PDGFR-induced autocrine SDF-1 signaling in cancer cells promotes metastasis in advanced skin carcinoma

Advanced and undifferentiated skin squamous cell carcinomas (SCCs) exhibit aggressive growth and enhanced metastasis capability, which is associated in mice with an expansion of the cancer stem-like cell (CSC) population and with changes in the regulatory mechanisms that control the proliferation and invasion of these cells. Indeed, autocrine activation of PDGFRα induces CSC invasion and promotes distant metastasis in advanced SCCs. However, the mechanisms involved in this process were unclear. Here, we show that CSCs of mouse advanced SCCs (L-CSCs) express CXCR4 and CXCR7, both receptors of SDF-1. PDGFRα signaling induces SDF-1 expression and secretion, and the autocrine activation of this pathway in L-CSCs. Autocrine SDF-1/CXCR4 signaling induces L-CSC proliferation and survival, and mediates PDGFRα-induced invasion, promoting in vivo lung metastasis. Validation of these findings in patient samples of skin SCCs shows a strong correlation between the expression of SDF1, PDGFRA, and PDGFRB, which is upregulated, along CXCR4 in tumor cells of advanced SCCs. Furthermore, PDGFR regulates SDF-1 expression and inhibition of SDF-1/CXCR4 and PDGFR pathways blocks distant metastasis of human PD/S-SCCs. Our results indicate that functional crosstalk between PDGFR/SDF-1 signaling regulates tumor cell invasion and metastasis in human and mouse advanced SCCs, and suggest that CXCR4 and/or PDGFR inhibitors could be used to block metastasis of these aggressive tumors.

. Effect of CXCR4 inhibition and SDF-1 knockdown on PD/S cell proliferation.       *, significant differences between the compared groups (t-test; P ≤ 0.05). I, growth kinetics (mean ± SE of tumor size, mm 3 ) of tumors generated after hSCC11 cell engrafting in immunodeficient mice, which were treated with vehicle solution (control) and AMD3100 (seven tumors per group). P value (t-test) of the compared groups is indicated. J, mean of metastatic foci (± SE) per lung section (categorized by size, mm 2 ) developed in control and AMD3100-treated mice (six mice per group). *, significant differences between the compared groups (t-test; P ≤ 0.05). anti-rabbit-HRP 1/1000 DAKO P0448

Isolation of SCC cells
Excised mouse tumors and fresh human skin SCC samples were minced and incubated with collagenase I (60 U/ml; Sigma) and dispase (0.7 U/ml; Gibco) overnight at 37 ºC.
Cell suspensions were filtered and depleted of red blood cells using ACK lysis buffer (Lonza). For endothelial cell depletion, mouse cell suspensions were incubated with anti-CD31 antibody for 30 min at 4 ºC, and then with Dynabeads ® anti-rat for 30 min at 4 ºC. Isolated tumor cells were then plated and cultured as described in Methods.

Cell proliferation assays and treatments
To analyze cell proliferation, PD/S cells were plated (2 x 10 3 mouse cells and 4 x 10 3 human cells per well) in six replicates on 96-well plates in basic medium, with a previous withdrawal of puromycin for 48 h in the case of transduced cells. To evaluate the effect of AMD3100, 4 x 10 3 mouse and human cells were seeded as described above. Then, fresh basic medium without or with AMD3100 (1 μg/mL and 5 μg/mL; Chemscene LLC) or with murine or human SDF-1 (150 ng/mL; PeproTech) was added. To evaluate the effect of Imatinib on human hSCC11 cells, 4 x 10 3 cells were seeded per well in six replicates on 96-well plates in basic medium. Then, fresh basic medium without or with Imatinib (LC Laboratories) and without or with PDGF-AA (30 ng/ml), PDGF-BB (30 ng/ml) or PDGF-CC (100 ng/ml) was added. In these assays, fresh medium with factors and/or inhibitors was added every 3 days. Each assay was carried out 2-3 times. Cell proliferation/survival were measured by calculating the mean (± SE) absorbance at 560 nm after 6 days of treatment by using 3-(4,5-dimethyl-2thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay.
After transfer, proteins were incubated overnight with the antibodies anti-PDGFR; anti-PDGFRβ and anti-GAPDH-HRP (Supplementary Table 2). Antibody binding was detected with a secondary antibody coupled to horseradish peroxidase using enhanced chemiluminescence detection reagents (Amersham).

Reverse transcription and quantitative PCR
RNA samples were previously treated with DNAse (Sigma). Reverse-transcription reactions were carried out with the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Real-time PCR reactions were performed (three replicates for each sample) on an Applied QuantStudio5, using SYBR Green Mix (Applied Biosystems) and primers described in Supplementary Table 3. Gapdh/GAPDH, Ppia, HPRT and GUSB were used to normalize the gene expression for all human and mouse samples. mRNA levels were shown as relative to Gapdh/GAPDH mRNA, or alternatively as fold change. In this case, mean of mRNA levels relative to two or three housekeepings was calculated.

Statistical analysis
Statistical analysis was performed using Prism 5.0 Software (GraphPad software, San Diego, CA). Normal distribution of data and similarity of variances between compared groups were checked using the Shapiro-Wilk Normality test and F-Fisher test, respectively. Student's t-test (two-tailed) was used to determine the significance of differences between groups, as detailed in legend figures. Data are presented as the means ± SE. Scatter plot and Pearson coefficient were to correlate the relative gene expression of CXCL12 and PDFGRA/B in patient samples. No statistical methods were used to predetermine sample size in in vivo experiments, which was estimated based on our previous experience and similar experiments reported in literature.