Sox2 communicates with Tregs through CCL1 to promote the stemness property of breast cancer cells

Ethics Statement

All mouse experimentations were conducted in accordance with the standard operating procedures approved by the Institute Research Ethics Committee at the Nankai University.

Gene cloning

shRNA sequences for silencing mice Sox2 gene and Ccl1 gene were searched and blasted using RNAi designer from the invitrogen website (https://rnaidesigner.invitrogen.com/rnaiexpress/index.isp). shRNA targeting mice Ccl1 was designed and synthesized as shRNA-Ccll (GCCGTGTGGATACAGGATGTTTTGGATCCAAAACATCCTGTATCCACACGGC), shRNA targeting Sox2 and the scrambled control sequence were described before [24].

Cell culture

Wild type (Wt) of 4T1, 4TO7, and EO771 cells were kindly provided by Dr. Ralph A. Reisfeld (The Scripps Research Institute, CA, USA). 4T1 and 4TO7 cells were maintained in RIPM-1640 media supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 0.1 mg/mL streptomycin. EO771 cells were maintained in DMEM media supplemented with 10% FBS, 100 U/mL penicillin and 0.1 mg/mL streptomycin, 1% NEAA, and 1% sodium pyruvate.

EO771-Wt and 4T1-Wt cells were infected with lentivirus carrying pLV-H1-EF1α-shSox2-puro or pLV-H1-EF1α-sh-scramble-puro plasmid, followed by clonal selection using puromycin to generate stably polyclonal cells expressing shSox2 (EO771-shSox2 and 4T1-shSox2) or the control (EO771-sc, 4T1-sc); 4TO7-Wt cells were infected with lentivirus carrying pLV-EF1α-Sox2-IRES-Bsd plasmid or pLV-EF1α-IRES-Bsd plasmid, followed by clonal selection using blasticidin to generate stably polyclonal cells overexpressing Sox2 (4TO7-Sox2) or the control (4TO7-ctrl).

EO771-sc, EO771-shSox2, 4T1-sc, and 4T1-shSox2 cells were infected with lentivirus carrying pLV-EF1α-Ccl1-IRES-Bsd or pLV-EF1α-IRES-Bsd plasmid, followed by clonal selection using blasticidin to generate polyclone cell lines with stable overexpression of Ccl1 (EO771-sc-Ccl1, EO771-shSox2-Ccl1, 4T1-sc-Ccl1, and 4T1-shSox2-Ccl1) or the controls (EO771-sc-ctrl, EO771-shSox2-ctrl, 4T1-sc-ctrl, and 4T1-shSox2-ctrl); 4TO7-ctrl and 4TO7-Sox2 cells were infected with lentivirus carrying pLV-H1-EF1α-shCcl1-puro or pLV-H1-EF1α-sh-scramble-puro plasmid, followed by clonal selection using puromycin to generate polyclone cell lines with stable expression of shCcl1 (4TO7-ctrl-shCcl1, 4TO7-Sox2-shCcl1) or the controls (4TO7-ctrl-sc, 4TO7-Sox2-sc).

Real-time PCR

Reverse transcription polymerase chain reaction (RT-PCR) and Real-time PCR was performed following previous protocol [24]. For an equal loading control, mRNA of Gapdh (mus) was tested. Primers’ sequences are: Ccl1: F: 5′-ACTGATGTGCCTGCTGCTGG -3′, R: 5′- TGGGGGATCAGGACAGGAGG-3′; Foxp3: F: 5′- AGGCAGAGGACACTCAATGAAATC-3′, R: 5′ -CGAAACTCAAATTCATCTACGGTC-3′.

Western Blotting

Cell lysates were prepared by RIPA buffer in the presence of protease inhibitor cocktails (PIC), PIC2 and PIC3 (Sigma-Aldrich). Proteins (25 μg) were loaded into 5–12% Tris-Acrylamide gels and blotted with antibodies that included: anti-Sox2, β-actin, (Santa Cruz Biotechnology), anti-Nanog, Oct4, Ki67 (abcam), anti-CCL1 (R&D Systems), and horseradish peroxidase-conjugated secondary antibodies. Blotting results were detected by an ECL chemiluminescence kit (Millipore) and analyzed with the Image J software.

Cell sorting

C57BL/6-Foxp3-EGFP transgenic mice were kindly provided by Dr. Zhinan Yin (Jinan University, Guangzhou, People’s Republic of China). The mice were grafted with 5×10⁵ EO771 cells. After 30 days, mice were sacrificed and the spleens were collected and grinded. The cells from spleens were then collected, red blood cells were removed by using Red Blood Cell Lysis Solution (SolarBio, People’s Republic of China) following manufacturer’s instructions, and then subjected to fluorescence-activated cell sorting (FACS). Enhanced green fluorescent protein (EGFP) was excited by FL1 channel at 488 nm, EGFP⁺ cells were harvested and then primarily cultured at 1.5×10⁶ cells per mL in RIPM-1640 media supplemented with 10% FBS, 100 U/mL penicillin and 0.1 mg/mL streptomycin, 1% NEAA, and 1% sodium pyruvate for 48 hours, and then the supernatant was harvest as Tregs conditioned medium (Tregs CM).

Side population, ALDH activity, and sphere formation assay

4T1-Wt, 4TO7-Wt, and EO771-Wt cells were treated with the Tregs CM for 2 days (breast-cancer-cell culture medium:Tregs CM = 3:1 volume ratio), then were harvested and resuspend by using the mixed culture medium. For control purpose, the breast-cancer-cell culture medium mixed with the culture medium prepared for Tregs at 3:1 volume ratio was used as the control medium. The following assay were performed:

For the side population (SP) assay, the breast cancer cells were harvested and resuspended in prewarmed DMEM+ buffer (DMEM with 2% FBS) at a density of 1.0×10⁶ cells per mL and subjected to FACS analysis as described before [24]. 10 μM Verapamil was used as the blocker.

For the ALDH activity assay, the ALDH activity was detected by using ALDFLUOR™ kit (STEMCELL Technologies) and then subjected to FACS analysis. 7.5 μM Diethylaminobenzaldehyde (DEAB) was used as the blocker.

For the sphere formation assay, these breast cancer cells were then harvested and resuspended by the mixture of sphere formation medium and Tregs CM (3:1 volume ratio). Sphere formation medium contains RIPM-1640 (for 4T1 or 4TO7) or DMEM (for EO771) media supplemented with 20 ng/mL EGF, 20 ng/mL bFGF, and 1×B27 supplement (Life Technologies). The cells were palted at 1000 and 100 cells per well in the Clear Flat Bottom Ultra Low Attachment 96-well Microplate (Corning). After 5 days (for 4T1 and EO771 cells) or 7 days (for 4TO7 cell), the spheres (determined as >20 cell/spheroid) were recorded by Olympus BX51 Epi-fluorescent microscopy under a ×10 objective (Olympus) and were counted by using ImageJ software.

Cytotoxicity and apoptosis assay

4T1-Wt, 4TO7-Wt, and EO771-Wt cells were treated with the Tregs CM (breast-cancer-cell culture medium:Tregs CM = 3:1 volume ratio) for 2 days and then were harvested and resuspend by using the mixed culture medium. 5×10³ cells were plated in each well of the 96- well plate for overnight. 4T1-Wt, 4TO7-Wt, and EO771-Wt cells cultured in the control medium were used as the control.

For the cytotoxicity assay, doxorubicin (DOX, Sigma-Aldrich), cyclophosphamide (CTX), and methotrexate (MTX) were then added at various concentrations to the culture medium for 2 days. Cell activity were tested by applying CCK8 kit (Dojindo) following the manufacturer’s instructions.

For the apoptotic assay, 5×10⁵ cells were harvested and replated in each 60-mm dish for overnight by using the mixed culture medium. DOX, CTX, and MTX were then added to the mixed culture medium for 24 hours. The concentations of MTX used were 5 × 10⁻⁸ μg/mL for 4T1, 7.9 μg/mL for EO771, 6×10⁻⁶ μg/mL for 4TO7; the concentarions of CTX used were: 0.03 μg/mL for 4T1, 0.92 μg/mL for EO771, 0.5 μg/mL for 4TO7; the concentrations of DOX used were 0.0007 μg/mL for 4T1, 0.0001 μg/mL for EO771, 0.001 μg/mL for 4TO7. Proteins excepting those from the mitochondria were isolated and subjected to Western blotting to detect the expression of Cytochrome C released to the cytoplasma as described before [25].

The cells were also stained by Alexa Fluor 647 Annexin V apoptosis detection kit with PI (Biolegend) following the manufacturer’s instructions and then subjected to FACS analysis.

Limiting-dilution transplantation and bioluminescent imaging

Female C57BL/6 mice at 6–8 weeks age were separated randomly into different groups. EO771-Wt cells were treated with Tregs CM for 48 hours, then harvested. 2×10⁵ (200K), 5×10⁴ (50K), and 1×10⁴ (10K) cells (EO771-Wt mixed with primarily cultured Tregs at the cell number ratio of 20:1) were inoculated separately into the fourth mammary gland of each mouse. For the bioluminescent image, 2×10⁵ cells (EO771-luc mixed with Tregs at the cell number ratio of 20:1) were injected into each mouse through tail vein.

For control purpose, same amount of control-medium-pretreated EO771-Wt were inoculated into the fourth mammary gland or same amount of control-medium-pretreated EO771-luc cells were inoculated through tail-vein injection for each mouse. Tumor volume and bioluminescent image were calculated or obtained as described before [24, 26].

Isolation of nuclear proteins

The nuclear and cytoplasma proteins were seperated as described before [28]. The protein was subjected to Western blotting and the blots were detected by using anti-H3K27Me3 (Cell Signaling Technology) and p65 (Santa Cruz Biotechnology), Lamin A or Histone H3 was used as an equal nuclear protein loading control and was detected by using an anti-Lamin A antibody (Sigma-Aldrich) or anti-Histone H3 (abcam) antibody.

Immunohistochemistry

Immunostaining was performed on human breast tissue arrays (Alenabio Co.) and the tumor xenografts of BALB/c and C57BL/6 mice. These tissues were detected separately with antibody against Foxp3, Sox2, and Oct4 (abcam) at a 1:100 dilution. Expression levels of FOXP3, SOX2, or OCT4 in the tissue microarray were scored according to the percentage of positive-staining cells in the wole area of each breast tissues. The tissues with 0%, 0–3%, and > 3% FOXP3⁺ cell were defined as FOXP3 expression at the level of 0, 1, and 2 separately. The tissues with< 30%, 31%–49%, and ≥ 50% SOX2⁺ or OCT4⁺ cell were judged as tissues with weak, moderate and strong SOX2- or OCT4-expression separately. The images were recorded under a ×40 objective (Olympus).

Immunofluorescence staining and confocal imaging

The tissues were incubated with antibody against Foxp3, CCL1, CCR8 (abcam) at 1:100 dilutions at 4°C for overnight, followed by incubation with Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 594 goat anti-rabbit IgG and Alexa Fluor 546 goat anti-rat IgG (Invitrogen Inc.) at room temperature for 1 hour, then counterstained with DAPI for 2 minutes and subjected to image with a confocal microscope (Olympus).

Cytokine array

2×10⁵ 4T1 cells were cultured in each well of the six-well plate, their supernatant were harvested after 48 hours. Cytokine array was performed by using the Mouse Cytokine Array Kit (R&D Systems).

Transwell assay

Transwell chambers (polycarbonate filters with 3-μm porosity, Millipore) were used. mCCL1 cytokine were obtained from R&D Systems. 1×10⁵ Foxp3-EGFP⁺ cells were suspended in serum-free medium and plated in the upper chamber. Cells penetrated and attached to the bottom of the filter were quantified as described before [29].

Dual luciferase assay

EO771-SC and EO771-sc cells were cultured in 24-well plates at 2×10⁵/well for overnight and transfected with 30 ng of pRL-TK plasmid and 300 ng of pGL4.32[luc2P/NFκB-RE/Hygro] plasmid following the protocol described before [30]. The activation of the NFκB-RE (response element) were quantified as a ratio of firefly luciferase activity versus renilla luciferase activity (FL/RL) in each well.

Quantitative chromatin immunoprecipitation assay

1×10⁷ of cells were harvested. The chromatin-associated proteins were crosslinked to chromatin by 1% formaldehyde and then sonicated to ~500–1,000 bp DNA fragment following the protocol described before [29], and immunoprecipitated with anti-H3 antibody (abcam) as the internal control, or with the anti-H3K27me3 antibody (Cell Signaling Technology). The DNA fragments were purified and subjected to quantitative-PCR with different primers along the mouse (m) p65 and Ccl1 promoters (Supporting Information Table S1).

Statistical analysis

Data are expressed as mean + SEM and the statistical significance is determined by t-test unless otherwise specified. A value of p < .05 is used as the criterion for statistical significance.

Tregs-conditioned medium increases CSCs-like population of breast cancer cells

In order to understand the potential role of Tregs in regulating stemness phenotype of breast cancer cells, we freshly isolated spleen Tregs (Tregs) from breast-tumor-bearing C57BL/6-Foxp3-EGFP mice (Fig. 1A). Mouse breast cancer cell lines-EO771, 4T1 and 4TO7 were treated with conditioned medium (CM) obtained from freshly isolated Tregs, or control medium for 48 hours, then subjected to FACS analysis for SP and ALDH^br population changes, which represent the CSCs-like cell population. Compared with the control, Tregs-CM treatment enriched the SP of three breast cancer cells (Fig. 1B, 1D) and also increased ALDH^br cell population in EO77, 4T1 and 4TO7 (Fig. 1C, 1E). In addition, Tregs-CM treatment increased sphere-forming capability of EO771, 4T1, and 4TO7 cells at different dilutions in both 100 cells per well and 1000 cells per well test groups (Fig. 1F) and also increased expression of Sox2, Nanog, and/or Oct4 proteins in these three cell lines (Fig. 1G). Taken together, these data demonstrate the promotion effect of Tregs on CSCs-like population and the stemness property of breast cancer cells in vitro.

Tregs-conditioned medium promotes tumorigenesis of EO771 cells in vivo

To verify our findings in vivo, the limiting-dilution transplantation assay was performed, following the experimental procedure described in Figure 2A. Earlier tumor formation (Fig. 2B), more tumor incidence at lower number of tumor cells inoculation (50K and 10K, Fig. 2C) and greater tumor volumes (Fig. 2C, 2D) were observed in Tregs-CM-pretreated EO771 combined with Tregs co-inoculation groups when compared with those of EO771-inoculation control in each dilution test, indicating that Tregs promote the stemness and tumorigenesis property of breast cancer cells in vivo.

Tregs-conditioned medium promotes metastasis of EO771 cells in vivo

To further investigate the effect of Tregs on metastatic potential of EO771 cells, we established a stable EO771 cell line with firefly luciferase expression (EO771-luc), which showed a strong correlation (r² =.99) between firefly luciferase activity and cell numbers ex vivo (Fig. 3A, 3B). We then performed the tail vein injection experiment (Fig. 3C) and found that Tregs-CM-pretreated EO771-luc combined with Tregs coinjection significantly increased lung metastatic burden, as demonstrated by the increased luciferase activity (Fig. 3D, 3E), as well as more metastatic foci number and metastatic areas in the lungs (Fig. 3F, 3G).

Tregs-conditioned medium promotes chemoresistance of breast cancer cells

Previous studies suggest that intrinsic properties of CSCs promote their resistance to chemotherapy [31, 32]. We therefore hypothesized that breast cancer cells pretreated with Tregs CM would demonstrate resistance to chemotherapeutic agents. To test this notion, Tregs-CM pretrteated EO771, 4T1, and 4TO7 cells were exposed to different chemicals in the presence of Tregs CM for 48 hours. Compared with control cells, EO771 cultured with Tregs CM demonstrated a higher IC50 after DOX treatment (Fig. 4A), similar results were also found in 4T1 cells when treated with MTX and DOX (Fig. 4B) and in 4TO7 cells treated with CTX (Fig. 4C). Among these three cell lines, 4T1 and 4TO7 are triple negative cells derived from BALB/c mice and EO771 is a basal-like cell line derived from the C57BL/6 mouse [33–35]. 4T1 shows higher metastatic property than the 4TO7 [33]. These distinct gene and molecular backgrounds may account for their different sensitivity to chemotherapy after culturing with the Tregs CM. To further validate the chemoprotective effect of Tregs on tumor cells, Tregs-CM pretrteated breast cancer cells were incubated with DOX, MTX and CTX in the presence of Tregs CM for 24 hours and then subjected to Western blotting. Lower amount of cytosol cytochrome C were found in EO771 cells treated with DOX, 4T1 treated with MTX, DOX, and 4TO7 treated with CTX when compared with their controls separately (Fig. 4D). Consistently, decreased apoptotic cell populations were also detected in these cells by using FACS analysis (Supporting Information Fig. S1A, S1B),. These results demonstrate that Tregs promote the chemoresistant property of different breast cancer cells.

Tregs are recruited by Sox2-positive breast cancer cells

Since Tregs in TME play such important roles in enhancing the stemness population of breast cancer cells, we then asked whether CSCs affect the recruitment of Treg to TME. To test this hypothesis, immunohistochemistry (IHC) staining was performed in tissue microarrays. We found higher expressions of FOXP3 in human breast cancer tissues (Fig. 5A). Most importantly, statistical analysis showed the expression of FOXP3 strongly correlated with that of SOX2 (Fig. 5B, left panel), supporting the stemness promotion effect of Tregs on cancer cells and also suggesting a possible recruitment effect of SOX2 on Tregs. To further confirm the recruitment effect of CSCs on Tregs, the expression of another important stemness marker-OCT4 was detected. A correlated expression between FOXP3 and OCT4 was also found (Fig. 5B, right panel).

Previous study demonstrated the regulatory effect of Sox2 on the cytokine-related signal pathway [28], to further verify the hypothesis that Sox2 may recruit Tregs to TME, transwell assay was performed in vitro. We first tested the expression of Sox2 in three mouse breast cancer cell lines and found relative high expressions of Sox2 in EO771 and 4T1, and relative low expression of this gene in 4TO7 (Fig. 5C). So we overexpressed Sox2 in 4TO7 and silenced its expression in EO771 and 4T1 (Fig. 5D). We found CM of 4TO7-Sox2 recruited more Tregs, but that of EO771-shSox2 and 4T1-shSox2 cells recruited less Tregs when compared with their controls separately (Fig. 5E), thus demonstrating the recruitment effect of Sox2-positive breast cancer cells on Tregs.

Sox2 recruits Tregs via the secretion of CCL1 in vitro

To investigate the cytokines that mediate recruitment of Sox2 on Tregs, cell culture supernatants from 4T1-sc and 4T1-shSox2 cells were collected and subjected to mouse chemokine array. We found that CCL5, CCL1, G-CSF, and CM-CSF were the four downsecreted cytokines in 4T1-shSox2 cells (Fig. 6A, 6B). We further examined expression of Ccl1 gene in 4T1, EO771, and 4TO7 and found its expression increased in 4TO7-Sox2, but decreased in 4T1-shSox2 and EO771-Sox2 cells at both mRNA and protein levels when compared with the controls, respectively (Supporting Information Fig. S2A; Fig. 6C). To examine whether CCL1 plays an important role in mediating the recruitment effect of breast cancer cells on Tregs, transwell assay were performed. We found that application of anti-CCL1 antibody to lower chamber inhibited the recruitment effect of CM of EO771, 4T1, and 4TO7 cells on Tregs (Fig. 6D). In contrast, adding recombinant mouse CCL1 cytokine to the lower chamber increased the recruitment of Tregs (Fig. 6E).

To confirm the results in vivo, these three mouse breast cancer cells were injected into the fourth mammary gland of 6 to 8-week old female mice (5×10⁵ EO771 cells were inoculated to each C57BL/6 mouse, 5×10⁴ 4T1 cells, or 1×10⁵ 4TO7 cells were inoculated to each BALB/c mouse, respectively). The mice were sacrificed at designated time after the inoculation (30 days for EO771, 28 days for 4T1, and 24 days for 4TO7). The infiltration of Foxp3⁺ cells and expression of CCL1 in tumor tissues correlated with the expression of Sox2 (Fig. 6F). Moreover, the mRNA expression level of Foxp3 and Ccl1 decreased in EO771-shSox2 and 4T1-shSox2 xenografts, but increased in 4TO7-Sox2 xenografts (Supporting Information Fig. S2B).

Overexpression of CCL1 rescues the inhibited recruitment of Tregs induced by Sox2 silencing and promotes tumor growth in vivo

To further elucidate CCL1’s function in mediating the recruitment effect of CSCs-like population on Tregs, we established stable cell lines with Ccl1 overexpression in EO771, 4T1 cells with Sox2 silencing and their controls, and downregulated its expression in 4TO7-Sox2 and 4TO7-ctrl cells (Fig. 7A). Compared with the controls, no difference for the expressions of Ki67 were found in these breast cancer cells after Ccl1 overexpression or silencing, indicating CCL1 has no effect on cell proliferation in vitro (Fig. 7A). This conclusion was further confirmed by CCK8 measurement of cultured breast cancer cells (Fig. 7B). However, delayed tumor growths were observed in 4TO7-Sox2-shCcl1 tumor xenograft when compared with 4TO7-Sox2-sc (Fig. 7C). In addition, silencing of Ccl1 compromised the proliferation-promotion effect of Sox2 on tumor growth (Fig. 7C). In contrast, overexpression of Ccl1 in 4T1-sc and 4T1-shSox2 cells promoted the growth of tumor xenografts when compared with the 4T1-sc-ctrl and 4T1-shSox2-ctrl separately. In addition, overexpression of Ccl1 rescued the inhibited tumor growth rates induced by Sox2 silencing (Fig. 7C), suggesting the promotion effect of CCL1 on tumor growth and also supporting the important function of CCL1 in mediaiting the tumor-promotion effect of Sox2 in vivo. Most importantly, IHC or immunofluorescent staining of the primary tumor tisues showed increased expression of Ki67 and Foxp3 in 4T1 xenografts with Ccl1 overexpression when compared with the controls separately (Supporting Information Fig. S3A). Decreased expression levels of Ki67 and Foxp3 were observed in 4TO7 xenografts with Ccl1 downregulation when compared with the controls separately (Supporting Information Fig. S3B). In addition, overexpression of Ccl1 rescued the inhibited Ki67 expression and inhibited recruitment of Foxp3⁺ cells in TME induced by Sox2 silencing in 4T1 tumor xenografts (Supporting Information Fig. S3A). Correspondingly, downregulation of Ccl1 compromised the increased expression of Ki67 and the recruitment of Foxp3⁺ cells in TME induced by Sox2 overexpression in 4TO7 tumor xenografts (Supporting Information Fig. S3B). These observations support the concept that CCL1 mediates the tumor-promotion effect of Sox2 through remodeling TME, in which more Foxp3⁺ Tregs cells are recruited to tumor tissues. Kaplan-Meier curves for overall survival and disease free survival in the breast tumor patients from the GSE58812 and GSE61304 cohorts showed that FOXP3 and CCL1 correlated with the poor prognosis and disease free survival (Supporting Information Fig. S3C). To further investigate the regulatory mechanism of Sox2 on the transcription of Ccl1, Western blotting was applied to detect the expression level of p65. Downregulation of nuclear p65 was found in 4T1-shSox2 and EO771-shSox2 cells. Correspondingly, upregulation of this protein was detected in 4TO7 cells with Sox2 overexpression (Fig. 7D). These results indicate that Sox2 may promote the expression of p65. By applying pyrrolidine dithiocarbamate (PDTC), a NF-κB inhibitor to the culture medium, inhibited expression of CCL1 and nuclear p65 was found in 4TO7 cells when compared with the controls. Application of PDTC also inhibited the upregulated Ccl1 induced by Sox2 overexpression in 4TO7-Sox2 (Fig. 7E). Dual luciferase further showed decreased NFκB signaling activity upon Sox2 silencing in EO771 cells (Fig. 7F), demonstrating the important function of NF-κB in mediating the transcriptional regulation effect of Sox2 on Ccl1.

CCR8, which is the receptor for CCL1 [36] was detected in FOXP3 positive cells from human breast cancer tissues (Supporting Information Fig. S3D). SOX2 was reported to regulate the expression of H3K27Me3 in lung carcinoma cells [37]. We demonstrated reduced expression of nuclear H3K27Me3 in 4TO7-Sox2 and upregulation of this protein in 4T1-shSOX2 and EO771-shSox2 cells (Fig. 7G). Chromatin immunoprecipitation (ChIP) assay revealed the reduced binding level of H3K27Me3 on the promoter region of both p65 and Ccl1 in 4TO7-Sox2 cells when compared with the 4TO7-ctrl contols (Fig. 7H), thus unveiling the regulatory mechanism of Sox2 on the transcription of p65 and Ccl1. Taken together, these findings demonstrate that Sox2 promotes the expression of Ccl1 through both the epigentic regulation method and the NF-κB signaling, and disclose that CCL1 may promote the recruitment of Treg cells to TME through CCL1-CCR8 axis.