other two are selective agonists for CCR4 and CCR8, respectively (39, 105). Together, vMIP1–3 act as immune modulators that push the immune response in a Th2 direction, thereby preventing an efficient antiviral Th1-mediated response in the host. In summary, HHV8 have found elegant ways of regulating angiogenesis – by means of chemokine as well as chemokine receptor expression. The CXCR2 resemblance to ORF74-HHV8 and the activation of this receptor either constitutively or by angiogenic ELR CXC chemokines and inhibition by angiostatic non-ELR chemokines elegantly mimic the chemokine receptor-dependent regulation of angiogenesis in humans, whereas the angiogenic effects of vMIP1–3 seem less obviously, in the light of the known chemokine receptors, to be involved in angiogenesis and anti-angiogenesis (see above) and the known receptor-interaction pattern of vMIP1–3 (39, 104, 105) (Fig. 1). Thus, most likely the angiogenesis induced by vMIP1–3 – originally and to date exclusively shown by means of the chick chorioallantoic membrane assay – depends upon other mechanisms (receptor-dependent or possibly receptor-independent).

Figure 4.

Human herpesvirus 8 encodes a homolog of CXCR2 with strong angiogenic properties. Open reading frame 74 (ORF74) from HHV8 encodes a broad-spectrum CXC chemokine receptor with strong ligand-independent (constitutive) signaling. Growth factors (e.g. VEGF), angiogenic and pro-inflammatory cytokines are released upon activation, and the increase in VEGF stimulates angiogenesis (97–100). Transgenic mice expressing ORF74-HHV8 develop Kaposi's sarcoma-like lesions with macroscopic (A) and histologic (C) resemblance to Kapsosi's sarcoma in humans (B & D) (101–103). The histology of Kaposi's sarcoma is characterized by an angioproliferation with slit-like vascular spaces containing erythrocytes surrounded by spindle cells and infiltrating inflammatory cells (92) (D) and the transgenic mice show the same pathology (C). The mouse pictures are from a transgenic mouse expressing ORF74-HHV8 under control of a CD2 promoter (101) and were kindly provided by Sergio Lira.


The receptor-dependent and -independent mechanisms by which the chemokine system influences angiogenesis have been reviewed above. Not only have the CXC chemokines paramount roles in inflammatory reactions and in angiogenesis, but also in the coordinated fine-tuned regulation of both processes, as exemplified during wound healing. Lack of proper regulation of this complex chemokine network may result in chronic inflammatory reactions, and even more important may contribute to tumor development. Together with the direct influence of chemokines as growth factors on tumor cells, and as implicated in migration/metastasis, the chemokine system plays pivotal roles respecting several aspects of oncogenesis. The fact that this chemokine network is efficient in angiogenesis is strongly supported by the notion that γ2-herpesviruses (e.g. HHV8) have picked one of the key receptors in angiogenesis control (CXCR2), modified it in very many rounds of mutation (with virus survival as outcome), and used it to create highly vascularized tumors in the host. Despite the fact that the mechanistic background seems well characterized for the chemokine system-mediated regulation of angiogenesis, there is still a long way from the lab bench to the bedside. 7TM receptors per se are excellent drug targets and, as such, chemokine receptors should also be easy targets. Yet, the redundancy and promiscuity in the chemokine system makes it very difficult to predict the outcome of selective chemokine receptor blockade (or activation) by non-peptide compounds (or peptides). Despite these difficulties, intense efforts are underway to identify non-peptide antagonists for many chemokine receptors. Thus, selective antagonists have currently been presented against, for instance, CXCR2 (67) and CXCR4 (106, 107). These have not yet been tested as angiogenesis modulators and more studies will be needed to uncover the pharmacological and thereby clinical applications of chemokine-mediated control of angiogenesis.


  1. 1 In this review chemokine names will be given according to the “International Union of Pharmacology Nomenclature for Chemokines and Chemokine Receptors” (9) followed by their “old” names (Table 1).

  2. 2 The non-signaling chemokine “receptors” D6 and DARC are more correctly classified as chemokine-binding 7TM proteins according to (9).