Introduction
Tumor progression is associated with intratumoral hypoxia, which leads to an increase in vascular density. The increased vascular density often exhibits an abnormal architecture and provides heterogeneous perfusion within the tumor tissue [
1]. HIF-1α is a transcription factor that permits the adaptation of tumor cells to changing environment, such as hypoxia [
2]. Many studies have shown that HIF-1α is overexpressed at very high levels in colorectal tumors, particularly in the most aggressive tumors [
3]. HIF-1α protein plays a major role in regulating the expression of many genes involved in angiogenesis and erythropoiesis, metabolic adaptation to hypoxia, epithelial-mesenchymal transition (EMT), extracellular matrix degradation and chemotaxis through CXCR4 and the CXCL12/SDF-1 axis [
4].
The expression of chemo-attractant molecules and their receptors (such as CXCL12-SDF1/CXCR4 and VEGF/VEGFRs) induces tumor cell dissemination from primitive tumor sites to metastatic niches. In many tumor models, these molecules permit tumor cell survival in the metastatic microenvironment and the recruitment of hematopoietic and endothelial progenitors for neovascularization [
5]. Interestingly, recent studies have shown that overexpressions of the chemokine receptor CXCR4 and of VEGF were predictive of early distant relapse in stages II and III colorectal cancers [
6]. CXCR4 is a highly conserved G protein coupled receptor (GPCR) that binds CXCL12. Although CXCR4 is expressed in a wide range of tissues, its expression is low or absent in normal tissues and becomes important in malignant cells of many human cancers types, including breast cancer, ovarian cancer, melanoma, prostate cancer and colorectal cancer [
7]. Its ligand, CXCL12, is constitutively and physiologically expressed in the liver, lungs, lymph nodes and bone marrow [
8,
9]. CXCR7 is another GPCR which also binds to CXCL12, but with a ten fold greater affinity compared to CXCR4 [
10]. Although the role of CXCL12/CXCR7 signaling is not yet fully described, this receptor seems to be essential for the survival and growth of tumor cells [
11‐
14].
Due to the crucial role of HIF-1α and CXCR4/CXCL12 in the metastatic process of colorectal cancer, we determined CXCR4 and CXCR7 gene expression in human colon carcinomas and their modulation by hypoxia and HIF-1α in colon cancer cell lines. We found that the CXCR4 and CXCR7 expression levels increased proportionally to the clinical stage and that hypoxia differentially regulated the receptors. Furthermore, CXCR4 remained stably expressed at the cell membrane after transient hypoxia followed by 24 or 48 hours of normoxia. We also found that inhibition of CXCR4 with siRNA or with the CXCR4/CXCL12 neutraligand chalcone 4 significantly decreased the migration of these cells in vitro, an effect that was amplified by concomitant inhibition of HIF-1α. Taken together, these results indicate the potential of targeting HIF-1α and CXCR4/CXCL12 in colorectal cancer.
Discussion
Analysis of a cohort of colon polyps and chromosome-unstable carcinomas showed that the expression of CXCR4 and CXCR7 was similar to that of the normal mucosa in the early-stage but significantly increased from early to late stage carcinomas. Using three colon cell lines, we showed that hypoxia was a strong activator of CXCR4 expression, mainly through the involvement of HIF-1α, whereas CXCR7, only expressed in SW480 cells, was not modulated by hypoxia or HIF-1α. In addition, we showed for the first time that after transient passage in hypoxia, CXCR4 remained expressed at the cell membrane when exposed to normoxia for up to 48 hours. Finally, a novel combination of an HIF-1α inhibitor (irinotecan) and a CXCL12-CXCR4 interaction inhibitor (chalcone 4) significantly impaired the in vitro cell migration process. Although the migration inhibition is only partial (40%), the fact that a higher chalcone concentration (10 μM) inhibits migration by 80%, is clearly in favor of the involvement of CXCL12 via CXCR4 in the tumor cell migration process.
Recent studies have reported on the overexpression of the chemokine receptors CXCR4 and CXCR7 by several tumor entities and have shown that CXCR4 plays a crucial role in organ-specific metastasis formation [
18]. However, the precise mechanisms of chemokine receptor-driven homing of cancer cells to specific sites of metastasis remain unclear.
Angiogenesis is critical to the growth, invasion, and metastasis of human tumors [
19,
20]. Because targeting angiogenesis has emerged as a promising strategy for the therapeutic treatment of cancer, understanding the molecular mechanism linking tumor angiogenesis to the potential of a tumor to disseminate has become very important. Dysregulation of HIF and/or cytokines, such as the CXCR4/CXCR7/CXCL12 axis, is one probable cause of increased angiogenesis via the overexpression of tumor VEGF. This has led to the development of targeted therapies such as an anti-VEGF antibody, recently approved for clinical use [
21]. However, other mechanisms are most likely responsible for tumor progression and dissemination, and the interaction between CXCL12 and its receptor CXCR4 was shown to play a major role in the settlement of colorectal tumor cells in the liver [
22]. Although CXCR4 expression is low or absent in normal tissues, CXCR4 is overexpressed in many cancer types, including melanoma, breast, ovarian, prostate and colorectal cancers [
7,
18]. In contrast, the chemokine CXCL12, expressed at the surface of normal intestinal epithelium [
23], is decreased in tumor tissues, such as colon or breast carcinomas [
24,
25]. As previously shown in glioblastoma cells, hypoxia and HIF-1α can regulate the expression of CXCR4 in colon cancer cell lines [
26]. Others have shown that hypoxia increases CXCR4 expression through HIF-1α activation and that HIF-1α enhances the expression and function of CXCR4 in normal cells monocytes, macrophages and endothelial cells [
27] and in tumor cells [
28]. In our hands, siRNAs targeting HIF-1α prevented both HIF-1α and CXCR4 upregulation under hypoxic conditions.
In human colon carcinomas, we observed that CXCR4 expression significantly increased during tumor progression as it increased from stages 0-II to III-IV, whereas for CXCR7, a significant increase was observed between early stages and liver metastases. Knowing that metastases develop from circulating tumor cells escaping the primary site of cancer during their passage in the blood stream [
29], these cells switch from a hypoxic to a normoxic environment and escape regulation by HIF-1α. Fitting with this hypothesis, we demonstrated for the first time that after a transient passage through hypoxia, which leads to the upregulation of CXCR4 expression, the receptor protein level remains high at the cell membrane even when the cells returned back to normoxia. The maintenance of high CXCR4 level could help circulating cells to home in organs expressing high levels of the CXCL12 ligand, and with the resident CXCR7 may aid endothelial extravasation favoring metastasis development [
30,
31]. During embryogenesis, it has been shown that CXCR7 is only expressed in the trailing cells of the primordium and is required to provide migration directionality [
32].
The CXCR4/CXCL12 interaction provokes calcium mobilization and activation of multiple signaling pathways, including PI3K/Akt, PLC-γ/Protein kinase C and Erk/Ras [
33,
34]. We show that hypoxia alone rapidly activated the PI3K/Akt and Erk/Ras pathways and that this effect was amplified under short-term CXCL12 stimulation. Interestingly, part of the PI3K/Akt activation was induced by the interaction of CXCL12 with CXCR4, as it was blocked by siRNA targeting CXCR4, but not by its interaction with CXCR7. As PI3K/Akt activation could not be totally abolished with siRNA targeting CXCR4, other receptors may participate in the activation of this oncogenic pathway, although at present no other receptors have been shown to interact with CXCL12. Nevertheless, the short-term activation of the oncogenic pathway may be sufficient to initiate the migration process observed when cells are switched to hypoxia, and this activation could be blocked with a siRNA against CXCR4.
Although CXCR4 inhibition with siRNA or AMD3100 affected the PI3K/Akt pathway, no change in activity was observed for the Erk/Ras pathway. A number of the components of this PI3K/Akt pathway are mutated or deregulated in a wide variety of human tumors, highlighting the key role of this pathway in cellular transformation [
35]. Following Akt phosphorylation, the subsequent phosphorylation of its targets regulates a variety of critical cell functions, including glucose metabolism, cell proliferation and survival. PI3K also is likely implicated in the metastatic phenotype. Indeed, several molecules involved in cell migration and cell adhesion can regulate -or be regulated by- PI3K. Indeed, PI3K/Akt was shown to be essential for Matrix Metalloproteinase (MMP) production in several cell lines [
36] and clinical and animal studies revealed that PI3K/Akt activates MMP-2, MMP-9, and Urokinase-type plasminogen activator (uPA), leading to destruction of the extracellular matrix [
37]. Other data might explain our observation of inhibition of cell migration with CXCR4 inhibitors. Gassmann and colleagues for instance, demonstrated that colon cell line extravasation into the liver parenchyma is regulated
in vivo by CXCL12-activated CXCR4 [
30]. In contrast, we found that CXCR7 silencing did not modify the migration process or the activation of the PI3K/Akt or Erk/Ras pathway. This is consistent with recent studies providing alternative mechanisms through which CXCR7 can regulate CXCL12-directed cell movement. CXCR7 does not appear to induce cell migration directly but may enhance cell adhesion [
11], and the involvement of CXCR7 in CXCL12-mediated transendothelial migration of human renal multipotent progenitor cells has been demonstrated [
12]. Additionally, Zabel et al showed that CXCR7, through association with β-arrestin but without Ca
2+ mobilization, regulates the ability of human CXCR7
+/CXCR4
+ lymphoblastoid cells to migrate across an endothelial cell monolayer [
38].
HIF-1α is frequently upregulated at protein level in response to the hypoxic tumor environment and this overexpression has been associated with an aggressive phenotype, namely resistance to chemotherapy and poor outcome in a wide range of tumors [
3]. One hypothesis concerning the metastatic process is based on an increasing CXCL12 gradient from the primary tumor to secondary niches at metastatic sites. Immunohistochemical analyses have shown that CXCL12 is highly expressed in hepatic sinusoids including endothelial and Kupffer cells [
38] and that disseminating tumor cells express CXCR4 [
7]. Thus, the CXCL12/CXCR4 interaction permits extravasation of colon tumor cells in the liver parenchyma [
36]. Moreover, CXCR4/CXCL12 interaction increases the expression of proteins important for cell migration, motility and invasion, such as Rho and Rac [
39‐
41].
Altogether, our results demonstrate the potential value of inhibiting HIF-1α and CXCR4/CXCL12 to counteract the migration process. We have used an innovative approach to impair tumor cell migration by combining irinotecan and chalcone 4 that could be of therapeutic interest. We have previously shown that irinotecan inhibited HIF-1α protein accumulation in
in vitro[
42] and
in vivo models of colon cancer [
16,
42]. We hypothesized that irinotecan would inhibit CXCR4 expression by inhibiting HIF-1α. Chalcone 4 is a neutraligand of CXCL12 and impairs CXCR4/CXCR7/CXCL12 interaction. In addition, other studies have already shown that inhibition of CXCR4
in vivo inhibits the metastatic process and the migration of breast cancer cells. We have shown that the combination of the two drugs is more effective than each drug separately as migration was decreased by more than 40%.
Materials and methods
Tumor specimens
Human tumor specimens were obtained at the Gastrointestinal Surgical Department of the University Hospital Hautepierre (Strasbourg-France) according to the French Ethical Committee recommendations and the ethical standards of the 1964 Declaration of Helsinki. All patients provided written informed consent.
Cell culture and treatments
Human colon carcinoma HCT-116, HT-29 and SW480 cells were maintained at 37°C under normoxic (20% O2) and hypoxic conditions (94% N2, 5% CO2, 3 or 1% O2, Sanyo) in DMEM (1 g/L glucose) supplemented with 10% fetal bovine serum. The cells were treated during exponential growth conditions (30% confluence).
Irinotecan (Campto
®, irinotecan chlorydrate, Pfizer) was used at a concentration of 1 μM. AMD3100 (Sigma France), an antagonist of CXCR4, was used at 10 μM. Chalcone 4 was provided by JL Galzi (ESBS, Strasbourg, France); [
17].
SiRNA transfections
The cells were transfected in 6-well plates with siRNA anti-HIF-1α (20 nM; siRNA 1: Hs_HIF-1α_5: SI02664053, siRNA 2: Hs_ HIF-1α_6: SI02664431, Qiagen®) and anti-CXCR4 (20 nM; siRNA 1: Hs_CXCR4 7: SI02664235 and siRNA 2: Hs_CXCR4 8: SI02664242, Qiagen®) with the Lipofectamine® RNAiMAX, (Invitrogen®, Life Technologies) according to the manufacturer’s instructions. A non-specific, non-targeting siRNA was used as the control treatment (Eurogentec®). The cells were incubated for 48 h at 37°C in hypoxia (94% N2, 5% CO2, 3 or 1% O2, Sanyo®).
Migration tests
Boyden chambers (BD Biosciences) were used for the in vitro migration assay. The upper and lower compartments were filled with 1% FCS and 10% FCS, respectively. Cells (5 × 105) were added in the upper compartment. After 24 h, the cells were fixed with 4% paraformaldehyde for 15 min and stained with DAPI (1/30000; 4', 6'-diamidino-2-phenyl indole, Sigma®). Migrating cells were counted using an epifluorescence microscope.
Relative quantitative PCR
The mRNA expression of CXCR4, CXCR7, HIF-1α and control PDGF genes was evaluated by relative quantitative real-time PCR (RT-qPCR) analysis using the LightCycler system (Roche Molecular Biochemicals®) and FastStart DNA Master Mix SYBR Green I (Roche Diagnostics®). RNA was extracted with Trizol reagent (Invitrogen®) according to the manufacturer's protocol. Reverse transcription of 2 μg RNA was performed using reverse transcriptase and oligo(dT) primers (FinnZyme®). PCR was performed as follows: denaturation at 95°C for 10 min, followed by 40 cycles at 95°C for 20 s and 62°C for 20 s and elongation at 72°C for 20 s using the maximum temperature transition rate of 20°C/s. Fluorescence measurements were taken at the end of the elongation phase. The specificity of the PCR products was assessed by generating a melting curve. All quantifications were performed in duplicate for three independent experiments and normalized with respect to the endogenous PDGF mRNA levels. Target cDNA expression was quantified using the ΔΔCt method. Validated primers were obtained from Qiagen®.
Western Blot
Western Blots were performed with the following antibodies: anti-Akt total (1/1000), anti-Erk total (1/2000), anti-phospho Akt (1/1000), anti-phospho-Erk (1/1000) (from Cell Signaling Technology ®), anti-HIF-1alpha (1/1500, BD Biosciences®), anti-actin (1/15000, Millipore ®).
Flow cytometry
For each condition, 106 cells were washed in PBS 1X at 4°C. Fluorescence was analyzed on a FACScan cytometer (BD Biosciences®), and the data were analyzed with CellQuest (BD Biosciences®). The measurements were performed twice in two independent experiments.
Immunohistochemistry
Immunohistochemistry used standard procedures. Briefly, tumors were fixed in 4% paraformaldehyde and embedded in paraffin. Sections (4 μm) were deparaffinized and heated for 10 minutes in 10 mmol/L citrate buffer, pH 6.2, for antigen retrieval. They were stained with Harris solution and Eosin for histological examination and immunostained using the primary antibodies raised against CXCR4 (1:200; eBiosciences) and CXCR7 (1:75; ThermoScientific). Slides were then incubated for 30 min with secondary biotinylated anti-mouse antibody (dilution 1:200; Vector Laboratories Inc., Burlingame, CA). Immunostaining was developed with a liquid DAB substrate kit (Roche Diagnostics) according to the manufacturer’s instructions.
Statistics
The data were analyzed with the Mann–Whitney parametric test, and the significance level was set at 5%.
Competing interests
The authors declare they have no competing interests or other interests that might be perceived to influence the results and discussion reported in this paper.
Authors’ contributions
BR and MHH performed all experiments. BR, SR and CB contributed to the surgical specimen, patient consent and clinical data collection. JLG and MPG revised the paper critically for important intellectual content. EP and DG performed the conception and design of the study as well as the final approval of the version to be published. All authors have read and approved the final manuscript.