This article is mentioned in:

Introduction

Colorectal cancer remains one of the most prevalent malignancies. While primary tumors are often resectable leading to a better prognosis, the presence of metastatic lesions represent the worst scenario medicine has to cope with. Colorectal cancer metastasizes mainly to the liver and almost 70% of patients with this cancer present distant invasion at the time of diagnosis. The mechanisms leading to metastatic formation and growth remain poorly characterized, especially those linked with stromal cells, known to facilitate tumor cell escape from the immune system and support tumor growth (1,2). Therefore, a better understanding of the tumor microenvironment is required for a wider comprehension of metastatic disease. One of the receptors expressed in stromal cells, along with tumor cells, is CXCR4, one of the two known receptors for CXCL12, also known as stromal cell-derived factor 1 (SDF-1) (3).

CXCR4 has been described to be upregulated in human melanoma metastasis or colorectal cancer lymph node and liver metastases and has been linked with poor survival (4–6). This receptor has been shown to mediate different pro-metastatic events in tumor cells in vitro, such as invasion and migration and cancer cell extravasation in vivo (7–9). Interestingly, this receptor is not only expressed by tumor cells but also has been found in hepatic stellate cells (HSCs), liver resident cells. Under inflammatory conditions, these cells become activated and trans differentiate into myofibroblast-like cells, supporting fibrosis, liver cancer and liver metastasis (10–12). In the liver, HSCs are the main population of cancer-associated fibroblasts (CAFs), which are the main component of the tumor microenvironment (13) and arise as a modulator of the tumor response. Intriguingly, tumor/fibroblast interaction has been postulated as a relevant event during the progression of cancer with the CXCR4/CXCL12 chemokine axis one of the main catalysts of malignancy (14). Furthermore, HSCs, along with liver sinusoidal endothelial cells (LSECs), are one of the main sources of CXCL12 in the liver where they mediate not only the recruitment of CXCR4-expressing tumor cells, but also CXCR4-expressing immune cells.

During cancer development and metastasis formation, the cytotoxicity against cancer cells is compromised due to immune suppression. This pathological process is in part orchestrated by myeloid-derived suppressor cells (MDSCs), CXCR4-expressing immature myeloid cells able to suppress the cytotoxic capacity of T lymphocytes (15). During liver metastasis, these cells are recruited into the organ (16) from the circulation leading to uncontrolled tumor growth and disease progression. In fact, this chemokine and its CXCR4 receptor seem to be relevant for tumor progression, as reported through the reduction of CXCL12 availability and CXCR4 blockage within the tumor microenvironment, reducing liver metastasis in nude mice (17). Moreover, CXCR4 blockage has been confirmed to be effective for reducing in vivo lung cancer metastasis (18), pointing out the interaction of CXCR4 with its lig and CXCL12 as a mediator in this process. However, the cellular and molecular mechanisms by which CXCR4 interaction with CXCL12 drives disease progression remain poorly understood, especially when it comes to stromal cells.

Therefore, the present study aimed to investigate the efficacy of treatment with the CXCR4 antagonist AMD3100 in an orthotopic model of colorectal cancer liver metastasis and the cellular mechanisms involved in the processes disrupted by this antagonist.

Materials and methods

Animals

Six-week-old male Balb/c mice were obtained from Janvier Labs (France). Housing, care, and experimental conditions were carried out in conformity with institutional guidelines and national and international laws for experimental animal care. The animals were fed a standard chow and had access to water ad libitum. All the proceedings were approved by the Basque Country University Ethics Committee (CEID) in accordance with institutional, national and international guidelines regarding the protection and care of animals used for scientific purposes.

Cell lines

All in vitro and in vivo experiments were conducted using the murine C26 colon adenocarcinoma (C26) cell line (also known as MCA-26, CT-26) syngeneic with Balb/c mice and purchased from ATCC (LGC Standards S.L.U. Barcelona, Spain). Cells were cultured in RPMI-1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS), penicillin (10,000 U/ml), streptomycin (10,000 µg/ml) and amphotericin B (25 µg/ml) all purchased from Thermo Fisher Scientific (Waltham, MA, USA).

Primary murine hepatic stellate cell isolation and culture

HSCs were isolated as previously described with some modifications (19,20). Briefly, the liver was perfused with collagenase from Clostridium histolyticum (Sigma-Aldrich, St. Louis, MO, USA) and the resulting cell suspension was subjected to isopycnic centrifugation through a Percoll gradient (GE Healthcare, Chicago, IL, USA). The fraction enriched in HSCs was cultured in culture medium without serum supplemented with antibiotics and antimycotics incubated at 37°C in 5% CO2 overnight before the experimental procedures.

Preparation of conditioned culture medium

Tumor cells were seeded at a concentration of 5×104 cells/cm2 and cultured overnight in RPMI-1640 supplemented with 10% FBS and antibiotics. Then, the culture medium was substituted with fresh medium supplemented with 1% FCS. After 24 h, the cell culture supernatant was collected and used as C26-conditioned medium (C26-cm) for further studies. HSC conditioned medium was also prepared. To do so, HSCs were cultured at a concentration of 7,5×104 cells/cm2 as previously described (10). Then, culture medium was replaced with fresh medium supplemented with 1% FBS and HSCs were cultured for an additional 24 h. The supernatant was collected and used as HSC-conditioned medium for further studies. In some cases, HSCs were exposed to C26-cm diluted 2-fold with fresh medium for 24 h. Then, the supernatant was collected and used as tumor-activated HSC-cm (taHSC-cm).

Western blot analyses

CXCR4 expression was analyzed by western blotting in C26 cells and HSCs under several conditions. To do so, the tumor cells were either untreated or treated with HSC-cm or taHSC-cm, and HSCs were either untreated or treated with C26-cm. After 48 h of treatment, the cultured medium was removed and cell lysates were obtained. To obtain cell lysates for protein extraction, the cells were washed twice with cold 1X phosphate-buffered saline (PBS) before adding 100 µl/106 cells of 1X Laemmli sample buffer (Bio-Rad, USA) with 1% β-mercaptoethanol and 1 M NaCl. Sample electrophoresis was performed using 10% SDS-PAGE gels and run for 1.5 hat 120 mV on ice. Proteins were transferred onto a nitrocellulose membrane in wet conditions at 380 mA for 3.5 hon ice. Then, the membranes were incubated with rabbit anti-mouse CXCR4 (1:1,000; cat. no. PA3-305; Thermo Fisher Scientific) overnight at 4°C in 5% milk containing TBS-T (2% 1 M Tris-HCL pH, 7.5 10% 5 M NaCl, 0.1% Tween-20), followed by a 1-h incubation with specific goat anti-rabbit biotinylated secondary antibodies (1:2,000; cat. no. 65-6140). Streptavidine-HRP conjugated (1:500) was used for detection. Antibodies and detection systems were purchased from Thermo Fisher Scientific. Finally, the bands were visualized using Luminata™ Crescendo Western HRP Substrate (Millipore, Billerica, MA, USA). SnapGene software (Syngene, Frederick, MD, USA) was used to measure the protein expression (Each experiment was performed three times).

Flow cytometric analyses

Flow cytometric analyses were carried out for the quantification of CXCR4 expression in C26 cells and quantification of MDSCs in the circulation of tumor-bearing mice. To do so, C26 cells and PMNCs (for collection of PMNCs see later on) were fixed in 70% ethanol for 10 min before the unspecific background was blocked by a 30-min incubation in PBS containing 5% FBS. The cells were incubated for 2 h with either rat anti-mouse CXCR4 monoclonal antibody (1:500; cat. no. MAB21651; R&D Systems Inc., Minneapolis, MN, USA), FITC-conjugated rat anti-mouse CD11b (1:500; cat. no. ab24874; Abcam, Cambridge, UK) and rat anti-mouse-Ly6G primary antibody (1:500; cat. no. NBP1-28168; Novus Biologicals, Littleton, CO, USA). Then, cells were incubated with Alexa-488 (excitation 488, emission 525) and Alexa-594 (excitation 594, emission 617) conjugated-secondary antibody for CXCR4 and MDSC detection, respectively. Immunolabelled cells were analyzed by flow cytometry (Gallios; Beckman Coulter, Brea, CA, USA). Flow cytometric analyses were performed three times for tumor cells and a pool of 5 mice were used for MDSC quantification.

Cell viability assay

C26 viability was analyzed using PrestoBlue™ Cell Viability reagent (Life Technologies Inc.). C26 tumor cells were seeded at a concentration of 5×103 cells/cm2 and let expand for 3 h. Afterwards, the tumor cells were incubated with increasing concentrations (0.1, 1 and 10 nM) of CXCL12 (OriGene, Rockville, MD, USA) with or without 10 µg/ml AMD3100 (from 2 h prior to the addition of CXCL12), for 48 h (AMD3100 was previously diluted in PBS for stock solution). Then, cell viability was quantified by adding PrestoBlue™ reagent for 120 min. The reaction was measured with the Ascent Multiskan (Labsystems). Cell viability was represented as percentage of the increase respect to cell viability at time 0 (Cell viability assay was carried out three times).

Gelatin zymography

MMP-2 and MMP-9 activity was measured by gelatin zymography in C26 cells or HSC supernatants. The cells were treated with either 1 nM CXCL12, or AMD3100 (10 µg/ml from 2 h prior to the addition of CXCL12) or the combination of both for 24 h. Cell culture supernatants were collected and run on 1% gelatin containing electrophoresis gel. After 2 washes in 2.5% TritonX-100, the gels were incubated overnight in a developing buffer containing divalent cations for the proper activity of MMPs. After wards, the gel was stained with Coomassie Blue to visualize degradation bands. Images were captured through Quantity One program (Bio-Rad Laboratories, Hercules, CA, USA) (MMP-2 and MMP-9 secretion was measured three times).

Migration assay

For the quantification of the migratory potential of C26 cells and HSCs, modified Boyden chambers were utilized. Briefly, cells were seeded onto type I collagen-coated 8-µm diameter pore membrane inserts (Greiner Bio-One GmbH, Frickenhausen, Germany). Cells were incubated in RPMI-1640 supplemented with 1% FBS and antibiotics and allowed to adhere and expand for 3 h before the addition of AMD3100 (10 µg/ml) in both upper and lower chambers. After incubation of the cells with AMD3100 for 2 h, the medium was replaced and the cells were treated with either AMD3100 (10 µg/ml), 1 nM CXCL12 or a combination of both for 18 h supplemented with 1% FBS. Then, the non-migrated cells were gently removed from the upper side of the insert and the remaining cells on the external side of the insert were subjected to 70% ethanol fixation and DAPI mounting medium for visualization of the nuclei. The number of migrated tumor cells and HSCs was quantified in 10 fields at ×10 magnification under Axioscope fluorescence microscope (Carl Zeiss, Oberkochen, Germany). The migratory potential is represented as the percentage of migrated treated cells in respect to the untreated cells (migration studies were performed three times).

Experimental development of liver metastasis

C26 tumor cells were in trasplenically injected at a concentration of 2×105 cells in 100 µl PBS. Briefly, a thin cut was made in the left flank of each anesthetized mouse and the spleen was exposed. Then, the tumor cell suspension of the grafted C26 colorectal cancer cells was slowly injected in the spleen distal pole. Afterwards, the spleen was relocated and the wound was closed. Mice were treated daily with 5 mg/kg AMD3100 or vehicle solution starting 24 h after tumor cell injection. All animals were sacrificed 14 days after tumor cell injection and livers were frozen and kept at −80°C for immunehistochemical analyses or paraffin embedded for hematoxylin and eosin staining and quantification of liver tissue area occupied by the tumor. In vivo metastasis assay was performed in duplicate with 5 animals in each group/experiment (n=10 each group).

Immunohistochemical analysis

Frozen liver sections (10-µm thick) were analyzed for the quantification of different immune cell populations. Liver sections were stained with specific primary antibodies: Ratanti-mouse CD11b as neutrophil marker (Abcam), ratanti-mouseLy6G as granulocytic marker (Novus Biologicals) and Cy3-conjugated mouse anti-human αSMA (1:500; cat. no. C6198; Sigma-Aldrich) as HSC activation marker. After blocking and incubation with primary antibodies, surface molecules were detected by the use of secondary antibodies conjugated either with Alexa-488 or Alexa-594. For αSMA staining, samples were incubated with rabbit anti-mouse IgG F (ab') (cat. no. 31192; Thermo Fisher Scientific) diluted in PSB for the blockage of endogenous mouse IgGs for 1 h prior to the primary antibody incubation (0.15 mg/ml). The number of CD11b+Ly6G+ cells along with in tratumoral αSMA expression was analyzed in the metastatic livers and quantified by image analyses with ImageJ software (NIH, Bethesda, MD, USA).

Isolation of blood PMNCs

For isolation of PMNCs, mice were anesthetized prior to their sacrifice and the abdominal cavity was opened exposing the cava vein. Sterile PBS-EDTA was inoculated into the cava vein right before extraction of blood using the same syringe. Erythrocytes were discarded using Lympholyte M (Cederlane, Canada) density gradient and PMNCs and lymphocyte fraction collected for flow cytometry analyses as described above. Results represent the pool of 5 mice/group.

Statistical analyses

Data are expressed as mean ± standard deviation (SD) of three independent experiments. Statistical analysis was performed using SPSS version 13.0 (SPSS, Inc. Chicago, IL, USA). Individual comparisons were performed using two-tailed, unpaired Student's t-test. Differences were considered to be significant for P<0.05.

Results

Tumor/stroma crosstalk stimulates CXCR4 expression in C26 cells and HSCs

Tumor/stroma interaction is known to modulate a wide variety of proteins in both cancer and stromal cells (10). As mentioned above, CXCR4 expression is localized in HSCs in the liver and tumor cells. We treated C26 cells with soluble factors derived from either quiescent HSCs or taHSCs, which resulted in a 4-fold increase in the CXCR4 expression in C26 tumor cells compared to untreated C26 cells when analyzed by western blotting (Fig. 1A) and FACS (Fig. 1B). In the same way, HSCs treated with soluble factors derived from C26 cells showed a 2-fold increased expression of CXCR4 when compared to that of control HSCs (Fig. 1C). Thus, CXCR4 expression was increased in the HSCs and tumor cells after their paracrine interaction.

Figure 1. Expression of CXCR4 in C26 cells and HSCs. CXCR4 protein expression was assessed by western blot analysis in tumor cells treated with either HSC-conditioned medium (cm) or tumor-activated HSC (taHSC)-cm supplemented with 1% FBS for 48 h (A) and on primary HSCs untreated or taHSCs (C). CXCR4 expression was further analyzed by flow cytometryin C26 cells (B).

CXCL12 stimulates the proliferation of C26 cancer cells through CXCR4

Since CXCL12 is known to induce proliferation in CXCR4-expressing cells the effect of CXCL12 on the viability of C26 cells after treatment with increasing concentrations of CXCL12 was quantified. As shown in Fig. 2A, 0.1 and 1 nM CXCL12 significantly enhanced C26 viability after a 24-hincubation, while 10 nM did not alter tumor cell proliferation. Blocking of CXCR4 by addition of AMD3100 (10 µg/ml) before CXCL12 addition was able to revert the effect of the chemokine in regards to C26 cell viability (Fig. 2A). Therefore, CXCR4 mediated the proliferation of CXCL12-stimulated C26 cells.

Figure 2. Tumor cell proliferation and MMP secretion by tumor and HSC cells. (A) Effect of CXCL12 ontumor cell viability. (B) Effect of CXCL12 on MMP-2 and MMP-9 secretion by C26 cells. (C) Effect of CXCL12 on MMP-2 and MMP-9 secretion by HSCs. Differences between control and treated cells were considered statistically significant, *P<0.05.

Effect of CXCL12 on MMP-2 and MMP-9 expression in C26 tumor cells and HSCs

The remodeling of the ECM represents a key event during organ colonization and tumor growth. Tumor and stromal cells contribute to this process by secreting metalloproteases (MMPs), mainly gelatinocytic MMP-2 and MMP-9. Thus, the presence of these MMPs in the conditioned medium derived from C26 cells and HSCs was analyzed after the treatment with either CXCL12 (1 nM), AMD3100 (10 µg/ml) or a combination of both. While CXCL12 was unable to increase the levels of any MMP over basal levels, treatment with AMD3100 slightly decreased the MMP-2 and MMP-9 secretion in tumor and stromal cells compared to those present in the supernatants of the control and CXCL12-treated cell cultures (Fig. 2B and C).

CXCR4 mediates the migration of C26 and HSCs towards CXCL12

CXCL12 has been pointed out as a potent chemokine for CXCR4-expressing cells promoting their migration. Thus, we investigated the role of the CXCR4 receptor in the migration of C26 cells and primary HSCs after activation with CXCL12. As shown in Fig. 3A, CXCL12 induced a 4-fold increase in the migratory potential of C26 cells compared to that of the untreated cells. Blockage of CXCR4 by AMD3100 treatment inhibited the effect of CXCL12 in decreasing the percentage of migrated C26 cells (Fig. 3A). Similarly, CXCL12 activation of primary HSCs also enhanced their migratory potential by 2-fold (Fig. 3B). The effect of this chemokine was partially compromised by using AMD3100, which points to CXCR4 as one of the mediators of CXCL12-induced HSC recruitment into metastatic lesions (Fig. 3B). These results uncover the role of CXCL12/CXCR4 crosstalk during tumor cell and HSC migration.

Figure 3. Effect of CXCL12 and AMD3100 on the migratory potential of C26 cancer cells and HSCs. (A) The migratory ability of C26 tumor cells was quantified after CXCL12 treatment at 24 h. (B) Migration of control and CXCL12-treated HSCs. Differences were considered significant at *P<0.05 and **P<0.05.

CXCR4 blockage via AMD3100 administration reduces the number and size of metastatic foci in the liver

To further investigate the significance of CXCR4 during metastatic spread, in vivo liver metastasis assay was carried out using colorectal cancer C26 cell line. Mice were administered with a daily dose of AMD3100 (5 mg/kg) starting the day after tumor cell injection. After 14 days, mice were sacrificed and the livers were collected for tumor burden analysis by quantifying the liver tissue area occupied by metastatic foci. Daily treatment with AMD3100 antagonist resulted in a significant reduction in the tumor burden in the liver of tumor-bearing mice compared to those treated with vehicle solution (Fig. 4A). In fact, AMD3100-treated tumor-bearing mice showed a 2-fold decrease in the metastatic area occupied in the liver along with a reduced number of foci of every size, and a significant reduction of small foci counts (Fig. 4B and C). Therefore, blocking CXCR4 is linked with reduced liver metastatic growth of colorectal C26 cancer cells in mice.

Figure 4. Metastatic area of tumor-bearing vehicle- and AMD3100-treated mice. Mice injected with 2×105 tumor cells were sacrificed 14 days after tumor cell injection and livers were collected for histological analysis. (A) Histological images of liver tissue from untreated and AMD3100 (5 mg/kg)-treated tumor-bearing mice. (B) The liver area occupied by tumor cells was quantified and is represented as the percentage of liver area occupied by the tumor foci. (C) Quantification of the number of foci in relation to area. Differences were considered statistically significant at *P<0.05.

αSMA-expressing cell recruitment is decreased within liver foci in the AMD3100-treated mice

Activated HSCs express αSMA and help support angiogenesis (10). To confirm the obtained in vitro results, we analyzed the infiltration of αSMA-expressing cells in the tumor foci of the vehicle and AMD3100-treated mice 14 days after tumor cell injection. The immunohistochemical analyses revealed a significant 2-fold decrease in the area positively stained for αSMA within the tumor foci in the liver of the AMD3100-treated mice compared to that of the vehicle-treated mice (Fig. 5A), which corroborates the in vitro results showing reduced migration of HSCs uponantagonist treatment. Hence, HSC expression of CXCR4 appears to be involved in chemotaxis leading to the recruitment of HSCs into tumor areas.

Figure 5. Effect of AMD3100 on αSMA-expressing cell and CD11+Ly6G+ cell recruitment into tumor foci. Mice were sacrificed 14 days after tumor cell injection and livers were collected for immunohistological analysis. (A) The area occupied by tumor-infiltrating αSMA-expressing cells was quantified. Cell nuclei were stained using DAPI. (B) The number of double-positive CD11b+Ly6G+ cells was assessed in the peritumoral areas. The mean of 10 different ×10 fields is represented (10 livers/group). Differences were considered statistically significant at *P<0.05.

AMD3100 treatment is related to reduced CD11b+Ly6G+ MDSC recruitment into the metastatic liver

HSCs are known to promote immune suppression via different pathways (11). In fact, evidence supports their role in the induction of MDSCs (21), which may drive immune suppression in the tumor microenvironment. These immune cells express CXCR4, and their chemotactic migration towards CXCL12 has been reported. Therefore, we investigated the effect of AMD3100 on the recruitment of CD11b+Ly6G+ MDSCs into the metastatic liver. The quantification of CD11b+Ly6G+ MDSCs in tumor-bearing mice revealed a marked 2-fold decrease in the number of positive cells in the liver of the AMD3100-treated mice, compared to that of the vehicle-treated mice (Fig. 5B), which is in line with the decrease in αSMA-positive cells in the tumor foci in the AMD3100-treated tumor-bearing mice. This observation lets us conclude that the infiltration of MDSCs into the metastatic liver may be mediated by expression of CXCR4.

Circulating CD11b+Ly6G+ number is reduced in the AMD3100-treated mice

The increase in MDSCs in the liver could represent an induction in the target organ but could also reflect an increase in the circulating number of MDSCs. Giving the influence of HSC-derived factors in these cells, the differentiation of the circulating pool could be compromised due to attenuated HSC activation and infiltration into the metastatic lesions. In order to assess the rate of blood born MDSCs, tumor-bearing mice treated with AMD3100 or vehicle solution were sacrificed after 14 days of tumor cell inoculation and blood was collected from 4 animals per experimental group for PMNC isolation. Then, the presence of double-positive CD11b+Ly6G+ cell percentage was calculated by flow cytometry. The results revealed a significant 2-fold decrease in CD11b+Ly6G+ cell population in the AMD3100-treated tumor-bearing mice, compared to the mice treated with vehicle where almost 98% of Ly6G+ cells were positive for the monocytic marker CD11b. In contrast, only 43.8% of Ly6G+ cells exhibited a double-positive phenotype in the AMD3100-treated tumo-bearing mice (Fig. 6). Consequently, blockage of CXCR4 and the subsequent response related to CXCL12 ligation promoted a reduced MDSC percentage in the blood of the AMD3100-treated mice.

Figure 6. Effect of AMD3100 treatment on the count of CD11b+Ly6G+ cells in the blood of tumor-bearing mice. Flow cytometric analyses from mouse blood revealed differences in the percentage of double-positive cells in the blood of untreated and AMD3100-treated mice. Double-positive cells for CD11b+Ly6G+ in tumor-bearing untreated mice are represented in the dark grey column while AMD3100-treated mouse double-positive cells are represented in the light grey column.

Discussion

CXCR4 has been reported in several human and mouse cancer cell lines (9,22,23). However, modulation of the expression of this receptor in the tumor microenvironment remains poorly characterized. CAFs have been shown to promote the expression of a wide variety of cell receptors and adhesion molecules.

Herein, we showed that soluble factors derived from control or tumor-activated HSCs increased the expression of CXCR4 in tumor cells. This increase may be mediated by the production of IL-1 and IL-6 by HSCs, which has been shown to promote CXCR4 upregulation in different cell types (24,25). Moreover, both human and mouse HSCs express CXCR4. In fact, the increased expression of this receptor has been linked to progressive HSC activation under culture conditions. Furthermore, the upregulation of CXCR4 has also been reported in fibrotic livers, primarily related to activated HSCs (26). Interestingly, HSC activation has also been observed during metastatic colonization of the liver (10). Consistent with these findings, we reported upregulated CXCR4 expression in tumor-activated HSCs in vitro.

The metastatic process is orchestrated by circulating tumor cells in concert with cells of the target organ. Tumor cells migrate towards soluble stimulus, which attract them to the colonizing organ. Among others, CXCL12, the main lig and for CXCR4, has been reported to drive tumor cell migration and has been shown to be upregulated in the liver under pathological conditions, including cancer (27). In line with several reports, our results showed that C26 migration was increased after CXCL12 stimulation (25) whereas C26 treatment with CXCR4 antagonist AMD3100 led to attenuated response to this chemokine. This phenomenon could be in part mediated by Rho and Rac GTPases, involved in metastatic colonization, as observed in colon cancer and hepatoma cells after CXCR4-mediated CXCL12 stimulation (28), therefore, promoting cancer cell extravasation. Indeed, the formation of pseudopodia was described in CXCL12-treated multiple myeloma cells (29), which may also account for the facilitation of tumor migration by CXCL12-treated C26 cells.

Once in a target organ, cancer cells begin to proliferate in an uncontrolled manner driving tumor growth. Target organ cells support the tumor growth by the secretion of a wide variety of growth factors, cytokines and chemokines. In the liver, CXCL12 is produced by HSCs and LSECs in addition to malignant cells (30). Activated LSECs and HSCs secrete CXCL12, which may have an effect on CXCR4-expressing metastasizing colorectal cancer cells. Our results showed, however, that CXCL12 slightly increased the proliferation of C26 cells, a response that could be more pronounced in vivo, since stromal cells stimulate CXCR4 expression in C26 cells. The activation of the MAPK pathway could be responsible for this proliferation enhancement, as previously reported in other cell types (31). Moreover, this chemokine switches on the PI3K/AKT pathway, driving tumor growth (32), which could account for the increased C26 cancer cell viability.

Tumor growth is usually accompanied by ECM remodeling, a required step for disease progression. ECM remodeling is carried out by both tumor and stromal cells, such as HSCs in the liver. Intriguingly, the migration of HSCs was also fostered after treatment with this chemokine, which was partially abrogated after blocking CXCR4 using AMD3100. After interaction of C26 cells with liver sinusoidal endothelial cells (LSECs), Valcarcel et al observed an increase in VEGF, known to be an upstream inductor of CXCL12 in the liver (33), along with the initiation of an inflammatory response (20) that may account for HSC recruitment in tumor foci. This observation agrees with previous reports linking stimulated migration ability of human primary HSCs treated with CXCL12, along with CXCL12-mediated contraction mechanism in mouse primary HSC line, which may play a significant role in this process (34,35). Surprisingly, we observed no changes in MMP-2 or MMP-9 activation in both C26 and HSCs treated with CXCL12, in line with other groups reporting no effect in CXCL12-treated cells (36). This result could be explained by the fact that HSCs do secrete CXCL12, which could activate in an autocrine manner the secretion of MMPs (37). In concordance with these results, we also observed a decrease in the expression of MMP-2 in these cells after AMD31000 treatment, which avoids the interaction of CXCL12 with its receptor CXCR4.

The production of CXCL12 in the liver is elevated (38) and this organ becomes a perfect collecting site for CXCR4-expressing circulating cells. Thus, this might be one of the reasons why the liver represents one of the main target organs for metastatic colorectal cancer cells. In fact, the administration of CXCR4 antagonist AMD3100 in colorectal liver metastasis model reduced by 2-fold the metastatic area in the liver. Several reports have shown similar results by hijacking the circulating CXCL12 or administering AMD3100 into nude mice, therefore, impairing the metastatic colonization of the liver by tumor cells. The CXCR4-CXCL12 interaction mediated proliferative response observed in colorectal cancer cells (39,40) can be blocked by AMD3100 administration in the tumor-microenvironment, contributing to the reported decrease in the metastatic area of AMD3100-treated tumor-bearing mice. This observation may be further mediated by reduced αSMA-expressing cells into metastatic foci, which impairs the angiogenic response and, consequently, the blood-mediated oxygen and element supply to the tumor mass, a required step for disease development. In addition, CXCR4-CXCL12 interplay has been shown to be responsible for elevated VEGF availability in tumor areas, driving new vessel formation (41). Interestingly, this group reported that HSCs mediated VEGF upregulation during liver metastasis, which may facilitate LSEC recruitment and transition from avascular to vascular stage of metastatic foci (10). Therefore, AMD3100 administration might interfere in the VEGF secretion by CXCL12-activated HSCs, leading to impaired neovascularization due to deficient LSEC recruitment.

These activated myofibroblast-like cells secrete a wide array of cytokines, such as, IL-1, IL-6 or IL-10 (42). IL-1 cytokine, along with VEGF, promote the differentiation and generation of MDSCs from the bone marrow, a highly relevant immunosuppressive cell population, profoundly linked with cancer progression and immune tolerance. Chou et al showed not only that HSCs can promote generation of MDSCs but also that these MDSCs possess a potent immune inhibitory activity (43). Our results showed that the proportion of circulating CD11b+Ly6G+ cells was elevated in untreated tumor-bearing mice, compared to those mice treated with AMD3100. This decrease in MDSCs may be due to the down regulation of differentiation stimuli secreted by both tumor cells and HSCs when blocking CXCR4, hence, leading to lower numbers of MDSCs. In fact, these CD11b+Ly6G+ cells express CXCR4 and migrate towards CXCL12 (44), therefore, accumulating in tumor areas. Our in vivo studies revealed that blocking CXCR4 drove to an attenuated recruitment of this suppressive cell subset into metastatic liver, in concordance with several studies reporting decreased numbers of these cells in breast carcinoma or subcutaneous lesions of hepatoma and breast carcinoma cells using CXCR4 antagonists (45,46). This is also in close relation with the reduced number of HSCs recruited to the tumor foci after blockage of CXCR4 signaling by AMD3100. Interestingly, the soluble factors produced after C26 interaction with LSECs recruit HSCs through a COX-2 mediated pathway (data not shown), and the same pathway is required for MDSC accumulation driven by recruited activated HSCs (21).

Taken together, our data uncovered the relevance of tumor and stromal crosstalk driving CXCR4-CXCL12 interplay stimulation, which in turn favors the recruitment of tumor cells and MDSCs into the liver. This is in relation to the infiltration of αSMA-expressing activated HSCs into metastatic foci which fosters the recruitment and accumulation of differentiated CD11b+Ly6g+ cell subsets from the circulation of tumor-bearing mice, driving liver metastasis. All of these processes are reduced by the administration of AMD3100, implying CXCR4/CXCL12 interaction as a perfect target for the inactivation of multiple pro-tumoral processes involving both tumor and host cells.

Acknowledgements

This study was financially supported in part by a postdoctoral fellowship from the University of the Basque Country to A.B. and by funds from the Basque Government-Saiotek to B.A.

Competing interests

The authors declare that they have no competing interests.

Glossary

Abbreviations

Abbreviations:

References