Background
CXCR4 is a well-known G-protein coupled receptor (GPCR) for the small chemokine stromal-derived factor (SDF)-1α, which is also known as CXCL12. Another GPCR, CXCR7, has been identified as a second receptor for SDF-1α. This receptor was originally cloned based on its homology with conserved domains of GPCRs and named as “RDC1” [
1]. At the beginning, it was believed to be a receptor for vasointestinal peptide, but later reports dismissed this possibility [
2]. Combined phylogenetic and chromosomal location studies revealed the structural resemblance of the orphan receptor RDC1 to CXC chemokine receptors and implicated CXC chemokines as potential ligands [
1]. It was shown that RDC1 could serve as a co-receptor for human immunodeficiency virus and simian immunodeficiency virus strains, just like CXCR4 [
3]. Soon afterwards, SDF-1α was shown to bind with high affinity to and signal through the orphan receptor RDC1 [
2], leading to the designation of the receptor as “CXCR7”.
CXCR7 is expressed on vascular endothelial cells, T cells, dendritic cells, B cells, brain-derived cells and tumor cells, including human glioma cells [
2‐
4]. Its expression is upregulated by hypoxia in human microvascular endothelial cells [
5]. CXCR7 plays an important role in several carcinomas, including breast cancer, lung cancer, and prostate cancer [
6,
7]. Immunohistochemical staining of metastatic melanoma sections demonstrated CXCR7 staining on tumor cells [
5]. This receptor is believed to play a pivotal role in growth, adhesion, survival, angiogenesis, and invasion of tumor cells [
2,
6,
7]. Administration of a small molecule antagonist of CXCR7 correlated with reduced tumor size in both xenograft and syngeneic
in vivo tumor growth studies [
6]. Ectopic expression of the receptor has been shown to enhance tumor formation in nude mice
in vivo[
8]. A recent study demonstrated that in prostate cancer, CXCR7 potentially promotes invasion through its downstream targets of CD44 and cadherin-11 [
7]. Balabanian and colleagues showed that SDF-1α-induced T cell migration was dependent on both CXCR4 and CXCR7, and combined inhibition of these two receptors resulted in additive inhibitory effects on the migration of T cells [
2].
Hypoxia is a major player in the microenvironment of gliomas that orchestrates adaptive responses by stimulating the expression of several genes involved in tumorigenesis. However, despite accumulating data, the regulation of CXCR7 by hypoxia and its contribution to glioma migration have not been fully elucidated yet. Here, we show that U87MG, LN229 and LN308 glioma cells express CXCR7 and exposure to hypoxia upregulates CXCR7 protein expression in these cell lines. CXCR7-expressing U87MG, LN229 and LN308 glioma cells migrated towards SDF-1α in hypoxic conditions in the Boyden chamber assays. While shRNA-mediated knockdown of CXCR7 expression did not affect the migration of any of the three cell lines in normoxic conditions, we observed a reduction in the migration of LN229 and LN308, but not U87MG, glioma cells towards SDF-1α in hypoxic conditions. In addition, knockdown of CXCR7 expression in LN229 and LN308 glioma cells decreased levels of SDF-1α-induced phosphorylation of ERK1/2 and Akt. Inhibiting CXCR4 in LN229 and LN308 glioma cells that were knocked down for CXCR7 did not further reduce migration towards SDF-1α in hypoxic conditions and did not affect the levels of phosphorylated ERK1/2 and Akt. Analysis of immunoprecipitated CXCR4 from LN229 and LN308 glioma cells revealed co-precipitated CXCR7. Taken together, our findings indicate that both CXCR4 and CXCR7 mediate glioma cell migration towards SDF-1α in hypoxic conditions.
Discussion
Our findings demonstrate that (1) hypoxia upregulates CXCR7 protein expression in glioma cells, (2) CXCR7 mediates the migration of LN229 and LN308 glioma cells towards SDF-1α in hypoxic conditions, (3) SDF-1α induces CXCR7-mediated phosphorylation of ERK1/2 and Akt in LN229 and LN308 glioma cells, (4) inhibiting CXCR4 in glioma cells that are knocked down for CXCR7 does not further reduce either the migration towards SDF-1α or the levels of SDF-1α-induced phosphorylation of ERK1/2 and Akt, and (5) CXCR4 and CXCR7 bind in glioma cells. Collectively, our findings indicate that both CXCR4 and CXCR7 mediate glioma cell migration towards SDF-1α in hypoxic conditions.
The presence of HIF-1α binding sites beginning at −155, -1012 and −1350 base pairs upstream of the transcription initiation site of CXCR7 suggests that its expression could be regulated by hypoxia. Indeed, hypoxia-induced upregulation of CXCR7 has been reported previously in microvascular endothelial cells [
5]. Our data show that the expression of CXCR7 is upregulated under hypoxic conditions in glioma cell lines. While the upregulation is evident at earlier time points of exposure to hypoxia in LN229 and LN308 glioma cells, it is not noticeable until 18 h in U87MG glioma cells. Hypoxia-mediated upregulation of CXCR7 is significant, because hypoxia is a common pathological feature of gliomas that controls the expression of many genes essential for acquisition of invasive phenotype. The invasive nature of gliomas hinders effective therapy and thus molecular mechanisms governing invasion represent attractive therapeutic targets [
9]. Although many hypoxia-induced molecules that are involved in glioma biology have been elucidated, more effective design of treatment strategies warrants further identification of novel hypoxia-responsive genes that drive invasion.
Although the key role of CXCR4 in mediating SDF-1α-induced migration of glioma cells is well established [
9‐
12], that of CXCR7, to our knowledge, has still not been confirmed. However, the discovery of CXCR7 as a second SDF-1α receptor brings to mind the possibility that CXCR7 might contribute to SDF-1α-induced migration. In a report by Balabanian et al., CXCR7 was described as a receptor that enhanced SDF-1α-dependent chemotaxis of T lymphocytes together with CXCR4 [
2]. Our data support a role for CXCR7 in mediating SDF-1α-induced glioma cell migration in hypoxic conditions. Knockdown of CXCR7 expression by two independent shRNA sequences resulted in a consistent reduction in the number of LN229 and LN308, but not U87MG, glioma cells that migrated towards SDF-1α. The discrepancy observed for the U87MG cell line is attributable to the lack of hypoxia-mediated CXCR7 upregulation at 8 h of exposure to hypoxia (which is also the timeframe for the migration assays). It should also be noted that LN229 glioma cells migrated towards SDF-1α only in hypoxic conditions, where levels of CXCR4 and CXCR7 were higher.
CXCR4 activation has been linked to ERK1/2, Akt, and FAK phosphorylation [
9], which are important pathways regulating the survival, proliferation and invasion of tumor cells. Our data demonstrate that SDF-1α induced the phosphorylation of ERK1/2 and Akt in LN229 and LN308 glioma cells that displayed CXCR7-mediated migration towards SDF-1α. This was mediated by CXCR7, as knockdown of CXCR7 expression decreased the levels of SDF-1α-induced phosphorylation of ERK1/2 and Akt. These data have important implications, because ERK1/2 and Akt pathways are frequently upregulated in several cancers and there are ongoing efforts exploring both pathways as potential therapeutic targets. For instance, positive staining for phosphorylated ERK1/2 is observed in a large percentage of gliomas, but not in normal brain. Indeed, inhibition of MAPK signaling by the inhibitor sorafenib suppressed development of malignant glioma in an orthotopic mouse model [
13].
Functionality of CXCR7 has long been the source of controversy. To date, several studies have yielded puzzling results. While some reports suggest a decoy activity, others indicate a signaling activity for CXCR7. Burns and colleagues showed that ligand activation of CXCR7 failed to induce typical chemokine responses, such as cell migration and calcium mobilization [
8]. This was supported by studies in zebrafish that showed CXCR7 functions primarily by sequestering SDF-1α to shape the extracellular chemokine gradient and provide directional migration [
14]. By contrast, Wang and coworkers provided evidence that CXCR7 induces invasiveness of prostate cancer cells and activates Akt [
7]. Invasiveness of hepatocellular carcinoma cells is also mediated by CXCR7 [
15]. There is evidence that ligand binding to CXCR7 activates MAPK through β-arrestin and thus the receptor is functional [
16]. CXCR7 is implicated in survival and proliferation of breast and lung cancer cells [
6]. Moreover, studies have unraveled that CXCR7 regulates interneuron migration [
17], and is involved in transendothelial migration [
18]. A recent study reported that CXCR7 modulates chemokine responsiveness in migrating neurons by regulating CXCR4 protein levels [
19]. CXCR7 is also a functional receptor in primary rodent astrocytes and controls proliferation and migration towards SDF-1α through G
i/o proteins [
20]. CXCR7 is involved in mediating anti-apoptotic events in glioma cells as well [
21,
22]. A functional interaction is evident between CXCR4 and CXCR7. In GBM cell lines, CXCR7 controls proliferation through a functional cross-talk with CXCR4 [
23], and in the developing rat brain, a cross-talk between CXCR4 and CXCR7 might account for the regulation of SDF-1α-dependent neuronal development [
24]. In breast cancer cells, inhibition of CXCR7 was shown to reduce the growth and metastasis of CXCR4-positive cells [
25]. Targeting of CXCR7 also inhibits SDF-1α/CXCR4-mediated transendothelial migration of human tumor cells [
26].
We now provide evidence that CXCR7 is induced by hypoxia, and mediates the migration of glioma cells towards SDF-1α in hypoxic conditions. Our data reveal that both CXCR4 and CXCR7 are required for migration towards SDF-1α and SDF-1α-induced phosphorylation of ERK1/2 and Akt. In LN229 and LN308 glioma cells, both inhibition of CXCR4 by AMD3100 and shRNA-mediated knockdown of CXCR7 expression diminished migration towards SDF-1α and reduced levels of SDF-1α-induced phosphorylation of ERK1/2 and Akt.
It is interesting that while both CXCR4 and CXCR7 are required for SDF-1α-induced migration of hypoxic glioma cells, blocking both CXCR4 and CXCR7 does not provide an additive effect, either with regards to migration assays or phosphorylation of ERK1/2 and Akt. Furthermore, CXCR7 can be co-immunoprecipitated with CXCR4-HA. It is probable that CXCR7 is part of a functional heterodimer, together with CXCR4, which mediates the migration of glioma cells towards SDF-1α under hypoxic conditions. Functional CXCR4/CXCR7 heterodimerization has previously been reported in HEK293T cells and glial cells [
27‐
29].
GPCRs can exist as monomers, homodimers or heterodimers and these conformations might have important implications in downstream signaling and the design of pharmacological inhibitors. It has been demonstrated that heterodimers can activate signaling pathways that differ from those activated by homodimers [
30]. Our previous data showed that CXCR4 inhibition by AMD3100 decreased the levels of SDF-1α-induced phosphorylation of FAK in LN308 glioma cells [
9]. Conversely, the data that we present here show that knockdown of CXCR7 expression in LN308 glioma cells did not affect the levels of SDF-1α-induced phosphorylation of FAK. Activation of FAK following exposure to SDF-1α might therefore depend on CXCR4 alone. This scenario has obvious implications for drug discovery. Heterodimers may be considered as distinct structural and functional entities, which might influence drug affinity and efficacy. A better understanding of how heterodimers are regulated, their function, and pathophysiological significance may help us exploit them as novel drug targets for improved therapeutics.
It is of note that, as mentioned above, CXCR4 and CXCR7 are present on both tumor cells and vascular cells. This suggests that paracrine signaling mechanisms between these two cell types might be in effect. Such mechanisms could affect several aspects of tumor biology, including angiogenesis, migration, survival and proliferation.
Competing interests
The authors declared that they have no competing interest.
Authors’ contributions
ME designed and did the experiments and drafted the manuscript. DZ conceived the study and critically revised the manuscript. YS assisted ME and DZ with the response letter. All authors read and approved the final version of the manuscript.