Contribution of CXCR3-Mediated Signaling in the Metastatic Cascade of Solid Malignancies

Abstract

Metastasis is a significant cause of the mortality resulting from solid malignancies. The process of metastasis is complex and is regulated by numerous cancer cell-intrinsic and - extrinsic factors. CXCR3 is a chemokine receptor that is frequently expressed by cancer cells, endothelial cells and immune cells. CXCR3A signaling in cancer cells tends to promote the invasive and migratory phenotype of cancer cells. Indirectly, CXCR3 modulates the anti-tumor immune response resulting in variable effects that can permit or inhibit metastatic progression. Finally, the activity of CXCR3B in endothelial cells is generally angiostatic, which limits the access of cancer cells to key conduits to secondary sites. However, the interaction of these activities within a tumor and the presence of opposing CXCR3 splice variants clouds the picture of the role of CXCR3 in metastasis. Consequently, thorough analysis of the contributions of CXCR3 to cancer metastasis is necessary. This review is an in-depth examination of the involvement of CXCR3 in the metastatic process of solid malignancies.

1. Introduction

Malignancies are the second leading cause of death in the United States [1]. Metastasis is not only a defining feature of cancer but is also a significant cause of cancer-related mortality. Metastatic disease contributes to organ dysfunction, which can be fatal depending on the secondary site and the extent of involvement. Moreover, metastases, almost categorically, preclude surgical management of cancer, thereby limiting utilization of the most effective treatment modality and necessitating the use of chemotherapies. The multifaceted nature of the contribution of metastases to cancer-related mortality makes it challenging to quantify the impact of metastasis on patient outcomes; however, conservative and liberal estimates indicate that between 66% and 90% of cancer deaths are related to metastatic disease progression, respectively [2–4]. These statistics highlight the crucial role that metastasis plays in cancer biology and patient outcomes.

Metastasis in solid malignancies proceeds through a long cascade involving invasion, intra- and extravasation, and colonization. This process begins with the invasion of cancer cells into the surrounding tissue [5]. As they invade, cancer cells encounter blood and lymphatic vessels, which act as conduits for cancer cells to secondary sites [5]. Upon entering the vessels, cancer cells must survive the loss of attachment [6, 7], fluid shear stress [8, 9], and the immune cell-rich environment [10, 11] and ultimately attach to the endothelium at a distant site. Once arrested at the distant site, cancer cells must escape from the vasculature through a process that is believed to mirror leukocyte extravasation [5]. Finally, cancer cells must adapt to the microenvironment of the secondary site and expand to produce a clinically detectable metastatic lesion [5]. Each of these steps of metastasis presents a unique set of challenges to the metastasizing cancer cells. Moreover, the complex nature of the metastatic process dictates that molecules may have competing effects during different phases of the metastatic process, making a thorough analysis of metastasis data both difficult and requisite for understanding the molecular mechanisms of metastasis.

CXCR3 is the receptor of the C-X-C domain-containing chemokines PF4 (CXCL4), PF4V1 (CXCL4L1), CXCL9, CXCL10, and CXCL11. Initially, CXCR3 was characterized as a chemokine receptor expressed predominantly on T-lymphocytes and subsets of natural killer (NK) cells [12]; however, subsequent studies revealed that CXCR3 is expressed on a wide variety of cells including endothelial cells and cancer cells. Like many other chemokine receptors, CXCR3 plays critical roles in the progression of numerous malignancies. These functions include modulating tumor immune infiltrate, promoting tumor growth, and influencing tumor angiogenesis, which collectively facilitate the metastatic process. However, because of the diversity in its expression and functions, the role of CXCR3 in metastasis is multifaceted, resulting in competing effects between the functions of CXCR3 splice variants and CXCR3 expression on different cell types. This review presents a detailed analysis of the nuances of direct and indirect involvement of CXCR3 in the metastatic processes of solid malignancies.

2. Biochemistry of CXCR3

An understanding of the biochemistry of a molecule is fundamental in developing an overall framework of its contributions to pathology. Currently, there are three described splice variants of CXCR3: CXCR3A, CXCR3B, and CXCR3 Alt. CXCR3A and CXCR3B constitute the vast majority of CXCR3 expression, and due to the lack of studies on CXCR3 alt, this review focuses solely on CXCR3A and B. Both CXCR3A and CXCR3B encode seven predicted transmembrane domains and both proteins couple with heterotrimeric G-proteins (HTGPs), which mediate their downstream signaling. In terms of sequence, the only difference between CXCR3A and B is in the N-terminus [13]. The coding sequence of CXCR3A includes three amino acids derived from exon 1, with the remainder being derived from exon 2. In contrast, the coding sequence of CXCR3B is derived from exon 2 and a retained intronic sequence that is contiguous with the 5’ end of exon 2 (Figure 1). By comparison, CXCR3B has 48 amino acids at the N-terminus that are not present in CXCR3A. Despite their sequence similarities, the behaviors of CXCR3A and B are vastly different, and appreciating these differences is paramount to understanding CXCR3 in the context of metastasis. For instance, CXCR3A is a high-affinity receptor for CXCL9, CXCL10, and CXCL11 [13], whereas CXCR3B has high affinity for CXCL4 and CXCL10 (roughly four-fold higher than its affinity for CXCL9 or 11) [14]. Interestingly, CXCL4 does not appear to bind or activate CXCR3A, but CXCL4L1 has intermediate affinities for both CXCR3A and B [14]. Despite having lower affinities for CXCL9 and CXCL11, CXCR3B can be activated by these ligands, though only at very high concentrations. It is essential to understand the differences in ligand binding between CXCR3A and B as these variants have different functions during metastasis, and thus, the ligands of each receptor produce different, sometimes conflicting, outcomes.

Figure 1.

Schematic of CXCR3 chromosomal organization and signaling. Top). CXCR3 is located on the X chromosome and is composed of two exonic segments. Differential splicing of CXCR3 gives rise to three described splice variants A, B and Alt. CXCR3B is encoded entirely by a retained intronic sequence and exon 2. Bottom) CXCR3 A and B signaling is mediated by heterotrimeric G proteins (HTGP). CXCR3A couples with Gai/q and results in downstream activation of Ras/Raf/ Erk as well as PI3K/Akt signaling leading to cell proliferation, survival, and migration/invasion. CXCR3B signals through Gas resulting in activation of adenyl cyclase and PKA as well as p38 and p21 activation leading to sensitization of cells to death.

In addition to ligand binding, the differences in splicing between CXCR3A and B alter the coupling of the two receptors to HTGPs (Figure 1). CXCR3A is believed to couple to Gαi and Gαq (though credible data for Gαq coupling is limited) HTGPs on the basis that treatment of cells with pertussis toxin and knockout of Gαi both inhibit the cellular effects of activation of CXCR3A [14, 15]. Downstream of Gαi and Gαq, the understanding of CXCR3A signaling is fragmented. However, it is believed that activation of Gαi results in downstream activation of SRC kinase, and subsequently, MAPK and PI3K/AKT signaling, leading to increased proliferation of cells and resistance to apoptosis, respectively. Additionally, recruitment of β-arrestin 2 to activated CXCR3 contributes significantly to the non-canonical activation of SRC and subsequent activation of pro-proliferative and anti-apoptotic pathways. Activation of Gαq downstream of CXCR3A initiates phospholipase C-β (PLC-β) signaling leading to phosphatidylinositol 4,5-bisphosphate (PIP2) cleavage to form diacylglycerol (DAG) and inositol-triphosphate (IP3). DAG and IP3, in turn, stimulate release of calcium from the endoplasmic reticulum resulting in pleiotropic cellular effects, including activation of calpains, which can contribute to cell migration. While the signaling downstream of CXCR3A is mainly consistent across ligands, recent reports have suggested the possibility of biased signaling downstream of activation of this splice variant [16, 17].

In general, the signaling pathways downstream of CXCR3A appear to be consistent regardless of the cell type in which the signaling is occurring. However, the cellular response to the activation of these signaling pathways appears to be cell type specific. In cancer cells, signaling downstream of CXCR3A appears to promote proliferation, as well as migration and invasions thereby contributing to the metastatic cascade. In contrast, the signaling in T-cells largely culminates in the homing of T-cells to sites of inflammation or secondary lymphoid organs. The process of T-cell homing is multifaceted and within this process, CXCR3 has roles in both directing migration and facilitating arrest of T-cells at endothelium. Initial studies of the molecular mechanism of involvement of CXCR3 in T-cell homing demonstrated that the CXCR3-mediated migration was sensitive to pertussis toxin, suggesting the involvement of Gαi [18]. Further work by Thompson et al. demonstrated that knockout of Gαi2 in mice markedly decreased the migratory capacity of T-cell in response to CXCR3 ligands [15]. Additional studies have demonstrated that PLC activation is an additional factor in CXCR3-mediated chemotaxis, which suggests that Ca²⁺ signaling may be involved [18]. Interestingly, the activation of PLC appears to link CXCR3 activation with T-cell receptor signaling through ZAP70 and Lck [19, 20]. Finally, using biased CXCR3 agonists, Smith et al. were able to demonstrate that β-arrestin signaling is a functional requirement of CXCR3-mediated T-cell chemotaxis [21]. It is important to note that CXCR3 function in the immune system is not restricted to T-cells, and that new roles in B-cells, NK-cells as well as myeloid lineage cells are an area of active investigation. Similarly, recent studies have demonstrated that CXCR3 may have important functions in the maturation, polarization, and effector functions of T-cells [22]. However, the cellular and molecular basis of these functions remain poorly characterized and largely beyond the scope of this review.

In contrast to CXCR3 A, CXCR3B is thought to couple to Gαs based on findings that activation of CXCR3B in cells ectopically expressing CXCR3B, specifically, results in a rise in intracellular cAMP production, which is augmented in the presence of forskolin [14]. However, this evidence does not rule out the possibility that CXCR3B couples with Gα_olf HTGPs, which also stimulate adenylate cyclase activity [23]. Given the opposing nature of the Gα subunits transmitting their respective signals, the signaling downstream of CXCR3B is quite different from that induced by CXCR3A. CXCR3B-mediated signaling proceeds through activation of PKA. While the roles of PKA have been widely researched as a whole, its contribution to metastasis is less well defined, though it is generally believed to suppress the malignant features of cancer cells. Additionally, CXCR3B activates p38, which, in turn, activates Bach1 and p21, thereby sensitizing cells to redox stress and suppressing progression through the cell cycle, respectively [24, 25]. Furthermore, the signaling downstream of CXCR3B activation is consistent irrespective of the activating ligand. It must also be mentioned that activation of any HTGP results in the release and activation of Gβγ subunits from the trimeric complex resulting in the activation of the PLC-β and, ultimately, the release of calcium from the endoplasmic reticulum. Consequently, both CXCR3A and B signal through this mechanism [26].

3. Role of CXCR3 in Metastasis

The expression of CXCR3 by diverse cell types in the tumor microenvironment allows CXCR3 to play several roles in the metastatic process. Malignant cells of numerous cancer types express both splice variants of CXCR3, and activation of CXCR3 directly modulates the metastatic behavior of these cells. In addition to this direct role, the expression of CXCR3A on lymphocytes indirectly affects the ability of malignant cells to metastasize. Similarly, the expression of CXCR3B on endothelial cells has profound effects on angiogenesis, which alters the availability of vessels for the transport of cancer cells to secondary sites. The following section reviews the data regarding the direct and indirect roles of CXCR3 in the metastatic processes of several malignancies, and Figure 2 presents a schematic of the trends in the roles of CXCR3 in the metastatic process.

Figure 2.

Schematic of the involvement of CXCR3 in the invasion-metastasis cascade. In the primary tumor CXCR3 ligands are primarily derived from surrounding stromal cells including macrophages and fibroblasts. Here, CXCR3A and B have opposing roles with respect to cancer cell invasion. CXCR3A promotes tumor cell invasion, migration, and intravasation largely through increased expression of matrix metalloproteinases. In contrast CXCR3B has an inhibitory effect on these processes and limits access of cancer cells to vasculature through its angiostatic activity. Additionally, CXCR3 is involved in the recruitment of T-cells and NK-cells to the TME as well as the differentiation of these cells into functionally important subsets. The effect of immune cell recruitment and differentiation on the overall metastatic process appears to context dependent. During the intravascular phase of metastasis, CXCR3 ligands are largely derived from platelets in the form of PF4. CXCR3B may augment cancer cell survival by augmenting stem-like phenotype in cancer cells. Finally, at the secondary site, the pro-invasive and immunomodulatory effects of CXCR3A can play critical roles in the establishment of a metastatic niche and ultimately colonization of the distant site.

3.1. Direct Contribution of CXCR3 to Metastasis

While the role of CXCR3 in metastasis is multifaceted, prostate cancer (PCa), melanoma, and colorectal cancer (CRC) serve as prime examples in which CXCR3 contributes directly to the metastatic process. In these settings, the activation of CXCR3 promotes the invasive phenotypes of cancer cells. Importantly CXCR3A is the key driver of this invasive phenotype, as demonstrated by a study of CXCL11-induced chemotaxis. Here, CXCR3 inhibition suppressed cancer cell chemotaxis, specifically in cancer cells that had comparatively high CXCR3A: CXCR3B ratios [27]. Interestingly, in those cells that had low CXCR3A: CXCR3B ratios, CXCL11 still promoted chemotaxis, which was augmented by inhibition of CXCR3 [27], thereby highlighting the importance of differentiating between CXCR3 splice variants. Corresponding to these changes in malignant cell phenotype, high expression of CXCR3, or its ligands, in cancer cells is associated with poor prognosis in patients and increased stage at diagnosis [28–35]. Moreover, this mechanism of CXCR3A’s involvement in metastasis has been observed in several other cancers. In pancreatic cancer (PC) [36, 37], oral squamous cell carcinoma [38], malignant glioma [39], and osteosarcoma [40], CXCR3 activation promotes the invasion and/or migration of cancer cells. Additionally, these studies independently demonstrate that CXCR3 functions to promote this activity through pathways including activation of AKT [38] and ERK [36], as well as increased expression of matrix metalloproteinases (MMPs) [40] further highlighting that these pathways are primary signaling events involved in CXCR3A’s function in cancer cells. In the settings of PC [36], osteosarcoma, and oral squamous cell carcinomas, CXCR3 was further connected to metastatic phenotypes through demonstration of robust CXCR3 expression in metastatic lesions, and suppressed metastasis in a tail vein injection model with CXCR3 loss [40], and association of primary tumor CXCR3 expression with the presence of metastatic disease at diagnosis [38].

Additionally, because of the role of CXCR3 in the homing of T-cells and its notable contribution to the dermato-tropism of cutaneous T-cell lymphomas, CXCR3 has long been suspected of being involved in the tropism of metastasizing tumor cells. Studies by Kawada et al. demonstrated that CXCR3 knockout significantly suppressed the formation lymph node metastasis in melanoma models [41]. Notably, lymph node metastasis was augmented by administration of adjuvants that culminated in the upregulation of CXCR3 A ligands. Similar findings were noted for CXCR3 in colorectal cancer by the same group. Here CXCR3 expression was associated with lymph node metastasis, but not with those of the liver or lung metastasis [29]. Cumulatively, these findings suggest that CXCR3 may be involved in the tropism of tumor cells to lymph nodes. Other studies have investigated the potential of CXCR3-mediated organotropism in hematogenous dissemination. Here, a study noted that inhibition of CXCR3 suppressed pulmonary metastasis in a tail vein injection model but did not affect liver metastasis following splenic injection [42]. The conclusions that can be drawn from this study are limited by the use of two different models to demonstrate a differential effect of CXCR3 inhibition on organ-specific metastasis. Finally, Doron et al. demonstrated that knockdown of CXCR3 in melanoma cells suppressed the formation of brain metastases in murine models. Importantly, they found that CXCL10 expressed by activated astrocytes was critical for driving the migration of melanoma cells [43]. These findings support the potential role of CXCR3 signaling in driving metastasis of CXCR3-positive cancer cells to inflamed sites. However, the extent to which this mechanism is involved in melanoma brain metastasis remains difficult to determine primarily due to a limited understanding of the underlying mechanism driving CXCL10 expression by astrocytes. Overall, the role of CXCR3 in organotropism will be a key area of future research to comprehensively define the functions of CXCR3 in metastasis.

Despite several studies providing strong evidence for CXCR3-mediated promotion of cancer cell invasion and/or migration, it must be noted that these effects were not observed in all studies, indicating that this role of CXCR3 is potentially context-dependent [44–46]. Such disparate findings regarding the function of CXCR3 in metastasis raises important considerations for studying and ultimately understanding the role of CXCR3 in complex biological systems (Table 1). Most notably, it is critical to differentiate between the expression profiles and function of the two CXCR3 splice variants. Cancer cells can express any combination of CXCR3A and B isoforms, ranging from no expression of CXCR3, to expression of a single splice variant, to mixed expression of both variants. This fact makes determining the expression pattern particularly critical for understanding the broader role of CXCR3 in an experimental system.

Table 1.

Summary of the Direct Roles of CXCR3 in Metastasis.

Cancer	Splice Variant	Cellular Function	Citations
Prostate	CXCR3A	Augment invasion and proliferation	[46,47]
	CXCR3B	Suppress invasion and proliferation	[46,47]
Melanoma	CXCR3A	Augment invasion and brain metastasis in vivo	[41,50–52]
Colorectal	CXCR3A	Increased proliferation invasion, migration, lymph node and tail-vein metastasis	[29,42,59–62]
	CXCR3B	Inhibition of proliferation, invasion and migration	[63]
Breast	CXCR3A	Increased migration, and cardiac -injection bone metastasis	[72, 74, 94]
	CXCR3B	Increased Cancer Stem Cell Features and in vivo metastasis	[65, 95]
Ovarian	CXCR3	Increased migration to malignant ascites	[98]
Gastric	CXCR3A	Increased migration, invasion, and liver metastases	[28, 102]
Hepatocellular Carcinoma	CXCR3	Increased invasion, migration, tail-vein lung metastases	[103–105]
Pancreatic	CXCR3A	Increased invasion	[36,37]
Lung	CXCR3	Stimulate migration	[125]
Renal Cell Carcinoma	CXCR3A	Induced migration	[32]
	CXCR3B	Suppressed proliferation Increased Apoptosis	[24,129]

3.1.1. Prostate Cancer

Most studies of CXCR3 in PCa have focused on CXCR3 function in malignant cells. In this context, the functions of CXCR3 in promoting metastatic spread are divided between the splice variants. One study showed that DU-145, a metastatic PCa cell line, had higher CXCR3A expression relative to CXCR3B as compared to RWPE-1, a non-cancerous, prostate epithelial cell line, and LNCaP, a primary tumor-derived PCa line [47]. For DU-145, CXCL10 and CXCL4 treatment increased cellular motility and invasiveness in vitro, which was not observed with RWPE-1 cells. Interestingly, overexpression of CXCR3B in DU-145 abrogated motility and invasiveness suggesting that the pro-metastatic effects of CXCR3 in PCa cells is mediated by CXCR3A and is opposed by CXCR3B signaling. Consistent with these findings, knockdown of CXCR3A in PC-3 PCa cells and overexpression of CXCR3B inhibited proliferation and invasion in vitro [48], further highlighting the disparate roles of CXCR3A and B in PCa metastasis. In patient samples, low CXCL4L1 expression, which can activate both CXCR3A and B, was associated with higher pathologic stage, greater Gleason score, and poor overall survival (OS) [49], suggesting that the anti-metastatic role of CXCR3B may predominate in human PCa; however, this conclusion is confounded by potential changes in tumor immune response and angiogenesis produced by CXCR3 activation.

The findings of the few studies conducted on CXCR3-mediated immune modulation provide important insights into the overall role of CXCR3 in the metastatic process, which may apply to multiple cancers. One study by Hu et al. demonstrated that PCa cells recruit CD4⁺ T-cells through the secretion of CXCL9. In turn, the recruitment of CD4⁺ cells resulted in the upregulation of FGF11 and downregulation of androgen receptor [50]. Together, the changes in the expression of these molecules increased the invasion and migration of PCa cell lines in vitro and increased the metastatic spread of cancer cell lines implanted in the prostates of nude mice co-transplanted with CD4⁺ T-cell lines. These findings are consistent with those from breast cancer, further suggesting that CXCR3 may promote the spread of cancer cells through an immune mechanism. However, it is difficult to know if this metastasis-promoting mechanism would outweigh the effect of an increased immune cell infiltrate in the presence of a complete host immune system.

3.1.2. Melanoma

Melanoma represents one of the most thoroughly researched malignancies in terms of the function of CXCR3. Interestingly, it is also a disease in which metastatic dissemination has a tremendous prognostic impact. Therefore, the contribution of CXCR3 to the metastatic process may be particularly critical in melanoma biology and patient outcomes. In a seminal study by Kawada et al., knockdown (KD) of CXCR3 in B16F10 melanoma cells followed by subcutaneous implantation showed that the CXCR3 KD cells produced fewer lymph node metastases compared to wild-type, CXCR3-expressing cells [41]. Consistently, treatment of mice with Freund’s adjuvant resulted in upregulation of CXCL9 and 10 in tumor-draining lymph nodes and, in turn, caused a dramatic increase in the number of tumor-positive lymph nodes, which was abrogated by the neutralization of CXCL9 and CXCL10 [41]. Subsequent studies showed that CXCR3 induced activation of the ERK signaling pathway, which is responsible for cellular invasion, as well as increased lymph node metastasis [51]. Similarly, another group studied highly metastatic and non-metastatic subclones of B16F10 cells and found a strong expression of CXCL10 in the metastatic subclones. Follow-up analyses demonstrated that silencing of CXCL10 or CXCR3 suppressed the metastatic ability of both sub-lines compared to non-targeted siRNA in tail vein models [52]. Finally, a very recent study suggested that CXCL10 expressed by glial cells of melanoma patients acts to promote brain-tropic metastasis of melanoma cells. Here, the knockdown of CXCR3 in melanoma cells abrogated brain metastasis in a murine model. Mechanistically, the function of CXCR3 in melanoma brain metastasis was theorized to occur through activation of integrins in a fashion parallel to that of T-cell migration. However, details of this mechanism and those regarding CXCL10 expression in the brain remain poorly characterized [43]. Analysis of correlations of clinicopathological features of melanoma patients with their CXCR3 expression levels may elucidate the potential effects that predominate in the human disease. In one study of melanoma cells, high CXCR3 expression in 82 patients was associated with increased Breslow depth and the presence of distant metastasis [53]. These findings suggest that CXCR3 function in melanoma cells may predominately function to promote metastatic dissemination.

In contrast, Antonicelli et al., demonstrated that treatment of B16F1 murine melanoma cells with CXCL10 modestly suppressed their invasive capacity in vitro [45]. Moreover, the investigators showed an association of high CXCL10 expressing peripheral blood mononuclear cells (PBMCs) with remission of melanoma, which was, in turn, associated with localized disease at time of diagnosis [45]. It is unclear, however, if this association is the result of CXCL10/CXCR3 signaling in cancer cells or immune cells as tumor-infiltrating lymphocytes were not assessed in these patients nor were similar studies conducted in mice lacking critical immune components. Similarly, treatment of lymphatic endothelial cells and melanoma cell lines with IFN-β caused marked upregulation of CXCL10 and CXCL11 [46]. In turn, CXCL10 suppressed melanoma proliferation and invasion in both IFN-β-sensitive and resistant cells [46]. Moreover, loss of CXCL10 through genetic ablation decreased the sensitivity of melanoma cells to IFN-β; notably, these effects were thought to be mediated by CXCR3B [46].

In addition, CXCR3 plays a crucial role in modulating the immune system in melanoma. Cell lines that expressed CXCR3 ligands recruited CD8⁺ T-cells both in vitro and when transplanted into mice [54]. Moreover, in CXCR3^−/− mice bearing melanoma, the T-cell infiltration into tumors was diminished, and immune checkpoint therapy failed to reduce tumor progression [55]. Similar findings were confirmed by TCGA dataset analysis and histological correlations with cytokine expression in melanomas in which high CXCR3 ligand expression was associated with improved patient survival [56, 57]. These findings implicated CXCR3 in a robust anti-tumor immune response in melanoma. Subsequently, administration of a CXCL10 expression plasmid as DNA-based therapy to mice bearing B16F10 melanoma tumors caused loss of pulmonary metastases in a NK cell-dependent manner [58]. Finally, the treatment with human adipose mesenchymal stem cells ectopically expressing CXCL10 resulted in the disappearance of metastatic lesions from the lungs in a tail vein injection model. Here, CXCL10-overexpressing cells were shown to potentially affect numerous factors in the metastatic microenvironment, including augmented melanoma cell apoptosis, suppressed T-regulatory cell (T-reg) infiltration, increased activated T-cell infiltration, and reduced angiogenesis in lung colonies [59]. In the setting of established metastases, a combination of angiostatic effects and augmented immune response likely mediate these changes. However, CXCR3 expression in melanoma cells is associated with decreased lymphocytic infiltrate, which is consistent with CXCR3-mediated PD-L1 regulation, as observed in gastric cancer [16, 53]. This distinction highlights the significance of compartment-specific CXCR3 expression.

3.1.3. Colorectal Cancer (CRC)

As with melanoma, CXCR3 has been intensively studied for its involvement in CRC metastasis, however, most studies have focused on CXCR3 signaling in cancer cells. These studies, combined with analysis of multiple CRC patient cohorts, provide significant insight into the overarching roles of the CXCR3 axis in the CRC metastatic process. Initially, Kawada and colleagues demonstrated that CXCR3-overexpressing DLD-1 cells had more rapid dissemination through the lymphatic system in rectal transplantation models [29]. Interestingly, metastasis to the liver and lung were infrequent during the reported study and was not different between CXCR3 overexpressing and wild-type control cells. In patient tumors, CXCR3 expression was associated with lymph node metastasis and poor overall survival compared to absence of CXCR3 expression in the tumors [29]. A follow-up study demonstrated that CXCR3 and CXCR4 expression was higher in lymph node and liver metastases in patient samples and that activated CXCR3 cooperated with CXCR4 to increase migration in a manner that was dependent on CXCR4 [60, 61]. Later, the association of these two receptors was shown to occur through an atypical interaction in which CXCR3 increased CXCR4 surface retention, thereby mediating augmented signaling downstream of CXCR4 [61]. Under in vivo conditions, the loss of CXCR3, CXCR4, or both the receptors abrogated lymph node and liver metastasis in rectal transplantation models [60]. Similarly, knockdown of CXCL11 in CRC cell lines resulted in a loss of invasive/migratory potential corresponding to a loss of epithelial-mesenchymal transition (EMT) phenotype in vitro, which was mirrored in vivo with decreased tumor growth in subcutaneous models (in nude mice) as well as decreased metastasis in a tail-vein injection model [62]. These in vivo findings were shown to occur through the EMT phenotype as rescue of N-cadherin expression in CXCL11-knockdown cells reduced cell proliferation and invasion/migration in cell culture [62]. Partially consistent with these findings, another group demonstrated that treating CRC cells with CXCL9, 10, or 11 increased the proliferation and migration of these cells, which was lost with inhibition of CXCR3 [42, 63]. Together these findings support multiple metastasis-promoting roles for the CXCR3 axis in CRC, which may be important for the overall metastatic process in the early and the late phases of the disease.

Interestingly, in vivo studies using tail-vein and portal-vein injections of CRC cell lines with and without inhibition of CXCR3 demonstrated that CXCR3 inhibition resulted in suppression of pulmonary but not hepatic metastases [42]. While the authors interpreted this as CXCR3-mediated organotropism, this conclusion is not well supported by experimental data. Most notably, the use of different metastatic models to demonstrate organotropism is challenging. In these models, there are numerous differences, which include not only the final site but also the path of the cells to arrive at the indicated sites. For example, compared to a short journey through portal circulation for a cell to arrive in the liver, for a cell to colonize the lung, it must travel through the entire venous system, pass through the high-pressure system of the right heart, and finally reach the lung capillary bed. Thus, while it is explicitly clear that CXCR3 inhibition suppressed the colonies formed following tail vein injection, it is not clear that this was related specifically to the target tissue. To bolster such claims of CXCR3-mediated tropism, left ventricular cardiac injection may be a better model, and these studies would benefit greatly by demonstrating that cells expressing CXCR3 show a predilection for lung colonization (or fail to colonize the liver). In contrast, cells lacking CXCR3 either have no predilection regarding the final site of metastasis or favor the liver.

Finally, a single study has examined potential differences in the function of CXCR3 splice variants in CRC metastasis. Here, the authors demonstrated that while CXCR3 (total protein) was upregulated in CRC tissues, and positively associated with TNM staging, CXCR3B was under-expressed in CRC tissue (at the mRNA level) compared to adjacent normal tissue and correlated inversely with tumor stage [64]. Studies using overexpression of CXCR3 variants in vitro, showed decreased proliferation, invasion, and migration with CXCR3B expression and opposing effects for CXCR3A. While this study did not explicitly study metastasis in animal models, they did evaluate tumorigenicity, a characteristic feature of cancer stem cells (CSCs). In this setting, CXCR3A again was associated with increased tumorigenicity, whereas CXCR3B appeared to inhibit tumor formation [64]. These findings contrast with findings in breast cancer that suggested that CXCR3B-positive cells had CSC phenotypes. Though, in this report, investigation of CSC phenotype and association with known CSC markers was lacking [64, 65].

In addition to mechanistic studies, there are numerous reports of the associations of CXCR3 with clinical features in CRC patients. In multiple studies, CXCR3 expression was positively associated with recurrence [30, 66] as well as with the presence of lymph node and distant metastases [30]. CXCR3 ligands have also been thoroughly studied in human CRC tissues, but the outcomes of these studies are conflicting. One study demonstrated that CXCL9 was associated with decreased metastases as well as improved OS in Kaplan-Meier and Cox Proportional hazards analyses [67]. These findings were supported by additional studies of human CRC tissues, which demonstrated that low CXCL10 expression in stage II and III CRC patients was associated with an increased likelihood of recurrence as well as worse OS [68]. These findings suggest that CXCR3 ligands may have a metastasis-suppressing role in CRC, but molecular associations with CXCR3 were not investigated, leaving considerable questions regarding the biology underlying these associations. However, two studies focusing on CXCL10 showed that high CXCL10 expression in the tumor and levels in the blood were both positively associated with metastatic disease, indicating a potential metastasis-promoting role for CXCR3 ligands [31, 32]. The underlying reason for these discrepancies is unclear; however, further analysis of patient data is inevitably needed to better understand how the expression of these molecules and the resulting activation of CXCR3 affect the CRC metastatic process.

CXCR3 signaling is also involved in anti-CRC immune responses, which suppress the formation of metastasis. Initial studies by Musha et al. showed that CXCR3, along with CCR5, mRNA expression correlated with CD3⁺ T-cell infiltration in CRC tumors, particularly at the invasive edge of the tumor [69]. Subsequent flow cytometry of immune cells from patient tumors revealed that, on average, 75% of tumor-infiltrating CD8⁺ T-cells within tumors were CXCR3⁺, whereas 28% of CD4⁺ T-cells were CXCR3⁺ [69]. Similarly, CXCL10 and CCR5 ligands were associated with more robust Th1-type immune signatures in CRC tissues, and this high Th1-type response was associated with high expression of IFN-γ and Granzyme B in CD8⁺ T-cells. Moreover, high, compared to low, Th1 signature and associated changes in CD8⁺ T-cell gene expression were associated with reduced Tumor/Nodes/Metastases (TNM) stage, particularly the presence of distant metastases at time of diagnosis [70]. Finally, forced expression of CXCL10 in CT26 murine CRC cells suppressed the growth of subcutaneous tumors as well as metastases arising from splenic injection of cancer cells; these suppressive effects were abrogated by NK-cell depletion [71]. Overall, these findings suggest that CXCR3 is crucial for the recruitment and function of T-cells and NK-cells within tumors and that CXCR3-mediated immune modulation suppresses CRC metastasis.

3.2. Immunological Role of CXCR3 in Metastasis

Currently, the contributions of CXCR3 to the anti-tumor immune response and the resultant effects on metastasis are nebulous (Table 2). In breast [72–74], ovarian [35], and gastric cancer [75, 76], CXCR3 has been shown to have roles that promote the anti-tumor immune response, thereby limiting metastasis. Similarly, CXCR3 expression is associated with delayed progression of Barret’s esophagus to esophageal cancer through increasing Th1-type cell recruitment [77]. CXCR3 signaling promotes an anti-cancer immune response in the setting of sarcomas as well. In hemangiosarcoma, treatment of cancer cells with parvovirus modified for the transduction of CXCL10 increased survival and reduced the number of metastases formed in comparison to an unmodified form of the virus likely through an immune-mediated mechanism [78]. Similarly, in osteosarcoma patients, CXCR3 expression correlated positively with survival [79]. Additional analysis of the TCGA osteosarcoma dataset indicated that CXCR3 expression correlated strongly with immune cell-related pathways and, in CIBERSORT analysis, with increased CD8⁺ T-cell signatures.

Table 2.

Summary of Immunologic functions of CXCR3.

Cancer	Function	Consequences	Citations
Prostate	CD4+ T-cell recruitment	Increased cancer cell invasion	[49]
Melanoma	CD8+ T-cell Recruitment, Suppress T-reg function	Decreased metastasis	[54–58]
	Increased PD-L1 expression in cancer cells	Decreased immune infiltrate	[53]
Colorectal	Associated with increased Th1 and CD8+ signatures	and associated with reduced TNM stage	[69–71]
	Increased NK-cell infiltrate	Reduced tumor growth and metastasis in mice	[71]
Breast	Increased NK-cell recruitment, Decreased myeloid cell infiltrate	Increased tail-vein metastasis	[72–74]
	Increased CD4+ T-cell and NK-cell recruitment	Reduced tumor growth and metastasis	[89–91]
Ovarian	Recruitment of Tregs	Undetermined	[99]
	Associated with increased number of T-cells and NK cells	Associated with increased survival	[100]
Gastric	Positively associated with increased CD4+ and CD8+ infiltrates	Associated with increased overall survival	[75, 76]
	Increased PD-L1 expression in cancer cells	Undetermined	[16]
Hepatocellular Carcinoma	Increased Treg recruitment	Increased likelihood of tumor recurrence	[81]
	Increased alternative macrophage activation	Suppressed anti-tumor immune response	[82]
	Alternate macrophage polarization	Suppressed anti-tumor immune response	[83]
Pancreatic Cancer	T-reg recruitment	Association with poor prognosis	[85,86]
	Suppressed CD8+ T-cell function	Association with poor outcome and more rapid tumor progression in mice	[84, 87]

However, CXCR3 also has immunosuppressive functions in many cancers, including breast [80], ovarian, gastric [16], hepatocellular carcinoma [81–83], and pancreatic cancer [84–87]. These disparate results regarding the immunological functions of CXCR3 highlight a natural balance present in the immune response. Moving forward, it will be critical to determine if and how CXCR3 and its signaling affects this balance. In pursuit of this goal, it will be essential to elucidate the differences in CXCR3-mediated recruitment of lymphocytes to tumors, non-recruitment functions of CXCR3 in lymphocytes, and finally, key differences between CXCR3 expression and activity in lymphocytes. Regarding the latter goal, recent findings have indicated that CXCR3-mediated T-cell recruitment may be, in part, independent of CXCR3 activation [88]. Regardless, further study of the CXCR3 concerning the cancer-specific immune response is required to understand the observed associations with this axis better.

3.2.1. Breast Cancer

In breast cancer, CXCR3 has robust expression on cancer cells as well as immune cells. However, there is controversy as to whether CXCR3 promotes or suppresses the metastatic dissemination of breast cancer cells. Inhibition of CXCR3 in mice reduced lung colonization in tail vein injection models of breast cancer metastasis [72, 73]. This suppression of metastatic colony formation caused by CXCR3 inhibition was slightly diminished by depletion of NK cells from recipient mice. However, this was a small percentage of the total change and the primary effect of CXCR3 in this model is likely mediated by signaling in other cell types besides NK cells [72, 73]. Similarly, CXCR3 inhibition did not abrogate metastasis in mice lacking IFN-γ further implicating the immune system [73]. Notably, high CXCR3 expression was associated with poor OS in a subset of breast cancer patients with local disease [73]. These findings in both humans and mice were confirmed by another study utilizing a 4T1 model of breast cancer in which the KO of CXCR3 form mice implanted with tumors resulted in decreased metastases, due to an augmented immune response secondary to a loss of myeloid cell-mediated immune suppression [74]. Again, this decrease in metastasis was dependent upon the function of IFN-γ.

In a subsequent study, however, 66.1 cells stably overexpressing CXCL9 had reduced metastasis and improved survival compared to parental cells with vector control. Moreover, these changes in metastasis and survival were strongly associated with increased infiltration of T-cells and NK cells, and the changes were not observed in genetically immunocompromised mice or mice depleted of NK cells [80]. Notably, in this model, CXCL9-overexpressing tumors were significantly smaller than non-overexpressing tumors, indicating that the suppression of metastasis may be related to the suppression of the primary tumor growth rather than directly affecting breast cancer metastasis. Because of these findings, gene therapies involving the overexpression of CXCR3A ligands have been considered as potential therapeutic avenues. Mice vaccinated with CXCL11-overexpressing 4T1 cells had reduced metastasis compared to mice that were administered vector control-transfected cells. The reduction of metastasis was linked to an augmented immune response against implanted tumors, as indicated by increased IFN-γ and TNF-α expression [89]. This immune response against tumors in vaccinated mice resulted in a significant improvement in survival. Similar findings were reported in a study where CXCL10, instead of CXCL11, was overexpressed [90, 91]. The metastatic suppression resulting from CXCR3 ligand overexpression was abrogated by depletion of CD8⁺ T-cells and partially reduced by the depletion of CD4⁺ T-cells and NK cells, suggesting the involvement of immune responses. Importantly, microvessel density was also decreased in this model, and thus, inhibition of angiogenesis may also contribute to the observed effect [91]. Nonetheless, there is substantial controversy regarding the immunological function of CXCR3A in terms of suppressing metastasis. Several studies point to a CXCR3-mediated immunosuppressive effect that, when activated, permits metastatic dissemination. However, overexpression of CXCR3 ligands suppresses both tumor growth and metastatic dissemination.

In addition to immunological functions, CXCR3 signaling in cancer cells plays a critical role in breast cancer metastasis. In this setting, the function of CXCR3 is more consistent. Treatment of breast cancer cell lines with CXCL9, 10, or 11 enhanced CXCR3 signaling, which promoted the migration of these cells in vitro [72, 74]. Moreover, inhibition of CXCR3 activity in cancer cells resulted in decreased expression of RANKL while activation of CXCR3 resulted in upregulation of cathepsin B; both of these molecules have been implicated in breast cancer metastasis [92, 93]. Additionally, and consistent with its regulation of RANKL, CXCL10 was found to be a critical mediator of osteoclastogenesis in breast cancer and melanoma bone metastasis [94]. Here the loss of CXCR3 from cancer cells, or neutralization of host-derived CXCL10, lead to fewer osteolytic lesions in cardiac injection models. Subsequently, CXCL10 signaling was shown to be critical for the recruitment of cancer cells to the bone and the resulting generation of osteoclasts [94]. Moreover, loss of CXCR3 from 4T1 cells inhibited the ability of these cells to produce lytic bone lesions.

With regard to CXCR3 splice variants, activation of CXCR3B suppressed bulk cancer cell invasion and proliferation; however, it promoted stemness properties, including mammosphere formation by breast CSCs [65]. Additional studies demonstrated increased expression of CXCR3B on breast CSCs, and that silencing of CXCR3B resulted in decreased aldehyde dehydrogenase activity and suppressed metastasis in an experimental model [95]. In human samples, a large cohort study showed that CXCR3B was associated with increased tumor grade, and expression of CXCR3B and CXCL4 were associated with poor prognosis [96]. In contrast to the metastasis promoting activities of CXCR3B, knockout of PF4 increased the formation of metastatic lung lesions [97]. This increase in metastasis was associated with increased vascular permeability and increased recruitment of myeloid suppressor cells to the pre-metastatic niche in the lungs, suggesting that the direct and indirect functions of CXCR3B in breast cancer may have conflicting outcomes. Consequently, more detailed studies with modulation of CXCR3B in multiple cell populations will be required to delineate the balance between the conflicting functions of CXCR3B in different cell compartments and phases of metastasis.

In sum, CXCR3 plays diverse roles in breast cancer metastasis. Immunologically, data from multiple groups suggest that CXCR3 may promote the suppression of anti-tumor immune responses, which permits metastatic dissemination. Curiously, CXCR3A ligands had the opposite effect. While the function of CXCR3 in the immune response is perplexing, the role of CXCR3 signaling in breast cancer cells is more apparent. In cancer cells, CXCR3A promotes the migration/invasion of breast cancer cells, and despite its opposite signaling effects, CXCR3B may also promote breast cancer metastasis through the promotion of stemness in cancer cells.

3.2.2. Ovarian Cancer

In ovarian cancer, expression of CXCR3 ligands is induced in cancer-associated fibroblasts by Lymphotoxin B/ Lymphotoxin B Receptor signaling [35]. The in vitro treatment of OVCAR3 or SKOV3 ovarian cancer cells with fluid from malignant ascites resulted in increased migration towards ascitic fluid, which was abrogated by CXCR3-neutralizing antibodies [98]. Moreover, in patient samples, high expression of CXCR3 was associated with high-grade tumor, positive lymph node status, and reduced OS and progression free survival (PFS) [35]. CXCR3 is also associated with a suppressive immunological role in ovarian cancer through the recruitment of T-regs to the primary tumor [99]. However, the immunosuppressive effect of CXCR3 in ovarian cancer may be counterbalanced by its immune-promoting effects, since high expression of CXCL9 and CXCL10 was associated with increased numbers of tumor-infiltrating T-cells and NK cells. Furthermore, these changes in immune infiltrate were associated with increased survival [100]. Importantly, these studies of the immunological role of CXCR3 in ovarian cancer did not assess the effects of CXCR3 on metastasis; such studies will be necessary for understanding the intricate balance of effects of CXCR3 on the immune response to ovarian cancer.

3.2.3. Gastric Cancer

The contribution of CXCR3 to the gastric cancer metastatic process has only recently been elucidated. Despite the limited number of studies, CXCR3’s functions in the anti-tumor immune response and direct promotion of metastasis have already been partially characterized. In immune cells, correlative studies in human gastric cancer samples demonstrated that both CXCR3A and CXCR3B were upregulated in tumor tissues compared to normal adjacent tissue [75, 76, 101]. Comparison of CXCR3 staining in tumors with infiltrating immune cells demonstrated a positive association of CXCR3 expression with the presence of dendritic cells, CD4⁺ T-cells, and CD8⁺ T-cells [75, 76]. Consistently, high CXCR3 expression was associated with decreased invasive depth, TNM stage, lymph node-negative disease, and well-differentiated histology [75, 76]. These associations translated into significantly improved overall survival for patients with high CXCR3 expression [75, 76]. Cumulatively, these findings suggest that CXCR3 may have a role in promoting metastasis-limiting immune responses. Interestingly, the outcomes of these studies contradict other data showing that CXCR3 signaling in gastric cancer cells promotes PD-L1 expression and, as a result, suppresses anti-tumor immune response [16]. Together, these findings highlight the differences in CXCR3 activity in different tissue compartments.

Initial studies of CXCR3 in gastric cancer cells demonstrated that treatment with CXCL10 increased MMP2 and MMP9 production in gastric cancer cells; knockdown of CXCR3 with siRNA attenuated this effect. Functionally, overexpression of CXCR3 resulted in increased invasion, migration, and MMP production in response to CXCL10. Overall, these effects of CXCL10 treatment were shown to be dependent upon PI3K/AKT signaling stimulated downstream of CXCR3 [28]. Interestingly, and in contrast to studies examining the association of CXCR3 expression with immune infiltrates and survival, this study found that CXCR3 staining in gastric cancer tissue was associated with poor OS and advanced stage at diagnosis [28]. This disparity is likely related to methodological differences in CXCR3 quantification and subsequent stratification of patients. While neither study reported IHC methods in sufficient detail to confirm this, presented IHC data suggests that these studies quantified CXCR3 staining in different tissue compartments.

While the above studies did not account for CXCR3 splice variants, this aspect of CXCR3’s function in gastric cancer was analyzed in a separate study. Here, the authors demonstrated that CXCR3A was upregulated in gastric cancer cell lines and tissues, whereas CXCR3B was downregulated [102]. Analysis of CXCR3 splice variant functions in a gastric cancer cell line showed that knockdown of CXCR3A inhibited CXCL10-stimulated migration and invasion, whereas knockdown of CXCR3B had little effect on the migration and invasion of gastric cancer cells. Furthermore, the loss of CXCR3A inhibited CXCL10-induced phosphorylation of ERK1/2, as well as expression of MMP13 and 16. Importantly, in a subcutaneous tumor model, mice bearing tumors derived from CXCR3A-knockdown cells had significantly fewer liver metastases compared to mice with WT tumors [102].

3.2.4. Hepatocellular Carcinoma (HCC)

As with gastric cancer, investigation of the role of CXCR3 in HCC is a recent development. The first study reporting the involvement of CXCR3 in HCC samples came from a study of tumor recurrence following liver transplantation [81]. In this study, patients who received small-for-size liver grafts had increased tumor recurrence accompanied by an increase in circulating endothelial progenitor cells and CXCL10. CXCL10- and CXCR3-KO animals that received similar liver transplants had significantly reduced recurrence. Most importantly, in an orthotopic nude animal model, administration of CXCL10 into the portal vein augmented the number of lung metastases formed. In this study, the authors claim that CXCL10 administration augmented angiogenesis in tumors, resulting in increased metastatic spread [81]. However, this contradicts the classical understanding of CXCR3’s functions, and while there were more CD34-positive cells within tumors of CXCL10-treated mice, these areas appeared to be smaller in comparison to untreated mice. Moreover, mice injected with endothelial progenitor cells had larger tumors and presence of multiple lesions in the liver but did not appear to have increased rates of metastases to the lungs. Because of these findings, the involvement of angiogenesis in metastases observed in this study is questionable and likely requires further investigation [81]. While this study was conducted in nude mice, thereby likely limiting the involvement of the immune system in mediating the effects of CXCL10, a separate study showed that CXCL10 in this same setting was associated with increased T-reg recruitment to the liver graft [82]. Similarly, the expression of CXCL10 by macrophages in HCC induced the differentiation of B-cells to IgG-secreting plasma cells, which, in turn, drove alternate activation of macrophages and suppression of the anti-tumor immune response [83]. Together, these studies compel further investigation of the role of CXCL10 in immune modulation as it relates to HCC metastasis.

Additional studies highlighted CXCR3 signaling in HCC cells and its effect on metastasis. These studies highlighted the ability of CXCR3 signaling to act through the AKT/PI3K pathways as well as ERK1/2 signaling to promote the invasive and migratory phenotype of HCC cells [103–105]. Notably, two studies found that CXCR3-mediated upregulation of MMP2 and MMP9 production, which augmented the in vitro invasive properties of cells and promoted colonization of the lung in tail-vein injection models [104, 105]. These findings parallel those observed in gastric cancer, suggesting that CXCR3-mediated regulation of MMPs may be an essential component of its function not only in metastasis but also in lymphocytes. Even though MMPs are an important part of the initial process of metastasis, additional studies are warranted to determine what fraction of CXCR3-induced invasion is mediated by MMPs. Finally, another study suggested that Gβγ-mediated activation of RAC downstream of CXCR3 increases expression of PREX2, leading to increased migratory/invasive behavior in cancer cells. This knockdown of PREX2 suppressed invasion in vitro, but it was not tested if, or to what extent, loss of PREX2 abrogates the invasion stimulated by activation of CXCR3 [103].

3.2.5. Pancreatic Cancer

Studies of the function of CXCR3 in PC are currently limited. However, from this work several important associations have emerged. In the setting of PC, it is hypothesized that CXCR3 ligands, mainly CXCL9 and CXCL10, are derived from the tumor stroma composed of activated pancreatic myofibroblast. This hypothesis is supported by observations of increased CXCR3 ligand expression, by RNA-Seq, in the stromal component of microdissected PC tumors [84]. Moreover, Lunardi and colleagues found that co-culture of PC cells with pancreatic myofibroblast resulted in expression of CXCL10 from the myofibroblast cells [85]. Further studies examining the function of a CXCR3A signaling axis in PC demonstrated clear associations between CXCR3 ligands and markers of immunosuppression and T-cell exhaustion [84–86]. Importantly, both studies found that high expression of CXCR3 A ligands were associated with worse OS in PC patients [84–86]. Consistent with these results, Gao and colleagues found that treatment of T-cells in vitro with CXCL9 reduced the activity of CD8⁺ T-cells and that this contributed to more rapid progression of PDAC in mouse models[87]. Cumulatively these results strongly suggest that activation of CXCR3 A signaling may result in suppressed immune function and worsened patient prognosis in PC. In contrast to CXCR3 Ligands, the expression of CXCR3 in PC samples had inconsistent effects across the two studies. The initial study demonstrated that CXCR3 was also associated with immunosuppression and worse overall outcomes [85]. In contrast, work from our lab in a larger number of patients demonstrated that high CXCR3 expression appeared to be associated more closely with increased markers of T-cell infiltrate with weaker associations with markers of immunosuppression suggestion the CXCR3 itself is a marker of more robust anti-tumor immune response. Accordingly, high CXCR3 expression in this study was associated with improved patient outcomes [84].

3.3. CXCR3-mediated Angiostasis as a Suppressor of Metastasis

Cancer, like any other human tissue, requires oxygen to carry out metabolic functions. The rapid growth rates and high metabolic demands of cancer cells require that they develop a vascular supply through angiogenesis [106]. In addition to hypoxia-driven VEGFA expression [107], cytokines and chemokines, including CXC chemokines, are important regulators of angiogenesis by acting as promoters or suppressors of the process [108, 109]. Mechanistically, it is crucial to understand the angiostatic effects of CXCR3 are mediated by CXCR3B [14]. Downstream of CXCR3B, the fact that forskolin augments the angiostatic effects of PF4 and CXCL10 treatment strongly suggests the involvement of Gαs [14]. Following Gαs activation, the angiostatic signaling of CXCR3B is enigmatic. Several studies suggest that PKA-induced endothelial cell apoptosis is the underlying mechanism [110, 111], while others cite CXCR3-induced P21 expression and resulting cell cycle arrest as the predominate signaling involved in the angiostatic activity of CXCR3 [112–114]. Consequently, the molecular mechanism of CXCR3-mediated angiostasis is a critical topic for future research.

Since access to lymphatic and blood vessels is an integral feature for metastatic dissemination, the ability of CXCR3 to suppress angiogenesis has important implications for metastasis and tumor growth. In lung cancer tumors derived from A549 cells, CXCL10 or CXCL4L1 treatment via intratumoral injection markedly reduced tumor angiogenesis, which was associated with decreased numbers and sizes of lung metastases, decreased numbers of cancer cells at secondary sites, and improved survival in treated mice [115–117]. While not conclusive, increased rates of cancer cell apoptosis in primary tumors, but not in cell culture or metastatic sites suggest that reduced metastasis may be a reflection of decreased access to vasculature and diminished malignant cell population at the primary site, which are expected due to intratumoral administration of CXCL10 and CXCL4L1 [117]. Similarly, Lewis lung carcinoma cells ectopically overexpressing PF4 produced fewer lung lesions and decreased lung mass following tail vein injections and a 28-day growth period; though, this study did not investigate associations with angiogenesis at the metastatic site [118]. Importantly the studies of intratumoral injection of CXCL10 and CXCL4L1 demonstrate that suppression of angiogenesis at the primary site is detrimental to the overall metastatic process. Similarly, the studies of PF4, which rely on ectopic overexpression of PF4 and tail-vein injection models of metastasis, show that angiogenesis is also critical at the end of the invasion-metastasis cascade.

Angiostasis as a mechanism of CXCR3’s metastasis suppressing activity is further supported by observations in RCC, PC, and CRC. In the former, intratumoral injection of CXCL9 induced necrosis in RCC tumors, which strongly suggests a hypoxic insult over immune-mediated cell death. In PDAC, CXCL4L1 was shown to inhibit both angiogenesis and lymphangiogenesis [36, 119]. While the role of CXCL4L1-mediated suppression of angiogenesis in the PC metastatic process was not specifically investigated, its suppression of lymphangiogenesis was shown to reduce the formation of lymph node metastasis in subcutaneous PC models [119]. Finally, treatment of CRC xenografts with DMXAA, a potent inducer of CXCL10 expression, results in vascular collapse within tumors and necrosis in 90% of the tumor area [120, 121]. Moreover, DMXAA suppressed bFGF-induced angiogenesis in Matrigel plugs in a CXCL10-dependent manner, further supporting the involvement of CXCL10-mediated angiostasis in the activity of the drug [120]. Finally, the direct administration of CXCL10 to CT26 derived colon cancer models suppressed alginate bead vascularization assays as well as CD31 expression in tumor samples [122].

3.3.1. Lung Cancer

In contrast to the previous studies which focused on the role of CXCR3 in malignant and immune cells, the studies of CXCR3 in lung cancer metastasis demonstrate yet another mechanism through which the CXCR3 signaling axis affects metastatic dissemination. Before the identification of CXCR3 as the receptor for CXCL4, 9, 10, and 11, the angiostatic functions of these chemokines were well characterized [112, 123]. In lung cancer, CXCL10 is highly expressed in squamous cell carcinomas compared to normal lung tissue and adenocarcinomas [116]. These findings are consistent with the strong association of squamous lung cancer with smoking and the profound inflammatory effect of cigarette smoke on lung tissue. In primary squamous lung cancer samples, neutralization of CXCL10 augmented angiogenic activities of cancer cells in vitro. Furthermore, tumor growth in a murine model of non-small cell lung cancer in severe combined immunodeficient showed that tumor growth was inversely correlated with plasma and tumor levels of CXCL10, although CXCL10 did not influence tumor cell proliferation in vitro. Importantly inhibition of CXCL10 augmented tumor growth and metastasis [116] and was associated with increased vascular density. Similarly, PF4-overexpressing Lewis lung carcinoma cells also demonstrated fewer lung metastasis in tail vein injection and subcutaneous models compared to vector controls, which was thought to occur through suppression of angiogenesis [118]. It should be noted that mice were sacrificed 28 days after tail vein injection of cancer cells. This time frame is consistent with detecting differences in the number of cells capable of successfully colonizing the lungs and being observed macroscopically. However, it is not sufficiently small to understand the effect of PF4 on the intravascular, extravasation, and early colonization phases. Moreover, this extended growth period allows tumors to become sufficiently large that they require angiogenesis. The result of this is that differences in the number of grossly observable metastatic lesions may be the product of decreased growth of lesions rather than a difference in the ability of cells to colonize the lungs successfully. Similar findings were demonstrated using subcutaneous injection models A549 and Lewis lung carcinoma cells that were treated with CXCL4L1. Notably, in these cases, both primary tumor growth and the number of metastatic cells in distant organs were suppressed by CXCR3 ligand treatment [115],[124]. Suppression of angiogenesis was suggested to be the mechanism by which CXCR3 inhibited metastasis by decreased intratumor microvessel density. However, rescue of angiogenesis in these tumors was not performed to confirm the anti-angiogenic activity CXCL4L1 as the underlying cause of reduced metastatic spread. Regardless, these findings are consistent with CXCR3-mediated suppression of angiogenesis as a predominant mechanism underlying CXCR3’s ability to suppress metastasis in lung cancer as well as with the current understanding of angiogenesis as a critical factor in metastatic dissemination.

While CXCR3 appears to inhibit lung cancer metastasis through signaling in endothelial cells, it is also expressed on malignant cells in ~90% of lung cancer patients. For these cells, the function of CXCR3 appears to oppose that for endothelial cells. For instance, A549 cells were shown to migrate towards CXCL10 in a CXCR3-dependent manner. Despite these in vitro findings and the expression of CXCR3 on lung cancer cells in patients, CXCR3 expression was not associated with lymph node status in these patients suggesting that of the two studied roles of CXCR3 in lung cancer the angiostatic role may predominate [125].

3.3.2. Renal Cell Carcinoma (RCC)

In RCC, CXCR3 plays roles in cancer cells, endothelial cells, and immune cells, and each of these functions may impact RCC metastasis. In a murine model of RCC, Pan et al. found that systemic administration of IL-2 induced the expression of CXCR3 and its ligands in PBMCs and serum, respectively and suppressed tumor growth over four weeks [126]. When the same study was conducted in CXCR3-null mice, there was minimal effect on tumor growth and a complete abrogation of tumor necrosis induced by IL-2 treatment [126]. Intratumoral injection of CXCL9, on the other hand, suppressed tumor growth, as a monotherapy and in combination with IL-2. Further analysis demonstrated that CXCL9 and combination therapy increased necrotic tumor area and decreased the percentage of cells positive for endothelial cell marker MECA-32. Despite these data, which convincingly implicate angiostasis as a mechanism of local tumor control in this model, CXCL9 also augmented the antitumor immune response [126]. Thus, it is unclear to what extent the effects of CXCL9 were mediated by angiostasis. The fact that CXCL9 treatment increases, and CXCR3 KO almost completely abrogates, the presence of necrosis in the tumor indicates the involvement of angiogenesis, since prominent necrosis is a salient feature of poor angiogenesis. Study of patient samples demonstrated that patients with advanced-stage disease had decreased CXCL10 expression in primary tumors compared to patients with local disease. These changes were in association with the loss of other cytokines with immunological functions, including SDF-1 (CXCL12) and IFN-γ, further suggesting an immunological mechanism [127]. However, a second study noted that metastatic RCC patients on high dose IL-2 had higher expression of CXCR3 ligands in PBMCs and, as a result, had a relative abundance of angiostatic cytokines in circulation [128]. Unfortunately, neither study examined changes in angiogenesis, or immune cell infiltrates associated with high CXCR3 ligand expression.

In contrast to the studies regarding CXCR3 in endothelial and/or immune cells, the studies of the CXCR3 axis in cancer cells delve into the cellular and molecular mechanisms that give rise to a metastatic phenotype. In RCC tissue, CXCR3 ligands were overexpressed, compared to normal tissue, and the CXCR3A: B ratio was 1.5 times higher in cancer compared to normal kidney tissues. Moreover, CXCL10 treatment of RCC cell lines induced migration and invasion in these cells, which was presumably mediated by CXCR3A [33]. These effects were later shown to occur through the activation of RhoA and downstream production of MMP9. Importantly, activation of HIF-1α through hypoxia or cobalt chloride upregulated CXCR3A expression. In contrast, treatment of the same cell lines with calcineurin inhibitors caused downregulation of CXCR3B, and this change augmented the invasive and proliferative capacity of these cells [129]. These results cumulatively suggest that CXCR3A expression and downstream signaling promote the initial stages of metastasis, while signaling mediated by CXCR3B may suppress them. Further studies confirmed the effect of CXCR3B as suppressing the malignant properties of RCC cell line by demonstrating that CXCR3B overexpression suppressed tumor cell proliferation and promoted apoptosis through downregulation of heme-oxygenase 1 [24]. Consistent with the findings that CXCR3A is the predominant form of CXCR3 expressed by RCC cells and that this variant of the protein promotes metastasis, two studies of RCC patient samples showed that CXCR3, specifically CXCR3A expression in cancer cells was significantly associated with metastatic RCC [33, 34]. Finally, it is essential to note that in the absence of animal models, and metastasis assays, it is difficult to understand how either CXCR3 splice variant or their ligands affect metastasis in general. Nevertheless, the in vitro studies and association of CXCR3 with patient clinicopathological features provide a framework for understanding that CXCR3A potentially promotes and CXCR3 B likely inhibits the metastatic spread of RCC.

4. Targeting CXCR3 in Cancer Therapy

The roles of CXCR3 in immune cell homing have made it an attractive target for therapy, especially in light of recent advances in immunotherapy. In this setting, augmentation of CXCR3 signaling through the administration of ligands or viral particles designed to induce expression of CXCR3 ligands has shown some promise. In melanoma, transfection with CXCL10 expressing plasmid or administration of mesenchymal stem cells overexpressing CXCl10 resulted in a regression of metastatic lesions in part through immune mediated and angiostatic mechanisms [58, 59]. Moreover, direct intratumoral administration of CXCL9 or 10 in combination with administration of anti-PD-1 antibody augmented immune response [130]. Similarly, expression of CXCL10 from melanoma cells and intratumoral injection of IFNγ was shown to augment the recruitment of adoptively transferred CXCR3 positive NK cells to tumors, resulting in decreased tumor size and increased survival of mice [131]. In RCC, intratumoral injection of inactivated sendai virus caused a marked increase in CXCL10 expression with a corresponding increase in NK cell infiltration leading to regression of tumor by 50 percent [132]. Furthermore, RCC tumors treated with CXCL9 showed immune and vascular mediated tumor regression. In this study the effects of CXCL9 were augmented by coadministration of IL-2, exceeding the effects of CXCL9 or IL2 alone. These findings suggest that CXCL9 acts synergistically with IL-2 [126]. Similar effects for CXCR3 ligand overexpression were observed in mesothelioma in which CXCL11-expressing vaccinia virus augmented the effects of CAR T-cell therapy in mice [133]. In contrast, the findings from CXCR3 targeted therapy in breast cancer are more conflicted. On the one hand, the overexpression of CXCL9, 10 and 11 produced markedly reduced metastasis in various breast cancer animal models [80, 89–91]. Moreover, CXCR3 signaling was required for the anti-tumor effect of a HER2/ CD3 bi-specific antibody [134]. In contrast to this, genetic ablation of CXCR3 significantly suppressed breast cancer metastasis [72, 74]. Consistently, Zhu et al. demonstrated that CXCR3 inhibition with AMG487 suppressed metastasis and augmented anti-tumor immune response [74]. This contrast in the setting of breast cancer highlights the complexity of the role of CXCR3 in cancer dissemination. Therapeutic approaches of targeting CXCR3 moving forward must account for the differential functions of CXCR3 in different tumors, as well as the significance of CXCR3 expression in different cell compartments within the same tumor. Finally, the development of splice variant-specific inhibitors and agonist of CXCR3 will be an essential component for unlocking the therapeutic potential of the CXCR3 axis.

5. Conclusion

The functions of CXCR3 in metastasis are incredibly nuanced. This complexity arises from the opposing roles of CXCR3 splice variants coupled with diverse patterns of expression in multiple different cell types (Figure 2). In experimentation, such complexity manifests itself in the form of seemingly incongruous findings from studies conducted in similar patient populations and models. However, analysis of the mechanism governing the observed phenotypes in a splice variant-, cell type-, and metastasis phase-specific manner reveals trends that appear to be consistent across cancers. From these studies, three guiding principles emerge. 1) CXCR3A signaling in cancer cells promotes the invasion and migration of these cells, thus representing a metastasis-promoting function of CXCR3, 2) CXCR3B-mediated inhibition of angiogenesis suppresses the formation of metastatic lesions, and 3) immune responses associated with CXCR3A attenuate the formation of metastasis. Thus, the role of CXCR3 in cancer metastasis is a balance of pro- and anti-metastatic activities and the extent of involvement of these activities in metastasis. This context-dependence is frequently magnified in model systems used to study metastasis. The presence of an intact immune system, use of tail-vein injection models versus spontaneous or orthotopic implantation models, and variable CXCR3 splice variant expression in cell lines exemplify features that may cause outcomes to differ based on the alterations of the balance of pro- and anti-metastatic effects of CXCR3 or host features that alter the relative importance of those effects.

Additional research on the contributions of CXCR3 to the metastatic process is still needed. For example, studies of the contribution of CXCR3 to tumor cell intravasation, survival in circulation, extravasation, and early events in colonization are underrepresented in the literature. The corollary of this is that CXCR3 may have roles in each of these aspects that represent critical aspects of its contribution to a cancer’s metastatic process. Similarly, the differential roles of CXCR3 splice variants remain poorly characterized in numerous cancers. As a result, the understanding of the roles of CXCR3 in those cancers suffers due to unexplained conflicting results.

Finally, one must ask if the CXCR3 axis is a therapeutic target. It is undeniable that the studies discussed here show some promise for targeting CXCR3 in the setting of malignancy. Despite this, CXCR3 and its ligands represent a complex biological system with pleiotropic effects, including pro- and anti-cancer activities. Moving forward, research must elucidate the specific contexts in which the pro- and anti-cancer properties of CXCR3 predominate. These studies must define not only the disease settings but also specific patient populations within a disease setting in which the CXCR3 activity acts to suppress malignant progression. Moreover, understanding the contributions of the CXCR3 splice variants is critical as targeting a specific variant will likely to increase therapeutic opportunities by promoting anti-cancer activity or suppressing pro-cancer activity while leaving the other functions of the signaling axis intact.

In conclusion, the CXCR3 signaling axis is intimately involved in several aspects of metastasis in a wide variety of cancers. Despite extensive study, there is no overarching conclusion as to whether CXCR3 promotes or inhibits metastasis. This ambiguity results from multiple factors that influence CXCR3 activity and the importance of those activities within tumors. Mechanistically, CXCR3 has several functions that appear consistently throughout the literature. These include the pro-invasive and/or migratory effects of CXCR3 that can promote malignant cell dissemination, while augmented anti-tumor immune response and suppression of angiogenesis may inhibit this same process. Moving forward, specific experimentation to further delineate the roles of CXCR3 splice variants in multiple phases of metastasis is required to further our understanding of the axis and reconcile seemingly opposing results present in the literature.