The chemokine system and cancer


  • Frances R Balkwill

    Corresponding author
    1. Centre for Cancer and Inflammation, Barts Cancer Institute, London, UK
    • Centre for Cancer and Inflammation, Barts Cancer Institute, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London EC1M 6BQ, UK.
    Search for more papers by this author

  • Conflict of interest. The author has a current research project in her laboratory relating to a chemokine receptor that is funded by Affitech AS.


Chemokines (chemo-attractant cytokines) are a group of small proteins that act together with their cell surface receptors, in development, normal physiology and immune responses, to direct cells to specific locations throughout the body. Cancer cells acquire the ability to subvert the chemokine system, such that these molecules and their receptors become important regulators of cell movement into and out of the tumour microenvironment and major players in cancer biology. Copyright © 2011 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.


Chemokines (chemo-attractant cytokines) are a group of small proteins that act together with their cell surface receptors, in development, normal physiology and immune responses, to direct cells to specific locations throughout the body 1, 2. Gradients of extracellular matrix-bound or soluble chemokines control leukocyte migration and positioning within tissues and direct their patterns of recirculation by inducing extra- or intravasation. They may also control movement of other cell types, such as adult stem cells and endothelial cells.

The chemokine system evolved with the vertebrates and there are nearly 50 human genes that encode chemokine ligands, with more than 20 corresponding human chemokine receptor genes, the latter being seven-transmembrane G-protein-coupled receptors, GPCRs. Chemokines are divided into four different groups, CXC, CC, CX3C or C, depending on the position of the conserved cysteine residue, and receptor nomenclature essentially follows that of the chemokines, ie CC chemokines bind to CC chemokine receptors, CXC ligands bind to CXC receptors (see Figure 1 for a summary) 3.

Figure 1.

The chemokine wheel. This cartoon shows major constituents of the chemokine systems, the G-protein-coupled cell-surface receptors and the chemokine ligands. The ‘inflammatory’ chemokines are inducible and involved in all aspects of the immune response. The ‘homeostatic’ chemokines are involved in development and normal physiology. Atypical chemokine receptors are generally ‘silent’ and act as negative regulators of the systems, ‘decoys’ that reduce chemokine levels. Viral chemokines and receptors allow the pathogens to modulate immune responses to infection. Inflammatory, homeostatic and atypical chemokine receptors are all found in the tumour microenvironment. Another chemokine receptor, CXCR7, which also binds to CXCL12 and CXCL11, is shown in the homeostatic group, but there is some evidence that it is an atypical chemokine receptor

The fact that there are more chemokines than receptors, and that some chemokines can bind to several different chemokine receptors, gives an impression of redundancy 2. However, as we learn more about the chemokine system and the complexity of leukocyte subtypes that bear different chemokine receptors, it is clear that this is an intricate and tightly regulated system with spatial and temporal control giving specificity to the receptor–ligand interactions 1, 2. The chemokine system has also evolved to combat many different infectious agents that in turn have developed mechanisms to subvert chemokine function. Many of the larger viruses, for instance, have chemokine or chemokine receptor-like molecules that help them inhibit host responses 4. This is probably another reason for the apparent complexity of the chemokine system.

In cancer, chemokines and their receptors are important for cell trafficking into and out of the tumour microenvironment 5, 6. For instance, chemokines made by malignant and stromal cells contribute to the extent and phenotype of the tumour-associated leukocyte component, to angiogenesis and to the generation of the fibroblast stroma 6, 7. These stromal chemokines may also directly aid survival of the malignant cells 5, 6. This is because malignant cells gain functional chemokine receptors that are often not found on their normal counterparts. These receptors may contribute to metastatic activity; malignant cells become like leukocytes, able to respond to chemokine gradients at sites of metastasis 8, 9.

This review provides a perspective of the chemokine system, then focuses on the ways that chemokines and their receptors influence the composition of the tumour microenvironment and metastatic spread. Finally, the potential of the chemokine system as a therapeutic target in cancer is considered.

Chemokine receptor signalling

The binding of a chemokine to its GPCR receptor activates a series of downstream effectors that facilitate internalization of the receptor and signal transduction 1. This leads to two main responses: integrin activation, which causes adhesion of cells, and polarization of the actin cytoskeleton. The latter allows directional sensing, cell polarization, accumulation of small GTPases, Rac, Cdc42 and PI3K at the leading edge, resulting in actin polymerization and F-actin formation. PTEN tyrosine/PIP3 phosphatase is found at the edges and posterior of the migrating cells, RhoGTPase and its effectors accumulate at the trailing edge. Taken together, these changes cause actomyosin contraction and tail retraction—and the cell migrates.

Chemokine GPCRs can homo- or heterodimerize; this has significant functional importance and adds to the complexity of the system. Another feature of chemokine receptors that is significant to their role in cancer biology is transactivation of other signalling systems, eg tyrosine kinase receptors and the Jak–STAT pathway.

An interesting insight into the dual actions of chemokines on adhesion and migration comes from a recent time-lapse video microscopy study of live CCR7-expressing dendritic cells (DCs) responding to the CCR7 ligands CCL19 and 21 on cryosections of mouse lymph nodes 10. Distinct cellular responses depended on the mode of presentation of the chemokine. The DCs adhered to CCL21 bound to the extracellular matrix, and exhibited random migration. However, the same DCs were able to truncate anchoring residues of CCL21, releasing it from the solid phase. Soluble CCL21 showed functional resemblance to the other CCR7 ligand CCL19 that does not anchor to the extracellular matrix. Both soluble chemokine ligands triggered directed chemotactic movement but not adhesion. This elegant study (viewing of the supplementary movies in this paper is recommended) shows that, at least in secondary lymph nodes, a combination of adhesive random migration and directional steering is mediated by chemokines to give the dynamic but spatially restricted patterns of locomotion 10.

Some chemokine receptors are labelled ‘atypical’ (see Figure 1). D6 and Darc are well-characterized members of this group. CXCR7, shown in the ‘homeostatic’ group of Figure 1, also has features of an atypical chemokine receptor. The major function of this subgroup may be to sequester chemokines, thus acting to dampen responses 11, 12, but they may also have some atypical signalling activity 13.

Chemokines in development and physiology

Chemokines and their receptors are very important in development of vertebrate embryos (reviewed in 1). Chemokine signalling between trophoblast and endometrium may be important in successful implantation of the embryo. The chemokine receptor CXCR4 guides gonadal stem cells in the developing zebrafish embryo 14 and both CXCR4 and CXCR7, together with their CXCL12 ligand, direct homing of neuronal precursors to the relevant area of the developing brain.

As summarized by Zlotnik et al9 in their recent review, chemokines play other important roles in organogenesis. They direct stem cells to the site of organ development, promote angiogenesis, maintain stem cells in a niche until they are needed to produce progeny, and provide a framework for organ development. As discussed later, these roles appear to have been corrupted during cancer development and metastases.

Several chemokines and their receptors are critical to the development of secondary lymph node tissues. Especially important are CCR7 and its ligands CCL19 and CCL21 (discussed above in DC migration in adult lymph nodes), and CXCR5 and its ligand CXCL13. CCR7 is needed for recruitment of naïve T cells into lymph nodes and CXCR5 for B cells. Other chemokine receptors recruit leukocyte subsets to different areas of the body: CCL25 recruits CCR9+ progenitor T cells to the thymus and α4b7 + CCR9+ T cells to the intestine; whereas CCL27 recruits a CCR10+ subset of T cells to the skin and CCL28 in the mammary gland recruits CCR10+ IgA+ plasmablasts. Zlotnik et al9 proposed a model of ‘cellular highways’ by which cells that express specific chemokine receptors are directed to specific locations in the body where their ligands are found.

All the chemokines/receptors described in this section can be classified as homoeostatic—playing an important role in the genesis and organization of the immune system and other organs by recruiting precursor cells, promoting tissue formation, then maintaining the trafficking of the leukocyte subtypes, and probably other cells such as adult stem cells, to the relevant healthy organ.

Inflammatory chemokines

While there is certainly some overlap in classification, the other functional subgroup of chemokines is referred to as the inflammatory chemokines, induced by cytokines during the processes of acute and chronic inflammation, as well as during adaptive immune responses. In general, CXC chemokines are attractants for neutrophils and B and T lymphocytes, while CC chemokines are attractants for neutrophils, basophils, mast cells, cells of the myeloid lineage, DCs and NK cells 15.

During inflammation, pathogens or tissue damage trigger pattern recognition receptors, such as Toll-like receptors in epithelial and dendritic cells, and this induces cytokine production. These cytokines in turn induce and amplify chemokine production, generally of the inflammatory chemokine class. Both homeostatic and inflammatory chemokines are involved in the adaptive immune response. Apart from those homeostatic chemokines mentioned above that control lymphocyte and DC migration in lymph nodes, inflammatory chemokines recruit immature DCs to the site of immune responses.

Oncogenic change and the chemokine system

Why do cancers have such complex chemokine networks? Chronic inflammatory diseases such as prostatitis, hepatitis and pancreatitis certainly predispose to cancers, and cancers arising in such complex inflammatory environments might be expected to have a ‘built-in’ chemokine environment caused by the tissue damage 16. However, not only does inflammation cause cancer, but also cancers can be said to cause inflammation in that the de novo production of inflammatory mediators and certain chemokines, as well as their receptors, is downstream of oncogenic change 6 (summarized in Table 1). The first evidence for this came from Mantovani and colleagues [17] and concerned papillary thyroid carcinoma, PTC. Rearrangement of the RET tyrosine kinase (RET/PTC) in thyrocytes represents a frequent, early, causative and sufficient genetic event in the pathogenesis of PTC. In primary human thyrocytes, RET/PTC activates an inflammatory programme 17. The transcriptome of RET/PTC-activated cells includes a number of different inflammatory mediators as well as the chemokines CCL2 and CCL20, known to attract monocytes and dendritic cells, and chemokines implicated in angiogenesis, such as CXCL8. In addition, RET/PTC activation induces the chemokine receptor CXCR4 on the transformed thyrocytes. Key elements of the RET/PTC-activated inflammatory programme were found in biopsy specimens, and patients with lymph node metastasis showed higher levels of the inflammatory molecules in their primary tumours 17. Hence, this early, causative and sufficient genetic event involved in the pathogenesis of a human tumour directly promotes an inflammatory microenvironment 17, 18.

Table 1. Some oncogenic changes known to modulate the chemokine system
CXCR4, CCL2, CCL20, CCL8Human papillary thyroid cancerRET/PTC rearrangement induces17
CXCL1, CXCL8Mouse pancreatic cancerRas–raf activation induces19, 20
CC and CXC chemokinesMouse pancreatic cancerMyc activation induces21, 22
CXCR4Human renal cancerVHL mutation induces23
CXCL5, CXCL12, CCL20Breast cancerType II TGFβ receptor inactivation induces66
CXCR4Human breast cancerWild-type TP53 and p53 rescue drugs repress25
CXCR4Cancer stem cellsTP53 mutation induces26
CCL27Mouse skin cancerEGFR–ras signalling inhibits27
CCL22Human ATLLHTLV-1 transactivator Tax induces67

Ras is the most frequently mutated dominant oncogene in human cancer and activated oncogenic components of the EGFR–ras–raf signalling pathway also induce the production of tumour-promoting inflammatory cytokines and chemokines, including CXCL1 and CXCL8 19, 20. These chemokines can enhance tumour cell proliferation via autocrine loops and have paracrine actions on angiogenesis.

Myc is over-expressed in many human tumours—its activation initiates and maintains key aspects of the tumour phenotype. In addition to promoting cell autonomous proliferation, myc instructs remodelling of the extracellular microenvironment, with inflammatory cells and mediators playing key roles. In a pancreatic islet tumour model, the myc-activated cells produced chemokines that recruited mast cells capable of sustaining new vessel formation and tumour growth 21, 22.

Mutations in tumour suppressor genes can also regulate the chemokine system and other inflammatory pathways. Three examples are the von Hippel-Lindau (VHL) tumour suppressor gene, transforming growth factor-β (TGFβ) and TP53. VHL is part of a molecular complex that targets degradation of hypoxia-inducible factor (HIF-1α). VHL mutations increase HIF-1α protein in malignant cells and this increases the cell and tissue response to hypoxia; HIF-1α interfaces with NF-κB to induce CXCR4 in human renal cell carcinoma cells 23. Genetic inactivation of the type II TGFβ receptor unleashes production of the chemokines CCL20, CXCL5 and CXCL12 in breast cancer 24. These chemokines attract immune-suppressive myeloid-derived suppressor cells (MDSCs) that may inhibit any antitumour immune responses as well as facilitate metastasis. In breast cancer, wild-type but not mutant TP53 represses CXCR4 expression and p53 rescue drugs reduce CXCR4 expression 25. It was also recently reported that TP53 mutation in cancer stem cells leads to CXCR4 up-regulation 26.

Oncogenic change not only induces chemokines but may also suppress homeostatic chemokine production in a manner that promotes tumour development and growth. EGFR–Ras signalling in cutaneous tumour cells reduces their ability to express CCL27, resulting in impaired recruitment of lymphocytes with anti-tumour potential to sites of malignant disease, enhanced tumour growth and metastasis 27. Thus, oncogenes representative of different molecular classes and modes of action (tyrosine kinases; ras–raf; nuclear oncogenes; tumour suppressors) are all able to regulate the chemokine system during malignant disease, even at its earliest stages (Table 1). There is at the moment little evidence that genetic mutation or variation in chemokines or their receptors contributes to the evolution of malignant disease. However, a genome-wide linkage scan provided evidence for a chronic lymphocytic leukaemia (CLL) susceptibility locus at 2q21, the location of the CXCR4 gene. One truncating and two missense mutations of CXCR4, with possible functional consequences, were identified among 186 familial cases of CLL 28.

Chemokines and tumour-associated leukocytes

CC chemokines, especially CCL2 and CCL5, are major attractants of monocyte and macrophage precursors to the tumour microenvironment and levels of these correlate with the extent of the myeloid cell infiltrate 6, 29–31. Macrophages in tumours (TAMs) generally express an M2-type polarization; they have well-documented tumour-promoting activities and, in general, high tumour macrophage counts are a poor prognostic sign 32. Hence, cancer cell production of CC chemokines is advantageous to tumour growth and spread, and these chemokines also enhance survival of TAMs once they are in the tumour microenvironment 33. Genetic evidence for a non-redundant role of inflammatory CC chemokines in carcinogenesis has now been obtained in experiments involving the ‘silent’ atypical chemokine receptor D6 12, 29, 34. D6 acts as a chemokine ‘sink’ or decoy, preventing excessive chemokine signalling during inflammatory responses. Mice deficient in D6 show increased inflammation and increased inflammation-induced cancers.

Lymphocytes are the other major leukocyte population found in cancers; their recruitment and trafficking is also controlled by chemokines of the CC and CXC class. For instance, Th2 lymphocytes are abundant in human cancers and are thought to be tumour-promoting. A good example of the role of chemokines in Th2 infiltrates comes from a recent study in pancreatic cancer 35, in which De Monte et al hypothesized that tumour-resident dendritic cells, conditioned by local factors, may be able to favour differentiation of tumour-specific Th2 cells in the draining lymph nodes. In work conducted entirely with human cells and tissues, the authors found a significant association between tumour Th2 cell infiltrates and poor prognosis, with the ratio between Th2/Th1 cells in tumour biopsies being independently predictive of patient survival. They found that inflammatory cytokines produced by the malignant cells (probably downstream of oncogenic mutations) stimulate fibroblasts to make a factor called TSLP. TSLP activates/matures resident tumour dendritic cells that migrate to draining lymph nodes where they in turn activate CD4+ T cells to a Th2 phenotype. Th2-attracting chemokines CCL17 and CCL22 bring the CD4+ cells back to the tumour, where they have a major promoting influence.

Another example of cell cooperation, lymphocytes and chemokines in the tumour microenvironment, involves CCL28. Hypoxia, immune evasion and the formation of new blood vessels are key enabling characteristics of a progressing tumour microenvironment, and recent work 36 shows that, at least in intraperitoneal ovarian cancer, CCL28 might link this ‘vicious triangle’. CCL28 was frequently and strongly up-regulated when human ovarian cancer cells were exposed to hypoxia. CCL28 is normally associated with mucosal immunity but it also recruits immunosuppressive, T regulatory Treg cells during liver inflammation. This malignant cell-produced CCL28 recruited FoxP3-positive T regulatory cells that also expressed chemokine receptor CCR10, a receptor for CCL28. The Treg cells not only contributed to immune tolerance, they also produced the angiogeneic factor VEGFA. Hypoxia can induce a type of cell death that can trigger immune rejection of tumours; the induction of CCL28 seems to be a mechanism to counter this via recruitment of immune suppressive and angiogenic T regulatory cells. It will be interesting to see whether different chemokines have similar actions in tumours at other sites in the body or whether CCL28 has a more universal role.

In all the examples above, chemokine gradients are involved in attracting tumour-promoting cells, but there is another way in which malignant cells may modify chemokine gradients in their local environment. The chemokine CXCL14 is ubiquitously expressed in normal prostate but is absent from prostate cancer cell lines and biopsy material—and the reason for this is that the CXCL14 promoter has been hypermethylated in the malignant cells. As CXCL14 may control recruitment of circulating dendritic cell precursors and promote their in situ differentiation 37, it could be an advantage for malignant cells to lose the ability to produce this particular chemokine. Another example of this, also mentioned above, is how EGFR–Ras signalling in tumour cells reduces their ability to express the homeostatic chemokine CCL27 preventing migration of T cells that could have tumour-inhibitory actions 27.

De novo expression of chemokine receptors on malignant cells

As described above, malignant cells can acquire chemokine receptors not present on their normal counterparts, and hence migrate in response to chemokine gradients normally used for trafficking of leukocytes and adult stem cells, etc. Over the past 10 years a great deal of evidence from animal models supports the importance of chemokine receptors in metastasis, most of this relating to CXCR4 and CCR7. CXCR4 is also the principle chemokine receptor expressed on cancer stem cells 6.

There are two major areas of evidence concerning chemokine receptors and malignancy in the published literature (recently reviewed in 9, 38). First is that inhibition of chemokine receptor expression in experimental animal models can inhibit metastases and that de novo expression of chemokine receptors on malignant cells can increase invasiveness and metastases. The second evidence comes from retrospective studies of human metastatic cancers, in which expression of CXCR4 in breast, ovarian, pancreatic, prostate and many other cancers is almost invariably a poor prognostic sign and positively correlates with lung, liver and bone marrow metastases. In contrast, CCR7 expression generally correlates with increased lymph node metastases 8, 9, 38.

The reason for aberrant receptor expression on malignant cells may be a direct result of oncogenic mutations, as discussed above and summarized in Table 1, but also changes such as hypoxia in the tumour microenvironment 39. As part of its autocrine action on breast cancer cells, vascular endothelial growth factor (VEGF) can also induce expression of CXCR4 40. In addition, CXCL12 is activated in oestrogen receptor α-positive human ovarian and breast cancer cell lines 41; the hormone up-regulates CXCR4 expression and the mitogenic effects of oestradiol are neutralized by addition of CXCL12 antibody. Furthermore, in breast cancer cells, NF-κB activation up-regulates CXCR4 42. In ovarian cancer, CXCR4 is an important component of a malignant cell autocrine network and is co-regulated with its ligand and cytokines such as IL-6 and TNFα in the malignant cells 43 (Kulbe et al, manuscript under revision).

As described above and reviewed in 8, 9, 38, CXCR4 is the most widely expressed chemokine receptor on malignant cells, but other malignant cell chemokine receptors are also implicated in aspects of tumour growth and spread, including CCR4, CCR7, CCR9, CCR10, CXCR1, CXCR2, CXCR3, CXCR5, CXCR7 and CXC3CR1 6, 9 (Table 2). The concept of cellular highways was discussed above 9 and an obvious extension of this is that chemokine receptor-expressing malignant cells will hijack these ‘normal’ cellular highways during the development of organ-specific metastases.

Table 2. Chemokine receptor expression on malignant cells
Cancer typeChemokine receptors expressed by malignant cells
  1. Primary papers relating to the data summarized in this table can be found in three reviews 6, 8, 9 as well as 51, 59 and articles cited in this review.

BreastCXCR4, CXCR7, CCR4, CCR5, CX3CR1
High grade serous ovarianCXCR4
Prostate cancerCXCR4, CXC3R1
MelanomaCXCR4, CCR10, CCR7, CCR9, CXCR1, CXCR2, CXCR3
Oesophageal cancerCXCR4
Non-small cell lung cancerCXCR4, CCR7
Head and neck cancerCXCR4, CCR7, CXCR5
Bladder cancerCXCR4
Colorectal cancerCXCR4, CCR6, CCR7, CXCR5
Pancreatic cancerCXCR4, CX3CR1
Acute lymphoblastic leukaemiaCXCR4, CXCR3
Chronic lymphocytic leukaemiaCXCR4, CCR7, CXCR3
Stomach cancerCXCR4, CCR7, CCR4
Non-Hodgkin's lymphomaCCR7
T cell leukaemiaCCR7
Cutaneous lymphomaCCR3, CCR4, CCR10
Hodgkin's lymphomaCCR5
Multiple myelomaCCR1
GliomaCXCR4, CXCR7, CX3CR1

It is certainly true that expression of a single chemokine receptor can define the destination of a metastatic tumour cell in animal models 44, 45. A striking example of how a single chemokine receptor may determine organ-specific metastasis in humans comes from intestinal metastases of melanoma, a relatively unusual site for melanoma metastases. Intestinal metastases only occur in those patients whose original biopsies had CCR9+ malignant cells 46, mimicking what happens in homeostasis, where CCR9+ T cells traffic to CCL27 gradients in the intestine. Other examples would be CXCR5 in site-specific liver metastases and CX3CR1 in perineural invasion 6. The latter example deserves further mention. CXC3CR1 binds to a chemokine called CX3CL1 that has an unusual structure, being both soluble and cell surface, and is highly expressed in the central nervous system, where it regulates communication between neurons, glia and microglia. Tumours of neural origin, such as gliomas and neuroblastomas express CXC3CR1 and there is evidence that it is involved in adhesion, transendothelial migration and mobilization of the malignant cells 47. Malignant cells from other epithelial cancers, eg breast, pancreas and prostate, also have high levels of CX3CR1 and in pancreatic cancer this receptor is involved in invasion of cells along intra- and extrapancreatic nerves 48.

While there is strong evidence that malignant cells can follow chemokine gradients, it is also clear from many studies that some malignant cells, at some stages of cancer development and spread, produce the chemokines for which they also express the appropriate receptor. CXCR4-positive cells from colorectal 49 and breast 50 cancer cells have epigenetically-silenced CXCL12 and this increases their metastatic potential. One speculation could be that autocrine chemokine–receptor loops give malignant cells a survival advantage in a primary or metastatic tumour microenvironment. In order to use chemokine gradients elsewhere in the body, the malignant cells will have to repress expression of the chemokine while retaining expression of its receptor. There are also data in prostate, glioma and head and neck cancer that autocrine chemokine–chemokine receptor signalling in malignant cells can promote cell survival and resistance to chemotherapy-mediated apoptosis 51–53.

The roles of chemokines and their receptors in the evolution of malignancy are therefore many and varied. Figure 2 provides a summary of this. However, there are many unanswered questions concerning the roles of chemokine receptors on cancer stem cells, in early non-invasive malignancies and at later stages of metastases. Cancer cells grown in vitro seem to express several chemokine receptors, but what happens in vivo at different metastatic sites? Do human liver metastases express a distinct chemokine receptor repertoire from lung metastases of the same primary tumour? There is increasing evidence that metastatic spread is an early event in malignancy and, as summarized in Table 1, genetic changes found in early cancers can induce chemokine receptors on ‘initiated’ cells. What roles do these chemokine receptors play in the early stages of the disease? At this point, maybe chemokine receptors are necessary for a cell to spread into the surrounding tissue, vasculature or lymphatics, but not sufficient. Although there are many unanswered questions, there is no doubt that chemokine receptor expression is an important attribute for a metastatic cell.

Figure 2.

Chemokines and invasion

Chemokines and their receptors as therapeutic targets in cancer

There are a number of ways in which our understanding of the ways the chemokine system is involved in malignancy could be exploited in new approaches to treatment. Therapeutic targets include chemokines, which are major attractants for tumour-promoting myeloid cells or lymphocytes, and chemokine receptors on malignant cells. In addition, there may be potential in inducing chemokines that attract cells of the immune system that are capable of inducing helpful anti-tumour responses.

As described above, CCL2 and its CCR2 receptor are of major importance in the attraction of function of tumour-promoting myeloid cells. Antibodies against CCL2 or its cognate receptor CCR2 have been investigated in preclinical models, and a strong case for anti-CCL2 therapy has been made for prostate cancer 33. Administration of antibodies to CCL2 in mice bearing prostate cancer resulted in decreased tumour burden and bone resorption, with lower CCL2-induced VEGF release. Combination studies with anti-CCL2 and chemotherapy have also yielded improved survival in preclinical settings 54. Anti-CCL2 antibodies are currently being evaluated in humans in prostate and ovarian cancer 55.

An extensive study of CCL2 in breast cancer metastases has shown that CCL2, synthesized by metastatic tumour cells and by stroma at sites of metastases, is critical for continual recruitment of CCR2-positive monocytes that enhance the extravasation of tumour cells. Blockade of CCL2 with neutralizing antibodies inhibited monocyte recruitment, reduced metastases and prolonged survival of tumour-bearing mice 30.

Chemokine production in the tumour microenvironment can be triggered by tumour-promoting cytokines such as TNF-α, and in early trials of TNF-α antagonists in patients with advanced cancer, systemic levels of CCL2 were inhibited 56. In high-grade serous ovarian cancer, IL-6 stimulates inflammatory cytokine and chemokine production, tumour angiogenesis and the tumour macrophage infiltrate. Treatment with a therapeutic anti-IL-6 antibody inhibited these actions in preclinical experiments and a small clinical study 57. Patients who received the anti-IL-6 antibody for 6 months showed a decline in their plasma levels of CCL2, CXCL12 and CCL8 as well as VEGF. Hence, it may be possible to target chemokines by inhibiting their cytokine inducers with therapeutic antibodies that have already shown efficacy in chronic inflammatory diseases.

Cutaneous T cell lymphoma and T cell acute lymphoblastic leukaemia express the chemokine receptor CCR4. A humanized defucosylated antibody to CCR4 has anti-tumour activity in lymphoma-bearing mice 58. Anti-CCR4 antibody therapy was associated with increased numbers of tumour-infiltrating CD56+ NK cells mediating ADCC, and reduced the number of FOXP3+ T reg cells. Clinical trials are now under way with this antibody.

CCL3 and its receptors, CCR1 in particular, play a central role in the pathogenesis of multiple myeloma and multiple myeloma-induced osteolytic bone disease 59. CCL3 and CCR1 stimulate tumour growth directly and indirectly and modulate the osteoclast/osteoblast balance. In preclinical models, targeting either ligand or receptor reverses these effects, controlling tumour burden and preventing osteolysis 59.

As described above, CXCR4 is the most commonly over-expressed chemokine receptor in human cancer and this receptor has been targeted by a number of small antagonists, including the bicyclam AMD3100 and analogues and peptides designed to the amino-terminal region of the chemokine, such as T22, TN14003 and CTCE-9908. CXCR4 antagonists inhibited the primary tumour and metastasis in animal models of melanoma, osteosarcoma, breast and prostate tumours 60, 61.

AMD3100 was first developed as an anti-HIV agent. Unexpectedly, it was found to mobilize CD34+ stem cells from the bone marrow 62. At present AMD3100 is in clinical use for haematopoietic stem cell mobilization 63. By mobilizing malignant cells from the bone marrow niche, AMD3100 enhances the sensitivity of multiple myeloma or acute myeloid leukaemia blasts to the cytotoxic effect of chemotherapy in preclinical models 63.

Drugs that inhibit chemokine receptors are already approved for treatment of HIV (Miraviroc, anti-CCR5) and for stem cell mobilization (AMD3100) but, as recently discussed in detail by Schall and Proudfoot 2, none have yet been approved for the treatment of inflammatory, autoimmune or malignant diseases. They suggest that inappropriate target selection and ineffective dosing, rather than the redundancy of the chemokine system, is the reason for the failure of current approaches. It is also not clear whether chemokines themselves or their receptors would be the best targets in malignant disease. Another consideration is that, given the capacity of malignant cells to evolve and survive in different niches, inhibition of one component of the leukocyte infiltrate may just encourage outgrowth of malignant cells clones that produce a different repertoire of chemokines, that can bring in other tumour-promoting leukocyte lineages with different chemokine receptors. However, given the increasing success of immune checkpoint blockade approaches 64 that inhibit the tumour-suppressive effects of T-regulatory cells, there would seem to be a rationale for targeting chemokine receptors such as CCR4.

Similar arguments about clonal evolution could be made concerning targeting chemokine receptors on malignant cells—this could simply redirect metastases to other sites in the body. However, the information presented above would clearly suggest that chemokines or their receptors should be considered as targets in cancer, especially when given alongside other therapies. In this context, certain chemotherapy drugs, eg paclitaxel, can rapidly induce pro-angiogenic bone marrow-derived circulating endothelial progenitor (CEP) mobilization and subsequent tumour homing 65. This acute CEP mobilization is mediated, at least in part, by systemic induction of CXCL12, and in mouse models CEP mobilization can be substantially blocked by antibodies to CXCL12. In cancer patients treatment with paclitaxel, but not other chemotherapies, caused an elevation of plasma CXCL12 levels 65. Platelets may be one source of this and CEPs are known to express CXCR4. These results suggest a potential use for an anti-CXCL12 antibody in combination with some chemotherapy regimes.

Conclusions and future perspectives

Chemokines and their receptors are involved in all stages of cancer development, influencing the cellular composition of the tumour microenvironment, malignant cell survival and metastatic spread. From their earliest stages, cancers harness this intricate and tightly regulated network to generate a corrupt version of a system that, in healthy vertebrates, is necessary for embryonic development, tissue homeostasis and successful immune responses. A greater understanding of the chemokine system in malignancy can not only give us important new insights into cancer biology but also suggest new treatment approaches. If drugs that target the chemokine system show success in chronic inflammatory disease over the next few years, we have enough preclinical data to warrant trialling them in cancers.

Author contribution

FRB researched and wrote the article.