Physiology of chemokines in the cancer microenvironment

Abstract

Chemokines are chemotactic cytokines whose canonical functions govern movement of receptor-expressing cells along chemical gradients. Chemokines are a physiological system that is finely tuned by ligand and receptor expression, ligand or receptor oligomerization, redundancy, expression of atypical receptors, and non-GPCR binding partners that cumulatively influence discrete pharmacological signaling responses and cellular functions. In cancer, chemokines play paradoxical roles in both the directed emigration of metastatic, receptor-expressing cancer cells out of the tumor as well as immigration of tumor-infiltrating immune cells that culminate in a tumor-unique immune microenvironment. In the age of precision oncology, strategies to effectively harness the power of immunotherapy requires consideration of chemokine gradients within the unique spatial topography and temporal influences with heterogeneous tumors. In this article, we review current literature on the diversity of chemokine ligands and their cellular receptors that detect and process chemotactic gradients and illustrate how differences between ligand recognition and receptor activation influence the signaling machinery that drives cellular movement into and out of the tumor microenvironment. Facets of chemokine physiology across discrete cancer immune phenotypes are contrasted to existing chemokine-centered therapies in cancer.

CHEMOKINE ORCHESTRATION OF IMMUNE FUNCTION

The immune system consists of a highly diverse network of cells tasked with protecting the body from foreign substances and pathogens or removing dysfunctional cells and repairing damaged tissue (1). A common thread in immuno-oncology over the past two decades is that previous understanding of how cells or molecules function within the immune system underestimates the intricate mechanisms at work for an effective immune response against pathogens or cancers. This intricacy may provide an explanation for why emerging immune-targeted therapies for cancer often fail to overcome barriers in cancer heterogeneity. To exert their core functions, immune cells need to circulate through the body and tissues, rapidly mobilize, and contact other cells (1, 2). The directed migration of immune cells is orchestrated by a family of secreted chemotactic cytokines called chemokines (2, 3).

Chemokines are a family of 8–15 kDa molecular weight cytokines traditionally defined by their ability to stimulate cellular migration along a chemical gradient. Typically, chemokines adopt a highly conserved tertiary structure comprising a flexible N-terminus, a three-stranded β-sheet, and a C-terminal α-helix. Four chemokine subfamilies (CXC, CC, XC, and CX3C) are defined by the spacing of cysteine residues near the N-terminus. Typically, the N-terminus has fewer amino acid residues and is essential for activating the target receptor. CXC chemokines are subclassified based on the presence or absence of an N-terminal glutamic acid-leucine-arginine (ELR) motif. Chemokines were originally classified according to their structure, function, or expression patterns under specific conditions. For example, CXCL9 was originally identified as “monokine induced by gamma interferon (MIG)” (4). Similarly, CCL3 and CCL4 were previously known as “macrophage inflammatory protein (MIP)-1α” and “MIP-1β,” respectively, as they were produced in activated macrophages (5). Subsequently, ligands and their receptors were classified using a systematic nomenclature that was further subdivided by general chemokine functions in health and disease (6). Classification based on function defines homeostatic chemokines as constitutively expressed molecules that regulate circulation and tissue localization of receptor-expressing lymphocytes, while inflammatory chemokines are induced by inflammation and promote myeloid and lymphocyte trafficking into peripheral tissues (6–8) (Fig. 1). Typically, homeostatic chemokine ligands activate a single receptor. Inflammatory chemokines display more redundancy, where a single ligand may bind and activate multiple receptors, or many receptors may be recognized and bound by several disparate chemokines.

Figure 1.

Chemokine functions in leukocyte circulation and extravasation. Tissue injury or infection with microbial agents stimulate the production of chemokines in discrete epithelial cells, endothelial cells, and fibroblasts in peripheral tissues and organs. Glycosaminoglycans play an important role in sculpting and enforcing the formation of chemokine gradients needed for leukocyte extravasation and migration into tissues. The highly sulfated and acidic residues present on glycosaminoglycans bind negative residues on chemokines, creating localized concentrations of chemokines in discrete endothelial and tissue locations. Glycosaminoglycans may also influence oligomerization of chemokine ligands. Immune cells follow the chemical gradient, moving from areas of low concentration to higher levels of ligand.

The biological activities of chemokines are mediated by expression of 7-transmembrane G protein-coupled receptors (GPCRs) of the rhodopsin-like family (9, 10). The chemokine receptor family comprises ∼20 distinct class A GPCRs that activate intracellular signaling pathways on target cells. Although many class A GPCRs bind small molecules or peptide ligands that dock into an orthosteric binding site within the transmembrane domain of the receptor, chemokines bind to their receptors using a two-site mechanism. The first binding site entails the flexible N-terminal domain of the receptor wrapping around the chemokine to create an extensive protein-protein interface. The second binding site encompasses the ligands N-terminus docking into the transmembrane orthosteric site (11). Emerging data suggest this two-site model masks additional intermediate ligand-receptor interactions that may further modulate downstream signaling and influence chemokine physiology (11). Binding to chemokine GPCRs elicits a signaling cascade starting with heterotrimeric G proteins and recruitment of β-arrestin that combined result in intracellular calcium flux, a decrease in cAMP, and activation of intracellular signaling targets such as ERK that culminate in cellular movement (9, 12) (Fig. 2). Ligand-induced activation of G protein receptor kinases phosphorylate the intracellular C-terminus of the receptor, leading to recruitment of β-arrestin and internalization of the ligand-receptor-arrestin complex. Signaling downstream of the active receptor stimulates calmodulin kinases, phospholipase C, and monomeric GTPase signaling to coordinate the cytoskeletal rearrangements necessary for cell movement. Agonist-induced activation of chemokine receptors facilitates chemotaxis by initiating actin polymerization on the leading edge of the cell during migration. In total, there are ∼50 known chemokines, 20 chemokine receptors, and 4 atypical chemokine receptors (ACKRs), related GPCRs with different signaling responses (7).

Figure 2.

Signaling of chemokine-activated receptors. Chemokine receptors and ligands can either be balanced (green circles and shading), G-protein biased (blue circles and shading), or β-arrestin biased (red circles and yellow shading). Figure created with BioRender.com.

Chemokine signaling through its receptors plays key functions during development, homeostasis, inflammation, infection, and pathological processes. These functions largely reflect the cardinal roles for chemokines in directed cell migration and substrate adhesion (Fig. 1). Chemokines also possess significant functional pleiotropy, with demonstrated roles in cell type-specific proliferation and apoptosis. Chemotaxis directs cells to move along an increasing concentration gradient of chemokine ligands. The chemokine gradient is jointly regulated by production and stabilization of ligand by glycosaminoglycans (GAGs). GAGs are complex polysaccharides expressed on the surface of most cells. Highly sulfated and acidic GAGs bind negative residues within the chemokine body allowing for stable interstitial chemokine gradients as well as presentation on endothelial surfaces to promote leukocyte extravasation (13, 14) (Fig. 1). GAGs may also facilitate oligomerization of chemokines (15), which can evoke contradictory effects either enhancing or hindering functional activity (16, 17).

Physiologically the chemokine system consists of multiple ligands, receptors, and extracellular factors that together influence cellular outcomes. Traditionally, chemokines have been considered as a one-dimensional, on-off process wherein a ligand binds a receptor, and the newly activated receptor induces cell migration (Fig. 3A). A multidimensional physiological chemokine framework includes time and location within a tissue or cellular microenvironment, as well as multiple, stable, intersecting gradients of individual chemokines that together activate a receptor or receptors to control cell movement or nonmigration functions (Fig. 3B). Thus, functional outcomes in chemokine physiology result from a pharmacological “QR code” interpretation that integrates multidimensional inputs rather than a unidirectional “barcode” view wherein chemokine ligand and receptor expression dictate cell movement. Integrated multidimensional signaling by individual chemokine is only beginning to be understood, particularly within cancer, a context in which molecular dysfunction and changing conditions over time are the rule.

Figure 3.

Overview of chemokine physiology. A: the conventional wisdom is that chemokines function as single ligands binding to a cognate receptor and inducing chemotactic migration from least to highest ligand expression. This unidirectional functional model of chemokine function can be depicted as a two-dimensional barcode where migration reflects ligand recognition by its receptor. B: chemokine physiology increasingly recognizes multiparametric pharmacological and cellular variables impacting both ligand and receptor, suggesting functional outcomes require an integrated multidimensional QR-code-type interpretation of these varying inputs. B, top: variables include migration as a biphasic curve with little cell movement at the lowest and highest concentrations of ligand, the formation of homo- or hetero-oligomers, binding to nonreceptor glycosaminoglycans or amino-terminal proteolytic cleavage. B, bottom: variables include biased signaling through the G-protein-coupled receptors, homologous or heterologous receptor desensitization, activation of discrete signaling pathways, and “traditional” or “atypical” chemokine receptors that mediate calcium signaling or ligand internalization, respectively. Figure created with BioRender.com.

CHEMOKINES AND THE TUMOR IMMUNE PHENOTYPE

Reflecting the heterogeneity of tumors, the cancer immune microenvironment has been broadly classified into immune-inflamed, immune-excluded, or immune-desert subtypes (18–20) (Fig. 4). It is increasingly recognized that the tumor immune microenvironment is dynamic and can even include aspects of the “inflamed,” “excluded,” and “desert” characterization within the same tumor and timepoints in tumorigenesis (21). The immune inflamed designation is characterized by tumor-infiltrating leukocytes being present directly adjacent to tumor cells and suggests the presence of an ongoing or prior antitumor response that was suppressed or is being actively evaded by the cancer cells. In the immune-excluded phenotype, the immune cells are physically separated from tumor cells by intervening stroma. Lastly, the immune desert classification is characterized by an overall dearth of tumor-reactive T cells and the preponderance of suppressive immune cells. The classification of a particular tumor into one of these tumor-immune subtypes may reflect the differential response of the immune system to tumor-regulated manipulation of chemokine signals present within the tumor itself or systemically within the host (Fig. 4). To date, these phenotypes remain centered on cellular definitions that have yet to establish a global chemokine or cytokine profile within each immune microenvironment. The absence of a concrete chemokine profile within each of these immune phenotypes likely reflects their reliance on histopathological definitions that would benefit from more precise molecular metrics (22). However, it remains possible that there is no uniform chemokine landscape for each immune subtype, as there are multiple other factors such as time, metabolites, or genotype that modulate tumor immune responses (23).

Figure 4.

Chemokines in cancer immunophenotype. The cancer immune microenvironment has been broadly classified into immune-inflamed, immune-excluded, or immune-desert subtypes. The immune-inflamed microenvironment is characterized by populations of effector T cells, conventional dendritic cells, and inflammatory-type tumor-associated macrophages. Tumors with an immune desert-type environment are predominated by fewer effector T cells and a preponderance of wound repair-type tumor-associated macrophages and regulator T cells. Immune-excluded tumor microenvironments have fewer immune cells in close proximity with cancer cells and may have elevated levels of stromal cells and tumor-associated neutrophils. Figure created with BioRender.com.

Despite these caveats, patterns of chemokine ligand and receptor expression have begun to emerge (Fig. 4). Data from several forms of cancer suggest that CXCL16 as well as the CXCR3 ligands CXCL9, CXCL10, and CXCL11, where each of the latter three is induced by IFN-γ, are important in immune inflamed-type tumors (24–26). Paradoxically, while elevation of these ligands may be beneficial in mediating antitumor immunity observed in immune-inflamed tumor, they may also exacerbate the transformation of inflammation-associated cancers (27). In the immune-excluded phenotype, T cell-attracting chemokine ligands such as CXCL9, CXCL10, and CXCL11, may still be present but are less effective in promoting antitumor immunity. The spatial exclusion of tumor-reactive T cells in immune-excluded tumors may reflect the predominance of granulocytic myeloid-derived suppressor cells (MDSCs) in these tumors, which stimulate the deposition and formation of peritumoral stroma and restricts T cell entry (20, 26). Alternatively, TGF-β may directly reduce expression of T cell-attracting chemokines (28, 29). Elevated levels of the MDSC-attracting chemokine CCL2 may have an important role in counterbalancing CXCL9, CXCL10, CXCL11, and CXCL16, further limiting the recruitment of effector T cells into these tumors (20).

The immune-excluded phenotype has been noted for the higher presence of angiogenesis, implicating roles for the angiogenic ELR+CXC chemokines like CXCL1 in these tumors (26). However, other studies have noted no transcriptional differences between immune-inflamed and immune-excluded tumors (22). In the immune-desert phenotype, there is a lack of cytotoxic T cell-attracting chemokines such as CXCL16, CXCL9, CXCL10, and CXCL11 (Fig. 4). In contrast, CCL20, a chemokine with roles in attracting regulatory T cells and Th17 cells, may suppress CD8 T-cell proliferation and cytolytic functions (30). Another characteristic of the immune-desert phenotype is a lack of MHC I antigen presentation, which may be due to reduced dendritic cell-recruiting chemokines CCL3, CCL4, CXCL1, and CXCL2 (31). The spatial and temporal heterogeneity between different cancers (32) and even within a single tumor (21) regarding immune phenotypes highlights the need for detailed molecular interrogation to establish the chemokine profile within tumor immune phenotypes.

Tumor mutational burden has been used as a predictive biomarker for the success of immune checkpoint inhibition across many solid cancers (33). However, patients with low mutational load may have a strong anticancer response to immune checkpoint blockade, whereas other patients with high tumor mutational burden remain unresponsive to immune checkpoint blockade (34). These reports suggest that tumor mutational burden cannot be used as the sole biomarker to predict response to biological immune therapies or to predict the tumor-immune phenotype category to which the tumor belongs. Interestingly, select chemokines such as CCL5 have emerged as putative biomarkers for atherosclerosis (17) and new trials with chemokine receptor inhibitors are currently being evaluated for use as combinatorial therapy with immune checkpoint inhibitors (Table 1). Characterization of the chemokine ligands and receptors associated with the three subtypes of immune microenvironments would open possibilities to precisely target the tumors with other, non-checkpoint-related immunotherapies as well, especially in the two subtypes that are less often responsive to checkpoint inhibition. For example, immune-desert and immune-excluded tumors could benefit from STING agonists that increase the production of CXCL9 and CXCL10 to increase T cell infiltration (40).

Table 1.

Summary of chemokine-focused therapeutics used in clinical trials treating patients with solid tumors from 2012 to present

Molecule	Target	Cancer Type	Phase	Start Date	Estimated End Date	Study Info/Results
SX-682	CXCR1 + CXCR2	Melanoma NCT03161431	I	6/19	12/22	In combination with PD-1 inhibition
		CRC NCT04599140	Ib/II	10/20	1/24	In combination with PD-1 inhibition
		PDAC NCT04477343	I	11/20	12/24	In combination with PD-1 inhibition
Reparixin		TNBC (Goldstein et al., 35)	II	7/15	3/20	No increase in PFS over paclitaxel alone
AZD5069	CXCR2	Prostate NCT03177187	I/II	11/17	10/23	In combination with antiandrogen therapy
BL-8040	CXCR4	PDAC NCT02907099	IIb	12/16	12/22	In combination with PD-1 inhibition
LY2510924		scLC (Salgia et al., 36)	II	9/11	8/16	No increase in PFS or OS when added to standard chemotherapy
AMD3100		PDAC NCT04177810	II	11/20	11/24	In combination with PD-1 inhibition
		CRC, Ovarian, PDAC NCT02179970	I	6/15	12/18	Drug is safe, can increase circulating B and T cells
BMS-813160	CCR2 + CCR5	NSCLC + HCC NCT04123379	II	3/20	10/24	In combination with PD-1 inhibition or IL-8 inhibition
		PDAC NCT03767582	I/II	12/19	3/23	Anti PD-1 + BMS + GVAX post chemo/radiotherapy
MLN1202	CCR2	Bone metastases NCT01015560	II	3/10	12/12	Decreased marker of bone turnover, indicating beneficial response
PF-04136309		PDAC (Noel et al., 37)	Ib/II	5/16	10/17	Decrease monocytes in blood, but high pulmonary toxicity
Mogamulizumab (KW-07621)	CCR4	Solid tumors (Kurose et al., 38)	I	2/13	6/16	Decrease in circulating T-regs, Th2, and Th17 T cells
Maraviroc	CCR5	CRC (Haag et al., 39)	I	4/18	3/20	Nontoxic when used with anti PD-1
GS-1811	CCR8	Solid tumors NCT05007782	I	8/21	2/25	Monotherapy or with ICI
S-531011		Solid tumors NCT05101070	I/II	1/22	4/27	Monotherapy or with ICI

Some studies and information were excluded due to lack of information published at the conclusion of the trial. CRC, colorectal cancer; HCC, hepatocellular carcinoma; NSCLC, nonsmall cell lung cancer; PDAC, pancreatic ductal adenocarcinoma; scLC, small cell lung cancer; TNBC, triple-negative breast cancer. Numbers in square brackets refer the literature citation.

A recent report highlights the critical importance of orchestrating cell localization within solid tumors, with proximity of T cells to tumor cells enhancing responsiveness to checkpoint blockade (41). Experiments to directly test and define the chemokines responsible for a particular immune phenotype are technically challenging, given pronounced chemokine ligand-receptor redundancy, diverse pharmacological properties, dynamic expression mechanisms, and the presence of ACKRs and GAGs capable of modulating chemokine function through mechanisms that would not necessarily be detected by expression studies. The balance between these disparate functions can also be affected by the stage of tumorigenesis, the state of immune cell activation, the balance of effector and regulatory response, and the relative expression of chemokine receptors on effector and suppressive target cells.

HETEROGENEOUS MECHANISMS OF CHEMOKINE COMPLEXITY IN CANCER

Tumor cell production of chemokines was first described nearly 35 years ago (42–44). A series of reports at the turn of the century revealed that tumor cells not only produce chemokines but also are themselves functional targets through the expression of chemokine receptors (3, 45). Over the past 20 years, cancer research has increasingly explored multiple cellular hallmarks that promote tumor initiation, progression, and chemotherapeutic resistance (46–48). Conventionally, the governing elements for tumorigenesis and metastatic behavior were thought to be oncogenes or tumor suppressors. The true breadth of contributing factors to tumor metastasis is more varied and include altered cell-cell and cell-matrix adherence, increased ability to invade surrounding tissue, and enhanced migratory capacity. In addition, fibrosis, tumor-promoting inflammation, angiogenesis, immune suppression, and evasion are intra- and extratumoral interactions increasingly recognized to play key roles in tumor malignancy (49). Chemokine functional responses in cancer reflect the integration of multiple factors, including stage of tumorigenesis, immune cell activation, recruitment of immune-activating or immunosuppressive cells in the tumor microenvironment, and chemokine receptor expression on effector and suppressive/regulatory target cells. However, as detailed here, there are additional structural and pharmacological aspects of chemokine physiology that play outsize roles in determining the functional outcomes of chemokines in cancer.

Biased Signaling

Chemokines mediate their functions downstream of ligand-receptor binding, activating the Gαi subunit of the heterotrimeric complex, recruiting β-arrestin, and mobilizing intracellular calcium release from the endoplasmic reticulum (Fig. 2). GPCR pharmacology uses efficacy and potency as the key parameters to distinguish ligand action on a common receptor. A full agonist activates the receptor to maximal efficacy, whereas a partial agonist binds and activates with submaximal efficacy or potency (12). One mechanism by which partial agonists have reduced potency compared with full agonists was demonstrated by recent work from our group showing that the flexible extended C-terminus of CCL21 interacts with the chemokine body and alters calcium flux, cAMP inhibition, β-arrestin recruitment, and chemotactic migration potency (50–52). A balanced agonist activates the entire repertoire of G protein and arrestin signaling cascades, whereas a biased agonist preferentially activates either the Gα and Gβγ signaling or the β-arrestin-dependent intracellular signaling pathways originating from the same GPCR (12, 53) (Fig. 2). Although chemoattraction follows a biphasic response wherein movement occurs in a narrow concentration range and is absent at lower and higher concentrations, calcium mobilization occurs dose dependently and adheres to a saturable sigmoidal response curve (6, 54). Our group has shown that homotypic dimerization resulted in a new structural chemokine ligand that functions as a biased agonist (54–56). Our data demonstrate that increasing concentrations of two different ligands, CXCL12 or CCL20, in the presence of their known binding partners, promotes dimerization and that homodimer binding to the cognate receptor produces a nonmotile signaling cascade that we have termed “ataxis” through their cognate receptors (57) (Fig. 5). These discoveries provide a new mechanism for the biphasic dose-response curve of chemokine migration and promiscuity of ligands.

Figure 5.

Two illustrative examples of chemokine-biased signaling. A: CCL19, CCL21, and CCR7 exhibit cell (or tissue) bias. CCL19 at low concentrations maintains migration of CCR7-expressing dendritic cells into draining lymph node. Although both bind CCR7 with comparable efficacy, CCL21 chemoattracts T cells with less potency than CCL19. CCL21 also exhibits a greater ability to bind glycosaminoglycans found in high endothelial venules compared with CCL19. B: optimal concentrations of the chemokine CXCL12 stimulate chemotaxis and balanced agonist signaling through its cognate receptor CXCR4. At this narrow concentration, CXCL12 primarily maintains a monomer configuration. As CXCL12 concentrations increase, or in the presence of receptor or GAG-binding partners, CXCL12 dimerization is enhanced. Dimerized CXCL12 binding to CXCR4 activates a G-protein-biased signaling pathway that inhibits cell movement. Figure created with BioRender.com. GAGs, glycosaminoglycans.

CCL19 and CCL21 are additional chemokine-biased agonists. Each bind to CCR7 and while both stimulate G protein activation, only CCL19 induces receptor phosphorylation and recruitment of β-arrestin (58). Biased signaling extends to cell bias and receptor bias. Cell bias occurs when the same chemokine ligand binds the same receptor, but on different cell types. An example of cell-biased signaling is CCL19 inducing chemotaxis in CCR7-expressing dendritic cells (DCs), while CCR7-expressing naïve T cells migrate to lymph nodes in response to CCL21 and not CCL19 (59) (Fig. 5). Receptor bias can occur when two different receptors bind the same ligand, as is the case for CXCL12 binding with both CXCR4 and ACKR3. In contrast to selective signal pathway activation, molecular antagonists bind to the receptor, often competing for the agonist binding site with comparable affinity and yield no receptor activation or signaling. Taken together, these emergent pharmacological properties highlight the utility of using structure-function studies with molecular pharmacology approaches to better understand how cells decipher the multitude of signals within the tumor microenvironment or are therapeutically targeted (60).

Homologous and Heterologous Receptor Desensitization

Cellular migration signaling is a multistep process that relies on the tight spatial and temporal control of G proteins and β-arrestin-dependent signaling pathways by chemokine-activated receptors (Fig. 2). Chemokine binding to their cognate receptors induces rapid desensitization, internalization into early endosomes, and sorting of the ligand-receptor complex into either a recycling pathway or degradative pathway (61). The tight control of chemokine receptor movement from cell surface to endocytic pools is essential to limit the magnitude and duration of chemokine signaling and controlling receptor access to ligands. Homologous chemokine receptor desensitization and internalization is mediated by agonist-induced phosphorylation of the intracellular C-terminus by G protein-coupled receptor kinases (GRKs) or protein kinase C (PKC). GRKs typically mediate receptor phosphorylation of agonist-bound receptor, leading to β-arrestin binding and subsequent G protein uncoupling, thereby terminating further receptor signaling. Heterologous receptor desensitization occurs following phosphorylation and internalization of the chemokine receptor C-terminus by second messenger-dependent protein kinases such as PKC in a cognate agonist-independent manner. The ability for chemokine receptors to be desensitized by either cognate or noncognate ligands suggests the potential for nonchemokine ligands to modulate directed migration.

Atypical Chemokine Receptors

Chemokine physiology is complicated by the presence of ACKR in lymphoid tissues, normal tissue, and heterogenous cell populations within the tumor microenvironment. Previously termed decoy receptors or chemokine scavengers, ACKRs bind chemokines and elicit disparate signaling through the heterotrimeric G protein complex (7). Although ACKRs retain the 7-transmembrane domains, they differ from canonical chemokine receptors with a modified or deficient DRYLAIV-motif within the second intracellular loop (7). The absence of the DRYLAIV-motif results in an inability to bind and signal through G proteins and is correlated with the lack of cell migration in response to ligand binding. Although ACKRs fail to activate G protein signaling, they are potent β-arrestin-biased receptors believed to fine-tune chemokine gradients through ligand depletion via arrestin-mediated internalization. ACKR1, previously termed “DARC,” is primarily expressed on erythrocytes and endothelial cells and is recognized and actively engaged by 20 chemokines (62). Interestingly, an ACKR1 polymorphism that prevents transcript expression results in an ACKR1-negative phenotype thought to have evolved to prevent the malarial parasite Plasmodium vivax from infecting erythrocytes (63). This polymorphism is predominately found in people of African descent. The ACKR1-negative phenotype was investigated as a possible causative factor in the increased incidence and mortality of prostate cancer in black men (64). Although initial data suggest that ACKR1 functions to bind and reduce the gradient of angiogenic chemokines and in turn blunt endothelial cell chemotaxis (65), subsequent case-control studies of ACKR1-positive and -negative men have not substantiated that result (66). ACKR2 predominately binds inflammatory CC chemokines whose expression by breast cancer cells has been associated with increased survival and decreased metastasis (67). ACKR3, previously defined as CXCR7, binds and internalizes CXCL12 (68). ACKR3 is expressed by hepatocellular cancer cells and has been linked with increased cell proliferation and metastasis (69). The current working model for ACKR3 function states that the receptor prevents excessive CXCL12 from mediating CXCR4 internalization and degradation, sculpts the chemical gradient to ensure adequate levels of CXCL12-CXCR4 signaling, and/or it prevents formation of nonmigration inducing, dimerized CXCL12 (70, 71). ACKR4, also known as CCRL1, binds CCL21 and has been shown to disrupt DC tumor infiltration, with a corresponding decrease in antitumor T cell responses (72). ACKR5, previously defined as CCRL2, and ACKR6 are additional ACKRs whose functions in cancer or other diseases remains understudied. ACKRs are therefore capable of modulating the full breadth of classical chemokine signaling, fine-tuning as “gain-of-function modulators” on the intensity of a particular chemokine signal or gradient.

CHEMOKINES IN CANCER

The metabolic, genetic, and epigenetic mechanisms responsible for generating the immune-inflamed, immune-excluded, and immune-desert phenotypes continue to be uncovered. Although much has been learned about the importance of chemokines in cancer, monotherapy treatments targeting chemokines have yet to demonstrate clear benefits in patients (Table 1). A thorough review of chemokine-based therapeutics used in hematological cancers, not covered in this review, has previously been published (73). The lack of durable benefits from these clinical trials likely reflects distinct complexities in chemokine physiology. What follow are illustrative examples of chemokine ligand and receptor complexity, including discussion of how biased signaling, desensitization, redundancy, and ACKR receptor expression may influence chemokine functions in cancer.

CXCR4 and CXCL12

Originally defined as an HIV-coreceptor (74), CXCR4, is widely expressed on hematopoietic progenitor stem cells, T cells, B cells, monocytes, macrophages, neutrophils, and granulocytes and is constitutively expressed in brain, lung, colon, heart, kidney, liver, and endothelial, epithelial cells, microglia, astrocytes, and neuronal cells. Prior work demonstrated expression of the HIV-coreceptors CXCR4 by human colon cancer cells and tissues (45). Muller and colleagues (3) subsequently demonstrated that elevated CXCR4 expression on malignant cancer cells plays a key role in directing metastasis of CXCR4-expressing tumor cells to organs that express its cognate ligand CXCL12. Subsequent reports validated the role for cancer cell expression of CXCR4 in the metastasis of over 20 different solid and hematological cancers (75). CXCR4-driven metastasis to the bone marrow provides a protective environment for tumor cells (76). CXCR4 inhibitors, such as the receptor antagonist plerixafor, were originally developed to inhibit HIV but were eventually Food and Drug Administration-approved for use in mobilizing hematopoietic progenitor stem cells from the bone marrow preceding apheresis and transplantation (77). More recently, plerixafor has been examined in clinical trials to test its efficacy in treating solid cancers (78). More potent and selective CXCR4 inhibitors are being investigated in multiple different cancer types, with an aim to use them in combination with immunotherapy and/or chemotherapy (79). Recent clinical trials of CXCR4 antagonists have focused on the potential role of stromal-produced CXCL12 to influence the formation of immune suppression within solid tumors (80). In this model, pathological silencing of CXCL12 in tumor cells (81, 82) was replaced by the upregulated production of CXCL12 by subsets of tumor-associated fibroblasts which act counterintuitively in preventing the entry of CD8+ cytolytic T cells into the tumor. In preclinical and clinical studies, plerixafor administration led to increased tumor-infiltrating effector CD8+ T cells (83). Current clinical trials have found mixed efficacy when used as a single agent or with traditional chemotherapies, but ongoing trials are looking into the effects of CXCR4 inhibition in combination with immune checkpoint inhibitors (Table 1) (36).

CXCL12 is nearly ubiquitously expressed, with the highest production in bone marrow, lymph nodes, and spleen, and plays essential roles in hematopoiesis, lymphocyte recirculation between lymphoid tissues, neural development, and angiogenesis (84). CXCL12 is the sole cognate ligand for CXCR4 and can also bind ACKR3 (68). Despite being essential for life, CXCL12 was initially considered protumorigenic, with its role in cancer centered on its role in metastasis of CXCR4-expressing cells. In conflict with those protumorigenic attributes, other research indicates homeostatic expression of CXCL12 functions as a tumor suppressor. Several reports have verified our discovery that constitutive epithelial CXCL12 is lost in several solid cancers, a pathological loss-of-function mutation that facilitated tumor dissemination and immune evasion (81, 82, 85–87). Restoring CXCL12 expression using gene expression or with exogenous chemokine administered as an injection decreased the metastatic potential of those cancer cells, supporting the notion that repression of the ligand was a key step in CXCR4-dependent tumor metastasis (55, 81, 82, 85). CXCL12’s ability to form homodimers and induce biased agonist signaling through CXCR4 provides a potential mechanism to explain the discrepancy in the ligands’ pro- and antitumorigenic properties (54, 56).

CXCR6 and CXCR3

Chemokine ligands and receptors directing the trafficking of antitumor effector cells have been abundantly studied. CXCR6 is expressed by natural killer (NK) cells and activated CD4+ and CD8+ T cells (88). The ligand for CXCR6 is CXCL16, one of the few transmembrane chemokines, and is produced predominantly by CCR7+ DCs and other antigen-presenting cells during inflammation (89, 90). Increased expression of CXCL16 was subsequently shown in solid tumors, with important consequences on the localization of CXCR6-expressing T cells, cytotoxic CD8+ T cells, NK cells, and stromal cells within the tumor (91, 92). A recent report has sought to highlight this signaling pathway to enhance the trafficking of chimeric antigen receptor T cells in solid cancer (93). CXCR6-expressing NK cells can also release the inflammatory chemokines CCL3, CCL4, and CCL5 capable of attracting DCs, further enhancing antitumor immunity (94). The CXCR3 ligands CXCL9, CXCL10, and CXCL11 are soluble inflammatory chemokines and play critical roles in antiviral and antitumor immune responses. In colorectal, ovarian, esophageal, and nonsmall cell lung cancer, high levels of these CXCR3 ligands are associated with increased infiltration of cytotoxic T cells and better prognosis (95–97). However, in renal cell carcinoma, high levels of the CXCR3 ligands predict a poor prognosis (98) and may indicate high levels of necrosis (99). In immune-excluded-type cancers, high tumor CXCR3 ligand levels may not correlate with improved prognosis as the T cells are denied entry by the protective stroma (100). Chemokine expression has emerged as a predictive marker for efficacy for anti PD-1, PD-L1, or CTLA-4 immune checkpoint inhibitor therapies (101), with the T cell recruiting chemokines CXCL9, CXCL10, and CXCL11 positively correlated with susceptibility. However, this presents a problem for immune desert-type tumors where expression of inflammatory chemokines remains diminished. One method to overcome the deficiency may entail the utilization of combined therapeutic approaches that may circumvent not only tumor heterogeneity but also the multifactorial nature of chemokine biology to improve cancer anti-immunity and limit tumor progression.

CCR7 and XCR1

CCR7 and XCR1 are receptors that play an essential role in dictating the localization of lymph node-homing naive T cells, parafollicular trafficking of B cells, and antigen-activated DCs (102, 103). Systemic deletion of CCR7 results in pronounced loss of peripheral immune tissues, including lymph nodes, Peyer’s patches, and spleen (104). CCR7 has two canonical ligands, CCL19 and CCL21, with the former responsible for immune cell trafficking into lymphatic vessels, whereas the latter directs movement within peripheral immune tissues. Although CCL19 typically regulates DC trafficking into draining lymph nodes (104), they have also been linked in a murine renal cell carcinoma model to recruit DCs directly into the tumor (105). Although CCR7 expression is upregulated in mature DCs (106), XCR1 has emerged as a key biomarker of conventional DC subset 1 (cDC1) that functions in the cross-presentation of antigens to naive T cells (103). Decreased expression of the ligands for XCR1, XCL1, and XCL2, has been correlated with increased tumor growth (107). Although XCR1 has shown utility as a marker for antigen cross-presenting DCs, its specific role in cancer immunity has been slower and may reflect XCL1s characterization as a metamorphic protein (108). Metamorphic proteins defy the chemokine folding paradigm and switch between different folds of the tertiary molecule, with consequent changes in functional signaling. XCL1 possesses two distinct folds, with one functioning as a canonical chemoattractant capable of binding to XCR1, and the second fold binding to GAGs (108–110). Even if limited to XCL1, this recent discovery represents a potentially paradigm-shifting change in our understanding of the behavior of individual chemokine molecules within the tumor microenvironment, and the associated complexity of individual chemokine axes, which will in turn greatly increase the complexity of the overall local chemokine network.

CXCR5 and Humoral Immune Responses

A primary chemokine receptor for B cells, CXCR5 is essential for naive B cell trafficking and establishment of lymphoid follicles in peripheral immune tissues through signaling by its sole ligand CXCL13 (111, 112). B cells play a complex and generally understudied role in solid tumors; they can exhibit pro- or antitumor properties depending on plasma cell subset and context-specific T cell interactions, and therefore variably correlate with prognostic value across different cancer types (113). The effects that B cells have on tumor immunity likely depend on their ability to form organized structures (114). Unorganized B cell infiltration into the tumor has been associated with lower survival in a mouse model of pancreatic cancer, whereas tertiary lymphoid structure formation is generally a positive prognostic factor for multiple other types of cancer (115). The protumorigenic effects of infiltrating B cells may reflect increased prevalence of regulatory B cells (Bregs) capable of suppressing effector T cell antitumor immunity through multiple mechanisms (116). Breg cells may also contribute to the tumor fibrosis characteristic of immune-excluded tumors. The mechanisms that regulate B cells organizing structures in nonlymphoid cancer tissues remain unknown.

CCR7 and CXCR5 in Inducible Lymphoid Aggregates

Tertiary lymphoid structures resemble secondary lymph organ formation in that both contain B cell follicles and germinal centers surrounded by a T cell region, and rely on the homeostatic chemokines CXCL13, CCL19, and CCL21 and their cognate receptors CXCR5 and CCR7, for their formation. Although tertiary lymphoid structures form in chronically inflamed tissues, their formation in tumors is inconsistent and generally associated with better prognosis and may predict a beneficial response to immune checkpoint inhibitor blockade (117). Just as in peripheral lymphoid organs, CXCL13, CCL19, and CCL21 are the chemokines most associated with tertiary lymphoid structure T cell and B cell spatial organization, with additional contributions linked to CCL2, CCL3, CCL4, CCL5, CCL8, CXCL9, CXCL10, and CXCL11 (118). To further complicate our understanding of the contribution of tertiary lymphoid structures to tumor behavior, regression of tertiary lymphoid structures after neoadjuvant chemotherapy has been linked with longer patient survival while corticosteroids used to manage side effects of chemotherapy may reduce their density and limit their beneficial anticancer effects (117).

CCR8, CCR10, and Tregs

Despite Tregs expressing multiple chemokine receptors, the specific receptor(s) that directly or redundantly control their infiltration into tumors is uncertain (119). CCR8, previously established as a receptor for CCL1, has recently been shown to bind the human chemokine CCL18 produced by antigen-presenting cells (120). CCR8 is expressed among multiple subtypes of T cells, including activated and memory Th2 CD4+ T cells, Tregs, as well as monocytes and macrophages. In addition to CCR8+ Tregs enforcing immune suppression, they may also facilitate tumor-immune tolerization (121). Given its potential role in immune suppression and evasion, preclinical studies inhibiting CCR8 have shown promise in a mouse model of colorectal cancer, with human-specific inhibitors undergoing clinical trials as monotherapy or in combination with immune checkpoint inhibition (Table 1) (122). Although initial phase 1 trials of CCR4 showed promise in blocking Treg movement into tumors, phase 2 trials demonstrated marginal efficacy (Table 1) (38, 123). Originally described as a regulator of effector T cell migration to the dermis (124, 125), CCR10 expression on Tregs was upregulated by hypoxia and evoked immune suppression in ovarian cancer (126). Further reports implicate CCR10 expression on melanocytes (91, 127), with a separate unbiased screen of human and murine pancreas cancer cell lines revealing increased expression of CCL28 production by tumor cells directing the trafficking of pancreatic cancer stellate cells (91).

Phagocyte Recruitment and Angiogenesis

CCR2, along with CXCR1 and CXCR2, play central role in recruitment of tumor-associated macrophages (TAMs), MDSCs, neutrophils, mast cells, and basophils. Neutrophils have many protumorigenic effects and a high neutrophil-to-lymphocyte ratio is associated with poor prognosis across multiple tumor types (128). CCR2 is expressed by monocytes and is responsible for recruiting monocytes to tissue sites where those cells undergo differentiation into resident macrophages. The receptor can be activated, with nearly equal potency, by CCL2, CCL7, CCL8, CCL13, and CCL16, and is among the most promiscuous GPCRs of the chemokine family. Consistent with its roles in host defense and innate immunity, CCR2 and its ligands are key participants in tumor-promoting inflammatory reactions early in cancer development as well as immune-inflamed type malignancies. TAMs may be recruited to the developing or established tumor as monocytes whereupon they differentiate into TAMs or they may also arise from tissue-resident macrophages (129). TAMs have previously been classified into M1 and M2 TAMs. Although this simplistic classification does not fully capture the heterogeneity seen in macrophage populations, it is sufficient to note that chemokines may functionally polarize TAMs into an inflammatory M1 phenotype or a wound-healing M2 phenotype independent of their roles in cell migration (130). MDSCs are protumorigenic due to their ability to suppress T cell activation and can be differentiated from TAMs by their low expression of F4/80 (131, 132). Canonically, ∼80% of MDSCs within solid tumors are polymorphonuclear/granulocytic MDSCs, with the other 20% being monocytic MDSCs subset (133). Congruent with reprogramming TAMs into more suppressive, wound repair-focused cells, CCR2 appears to also be sufficient for the recruitment of both granulocytic and monocytic MDSCs in both immune-excluded and immune-desert phenotypes (134). Complicating matters further, CCR2-expressing M1-type TAMs appear critical for effective antitumor immune responses (135). The heterogeneity of functions of these CCR2+ cells and its marked ligand redundancy may restrict the utility of therapeutically targeting the receptor (37). Current studies aiming to inhibit CCR2 do so in combination with CCR5 inhibition, a receptor that is also expressed on both types of MDSCs (39, 134). In addition, dual CCR2/CCR5 inhibition is done in tandem with multiple other treatments, such as with an anticancer vaccine or CXCL8 inhibition (Table 1).

CXCR1 is the receptor for the ELR-motif chemokines CXCL6 and CXCL8, whereas CXCR2 is the receptor for other ELR-motif chemokines CXCL1, CXCL2, CXCL3, CXCL5, and CXCL7. Although the causes of cancer neutrophilia are multifactorial, the ELR-motif chemokines CXCL1, CXCL2, and CXCL5 promote the release of CXCR2-expressing neutrophils from the bone marrow (136, 137). Protumorigenic inflammation results in upregulated expression of these ligands in neoplasia while tumor hypoxia may sustain their expression in malignant tumor (138). Activated neutrophils also release neutrophil extracellular traps composed of DNA histone complexes and proteins. These neutrophil extracellular traps are thought to increase the ability of cancer to proliferate and metastasize by activating dormant cancer cells and entrapping circulating cancer cells to enhance new tumor formation (139). However, not all tumor-associated neutrophils (TANs) are protumorigenic. Recent studies have classified antitumorigenic “N1”-type TANs and TGF-β-dependent protumorigenic “N2”-type TANs (140). N2-type TANs have increased levels of arginase, as well as the chemokines CCL2, CCL5, and CCL17 known to recruit MDSC and Tregs (141). N1 TANs, on the other hand, have increased TNF production and can activate CD8+ T cells, likely through a unique ability to cross-present tumor antigens (142). TANs typically do not circulate for extended periods of time and are thought to be polarized once they have infiltrated the tumor under the control of variably expressed chemokine receptors on N1 or N2 TAN subsets (142). N1-type TANs have been found to have higher levels of CCR5 and CCR7, chemokine receptors typically found on DC, whereas N2-type TANs largely express CXCR2 (143).

Mast cells express multiple chemokine receptors, such as CCR3, CXCR1, CXCR2, and CXCR4 (144). Once in the tumor microenvironment, mast cells are believed to promote tumorigenesis through the production of both proangiogenic molecules and immunosuppressive cytokines (145, 146). Contrasting reports indicate antitumor functions of histamine produced by mast cells (147). Basophils are similar to mast cells in both their chemokine receptor expression and conflicting information on their functions in tumorigenesis (148). The chemokine receptor most associated with eosinophils prototypically expresses CCR3, as well as CCR1 and CCR5 (149), and like the other granulocyte lineage mast cells and basophils produces angiogenic factors (150).

In addition to roles in MDSC, M2-type TAM and N2-type TAN trafficking, CXCR1 and CXCR2 also have tumor-promoting roles in angiogenesis. Increased angiogenesis is crucial for adequate nutrient delivery and waste removal for tumor cells, and it can also aid in promoting dissedata mination of metastatic tumor cells. Although CXCR1 and CXCR2 inhibitors have shown promise in preclinical studies, preliminary data from one recently completed study using dual CXCR1 and CXCR2 inhibition in combination with tubulin-targeted chemotherapy (Paclitaxel) showed no benefit (35). However, multiple ongoing studies are looking at the effects of single or dual inhibition of CXCR1 and CXCR2 with immune checkpoint inhibitors (Table 1). Indirect angiogenesis, where monocytes are recruited and secrete angiogenic factors such as VEGF, provides another avenue whereby chemokines contribute to neovascularization. Oxygen levels are important for chemokine regulation as hypoxia-inducible factor 1α upregulates expression of different chemokine receptors or ligands (121). These data provide a plausible mechanism for resistance to antiangiogenic therapy in cancer as the increased hypoxia resulting from antiangiogenic drugs may ultimately promote downregulation of inflammatory chemokines and upregulation of angiogenic and immunosuppressive chemokines. As multiple CXCR2 antagonists are currently in clinical trials, combinatorial with antiangiogenic therapy may both prevent resistance and overcome hypoxia-mediated immunosuppression (151–153). Although these receptors have key roles in leukocyte recruitment, the influence of genetic, epigenetic, metabolic, or damage repair mechanisms responsible for the spatial and temporal changes in their ligand production continues to be established. Moreover, the pronounced redundancy of ligands that activate CCR2, CXCR1, and CXCR2 makes therapeutic intervention challenging.

Mucosal Chemokines

Four chemokines, CCL25, CCL28, CXCL14, and CXCL17, have been termed “mucosal chemokines” because they are the predominant chemoattractants expressed at mucosal tissues. First classified by Zlotnik (154), there is still much to be discovered about the function of mucosal chemokines in inflammation and cancer. Although many chemokines play important roles in the spatial organization of immune cells within mucosal tissues, this chemokine subfamily is abundantly expressed in noninflamed conditions and is thought to have crucial roles for maintaining homeostatic lymphocyte trafficking and provides immediate host defense against infections. CCL28 attracts CCR10+ IgA plasma cells, activated T cells, and Tregs (155, 156). CCL25 recruits CCR9+ T cells to the small intestine (157). Although their cognate receptors remain unknown, both CXCL14 and CXCL17 recruit monocyte populations (158–160). CXCL17 stimulates macrophage migration and angiogenesis in culture models (161) and has been linked with inflammatory fibrosis (160) and tumorigenesis (162, 163). Consistent with mucosal tissues being contiguous with the external environment, another key facet of mucosal chemokines is that, at elevated concentrations, they may possess broad antimicrobial activity, a characteristic not typically seen in other chemokines (154).

CLOSING THOUGHTS

In an age where cancer is increasingly thought of as a disorder in tumor-promoting inflammation that evolves into immune suppression and evasion, studying how tumors hijack and disrupt the body’s natural chemokine gradients provides a pathway for improving existing or emerging anticancer agents. Similar to their roles in cancer progression, chemokines demonstrate a complicated relationship to antitumor chemotherapy. Different types of chemotherapy may increase expression of a diversity of chemokine ligands and receptors in cancers, a response that may potentiate the antitumor effects or, in contrast, weaken tumor reactivity and promote therapeutic resistance. Many cytotoxic therapies can promote macrophage and DC engulfment of dying tumor cells and subsequent antigen presentation in a dose-dependent manner, enhancing recruitment of tumor-reactive T cells, DCs, and macrophages. Transient lymphopenia resulting from some treatments can provoke an inflammatory response capable of a delayed effector T cell response days later. Reports also indicate that therapy-induced secretion of proinflammatory members of the chemokine family can increase pain hypersensitivity and peripheral neuropathy in treated patients (164, 165). Alternatively, CXCL1 and CXCL2 may become elevated after chemotherapy or radiation and function in recruiting immunosuppressive cells, triggering wound-healing responses, eliciting antiapoptotic signaling, and/or stimulating angiogenesis and cancer dissemination that participate in tumor recurrence.

Chemokines are a multidimensional system that includes ligand and receptor expression, redundancy in ligand-receptor utilization, and discrete pharmacology activated by structural nuances in ligand binding. Accurately understanding how chemokines influence cancer immunity will require examination of larger dose-response curves and definition of the receptor internalization and signaling. The ability to shift the tumor immune microenvironment from tumor-promoting to antitumorigenic through the combined use of chemokine-targeted therapies, designed from a comprehensive understanding of chemokine physiology, could both improve outcomes and lessen toxicity compared with traditional chemotherapy or immunotherapy regimens.

GRANTS

D.D. is a predoctoral fellow of the MCW Medical Scientist Training Program, T32 GM080202, which is partially supported by a training grant from NIGMS. B.T.C. is supported in part by an award from the Children’s Research Institute, Children’s Wisconsin, and an NCATS KL2 Mentored Career Development Award KL2TR001438. M.B.D. is supported in part by a grant from the National Cancer Institute, R01 CA226279, and continuing philanthropic support from the Hanis-Stepka-Rettig Endowed Chair in Cancer Research and the Bobbie Nick Voss Charitable Foundation.

DISCLAIMERS

The content is solely the responsibility of the author(s) and does not necessarily represent the official views of the NIH.

DISCLOSURES

M.B.D. is a co-founder and has ownership and financial interests in Protein Foundry, LLC, and Xlock Biosciences, LLC. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

D.D. conceived and designed research; M.B.D. prepared figures; D.D. drafted manuscript; D.D., B.T.C., and M.B.D. edited and revised manuscript; D.D., B.T.C., and M.B.D. approved final version of manuscript.