Stem cell factor produced by tumor cells expands myeloid-derived suppressor cells in mice

Immunotherapy is a novel treatment approach for cancers; however, its therapeutic effects are impeded by myeloid-derived suppressor cells (MDSCs). This study aimed to determine how MDSCs are expanded in cancer hosts. MDSCs were positive for Gr-1 and CD11b. Hepa1-6 hepatoma cells, EL4 lymphoma cells, and mice bearing Hepa1-6 hepatoma or lymphoma were examined. Following the inoculation of Hepa1-6 cells into the flanks of mice, a linear correlation was evident between the frequency of MDSCs in the spleen and tumor sizes. MDSC numbers diminished gradually and returned to the normal level within 3 weeks if the tumors were excised. To identify the cytokines produced by tumor cells that allowed expansion of MDSCs, cytokines in Hepa1-6 cell culture medium and murine serum were examined using a cytokine array. Stem cell factor (SCF) was implicated as the relevant cytokine. When recombinant SCF was added to the spleen cell culture medium, MDSC expansion could occur. In the presence of c-kit blockade, this effect of SCF was partially reversed. In conclusion, MDSCs can be expanded in tumor cells in a process that involves SCF released by tumor cells.

www.nature.com/scientificreports/ In humans, MDSCs have been identified in patients with advanced cancers 28,29 . Many factors may expand MDSC [30][31][32] , but there is no direct evidence. This study examined the possible factors associated with expansion of MDSCs. The findings could inform the design of strategies for treatment of malignancies.

Results
MDScs in the spleen of tumor-bearing mice. To determine MDSC induction, spleens were harvested from B10 mice harboring Hepa1-6 tumors and from B6 mice harboring EL4 tumors, when the tumors were approximately 15 mm in diameter. A suspension of single spleen cells was prepared. The spleen cells were stained using phycoerythrin (PE)-conjugated anti-Gr-1 and fluorescein isothiocyanate (FITC)-conjugated anti-CD11b, and analyzed by flow cytometry. Gr-1 + CD11b + cells were detected in the tumor-bearing mice (Fig. 1a). To determine whether the population of Gr-1 + CD11b + cells was correlated with tumor size, 10 B10 mice bearing

Reversal of MDSc frequency by removal of tumors. B10 mice were inoculated in their flank with
Hepa1-6 tumor cells. When the diameters of the tumors approached 15 mm, the tumors were excised completely. The spleens were harvested at postoperative week 1, 3, and 5. The frequency of MDSCs was 2.53 ± 1.44% in naïve mice (n = 11). According to the above equation relating MDSC frequency and tumor size, the frequency of MDSC in the spleen was 22.2% for 15 mm-diameter tumors. After excision of the tumors, the frequency of MDSCs in the spleen was 5.23 ± 2.42% (n = 3) at postoperative week one (p = 0.045, compared to naïve mice), 2.97 ± 1.86% (n = 3) at postoperative week 3 (p = 0.663, compared to naïve mice), and 3.12 ± 1.05% (n = 2) at postoperative week 5. These findings indicated that MDSCs were maintained by tumors. In this tumor model, the frequency of MDSCs diminished gradually and returned to normal within 3 weeks after the tumors were excised. expansion of MDScs by conditioned medium. The basis of the expansion of MDSCs is unclear.
The finding that tumor removal could eliminate MDSCs in the spleen indicated that some cytokines must be released by cancer cells to induce MDSCs. To confirm that such eligible cytokines were produced by cancer cells, conditional medium collected from Hepa1-6 cell culture was used to culture naïve spleen cells. Gr-1 + CD11b + MDSCs were analyzed by flow cytometry on day 1, 3, 5, and 7. Gr-1 + CD11b + cells were not increased in the first 3 days. However, beginning on day 5, increased numbers of Gr-1 + CD11b + were evident (14.11 ± 4.71% versus 2.21 ± 1.06% for the control, p < 0.001). On day 7, the proportion of Gr-1 + CD11b + cells among the total cells had increased significantly (23.97 ± 10.43% versus 2.66 ± 1.71% for the control, p = 0.006). These results clearly showed that cytokine(s) capable of stimulating expansion of MDSCs were produced by cancer cells, and that Gr-1 + CD11b + MDSCs could be expanded during culture in the conditioned medium (Fig. 2).

Release of cytokines by tumor cells.
To further identify the cytokines released by cancer cells, the conditioned medium was analyzed using a cytokine array. Compared with RPMI-1640 containing 5% fetal calf serum (FCS), several cytokines were detected at higher levels in the conditioned medium. These included  www.nature.com/scientificreports/ granulocyte-macrophage colony stimulating factor (GM-CSF), stem cell factor (SCF), monocyte chemoattractant protein 1 (MCP-1), soluble tumor necrosis factor-receptor 1 (sTNF-R1), pro-matrix metalloproteinase-9 (pro-MMP-9), and osteoporotegerin (Fig. 3a). Cytokines in the serum from naïve and Hepa1-6 tumor-bearing mice were identified using the cytokine array. The expression levels of several cytokines including SCF, MCP-1, granulocyte-colony stimulating factor (G-CSF), sTNF-R1, and pro-MMP-9 were increased in the serum of tumor-bearing mice (Fig. 3b). SCF, MCP-1, sTNF-R1, and pro-MMP-9 were increased in both the conditioned medium and serum. When SCF, MMP-9, or SCF and MMP-9 was added to spleen cell culture medium, MDSCs only appeared in the medium containing SCF. Therefore, SCF was considered as the most eligible cytokine capable of inducing MDSCs in conditioned medium or serum of tumor-bearing mice among the common cytokines. To further confirm that SCF was actually released by tumor cells, total protein was extracted from Hepa1-6 and western blotting was performed. Hepa1-6 cells produced soluble and membrane-bound forms of SCF (Fig. 3c). Furthermore, when serum SCF was measured by ELISA, the serum level of SCF tended to correlate with MDSC frequency in tumor-bearing mice [MDSC (%) = 4.357 + 0.131 × SCF (ng/ml); r = 0.577] (Fig. 3d). To determine whether human tumor cells could produce SCF similar to murine tumor cells, total protein was extracted from human Hep G2, 3B, SK, and Huh7 cells and western blotting was performed. These human tumor cells also produced SCF (Fig. 3e).

Expansion of MDSC by SCF in vitro.
To determine whether SCF or MMP-9 was the actual cytokine that stimulated expansion of MDSCs, 12.5 ng/ml of recombinant mouse SCF and/or 10 ng/ml of MMP-9 were added to the B10 spleen cell culture medium. Gr-1 + CD11b + cells detected by flow cytometry were only evident in the culture supplemented with SCF beginning at day 7 of culture (14.85 ± 4.03% versus 5.10 ± 1.35% in RPMI and 4.77 ± 1.76%, p = 0.002). On day 9 of culture, 26.66 ± 8.25% of the cultured spleen cells were Gr-1 + CD11b + (Fig. 4a). Further analysis of cells by staining with Ly-6C and Ly-6G showed that the frequency of CD11b + Ly-6C + cells was higher than that of CD11b + Ly-6G + cells (Fig. 4b). The immune suppressive function of these CD11b + Ly-6G + and CD11b + Ly-6C + cells was determined on the basis of arginase activity. Both the CD11b + Ly-6G + and CD11b + Ly-6C + cells displayed higher arginase activity than splenocytes (7.810 ± 6.211 U/l and 5.271 ± 2.794 U/l, respectively, versus 0.291 ± 0.426 U/l, p = 0.011). We further examined whether SCF was necessary to maintain Gr-1 + CD11b + cells. Spleen cells harvested from tumor-bearing mice were maintained in RPMI-1640 containing different concentrations of SCF. Gr-1 + CD11b + cell numbers diminished gradually if no SCF was present in the culture medium. When SCF was added (≥ 12.5 ng/ml), the population of Gr-1 + CD11b + cells could be maintained. At 7 days of culture in the presence of SCF, the frequency of Gr-1 + CD11b + cells were www.nature.com/scientificreports/ maintained at 38.79 ± 11.36%, compared to 17.46 ± 5.29% for the control (p = 0.333). The respective vales after 9 days of culture with SCF was 22.54 ± 5.09% and 7.15 ± 3.619% (p = 0.003, Fig. 4c).

Suppression of t-cell proliferation by MDScs.
To determine the suppressive effects of MDSCs cultured in the presence of SCF, Gr-1 + CD11b + MDSCs were isolated using a commercial kit and sub-grouped into CD11b + Ly-6G + and CD11b + Ly-6C + cells. These cells were added to a mixed lymphocyte reaction of T-cells activated by anti-CD3/anti-CD28. The proliferation of T-cells was suppressed by MDSCs using a 1:1 ratio of T-cells and CD11b + Ly-6G + or CD11b + Ly-6C + MDSC (Fig. 5). (a) Gr-1 + CD11b + cells appeared from day 7 of culture when 12.5 ng/ml recombinant mouse SCF was added to the culture medium. On day 9 of culture, approximately 25% of the cultured spleen cells were Gr-1 + CD11b + . Gr-1 + CD11b + cells were not increased when MMP-9 was added to the culture medium. (b) When large Gr-1 + CD11b + cells were gated, Ly-6C + cells were more prevalent than Ly-6G + cells. (c) The spleen cells harvested from tumor-bearing mice were cultured in RPMI-1640 and different doses of SCF were added. Gr-1 + CD11b + cells were diminished gradually if SCF was not added to the culture medium. The population of Gr-1 + CD11b + cells could be maintained when SCF (≥ 12.5 ng/ml) was added. www.nature.com/scientificreports/

Blocking SCF decreases the frequency of MDSCs.
To determine whether blocking of SCF would decrease the frequency of MDSCs, Hepa1-6 cells were transfected with pENTR/H1/TO short hairpin RNA (shRNA) targeting SCF. The conditional medium used to culture MDSCs was collected and analyzed. The expression of SCF could be effectively blocked by the silencing of SCF using pENTR/H1/TO shRNA (Fig. 6a). The frequency of MDSCs decreased when the spleen cells were cultured in the conditional medium collected from Hepa1-6 cells transfected with pENTR/H1/TO shRNA (Fig. 6b). As c-kit is the receptor of SCF and Imatinib blocks c-kit, we further examined whether Imatinib could block SCF and decrease the frequency of MDSCs. Spleen cells harvested from B10 mice were cultured in conditioned medium for 7 days to expand MDSCs. Subsequently, different doses of Imatinib (10, 20, and 30 ng/ml) were added to the culture medium. The frequency of Gr-1 + CD11b + cells tended to decrease using 20 or 30 ng/ml Imatinib (21.54 ± 2.68% and 20.07 ± 9.43% compared to 28.92 ± 6.97% for the control, p = 0.099) (Fig. 7a). To examine the effects in vivo, Imatinib (25 mg) was injected into peritoneum of tumor-bearing mice each day for 9 days, and tumor growth was measured. When the EL4 tumors on the backs of B6 mice were 5 × 5 mm in diameter, the mice were treated with Imatinib. The tumor growth in Imatinib-treated mice was slower than in control mice. On day 11 of treatment, the tumor volume was 927.7 ± 558.5 mm 3 in the mice treated with Imatinib, compared to 2050.4 ± 282.7 mm 3 in control mice (p = 0.023, Fig. 7b).

Survival benefit of splenectomy for tumor-bearing mice.
Most of the Gr-1 + CD11b + MDSCs accumulated in the spleens in the tumor-bearing mice. Removal of MDSCs might be beneficial for host survival. To assess the immunosuppressive effect of MDSCs in vivo, the spleens of tumor-bearing mice were excised to examine whether the survival of the mice could be prolonged. Twenty-four mice bearing tumors 15 mm in diameter were divided into two groups. Splenectomy was performed in 12 of the mice. The survival of the mice who received splenectomy was prolonged compared to the mice without splenectomy (16.4 + 8.36 vs. 9.25 + 6.02 days, p = 0.020) (Fig. 8).

Discussion
MDSCs are a type of immunosuppressive cell in tumor-bearing mice. In this study, the frequency of MDSCs in the spleen was correlated with tumor size. In Hepa1-6 mouse hepatoma model, the frequency of MDSCs in the spleen could be calculated. Diaz-Montero et al. reported elevated numbers of MDSCs in the circulating peripheral blood of cancer patients compared to health volunteers 29 . In human HCC, MDSCs could be detected in peripheral blood and displayed immunosuppressive activity by inducing regulatory T-cells 28 . Gabitass et al. also mentioned that the frequency of MDSCs was increased in pancreatic, gastric, and esophageal cancers 33 . The present findings further show that the frequency of MDSCs can be correlated with the tumor burden in the hosts.
MDSCs are induced and maintained by tumors. Presently, when tumors that had developed in the flank of mice injected with Hepa1-6 hepatoma cells were completely resected, the frequency of MDSCs returned to normal within 3 weeks. Salrador et al. demonstrated that the cell population in the spleen returned to normal following excision of fibrosarcoma in the mice 34 . These results imply the existence of a paracrine effect, where particular cytokines must be released by the tumor cells to maintain the population of MDSCs. In vitro, MDSCs appeared when naïve spleen cells were cultured in Hepa1-6 conditioned medium for 7 days. Particular cytokines capable of inducing MDSCs must be produced and released into the cell culture medium from tumors. However, www.nature.com/scientificreports/ the identity of these particular cytokines has been unclear. In this study, several candidates were identified using a cytokine assay. SCF was one of the cytokines capable of stimulating MDSC expansion. The cytokine assay revealed that several common cytokines were overproduced in the supernatant of cultured tumor cells as well as in the serum of tumor-bearing mice. Among these common overproduced cytokines, SCF was the most eligible cytokine to induce MDSCs. SCF appeared in the supernatant of Hepa1-6 cell cultures but not in medium of naïve cell cultures. SCF was overproduced in the serum of tumor-bearing mice, compared with the serum from naïve mice. SCF is a hematopoietic cytokine that promotes the survival, proliferation, and differentiation of hematopoietic cells. Western blot analysis also showed that Hepa1-6 could produce membrane-bound and soluble SCF. Direct measurement of SCF in the serum of Hepa1-6 tumor-bearing mice correlated the level of serum to tumor size. We further demonstrated that SCF could induce MDSCs. Supplementation of the RPMI-1640 medium with ≥ 12.5 ng/ml SCF resulted in the induction of a significant population of Gr-1 + CD11b + MDSCs after 9 days of culture. Eligible cytokines that could induce MDSCs have been reviewed 31 . However, no direct evidence that a single cytokine can induce MDSCs has been previously published. Pan et al. mentioned that block SCF enhanced immune-enhancing cancer therapy 35 . The present findings provide direct evidence that SCF can stimulate expansion of MDSCs.
C-kit is a cell surface receptor of SCF. Imatinib blocks c-kit. In the experiment in which SCF or conditioned culture medium were present during spleen cell culture, MDSC expansion was evident. When Imatinib was added to block c-kit, the SCF-c-kit downstream pathway could not be activated and the MDSC population decreased. When SCF was directly blocked by shRNA, the frequency of MDSCs cultured using conditional medium was decreased. These results provided conformation that SCF was necessary to maintain MDSCs. Several cancer therapy strategies have been employed to block or diminish MDSCs 36 . In one study, the Toll-like receptor ligand CpG was used to block the suppressive effect of MDSC on T-cells. The tumor size in tumor-bearing mice could be reduced significantly 37 . Gemcitabine in cancer therapy can reduce the number of MDSCs, increase the antitumor activity of CD8 + T-cells, and activate NK cells 38 . In this study, the tumor growth rate was decreased when EL4 tumor-bearing mice were treated by Imatinib. The collective results indicate that tumor sizes can be reduced when the MDSC population is decreased. However, the tumors were not treated successfully. The findings imply that directly targeting SCF might be helpful for cancer treatment.
MDSCs in tumor-bearing hosts are not only immunosuppressive cells. They also contributes to survival. In our animal model, when the spleen was removed, the survival of the mice was significantly prolonged. The spleen is the largest immune organ and plays a role in defense to infection or foreign antigens. However, many MDSCs accumulate in the spleen when a host harbors a large tumor. In this setting, the spleen is no more an immunedefense organ and contrarily contributes to immunosuppression. Removal of the spleen to diminish MDSCs and prolong survival clearly demonstrates that removal of the spleen removes the immunosuppressive effect of MDSC in tumor-bearing hosts. Hence, diminishing MDSCs could be a valuable strategy in the treatment of cancer. Western blot. Proteins from Hepa1-6, EL4, Hep G2, 3B, SK, and Huh7 cells were separated on 12% polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Amersham Biosciences, Uppsala, Sweden). These blots were incubated for 1 h at room temperature in Tris-buffered saline containing Tween (TBST; 20 mM Tris-Cl, 140 mM NaCl, pH 7.5, 0.05% Tween 20) also containing 5% skim milk. Primary antibody used was anti-SCF (diluted 1:1,000, Abcam, Cambridge, MA, USA). Blots were incubated with primary antibodies overnight at 4 °C. After washing three times in TBST, blots were incubated with horseradish peroxidase-conjugated secondary antibody (diluted 1:2000, Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h at room temperature. Immunoreactive complexes were visualized using enhanced chemiluminescence reagents (Millipore, Billerica, MA, USA).
tumor inoculation. Hepa1-6 cells (5 × 10 5 ) were subcutaneously inoculated in a flank of B10 mice to induce hepatoma. EL cells (1 × 10 5 ) were subcutaneously inoculated in the flank of B6 mice to induce lymphoma. The tumors were allowed to grow until they were 10-15 mm in diameter. The tumor volume was calculated as 0.52 × width 2 × length 39 . The mice were sacrificed when the diameter of tumor reached 20 mm.
Arginase activity. Arginase activity was measured in cell lysates using an arginase assay kit (Abnova Corp., Taipei, Taiwan). Cells (2 × 10 6 ) were lysed for 10 min in 100 μl of 0.1% Triton X-100. The cell lysate was centrifuged at 14,000g at 4 °C for 10 min. The supernatant was collected for the arginase assay performed according to the manufacturer's instructions.
Splenectomy. Under adequate anesthesia with isoflurane inhalation, an oblique incision was made over left flank of the mice to access the abdominal cavity. The spleen was located just beneath the wound and was pulled out. The splenic artery and vein were ligated and the spleen was removed. The wound was closed with 4-0 Dexon sutures. Tests if test number was < 5. The significance of the differences between different groups was determined by paired or unpaired Student's t test. Survival rates were calculated using the Kaplan-Meier method and compared between groups using the log-rank test (SigmaStat 3.0 software). P-values < 0.05 were considered significantly different.