The non-malignant human mammary epithelial cell line MCF-10A, human breast cancer cell line MDA-MB-231, mouse breast cancer cell line 4T1, human acute monocytic leukemia cell line Thp1 and mouse macrophage cell line Raw264.7 were obtained from the American Type Culture Collection. The identities of all these cell lines have been authenticated by short tandem repeat profiling. MCF-10A cells were cultured in DMEM/F12 medium supplemented with 5% horse serum, 1% penicillin and streptomycin (Gibco, Grand Island, NY, USA), 20 ng/ml recombinant human epidermal growth factor (EGF), 0.5 µg/ml hydrocortisone, 100 ng/ml cholera toxin, and 10 µg/ml insulin (Sigma-Aldrich, Shanghai, China). MDA-MB-231 and 4T1 cells were cultured in DMEM medium supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin (Gibco). Thp1 and Raw264.7 cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin (Gibco). All the cells were maintained at 37°C in a humidified incubator containing 5% CO₂. 100 ng/ml phorbol-12-myristate-13-acetate (PMA, Sigma-Aldrich) was used to induce attachment and differentiation of Thp1 monocytes into macrophages. The cell culture supernatant of breast cancer cell lines MDA-M-231 and 4T1 was collected as conditioned medium (CM), and was used to induce the transformation of Thp1 macrophages and RAW264.7 macrophages into M2 phenotype TAMs, respectively. For co-culture of breast cancer cells and TAMs, the 6-well or 24-well transwell co-culture system was used. In brief, transwell inserts were placed in 6-well or 24-well culture plates seeded with breast cancer cells. TAMs were placed in the upper transwell chamber. Transwell inserts were separated by a 0.4 µm permeable membrane allowed free exchange of media and soluble molecules.

XPS was extracted from a mixture of 10 herbs including Epimedium brevicornum Maxim. (Chinese name Yin Yang Huo), Cistanche deserticola Ma (Chinese name Rou Cong Rong), Leonurus japonicus Houtt. (Chinese name Yi Mu Cao), Salvia miltiorrhiza Bunge (Chinese name Dan Shen), Curcuma aromatica Salisb. (Chinese name Yu Jin), Curcuma longa L. (Chinese name E Zhu), Ligustrum lucidum W.T.Aiton (Chinese name Nv Zhen Zi), Reynoutria multiflora (Thunb.) Moldenke (Chinese name He Shou Wu), Crassostrea gigas (Chinese name Mu Li) and Carapax trionycis (Chinese name Bie Jia) by refluxing extraction method. Its quality control was carried out by detecting the high performance liquid chromatography fingerprints between different batches. The detailed preparation and quality control information have been previously reported (Liu, 2010; Wang and Huang, 2010; Wang et al., 2017).

The phenotype of macrophages was identified by analyzing the surface markers of macrophages using flow cytometer. Briefly, macrophages were harvested, washed, and resuspended in 100 µl PBS solution at a density of 5 × 10⁶ cells/ml. For Thp1 macrophages, cells were incubated with FITC-conjugated F4/80 antibody (SC-71085, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and PE-conjugated CD163 antibody (12-1639-42, eBioscience, San Diego, CA, USA) for 30 mins at 37°C. For Raw264.7 macrophages, cells were incubated with FITC-conjugated F4/80 antibody (SC-71085, Santa Cruz) and PE-conjugated CD206 antibody (141705, Biolegend, San Diego, CA, USA) for 30 mins at 37°C. For phenotype analysis of primary macrophages extracted from mouse 4T1-Luc xenografts, cells were incubated with CD45-PE-Cy7 (25-0451-82, eBioscience), FITC-conjugated F4/80 antibody (SC-71085, Santa Cruz) and PE-conjugated CD206 antibody (141705, Biolegend) for 30 mins at 37°C. After incubation, cells were washed once with PBS and subjected to analysis using a FC500 flow cytometry (Beckman Coulter, Fullerton, CA, USA) or a FACSAria III flow cytometer (BD Biosciences, San Diego, CA, USA). The F4/80⁺/CD163⁺ subpopulation, the F4/80⁺/CD206⁺ subpopulation or the CD45⁺/F4/80⁺/CD206⁺ subpopulation were quantified and defined as M2 phenotype macrophages.

MTT assay was applied to assess the cytotoxicity of XPS in mammary epithelial cells, breast cancer cells and macrophages. In brief, cells were seeded into 96-well plates at a density of 5×10³ cells/well. After cells attachment, cells were treated with serial concentration gradients of XPS for 24 h or 48 h. To investigate whether XPS still has an inhibitory effect on proliferation of breast cancer cells in the presence of TAMs, breast cancer cells and TAMs were co-cultured in the 24-well transwell co-culture system. TAMs were seeded in the upper transwell chamber at a density of 3 × 10⁴ cells per well while breast cancer cells were seeded in the lower chamber at a density of 5 × 10³ cells per well. After cells attachment, cells were treated with serial concentration gradients of XPS for 48 h. Cell viability was detected using MTT (MP Biomedicals, Shanghai, China) according to the manufacturer’s instructions. MTT assay was performed in independent triplicates.

To investigate the effect of XPS on colony formation of breast cancer cells, MDA-MB-231, and 4T1 cells were seeded in 6-well plate at a density of 2,000 cells per well. To investigate whether XPS still has an inhibitory effect on colony formation of breast cancer cells in the presence of TAMs, breast cancer cells and TAMs were co-cultured in the 24-well transwell co-culture system. TAMs were seeded in the upper transwell chamber at a density of 3s × 10⁴ cells per well while breast cancer cells were seeded in the lower chamber at a density of 200 cells per well. After attachment, cells were treated with serial concentration gradients of XPS for 48 h and then replaced with fresh medium. Then, cells were cultured for 2 weeks and the resultant colonies were fixed with 4% paraformaldehyde and then stained with coomassie blue solution.

Breast CSCs populations were analyzed by flow cytometry using the ALDEFLUOR Stem Cell Identification Kit (01700, STEMCELL Technologies, Vancouver, Canada) according to the manufacturer’s instructions. Briefly, breast cancer cells were seeded in a 6-well plate at a density of 4 × 10⁵ cells/well. For co-culture of breast cancer cells and TAMs, 4s × 10⁵ TAMs were seeded in the upper transwell chamber and the transwell inserts were placed in the 6-well culture plates seeded with 4×10⁵ breast cancer cells/well. Then, cells were treated with indicated concentrations of XPS, CXCL1 (Human CXCL1, C597, Novoprotein, Shanghai, China; Murine CXCL1, 250-11-20, Peprotech, Rocky Hill, NJ, USA) or XPS and CXCL1 combination for 48 h. After treatment, breast cancer cells were harvested and resuspended in 500 µl ALDEFLUOR™ assay buffer. Then, cells were incubated with ALDEFLUOR™ at 37°C for 30 min to mark ALDH⁺ cells. After incubation, cells were washed once with PBS and subjected to analysis using the FC500 flow cytometer (Beckman). The ALDH⁺ subpopulation was defined as breast CSCs. Diethylaminobenzaldehyde (DEAB), a specific inhibitor of ALDH activity, was used to control for background fluorescence in the ALDH staining assay. For breast CSCs population analysis in the mouse 4T1-Luc xenografts, primary breast cancer cells were firstly isolated from mouse mammary tumors by mechanical methods and then subjected to breast CSCs population analysis as indicated above.

Breast cancer cells were seeded in ultralow attachment 6-well plate at a density of 1 × 10⁵ cells/well. For co-culture of breast cancer cells and TAMs, 4 × 10⁵ TAMs were seeded in the upper transwell chamber and the transwell inserts were placed in the 6-well culture plates seeded with 1 × 10⁵ breast cancer cells/well. Then, the above cells were treated with XPS, CXCL1 or XPS and CXCL1 combination for 48 h as indicated. The complete culture medium for mammosphere formation assay was composed of DMEM/F12 medium supplemented with 1% penicillin-streptomycin (Gibco), 2% B27 supplement (Gibco), 20 ng/ml EGF, 5 µg/ml insulin and 0.4% bovine serum albumin (BSA, Sigma-Aldrich). The numbers of mammospheres were quantified microscopically.

Total RNA was extracted with Trizol and reverse transcribed to complementary cDNA using the PrimeScript™ RT reagent Kit (Takara, Shiga, Japan) in accordance with the manufacturer’s instructions. RT-PCR was performed using the SYBR Premix Ex Taq Kit (Takara) and the QuantStudio 6&7 Flex Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Primer sequences of human β-actin were 5′-CCAACCGCGAGAAGATGA-3′ (forward) and 5′-CCAGAGGCGTACAGGGATAG-3′ (reverse). Primer sequences of human β-catenin were 5′-GCTTTCAGTTGAGCTGACCA-3′ (forward) and 5′-CAAGTCCAAGATCAGCAGTCTC-3′ (reverse). Primer sequences of human OCT4 were 5′-CAATTTGCCAAGCTCCTGA-3′ (forward) and 5′-AGATGGTCGTTTGGCTGAAT-3′ (reverse). Primer sequences of human Nanog were 5′-ATGCCTCACACGGAGACTGT-3′ (forward) and 5′-CAGGGCTGTCCTGAATAAGC-3′ (reverse). Primer sequences of human CXCL1 were 5′-AGGGAATTCACCCCAAGAAC-3′ (forward) and 5′-ACTATGGGGGATGCAGGATT-3′ (reverse). Primer sequences of mouse β-actin were 5′-GGAGGGGGTTGAGGTGTT-3′ (forward) and 5′-GTGTGCACTTTTATTGGTCTCAA-3′ (reverse). Primer sequences of mouse β-catenin were 5′-ACAGCACCTTCAGCACTCT-3′ (forward) and 5′-AAGTTCTTGGCTATTACGACA-3′ (reverse). Primer sequences of mouse OCT4 were 5′-CTGTAGGGAGGGCTTCGGGCACTT-3′ (forward) and 5′-CTGAGGGCCAGGCAGGAGCACGAG-3′ (reverse). Primer sequences of mouse Nanog were 5′-AGGGTCTGCTACTGAGATGCTCTG-3′ (forward) and 5′-CAACCACTGGTTTTTCTGCCACCG-3′ (reverse). Primer sequences of mouse CXCL1 were 5′-GACTCCAGCCACACTCCAAC-3′ (forward) and 5′-TGACAGCGCAGCTCATTG-3′ (reverse).

Cells were treated as indicated and then lysed using RIPA (Beyotime Biotechnology, Shanghai, China). Protein concentration was quantified with the Bicinchoninic Acid Kit (Sigma-Aldrich) according to the manufacturer’s instructions. Equal amounts of protein (30 µg) were resolved on SDS-PAGE gel, transferred to a polyvinylidene fluoride microporous membrane (Millipore, Billerica, MA, USA) and probed with primary antibodies for β-actin (4970S, Cell Signaling Technology, CST, Boston, MA, USA), CXCL1(AF5403, Affinity Biosciences, Cincinnati, OH, USA), and β-catenin (51067-2-AP, Proteintech, Chicago, USA). The relative expression levels of proteins between groups were calculated by comparing optical densities of bands using the Gel-Pro analyzer four software (Media Cybernetics, Maryland, USA).

The effect of XPS on CXCL1 secretion from TAMs was detected by Elisa assay. Briefly, TAMs were cultured in a 6-well plate and treated with serial concentration gradients of XPS for 48 h. After XPS treatment, TAMs were harvested and seeded into 6-well plate again in a density of 5 × 10⁵ cells/well and cultured for 24 h. Then, the cell culture supernatant in each well was collected and CXCL1 concentration in the cell culture supernatant was detected using the Human CXCL1 Elisa Kit (SEA041Hu, USCN Business, Wuhan, China) or Mouse CXCL1 Elisa Kit (SEA041Mu, USCN Business) according to the manufacturer’s instructions.

The effect of XPS on CXCL1 promoter activity was investigated by double luciferase reporter gene assay. Briefly, the CXCL1 promoter plasmid (Genecopeia, Guangzhou, China) was transfected into Raw264.7-derived TAMs using Vigenefection (FH880806, Vigene Biosciences, Jinan, China) according to the manufacturer’s instructions. Then, the transfected cells were seeded into 96-well plate at a density of 2 × 10⁴ cells/well and treated with serial concentration gradients of XPS for 48 h. After XPS treatment, the cell culture supernatant in each well was collected. The CXCL1 promoter activity was detected by analyzing the Gaussia luciferase activity and secreted alkaline phosphatase activity in the cell culture supernatant using the Secrete-Pair™ Dual Luminescence Assay Kit (LF031, Genecopeia) according to the manufacturer’s instructions.

For immunofluorescence analysis, frozen tissue sections of breast tumor were fixed with 4% paraformaldehyde for 20 mins, washed three times with PBS, permeabilized with 0.25% Triton X-100 for 20 mins and then blocked in 5% BSA at room temperature for 1 h. For F4/80 and CXCL1 detection, tissue sections were incubated with F4/80 antibody (17-4801-80, eBioscience) and CXCL1 antibody (AF5403, Affinity) overnight at 4°C, followed by an incubation with the Alexa Fluor^® 555 conjugated-anti-rat IgG (4417S, CST) and the Alexa Fluor^® 488 conjugated-anti-rabbit IgG (4412S, CST) for 2 h. For ALDH1A1 detection, tissue sections were incubated with ALDH1A1 antibody (BF0220, Affinity) overnight at 4°C, followed by an incubation with the secondary antibody of Alexa Fluor^® 555 conjugated-anti-mouse IgG (A21422, ThermoFisher) for 2 h. 4’, 6-diamidino-2-phenylindole (DAPI, Sigma) was used to visualize the nuclei. Fluorescence images were obtained using the LSM710 confocal microscope (Zeiss, Jena, Germany).

All in vivo experiments were performed according to our institution’s guidelines for the use of laboratory animals and were approved by the Institutional Animal Care and Use Committee of Guangdong Provincial Hospital of Chinese Medicine (No.2018044). Six-week-old female Balb/c mice were raised at the Experimental Animal Center of Guangdong Provincial Hospital of Chinese medicine under specific pathogen-free conditions with ambient temperature of 20-25°C and 45%–50% relative humidity and given sterilized food and water. The rearing facility was maintained on a 12 h light-dark cycle. For 4T1-Luc xenografts establishment, 2×10⁶ 4T1-Luc cells were suspended in 200 µl PBS and inoculated subcutaneously into the mammary fat pads of mice. When tumors reached a mean diameter of 0.5 cm, mice were randomized into two groups (six in each group) and received either saline or XPS (1 g/kg/day) by intragastric perfusion. Throughout the treatment, mice were weighed and their tumors were measured with a caliper every 3 days. Tumor volumes (V) were calculated using the formula: V = (length) × (width)²/2. D-Luciferin (150 mg/kg weight, 122799, PerkinElmer, Boston, USA) was injected intraperitoneally and mice were imaged using the IVIS Lumina XR in vivo imaging system (PerkinElmer) every week to monitor tumor growth and metastasis. When tumors grew to a proper size, mice were euthanized and tumors were excised. Primary cells were isolated from fresh mammary tumors by mechanical methods and subjected for macrophages phenotype analysis or breast CSCs subpopulation analysis as indicated above. The remaining tumor tissue was stored at -80°C and used for tissue immunofluorescence assay as indicated above.

Data was presented as mean ± standard deviation (SD). All statistical analyses were performed using SPSS 19.0 software (Abbott Laboratories, Chicago, USA). Student’s t-test and one-way ANOVA were performed for comparisons among groups. P < 0.05 was considered to be statistically significant.

Firstly, the cytotoxicity of XPS in both nonmalignant mammary epithelial cells and breast cancer cells was investigated by MTT method. As shown in Figure 1A, XPS treatment for 24 h or 48 h could only moderately inhibit the proliferation of human breast cancer cell line MDA-MB-231 and mouse breast cancer cell line 4T1, while having no obvious cytotoxicity in MCF-10A cells. The IC₅₀ values of XPS in these three kinds of cells reached over 500 µg/ml. However, XPS could dramatically inhibit the colony formation abilities of breast cancer cell lines MDA-MB-231 and 4T1 during the 2-week culture period (Figure 1B), indicating a long-term inhibitory effect of XPS on breast cancer cell proliferation. In the tumor microenvironment, cancer cells coexist with multiple kinds of cells, of which the largest proportion is TAMs. TAMs often display an immune-suppressive M2-like phenotype that fosters tumor growth and promotes resistance to therapy. In the present study, the conditional medium (CM) of MDA-MB-231 cells significantly increased the F4/80⁺/CD163⁺ subpopulation in Thp1 macrophages while 4T1-CM significantly elevated the F4/80⁺/CD206⁺ population in mouse Raw264.7 macrophages. These results indicated that both human Thp1 macrophages and mouse Raw264.7 macrophages were successfully stimulated into M2 phenotype macrophages (Figure 1C). Breast cancer cells were co-cultured with these M2-like phenotype TAMs using the transwell co-culture system to investigate whether XPS could inhibit the growth of breast cancer cells in the presence of TAMs. As shown in Figure 1D, XPS treatment for 48 h could dramatically inhibit the proliferation of breast cancer cell lines MDA-MB-231 and 4T1 in the co-culture system.The IC₅₀ values of XPS were 402.05 µg/ml for MDA-MB-231 cells and 92.64 µg/ml for 4T1 cells respectively when they were co-cultured with TAMs. Meanwhile, XPS could also strongly suppress the colony formation abilities of breast cancer cell lines MDA-MB-231 and 4T1 in the co-culture system (Figure 1E). Altogether, XPS could significantly inhibit the proliferation and colony formation abilities of breast cancer cells particularly when they were co-cultured with TAMs.

Breast CSCs are considered to be the initiator cell which is responsible for breast cancer cell proliferation and growth. Therefore, we further investigated the influence of XPS on breast CSCs. Hyperactive aldehyde dehydrogenase (ALDH) activity is closely related to the physiological properties of CSCs (Wang and Lei, 2017) and the ALDH⁺ subpopulation exhibits enhanced stemness when compared with the ALDH^- subpopulation (Shao et al., 2016). As shown in Figure 2A, XPS could significantly decrease the proportions of ALDH⁺ subpopulations in both MDA-MB-231 cells and 4T1 cells when they were co-cultured with TAMs. Breast CSCs are enriched in non-adherent spherical clusters of cells, which are termed mammospheres (Wang et al., 2014). Therefore, we further investigated if XPS could suppress the mammosphere formation efficacy of breast CSCs in the breast cancer cells and TAMs co-culture system. As shown in Figure 2B, XPS treatment significantly decreased the mammosphere number in the breast cancer cells and TAMs co-culture system. It has been reported that multiple stem cell transcription factors are overexpressed in CSCs, such as octamer-binding transcription factor 4 (OCT4), Nanog, and β-catenin, which maintain pluripotency (Hattermann et al., 2016). To further verify the inhibitory effect of XPS on breast CSCs in the co-culture system, QPCR assay was applied to investigate the effect of XPS on the expression levels of stemness-related genes in these co-cultured breast cancer cells. As shown in Figure 2C, XPS strongly attenuated the expression levels of stemness-related genes, including β-catenin, OCT4, and Nanog, in the co-cultured breast cancer cells. As well known, β-catenin signaling is associated with increased stemness activity and chemoresistance of breast CSCs. In addition, it has been widely reported that β-catenin could directly upregulate OCT4 (Kelly et al., 2011; Lee et al., 2014; Qi et al., 2016; Yong et al., 2016) and Nanog (Takao et al., 2007; Li et al., 2013; Yong et al., 2016; Yu et al., 2017) gene transcription and protein expression in multiple cancer cells and CSCs. Therefore, we only further investigated whether XPS could attenuate the protein expression level of β-catenin. As shown in Figure 2D, western blot assay further convinced that XPS could also dramatically attenuate β-catenin protein expression levels in the co-cultured breast cancer cells. Altogether, XPS could inhibit the self-renewal activity of breast CSCs in the breast cancer cells and TAMs co-culture system.

Extensive researches have reported that TAMs in the tumor microenvironment could foster tumor growth. Therefore, we speculated that XPS could inhibit breast cancer growth by modulating TAMs. As shown in Figure 3A, XPS could dramatically inhibit the proliferation of Raw264.7-derived TAMs while exhibiting no obvious cytotoxicity in Thp1-derived TAMs. These results indicated that XPS might regulate TAMs through other mechanisms rather than direct cytotoxicity effects. TAMs often display a pro-tumorigenic M2 phenotype that fosters tumor growth and promotes resistance to therapy. Yet, macrophages are highly plastic and can also transform into anti-tumorigenic M1 phenotype (Parisi et al., 2018). As shown in Figure 3B, XPS could dramatically inhibit M2 phenotype polarization of TAMs induced by CM of breast cancer cells. Next, it is important to investigate how XPS-induced TAMs polarization variations brought into robust changes in breast CSCs population in the co-culture system. Since breast cancer cells and TAMs were separated by 0.4 µm permeable membrane in the transwell co-culture system, which only allowed free exchange of media and soluble molecules, but forbade direct cell interaction, we therefore speculated that some pro-tumorigenic molecules secreted by TAMs may be suppressed by XPS and subsequently weakened the promotion effect of TAMs on breast CSCs self-renewal. CXCL1 was the most-abundant chemokine secreted by TAMs (Palomino and Marti, 2015; Wang et al., 2018). Elisa assay indicated that XPS could significantly inhibit CXCL1 secretion from TAMs in a dose-dependent manner (Figure 3C). Western blot and QPCR results further verified that XIAOPI formula could dramatically inhibit CXCL1 protein expression levels (Figures 3D, E) and mRNA transcription in a dose-dependent manner in both human and mouse TAMs (Figure 3F). Double luciferase reporter gene assay suggested that XPS suppressed the promoter activity of CXCL1 gene in TAMs. Altogether, XPS could inhibit M2 phenotype polarization and CXCL1 expression and secretion of TAMs.

Considering that XPS could significantly inhibit CXCL1 expression and secretion from TAMs, it is important to further study whether CXCL1 is the key molecule involved in the inhibitory effect of XPS on breast CSCs. As shown in Figures 4A, B, CXCL1 administration could significantly increase the ALDH⁺ subpopulations and mammospheres numbers in breast cancer cells. Meanwhile, CXCL1 administration could abrogate the inhibitory effect of XPS on breast CSCs subpopulations and mammospheres formation abilities in the co-cultured breast cancer cells (Figures 2A, B and Figures 4A, B). QPCR assay further suggested that CXCL1 administration could elevate the mRNA expression levels of stemness-related genes, including β-catenin, OCT4, and Nanog in breast cancer cells while abrogating the inhibitory effect of XPS on mRNA expression levels of the above genes in the co-cultured breast cancer cells (Figure 4C). Western blot assay further convinced that CXCL1 administration could elevate β-catenin protein expression levels in breast cancer cells while reversing the inhibitory effect of XPS on β-catenin protein expression in the co-cultured breast cancer cells (Figure 4D). Altogether, XPS could inhibit the self-renewal activity of breast CSCs in the breast cancer cells and TAMs co-culture system by inhibiting TAMs/CXCL1 pathway.

Based on our in vitro findings, it is important to validate the anti-breast cancer activities and mechanisms of XPS in vivo. Breast cancer xenografts were established by implanting luciferase-labeled 4T1-Luc cells into the mammary glands of Balb/c mice. It was found that XPS administration could effectively suppress mammary tumor growth in the mouse 4T1-Luc xenograft model in vivo (Figures 5A, B). In addition, compared with the saline group, no treatment-correlated mortality or apparent decrease in body weight (Figure 5C) was observed, implying that XPS administration didn’t induce additional toxic and side effects. Notably, an in vivo imaging experiment also suggested that XPS could inhibit the metastasis efficacy of the breast cancer cell line 4T1-Luc (Figure 5A), suggesting XPS may also have anti-metastatic activity in vivo. Furthermore, XPS administration significantly decreased the infiltration degree of macrophages as well as their M2 phenotype polarization in the 4T1-Luc xenografts in vivo. Meanwhile, consistent with our in vitro findings, XPS administration strongly attenuated the expression of CXCL1 in mammary tumor tissue in vivo (Figures 5D, E). These results indicated that XPS might also inhibit M2 phenotype transformation and CXCL1 secretion of TAMs in vivo. Moreover, XPS administration significantly decreased the proportions of ALDH⁺ subpopulations and suppressed the expression of ALDH1A1 in the 4T1-Luc xenografts, indicating that XPS could also inhibit the activity of breast CSCs in vivo (Figures 5F, G). As shown in Supplementary Figure 1 and Supplementary Materials and Methods, TAMs co-injection significantly increased mammary tumor growth as well as breast CSCs population when compared with saline group. Furthermore, both XPS treatment and CXCL1 knockdown in TAMs could partly abrogate the promotion effect of TAMs on mammary tumor growth and breast CSCs population. More importantly, CXCL1 overexpression in TAMs partly reversed the inhibitory effect of XPS on mammary tumor growth and breast CSCs population. Taken together, XPS could suppress breast tumor growth and breast CSCs activity in vivo by modulating TAMs/CXCL1 pathway.