Enhancement of the CXCL12/CXCR4 axis due to acquisition of gemcitabine resistance in pancreatic cancer: effect of CXCR4 antagonists

Human pancreatic cancer (PaCa) cell lines MIA PaCa-2 and AsPC-1 and human dermal fibroblast were obtained from the American Type Culture Collection (Rockville, MD) and Kurabo Industries (Osaka, Japan), respectively. The cell lines were maintained at 37 °C with 5 % CO₂ in a humidified atmosphere. The following media were used: (1) MIA PaCa-2 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM), (2) AsPC-1 cells were incubated in Roswell Park Memorial Institute (RPMI-1640) medium (Sigma Aldrich, St Louis, MO, USA) supplemented with 10 % fetal bovine serum (FBS) and antibiotics and (3) fibroblasts (FB) were maintained in fibroLife S2 Comp kit (Kurabo Industries Ltd., Osaka, Japan) supplemented with 2 % FBS.

Gemcitabine (GEM) was purchased from Toronto Research Chemicals, Inc. (Toronto, Ontario, Canada). First, we determined the half maximal inhibitory concentration (IC50) of GEM for MIA PaCa-2 or AsPC-1 cells using the Premix WST-1 Cell Proliferation Assay System (Takara Bio, Japan) according to the manufacturer’s instruction. Briefly, MIA PaCa-2 or AsPC-1 cells were seeded at a density of 2 × 10³ cells per 100 μL in 96-well plates and allowed to adhere overnight. Then, cultures were re-fed with fresh media containing various concentrations of GEM. After 72 h of incubation, absorbance was measured at 450 nm in each well using a SpectraMax 340 spectrophotometer (Molecular Devices, CA, USA). The IC50 of GEM for each pancreatic cancer line was determined by constructing a dose-response curve. Each pancreatic cancer cell line was passaged in the cell lines’ IC50 concentration of GEM for 2 to 3 weeks. After passage, we again determined the cell lines’ IC50 value for GEM. Then, each pancreatic cancer cell line was passaged in the cell lines’ re-determined IC50 concentration of GEM for 2 to 3 weeks. The process was repeated at increasing doses of GEM until the cell lines demonstrated at least a 50-fold greater IC50 value for GEM than the parental cell lines. The resultant cell lines were resistant to GEM at a concentration of 20 μM.

The proliferation assay was conducted using the Premix WST-1 Cell Proliferation Assay System (Takara Bio, Japan) according to the manufacturer’s instruction. Briefly, GEM-resistant (GEM-R) and GEM-sensitive (GEM-S) MIA PaCa-2 or AsPC-1 cells were seeded at a density of 2 × 10³ cells per 100 μL in 96-well plates and allowed to adhere overnight. Then, cultures were re-fed with fresh media containing various concentrations (0–100 μM) of GEM. After 72 h of incubation, absorbance was measured at 450 nm in each well using a SpectraMax 340 spectrophotometer.

The expression of CXCR4 protein by GEM-R/S MIA PaCa-2 and AsPC-1 cells was examined using the CXCR4 ELISA kit (USCN Life Science Inc., Wuhan, China) according to the manufacturer’s instructions. A total of 1 × 10⁵ GEM-R/S cells were seeded in each 100 mm dish. Then, we added different concentrations of GEM (0 – 20 μM), and the cells were incubated for 72 h. After indicated treatments, cell lysates were prepared. A total of 150 μg of protein was taken for ELISA assay. Similarly, CXCL12 levels in the supernatant from FB co-cultured with GEM-R/S MIA PaCa-2 cells were determined using the CXCL12 ELISA kit (R&D, Minneapolis, MN, USA) according to the manufacturer’s instruction. To determine the synergistic effect of the tumor-stromal interaction, we cultured FB (1.0 × 10⁶ cells in 6-well plates) with or without GEM-R/S [1.0 × 10⁶ cells on inserts with 0.4-μm pores (Thermo Scientific, Rockford, IL, USA)] for 72 h using a double chamber method. After the incubation, the media were collected and microfuged at 1500 rpm for 5 min to remove particles. The supernatants were frozen at -80 °C until use. A total of 150 μg of the protein was taken for ELISA assay.

Total RNA was extracted from cell pellets using an RNeasy Plus Mini Kit (Qiagen, TX, USA), and RT-PCR was performed using Superscript III First-strand Synthesis SuperMix for qRT-PCR (Invitrogen, Carlsbad, CA, USA). The concentration of each cDNA was measured with a NanoDrop1000 (Thermo Fisher Scientific, DE, USA) and adjusted to 40 ng/mL with diethylpyrocarbonate (DPEC)- treated water. We performed real-time PCR with FAM-labeled TaqMan probes (CXCR4: Hs00607978_s1; CXCL12: Hs03676656_mH; GAPDH: Hs99999905_m1; CXCR7: Hs00664172_s1 (Applied Biosystems, Foster City, CA, USA)) and TaqMan Universal Master Mix (Applied Biosystems) using Chromo4 (BioRad, MA, USA). PCR was carried out by an initial incubation at 50 °C for 2 min, followed by denaturation at 95 °C for 10 min and 50 cycles of 95 °C for 15 s and 60 °C for 1 min.

The expression of CXCR4 protein in GEM-R/S MIA PaCa-2 cells was detected by immunostaining. Three days after treating with GEM, GEM-R/S PaCa cells were washed twice with ice-cold PBS, fixed in 4 % paraformaldehyde for 20 min at room temperature and washed twice with ice-cold PBS. The cells were then incubated for 15 min in PBS containing 0.5 % Triton X-100, washed with PBS, blocked in 1 % BSA in PBS for 30 min and incubated with rabbit anti-CXCR4 polyclonal antibody (1:100, Abcam, Cambridge, UK) at 4 °C overnight. Subsequently, the cells were washed with PBS, incubated with Alexa Fluor 488 goat anti-rabbit IgG (H + L) (1:100, Life Technologies, Carlsbad, CA) and mounted with Prolong® Gold Antifade Reagent with DAPI (Life Technologies, Carlsbad, CA).

In vitro invasion assays were performed using the BD Bio-Coat Matrigel invasion assay system (BD Biosciences, Franklin Lakes, NJ) according to the manufacturer’s instructions. Briefly, GEM-R/S cells (2.5 × 10⁴ cells) were seeded into the Matrigel precoated Transwell chambers consisting of polycarbonate membranes with 8.0 μm pores. The Transwell chambers were then placed into 6-well plates, into which we added basal medium only or basal medium containing various concentrations of recombinant CXCL12. After incubating GEM-R/S cells for 22 h, the upper surface of the Transwell chambers was wiped with a cotton swab and the invading cells were fixed and stained using Diff-Quick cell staining kit (Dade Behring, Inc., Newark, DE). The number of invading cells was counted in 5 random microscopic fields (200×). To confirm whether the invasive potency of PaCa cells was increased by FB-derived CXCL12 and inhibited by the CXCR4 antagonists, AMD11070 (AdooQ BioScience, Irvine, CA) and KRH3955 (Kureha Chemical Industry, Tokyo, Japan), we performed an invasion assay for GEM-R/S cells using a double-chamber method. Briefly, we co-cultured GEM-R/S cells (2.5 × 10⁴ cells in Transwell chambers) with FB (1 × 10⁴ cells in 6-well plates) blocking with or without CXCR4 antagonists, AMD11070 and KRH3955, at a concentration of 1 μM. After incubation for 22 h, invading cells were counted in the same manner.

All animal studies were conducted in accordance with the guidelines established by the internal Institutional Animal Care and Use Committee and Ethics Committee guidelines of Nagoya City University.

A total of 2 × 10⁷ MIA PaCa-2 cells were injected subcutaneously into mice. Tumors were measured weekly and tumor volume was documented. Tumors were allowed to grow until they reached a volume of 1 cm³, at which time the mice were sacrificed and the tumor tissue was harvested. For serial transplantation, the harvested tumor tissues were chopped into pieces approximately 1 to 2 mm³ in dimension. Tumor pieces were implanted subcutaneously into the mice. GEM and CXCR4 antagonists were administered 3 weeks after tumor implantation as follows: 25 mg GEM/kg body weight, 1 mg AMD11070, and 1 mg KRH3955/kg body weight were given intraperitoneally every week.

Mice were randomly assigned to 1 of the following 6 treatment groups (4 mice per group): group I was not given any drugs; group II was given GEM alone; group III was given AMD11070 alone; group IV was given KRH3955 alone; group V was given GEM plus AMD11070; group VI was given GEM plus KRH3955. Therapy was continued for 4 weeks, and the mice were sacrificed 2 weeks later. We calculated the tumor volume according to the following formula: tumor volume (mm³) = d² x D/2, where d and D were the shortest and longest diameter, respectively. Finally, the tumors were harvested from mice after the treatment and fixed in formaldehyde for further analysis.

Formalin-fixed, paraffin-embedded mouse tumor tissue sections were mounted on 3-amino-propyltriethoxylsilane-coated slides. Dewaxed paraffin sections were placed in a microwave (10 min, 600 watts) to recover antigens before staining. Antibodies used were as follows: rabbit anti-CXCR4 polyclonal antibody, rabbit anti-SDF-1α polyclonal antibody (1:50) and, rabbit anti-Hypoxia-Inducible Factor (HIF)-1α monoclonal antibody (1:100) (Abcam, Cambridge, UK), followed by secondary antibodies conjugated to biotin. Peroxidase-conjugated streptavidin was used with 3,3-diaminobenzidine tetrahydrochloride (DAB) (Biocare Medical, Concord, CA, USA) as the chromogen for detection. Hematoxylin was used for nuclear counterstaining. CXCR4-positive PaCa cells, CXCL12-positive stromal cells and HIF-1α-positive PaCa cells exhibited DAB-positive (brown) staining; negative cells were stained with the hematoxylin counterstain only. The number of CXCR4-immunoreactive cells in mouse specimens was expressed as a percentage of the total number of cells that were randomly counted in 10 fields at × 400 magnification. For each image, a color deconvolution method was used to isolate CXCL12-positive and HIF-1α-positive DAB-stained cells from CXCL12-negative and HIF-1α-negative hematoxylin-stained cells. DAB and hematoxylin were digitally separated using ImageJ software (version 1.46c; WS Rasband, National Institutes of Health, Bethesda, MD, USA, http://​rsb.​info.​nih.​gov/​ij/​) and an ImageJ plugin for color deconvolution that calculated the contribution of DAB and hematoxylin, based on stain-specific red-green-blue (RGB) absorption. Following deconvolution, the scale was set to the 200 μm scale bar on each image. The measurement parameter was integrated optical density (IOD). Optical density was calibrated and the area of interest was set as follows: hue, 0–30; saturation, 0–255; intensity, 0–255. Then, the values were counted. The IOD was log₁₀ transformed [27].

The activity of NF-κB was measured using NF-κB (p65) transcription factor assays.　A total of 1 × 10⁵ GEM-R/S cells of MIA PaCa-2 cells were seeded in 100-mm dishes and incubated with different concentrations of GEM for 72 h. After indicated treatments, nuclear proteins were extracted using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, IL, USA). The concentrations of nuclear proteins were measured using a Pierce BCA Protein Assay Kit (Thermo Scientific), and protein concentrations were adjusted for equal loading (200 μg/mL). The levels of NF-κB p65 protein detected with the NF-κB p65 ELISA kit (Invitrogen, USA) according to the manufacturer’s instructions.

All measurement data were expressed as means ± standard deviation (SD). They were calculated for experiments performed in triplicate (or more). Multiple group comparisons were performed by using one-way analysis of variance (ANOVA) followed by the Dunnett test, and Bonferroni tests were used for post hoc 2-sample comparisons. A two-sided p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, http://​www.​jichi.​ac.​jp/​saitama-sct/​SaitamaHP.​files/​statmedEN.​html; Kanda, 2012), a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria, version 2.13.0). More precisely, EZR is a modified version of R Commander (version 1.6-3) that was designed to add statistical functions frequently used in biostatistics.

We first determined how GEM affected the proliferation of PaCa cells that were sensitive or resistant to the drug. We used 2 GEM-R PaCa cell lines, MIA PaCa-2 and AsPC-1. With these 2 cell lines, we found that GEM significantly inhibited GEM-S cell proliferation in a dose-dependent manner (P < 0.01); however, it could not inhibit GEM-R cell proliferation at the doses used (Fig. 1a, MIA PaCa-2; Fig. 1b, AsPC-1).

The expression of CXCR4 protein by PaCa cells was examined by means of ELISA assays. In MIA PaCa-2, the expression of CXCR4 protein by GEM-S cells was significantly inhibited by GEM in a dose-dependent manner (P < 0.01) (Fig. 2a). In contrast, the expression of CXCR4 protein by GEM-R cells showed a significant dose-dependent enhancement by GEM (P < 0.01) (Fig. 2b). In AsPC-1, there was no change of expression of CXCR4 protein during GEM treatment of sensitive cells (Fig. 2c). However, the expression of CXCR4 protein by resistant cells significantly increased in the presence of GEM in a fashion that varied with the dose (P < 0.01) (Fig. 2d). Furthermore, in RT-PCR, there was no change of CXCR4 mRNA levels by GEM treatment of sensitive MIA PaCa-2 cells (Fig. 3a). However, the level of CXCR4 mRNA in GEM-R cells was significantly elevated by treatment with GEM in a dose-dependent manner (P < 0.01) (Fig. 3b).

In immunocytochemical assays, staining of CXCR4 protein was primarily found in the cell membrane of GEM-S and GEM-R cells. These cells were treated with GEM at concentrations of 0 μM, 1 μM, 10 μM and 20 μM. The staining of CXCR4 in GEM-R cells was enhanced by GEM as the dose was increased [CXCR4 (green) and DAPI (blue)] (Fig. 3c-j).

To investigate the function of the CXCL12-CXCR4 signaling axis, we determined whether the secretion of CXCL12 from FB was enhanced when FB were co-cultured with PaCa cells. The expression of CXCL12 mRNA in FB was significantly enhanced by co-culturing with GEM-R PaCa cells treated with GEM (P < 0.01) (Fig.4a). Furthermore, the secretion levels of CXCL12 protein from FB were significantly enhanced by co-culturing with GEM-R PaCa cells treated with GEM (P < 0.01) (Fig. 4b).

When GEM-R PaCa cells were exposed to GEM, the invasive behavior of these cells was significantly enhanced by stimulation with recombinant CXCL12 (P < 0.01) (Fig. 5a). Moreover, the enhanced invasive behavior of GEM-R PaCa cells was significantly inhibited by exposure to CXCR4 antagonists (P < 0.01) (Fig. 5b). Furthermore, when GEM-R PaCa cells were exposed to GEM, the invasive behavior of these cells was significantly elevated by co-culturing with FB (P < 0.01) (Fig. 5c). The activated invasive behavior of GEM-R PaCa cells was significantly inhibited by treatment with neutralizing CXCR4 antagonists (P < 0.01) (Fig. 5d). Photographs show alterations of invasive behavior of the PaCa cells in each treatment group (Additional file 1: Figure S1).

On the basis of these results, we asked whether CXCR4 antagonists affected the growth of GEM-R PaCa cells in vivo, either alone or in combination with GEM. Six experimental conditions were tested (see Methods, Experimental Protocol). Using mice implanted with GEM-R 4 weeks earlier, the final tumor volume of group II (GEM+) was significantly greater than that found in any of the other groups (P < 0.01) (Fig. 6a). With GEM-S PaCa cells, the tumor volume of group I (no treatment) was significantly greater than groups II (GEM+), V (GEM+ AMD+) and VI (GEM+ KRH+) (P < 0.01) (Fig. 6b).

In GEM-R PaCa cells, tumor volumes in group V (GEM+ AMD+) and groupVI (GEM+ KRH+) were significantly less than groups III (AMD+) and IV (KRH+) (P < 0.05) and group I (no treatment) (P < 0.01) (Fig. 6c). There was no significant difference among groups I (no treatment), III (AMD+) and IV (KRH+). However, tumor volume in group II (GEM+) was significantly smaller than groups III (AMD+) and IV (KRH+) (P < 0.05) (Fig. 6d).

The photographs show the differences of the final tumor volume of all groups in GEM-R PaCa cells (Fig. 6e) and in GEM-S PaCa cells (Fig. 6f).

CXCR4 protein was primarily identified in the cell membrane of PaCa cells. In contrast, it was not detected in normal stromal cells of noncancerous regions in PaCa tissue. Staining of CXCR4 protein in GEM-R cells treated with GEM was greatly enhanced (Fig. 6g-j). Significantly more CXCR4-positive cells were observed in GEM-R cells treated with GEM than GEM-R cells lacking such treatment (P < 0.05), GEM-S treated with GEM (P < 0.01) and GEM-S without GEM treatment (P < 0.01) (Additional file 2: Figure S2A). Staining of CXCL12 protein primarily occurred in the cytoplasm of stromal cells around PaCa cells, but it was not detected in PaCa tissues. Staining of CXCL12 protein was greatly enhanced in stromal cells around GEM-R treated with GEM (Fig. 6k-n). CXCL12 IOD values in stromal cells around GEM-R PaCa cells treated with GEM were significantly enhanced compared with other groups (P < 0.01) (Additional file 2: Figure S2B).

To examine the details of the molecular mechanisms, the activity of NF-κB in GEM-R/S PaCa cells was measured by NF-κB (p65) transcription factor assay. The activity of NF-κB in GEM-R MIA PaCa-2 cells was significantly higher compared to GEM-S MIA PaCa-2 cells (P < 0.01) (Fig. 7a). Moreover, the activity of NF-κB in GEM-R MIA PaCa-2 cells was significantly enhanced by GEM dose dependently (Fig. 7b).

Similarly, since HIF-1α might regulate the expression of CXCR4, we examined the expression of HIF-1α in implanted tumor tissue. Staining of HIF-1α protein in GEM-R cells treated with GEM was greatly enhanced (Fig. 7c). Significantly more HIF-1α-positive cells were observed in GEM-R cells treated with GEM than GEM-R cells lacking such treatment (P < 0.01), GEM-S treated with GEM (P < 0.01) and GEM-S without GEM treatment (P < 0.01) (Fig. 7d).

Since CXCR7 is another receptor of CXCL12, we examined the CXCR7 expression in GEM-R/S MIA PaCa-2 cells by RT-PCR. In RT-PCR, the expression levels of CXCR7 mRNA in GEM-R Mia PaCa-2 cells were significantly higher compared with GEM-S Mia PaCa-2 cells. There was no change of CXCR7 mRNA levels by GEM treatment of both GEM-R/S Mia PaCa cells (Additional file 3: Figure S3).