Tumor-Associated Neutrophils Regulate Breast Cancer Progression Through the AQP9/STAT3 Signaling Pathway.

Wuqin Xu: data curation, investigation, methodology, validation, writing – original draft. Guilu Zhu: investigation, methodology, resources, validation. Youjing Sheng: formal analysis, supervision, visualization. Wenjun Zhang: investigation, methodology, resources, validation. Shujing Wang: conceptualization, formal analysis, project administration, supervision, visualization, writing – review and editing. Qiang Wu: conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, supervision, writing – review and editing.

Approval of the research protocol by the Medical Ethics Committee of Wannan Medical College (No. 177).

Neutrophils isolated from the peripheral blood of healthy adults were labeled with common neutrophil markers CD11b and CD66b, demonstrating a purity of over 90% (Figure  1A ). The harvested neutrophils cultured with TCCM from MDA‐MB‐231, MCF‐7, or SK‐BR‐3 cells for 10 h were named MDA‐MB‐231‐TANs, MCF‐7‐TANs, and SK‐BR‐3‐TANs, respectively. The lifespans of MDA‐MB‐231‐TANs, MCF‐7‐TANs, and SK‐BR‐3‐TANs were extended compared with those of neutrophils without TCCM stimulation (Neu). MDA‐MB‐231‐TANs exhibited the longest lifespan extension (Figure  1B,C ). Hence, MDA‐MB‐231‐TANs were selected for subsequent investigations and are referred to as TANs in subsequent experiments. The conditioned media from TANs is referred to as TANCM. The expressions of surface markers CD11b, PD‐L1, and CD62L on activated TANs were significantly higher than those on Neu (Figure  1D,E ). Moreover, MDA‐MB‐231‐TANs exhibited morphological changes, i.e., they showed irregular shapes with extending cellular protrusions (Figure  S1 ). Neutrophils were activated by tumor‐conditioned medium(TCCM). (A) Flow cytometry analysis showed the purity of neutrophils derived from the peripheral blood of healthy adults through the detection of surface markers on neutrophils. (B,C) Flow cytometry analysis showed the viability of neutrophils treated with breast cancer (BC) cell conditioned medium. (D,E) Flow cytometry analysis showed surface marker expression differences between neutrophils without TCCM and tumor‐associated neutrophils (TANs). We subsequently evaluated the function of TANs in BC cells. CCK‐8 assay revealed that MCF‐7 or SK‐BR‐3 cell proliferation was significantly higher after treatment with TANCM than after treatment with conventional medium (NC) and NeuCM (Figure  2A ). Transwell assay showed that treatment with TANCM enhanced the migration and invasion of MCF‐7 and SK‐BR‐3 cells compared with the NC and NeuCM treatment groups (Figure  2B,C ). Similarly, wound healing assay showed that the treatment of MCF‐7 and SK‐BR‐3 cells with TANCM‐enhanced healing abilities compared with the NC and NeuCM treatment groups (Figure  2D,E ). These results indicate that TANs promoted the proliferation, migration, and invasion of BC cells. TANs promoted the proliferation, migration, and invasion of BC cells (A) CCK‐8 assay showed that TAN‐conditioned medium (TANCM) promote increased proliferation of MCF‐7 and SK‐BR‐3 cells. (B,C) Transwell assay showed that TANCM induced increased migration and invasion in MCF‐7 and SK‐BR‐3 cells. (D,E) Wound healing assay showed that TANCM induced increased migration in MCF‐7 and SK‐BR‐3 cells. Analysis of the TIMER2.0 database revealed a significant positive correlation between AQP9 expression and neutrophil infiltration in BC tissues (Figure  3A p  < 0.0001, R  = 0.502). IHC showed that CD66b + neutrophils infiltrated the parenchyma and stroma of BC tissues, and only cells located inside or loosely beside the parenchyma were identified as TANs (Figure  3B ). AQP9 was located in the cell membrane and cytoplasm (Figure  3B ). The density of TAN infiltration in BC tissues significantly and positively correlated with AQP9 expression (Figure  3C p = 0.0042, R  = 0.3150). In vitro, western blot and RT‐qPCR assays showed that the protein and mRNA expression of AQP9 was notably upregulated in MCF‐7, SK‐BR‐3, and MDA‐MB‐231 cells treated with TANCM compared with that in cells treated with NC and NeuCM (Figure  3D,E ). These results indicate that TANs promote AQP9 expression in BC cells. TAN infiltration positively correlated with AQP9 expression in BC. (A) Analysis of the TIMER2.0 database revealed that AQP9 expression in BC was positively correlated with neutrophil infiltration. (B) Immunohistochemistry (IHC) staining showed AQP9 protein expression in BC tissues and CD66b + neutrophil infiltration in parenchymal and stromal tissues. Scale bars: 50 and 100 μm. (C) AQP9 expression in BC tissues was positively correlated with neutrophil infiltration. (D) RT‐qPCR assay showed that the mRNA expression of AQP9 in BC cells increased after treatment with TANCM. (E) Western blot assay showed that the protein expression of AQP9 in BC cells increased after treatment with TANCM. RNA transcriptomic data were obtained from the TCGA database, comprising 1053  bc samples and 111 normal breast samples. AQP9 expression was significantly higher in BC samples than in normal tissues (Figure  4A ). Besides, the Bc‐GenExMiner database was used to analyze AQP9 expression across various BC subtypes. AQP9 was found to be substantially higher in basal‐like BC than in luminal A, luminal B, and HER‐2 overexpressing subtypes, as well as in normal breast tissue (Figure  S2A ). In clinical samples, AQP9 immunostaining intensity was significantly stronger in BC tissues than in paracancerous tissues (Figure  4B ). Moreover, AQP9 expression was significantly higher in TNBC tissues than in other subtypes (Figure  4B ). However, no significant associations between AQP9 expression and sex, age, lymph node metastasis, or TNM staging of BC patients were detected (Table  S1 ). AQP9 was highly expressed in BC. (A) TCGA data analysis showed increased AQP9 expression in BC tissues compared with normal breast tissues. (B) IHC showed that AQP9 expression increased in BC tissues, with the highest expression observed in triple‐negative BC tissues. (C) CCK‐8 assay showed that AQP9 promoted the proliferation of MCF‐7 and MDA‐MB‐231 cells. Western blot assay showed differential expression of AQP9 protein in BC cells of different molecular subtypes. AQP9 expression was higher in TNBC MDA‐MB‐231 and HER‐2 overexpressing subtype SK‐BR‐3 cells than in luminal subtype MCF‐7 cells (Figure  S2B,C ). MCF‐7 cells overexpressing AQP9 and SK‐BR‐3 and MDA‐MB‐231 cells with downregulated AQP9 expression were generated (Figure  S2D–G ). CCK‐8 assay revealed that the overexpression of AQP9 enhanced MCF‐7 cell proliferation, and AQP9 knockdown reduced MDA‐MB‐231 cell proliferation (Figure  4C ). Transwell and wound healing assays demonstrated that neither overexpression nor knockdown of AQP9 affected the migration and invasion of MCF‐7 and MDA‐MB‐231 cells (Figure  5A–F ). These findings suggested that AQP9 promotes BC cell proliferation but does not affect migration and invasion. High expression of AQP9 is associated with poor prognosis in patients with BC. (A‐C) Transwell assay showed that AQP9 did not alter the invasion and migration of MCF‐7 and MDA‐MB‐231 cells. (D‐F) Wound healing assay showed that AQP9 did not alter the migration of MCF‐7 and MDA‐MB‐231 cells. (G) AQP9 expression indicated reduced relapse‐free survival (RFS) and overall survival (OS) in patients with BC through the Kaplan–Meier Plotter database. (H) Kaplan–Meier survival curves showed that AQP9 expression was associated with poor RFS and OS in patients with BC. (I) ROC analyses showed that AQP9 had a high prognostic value for predicting RFS in patients with BC. (J) ROC analyses showed that AQP9 had a high prognostic value for predicting OS in patients with BC. The prognostic value of AQP9 was analyzed. The Kaplan–Meier Plotter database showed that higher AQP9 expression was statistically significantly correlated to poor RFS and OS in patients with BC (Figure  5G ). In clinical samples, high expression of AQP9 was correlated with poor RFS (Figure  5H p  = 0.0153) and OS (Figure  5H p  = 0.0135). ROC analysis revealed that the high expression of AQP9 exhibited a significant prognostic value in predicting poor RFS (AUC = 0.9816, p  < 0.0001) and OS (AUC = 0.9939, p  < 0.0001) in patients with BC (Figure  5I,J ). We further validated whether TANs promoted the progression of BC through the regulation of AQP9. CCK‐8 assay showed that the upregulation of AQP9 expression in MCF‐7 cells following the treatment of TANCM significantly enhanced proliferation. However, when AQP9 was knocked down in MDA‐MB‐231 cells, TANCM could not enhance proliferation (Figure  6A ). Transwell and wound healing assays showed that the overexpression of AQP9 in MCF‐7 cells, followed by the treatment of TANCM‐enhanced migration and invasion. Similarly, when AQP9 was knocked down in MDA‐MB‐231 cells, TANCM could not enhance migration and invasion (Figure  6B–E , Figure  S3A–D ). These data suggested that TANs may require the assistance of AQP9 to exert their effects on proliferation, migration, and invasion in BC cells. TANs promoted BC progression via upregulating the AQP9 pathway. (A) CCK‐8 assay showed that TANCM stimulated MCF‐7 cell proliferation following AQP9 overexpression and inhibited MDA‐MB‐231 cell proliferation after AQP9 knockdown. (B‐E) Transwell assay showed that TANCM stimulated MCF‐7 cell migration and invasion following AQP9 overexpression and inhibited MDA‐MB‐231 cell migration and invasion after AQP9 knockdown. (F‐I) The breast tumor growth rate, volume, and weight increased in mice after TAN injection, whereas they decreased after AQP9 knockdown. (J) Hematoxylin and eosin (H&E) staining revealed tumor tissues in the lungs of the mice. (K) IHC staining demonstrated Ki‐67 protein expression in tumor tissue, and H&E staining showed tumor tissue invading mouse bone. A BC xenograft model was established by subcutaneously injecting MDA‐MB‐231 cells with AQP9 knockdown into mice. Mice receiving TAN injections showed a higher tumor growth rate, volume, and weight than those that were not injected with TANs. However, after knocking down AQP9 in MDA‐MB‐231 cells, the tumor growth rate, volume, and weight were significantly reduced in mice receiving TAN injections (Figure  6F–I ). This result suggests that TANs may promote BC xenograft growth with the assistance of AQP9. H&E and IHC staining of tumor tissues showed two cases of lung metastasis: one featuring an isolated micrometastasis and one involving 64 microsatellite foci in the TANs group (Figure  6J ). IHC analysis was used to assess Ki‐67 expression in metastatic tumors. Ki‐67 is a marker of tumor cell proliferation. The Ki‐67 index was approximately 70% (Figure  6K ). One case in the TAN group showed tumor cell invasion into adjacent bone (Figure  6K ). These findings suggest that TANs facilitate BC progression. Several studies have demonstrated that TANs participate in tumor progression by secreting cytokines to activate the STAT3 signaling pathway [ 26 , 27 ]. AQP9 may also participate in tumor progression by activating the STAT signaling pathway [ 19 ]. We speculated that TANs may activate the STAT3 pathway by upregulating AQP9 in BC. Western blot assay showed a significant increase in STAT3 protein phosphorylation levels in MCF‐7 and SK‐BR‐3 cells treated with TANCM compared with the control group. Overexpression of AQP9 increased pSTAT3 level in MCF‐7 cells treated with TANCM (Figure  7A ). In contrast, pSTAT3 levels decreased in MDA‐MB‐231 cells treated with TANCM after AQP9 knockdown (Figure  7A ). Similar findings were observed in vivo. Western blotting indicated that pSTAT3 levels increased in xenograft tumors from the TANs group compared with pSTAT3 levels in tumors from the NC group, and pSTAT3 levels decreased in the shAQP9 and TANs+shAQP9 groups (Figure  7B ). These results indicate that TANs may enhance STAT3 phosphorylation via AQP9 in BC (Figure  7C ). TANs enhanced STAT3 phosphorylation via AQP9 in BC. (A) Western blot assay showed increased pSTAT3 levels in MCF‐7 cells treated with TANCM. In contrast, pSTAT3 levels were reduced in TANCM‐treated MDA‐MB‐231 cells after the knockdown of AQP9. (B) Western blot analysis indicated that the expression of pSTAT3 levels decreased in the shAQP9 and TANs + shAQP9 groups. (C) Schematic diagram of TAN‐mediated regulation of AQP9/STAT3 pathway in BC progression.

The signaling cross‐talk between tumor cells and immune cells can create a tumor microenvironment (TME) that facilitates tumor development. Neutrophils, one of the primary immune cell types, are involved in this process. A series of studies have demonstrated that activated neutrophils exhibit high expressions of CD11b, CD66b, and PD‐L1, while CD62L expression is reduced [ 28 , 29 , 30 ]. In this study, compared to non‐activated neutrophils, CD11b and PD‐L1 expressions were significantly increased in TANs. However, we also observed a notable increase in CD62L expression. CD62L, an adhesion marker that can identify active neutrophils. Fatma et al. found that MDA‐MB‐231 cell conditioned medium may trigger an inflammatory pattern with evidence of stronger adhesion (CD62L), thereby facilitating the activation of neutrophils [ 31 ]. Neutrophils normally have a short lifespan of 7–10 h in peripheral blood. However, Maas et al. demonstrated that up to 80% of TANs survive for 24 h, and approximately 20% survive after 48 h [ 32 ]. The prolonged lifespan is a marker of TANs activation. In this study, TCCM was used to mimic the TME and stimulate peripheral blood neutrophils to form TANs, significantly extending their lifespan. Interestingly, neutrophils survived much longer when exposed to TCCM derived from MDA‐MB‐231 cells than when exposed to TCCM derived from MCF‐7 and SK‐BR‐3 cells. Sengupta et al. reported that TCCM derived from highly invasive TNBC cells induced a polarized morphology, enhanced neutrophil migratory capacity, and an elongated shape upon activation. In contrast, TCCM harvested from poorly invasive ER+ BC cells did not trigger this response [ 33 ]. In our study, neutrophils stimulated by MDA‐MB‐231‐derived TCCM exhibited irregular protrusions, whereas neutrophils exposed to MCF‐7 or SK‐BR‐3‐derived TCCMs did not exhibit irregular shapes. These findings align with Sengupta's report and suggest that the morphology and functions of TANs in the TME depend on the invasive capabilities of tumor cells. We demonstrated that TANs promote BC cell proliferation, invasion, and migration, consistent with our previous studies on lung adenocarcinoma and BC, which showed that TANs were linked to poor patient prognoses [ 11 , 12 , 13 ]. We also demonstrated that TANs promote BC xenograft growth, osseous invasion, and lung metastasis in vivo. These findings were consistent with those of Tian et al., who discovered that TANs promote metastasis in colorectal cancer models [ 34 ]. Overall, these observations suggest that TANs play a crucial role in BC progression. A deeper understanding of the mechanisms by which TANs drive BC progression may reveal new therapeutic targets. AQP9 is a member of the water channel protein family involved in water, H 2 O 2 , and urea transport. AQP9 also promotes glycerol uptake by hepatocytes. Thus, AQP9 is essential for various physiological functions, including metabolism [ 35 ], energy homeostasis [ 36 ], and oxidative stress regulation [ 37 ]. Previous studies have implicated AQP9 in the progression of multiple malignancies, including lung cancer [ 38 ], colon cancer [ 16 ], glioma [ 39 ], and renal cell cancer [ 40 ]. However, the role of AQP9 in BC remains unclear. Kirkegaard et al. demonstrated that AQP9 is expressed in normal breast tissue, but is expressed at higher levels in bc [ 21 ]. We herein demonstrated that AQP9 was highly expressed in BC, particularly in TNBC, and its higher expression was correlated with poor RFS and OS in patients with BC. These findings suggest that AQP9 plays a role in BC progression. Our in vitro results showed that AQP9 promoted BC cell proliferation but had limited effects on BC cell migration and invasion. These results were contradictory to those of other studies showing that several AQPs promote cancer spread by increasing MMP secretion, leading to extracellular matrix degradation, which facilitates cancer cell migration [ 21 ]. In endometriosis, AQP9 promotes migration and invasion by inducing MMP2 and MMP9 expression [ 41 ]. AQPs also participate in tumor progression by regulating cell adhesion proteins [ 42 ], modulating signal transduction [ 43 ], and promoting cell motility [ 16 ]. Further investigation is needed to clarify the mechanisms by which AQP9 influences BC. Both TANs and AQP9 contribute to tumor progression. AQP9 affects immune cells within the TME. For example, AQP9 in macrophages supports colorectal cancer progression by transporting lactate to facilitate TAM polarization [ 20 ]. However, the interaction between AQP9 and TANs in tumors is not well understood, and no related reports have been published to date. In our study, the positive correlation between AQP9 expression and TAN infiltration in clinical samples was confirmed through bioinformatics analysis. Both in vivo and in vitro studies showed that TANs upregulated AQP9 expression in BC cells. Interestingly, while AQP9 did not significantly affect the invasion and metastasis of BC cells, BC cells overexpressing AQP9 showed enhanced invasion and metastasis when cocultured with TANs. To the best of our knowledge, this is a novel observation, but the underlying mechanism is unclear. Our findings suggest that TANs may rely on AQP9 to facilitate BC invasion and metastasis. TANs can activate the STAT3 pathway in gastric cancer through the secretion of IL‐17ɑ [ 26 ], and TANs promote STAT3 activation by secreting IL‐10 in lung cancer, which enhances metastasis [ 27 ]. In this study, western blot assay revealed that TANs enhanced STAT3 phosphorylation in BC following AQP9 overexpression. This suggests that TANs can activate the STAT3 signaling pathway via AQP9 in BC. However, there are several limitations to our study. We did not further investigate the mechanism for increased AQP9 expression and the subsequent activation of STAT3 by TANs in BC. TANs contain numerous signaling molecules. Auto‐immune regulator (AIRE) was recently discovered in TANs. AIRE may regulate tumor cell death and inflammation to promote tumor progression [ 44 ]. Future studies should focus on identifying the specific signaling molecules in TANs and their role in tumor progression. In conclusion, TANs enhance STAT3 phosphorylation by upregulating AQP9 to promote BC progression. AQP9 may be crucial for TAN‐mediated BC progression and is a novel target for immunotherapy in patients with BC.

The authors have nothing to report.

Breast cancer (BC) is the most commonly diagnosed malignancy and the leading cause of cancer‐related deaths among women worldwide; moreover, the incidence of BC is continuously increasing each year [ 1 , 2 ]. However, advances in targeted treatments and early diagnosis via screening mammography have decreased the overall mortality rate from BC by 44% since 1989 [ 3 ]. Despite these improvements, the 5‐year survival rate for metastatic BC is less than 30% [ 4 ]. Advances in treatment for patients with advanced‐stage BC have not kept pace with those for early‐stage BC, highlighting the urgent need for new therapeutic approaches or targets for advanced bc [ 5 ]. Neutrophils, which account for 50%–70% of circulating leukocytes in humans, are essential components of the immune system that respond to infectious agents and tissue damage. Neutrophils play a complex dual role in tumors. Tumor‐associated neutrophils (TANs) may hinder cancer development by directly killing tumor cells or activating innate and adaptive immunity [ 6 ]. However, TANs have been reported to promote tumor progression due to their functional plasticity [ 7 ]. Protumor TANs affect multiple stages of tumor development, including initiation, metastasis, and immunosuppression [ 8 ]. High TAN infiltration in solid tumors is usually associated with poor clinical outcomes [ 6 ]. Wang et al. reported that abundant TAN infiltration in glioblastoma was linked to poor overall survival (OS) [ 9 ]. In bladder cancer, increased CXCL1/8 secretion recruits TANs, which enhance vascular endothelial growth factor (VEGF) receptor and matrix metalloproteinase 9 (MMP9) expression to promote lymphangiogenesis and lymph node metastasis [ 10 ]. We conducted extensive studies on TANs. In lung adenocarcinoma, TAN infiltration promotes cancer invasion and metastasis via the Notch3 signaling pathway [ 11 ]. In BC, TANs activated by tumor‐derived C–C motif chemokine ligand 20 suppress T cell activity via programmed death‐ligand 1, and high TAN density correlates with shorter disease‐free survival [ 12 ]. Tumor‐derived granulocyte colony‐stimulating factor induces TANs to secrete RLN2, which activates the PI3K–AKT–MMP9 pathway, facilitating BC metastasis [ 13 ]. Despite these findings, the mechanisms by which TANs promote BC progression require further exploration. Aquaporins (AQPs) are a family of transmembrane proteins that facilitate water transport across cell membranes following an osmotic gradient. AQP9 transports water, glycerol, and urea and plays a role in glycerol uptake by hepatocytes, thereby regulating energy homeostasis via lipid synthesis and gluconeogenesis [ 14 ]. AQP9 exhibits diverse roles in different tumor types. In the hypoxic tumor microenvironment, AQP9 inhibits hypoxia‐inducible factor 1α expression, thereby suppressing hepatocellular carcinoma invasion [ 15 ]. In contrast, AQP9 promotes metastasis in colorectal cancer by binding and stabilizing disheveled 2, thereby leading to the activation of the Wnt/β‐catenin pathway [ 16 ]. Bioinformatics analyses suggest that AQP9 is associated with immune infiltration and prognosis of various cancers [ 17 , 18 , 19 ]. However, studies investigating the relationship between AQP9 and immune infiltration in tumors are limited. AQP9 facilitates lactate transport, induces polarization of tumor‐associated macrophages (TAMs), and promotes the secretion of VEGF, thereby accelerating colorectal cancer progression [ 20 ]. In BC, high AQP9 mRNA expression is associated with poor relapse‐free survival (RFS) and OS [ 21 , 22 ]. However, the exact mechanisms by which AQP9 contributes to TAN‐mediated BC progression are unclear. We herein aimed to determine the role of AQP9 in TAN‐mediated BC progression and explore the underlying mechanisms. Our results highlight the potential of AQP9 as a target for immunotherapy in patients with BC.

The authors declare no conflicts of interest.

BC gene expression data (counts: FPKM) were sourced from the Cancer Genome Atlas (TCGA) database ( https://portal.gdc.cancer.gov ). Perl ( https://www.perl.org ) was used to convert the Ensembl IDs of genes into symbols. The R4.2.0 software, combined with the “limma” [ 23 ] and “beeswarm” [ 24 ] packages, was employed for data normalization, processing, and analysis. The expression data of the AQP9 was extracted and visualized via a differential scatter plot. The Bc‐GenExMiner database ( http://bcgenex.ico.unicancer.fr ) was used to examine the variability in AQP9 expression across different histological subtypes of BC. The Kaplan–Meier Plotter database ( https://kmplot.com/analysis ) was used to analyze the prognostic significance of AQP9 in patients with BC. The TIMER 2.0 database ( https://timer.cistrome.org ) was used to assess the association between neutrophils and AQP9 in BC. The formalin‐fixed, paraffin‐embedded (FFPE) surgical specimens from 81 patients with BC, along with their paracancerous breast tissues 2 cm away from the tumor, were randomly collected at the Pathology Department of the Wannan Medical College Affiliated Hospital from 2014 to 2019. None of these patients had undergone radiotherapy or neoadjuvant chemotherapy before surgery. The patients, including survival, recurrence, and metastasis, were followed up from the date of surgery until the date of death or last follow‐up (February 2022). Clinicopathological data from all patients with BC were meticulously collected, with a median patient age of 48 years. The expression status of estrogen receptor (ER), progesterone receptor (PR), Erb‐B2 receptor tyrosine kinase 2 (HER‐2), and Ki‐67 was reassessed by two pathologists. According to the molecular classification of BC as redefined in the 15th St. Gallen International Breast Cancer Conference in 2017, all samples were categorized into the following molecular subtypes: Luminal A, Luminal B, HER‐2 overexpression, and triple‐negative BC (TNBC) [ 25 ]. Among the 81 BC patients, there were 11 Luminal A subtype, 33 Luminal B subtype, 6 HER‐2 overexpression subtype, and 31 TNBC patients. Table  S1 lists these clinicopathological details. FFPE tissue was cut into 4‐μm‐thick sections for hematoxylin and eosin (H&E) staining and IHC staining. The IHC procedure was carried out according to the product protocol. Primary antibodies mainly included AQP9 (1:100; Cat. A8540, ABclonal, China) and CD66b (1:100; Cat. 555,723, BD Pharmingen, USA). Subsequently, after secondary antibody (HRP‐conjugated goat anti‐mouse/anti‐rabbit IgG, MXB Biotechnologies, China) incubation, chromogenic reaction, and nuclear staining, a visualized image was obtained. The method for quantifying the infiltration density of CD66b + neutrophils in the tumor stroma was consistent with that used in a previous study [ 11 ]. In total, 10 HPFs (× 400) were randomly selected, and CD66b + neutrophils were counted. Based on the median count across all samples, CD66b + neutrophils were categorized into high‐ and low‐density groups. AQP9 expression was evaluated by considering both the staining intensity and distribution of deposited chromogenic agents. Intensity grading was defined as follows: 0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining. The classification of the extent of positive staining was as follows: 0, 0%–5%; 1, 6%–25%; 2, 26%–50%; 3, 51%–75%; and 4, 76%–100%. The final IHC score was determined by multiplying the staining intensity by the staining area: a score of ≥ 4 was positive and a score of < 4 was negative. Human BC cells MDA‐MB‐231 and MCF‐7 were purchased from Pricella Biotechnology (Wuhan, China). SK‐BR‐3 cells were donated by the Department of Immunology of Anhui Medical University. MDA‐MB‐231 cells were cultured in L‐15 medium (Biosharp, China), MCF‐7 cells in MEM medium (Gibico, USA), and SK‐BR‐3 cells in DMEM medium (Gibico, USA). All media were supplemented with 10% fetal bovine serum (FBS; Wisent, Canada) and 1% penicillin–streptomycin (Beyotime, China). The cell culture conditions for all cell lines were maintained at a constant temperature of 37°C under 5% CO 2 . Peripheral blood was obtained from healthy adults. The method for isolating neutrophils from peripheral blood was the same as that described in our previous study [ 11 ]. The resultant white precipitate represented neutrophils, which were then resuspended in 1640 medium with 2% FBS. Tumor cell conditioned medium (TCCM) was from the cultured MDA‐MB‐231, SK‐BR‐3, or MCF‐7  bc cells, respectively. Neutrophil‐conditioned medium (NeuCM) was obtained from the isolated neutrophils cultured for 10 h. TANs conditioned medium (TANCM) was obtained from the coculture system of TCCM and neutrophils for 10 h. Neutrophil viability was assessed using the Annexin V‐FITC/PI apoptosis detection kit (BB‐4101, BestBio, China) according to the product protocol. The expression levels of surface markers CD11b (Cat. 568,229, Clone No. ICRF44, 1:100; BD Biosciences), CD66b (Cat. 571,719, Clone No. G10F5, 1:100; BD Biosciences), PD‐1 (Cat. 569,074, Clone No. 29E.2A3, 1:100; BD Biosciences), and CD62L (Cat. 565,037, Clone No. SK11, 1:100; BD Biosciences) on neutrophils were respectively detected according to the instructions. The data were processed using FlowJo software (version 7.6.1). A short hairpin RNA plasmid for the knockdown of AQP9 (shAQP9, P29436 pLV3‐U6‐MCS‐shRNA‐EF1a‐CopGFP‐Puro) was purchased from Miaoling (Wuhan, China). A plasmid for overexpressing AQP9 (OEAQP9, pSIN‐EF2‐3Xflag‐puro‐EV0319) was purchased from Tongyong (Anhui, China). Table  S2 lists the sequence information. These plasmids were subjected to transformation, extraction, and transfection into BC cells using Lipofectamine 8000 (Beyotime, China) following the manufacturer's protocol. Transfected cells were harvested after 48 h. Stable cell lines were selected using puromycin (Beyotime, China) over a span of 2 weeks. Cell proliferation was assessed using the CCK‐8 assay and performed according to the product protocol. Absorbance was measured at 450 nm using a microplate reader. The resulting data were further processed and analyzed using Excel and GraphPad Prism (version 8.0). The invasion and migration assays were conducted in 24‐well plates using transwell inserts with 8‐μm pore size. The invasion assay mirrored the migration assay, with the exception of the upper chambers coated with Matrigel (BD Pharmingen, USA). Cell suspensions were introduced into the upper chamber, whereas the lower chamber was supplemented with the following media formulations: NC, NeuCM, and TANCM. The cells were cultured for 24–48 h. Following incubation, the cancer cells adhering to the underside of the inserts were fixed and stained, and five randomly selected microscopic fields were used for cell counting. Cells were seeded in 6‐well plates, and cell confluence was found to reach > 90% after adhering overnight. Wounds were induced by gently scraping the monolayer with a 100‐μL pipette tip, creating cell‐free zones. Subsequently, the cells were subjected to treatments with NC, NeuCM, and TANCM for 24 h or 48 h. The migration of cells into the cell‐free zones was monitored, and the migration rate was quantified as a percentage of the initial wound area healed. Data were analyzed using ImageJ software (version 1.42q). Total RNA was extracted from the lysed cells using TRIzol Reagent (Thermo Fisher Scientific, USA). The purity and concentration of RNA were assessed using the NanoDrop 2000 system (Thermo Fisher Scientific, USA). The purified RNA was reverse transcribed into cDNA using a Reverse Transcriptase Kit (Evo M‐MLV RT Master Mix; Accurate Biology, China) following the specifications of the manufacturer. The SYBR Premix ExTaq II real‐time PCR reagent kit (Accurate Biology, China) was used for RT‐qPCR. Data analysis was performed using LightCycler 480 software (Roche, Switzerland). The relative expression of mRNA was evaluated using the 2 − ΔΔct method and normalized against the internal control GAPDH. Table  S3 lists the sequence information of AQP9 primer (Sangon Biotech, China) and GAPDH primer (General Biology, China). Proteins were obtained by lysing cells or tissues using RIPA lysis buffer, quantified, and standardized using a BCA assay kit (Biosharp, China). Subsequently, the proteins were subjected to SDS‐PAGE and transferred onto PVDF membranes. The membranes were blocked with Protein Free Rapid Blocking Buffer (Epizyme, China) and incubated overnight at 4°C with the primary antibodies, including β‐ACTIN (Cat. 66,009‐1‐Ig, Clone No. 2D4H5, 1:2000; Proteintech, China), AQP9 (Cat. sc‐74409, Clone No. G‐3, 1:500; Santa Cruz, USA), signal transducer and activator of transcription 3 (STAT3; Cat. 30835, Clone No. D1B2J, 1:1000; Cell Signaling Technology, USA), and phospho‐STAT3 (pSTAT3; Cat. 9145, Clone No. D3A7, 1:1000; Cell Signaling Technology, USA). Then, the membranes were incubated at room temperature for 2 h with the peroxidase‐conjugated secondary antibody (Cat. SA00001, 1:10,000; Proteintech, China), and protein bands were visualized using the ECL luminol reagent (Cat. BMU102‐CN; Abbkine, China). Data were analyzed using ImageJ 1.42q software. Female NOD/SCID mice (6–8 weeks old) used for in vivo experiments were procured from GemPharmatech LLC (Jiangsu, China). A total of 1.0 × 10 6 cells were subcutaneously injected into the left axilla of the mice, which were subsequently divided into the following four groups: NC group, injecting BC cells only; TANs group, injecting the mixture of TANs and BC cells; shAQP9 group, injecting shAQP9 cells; and shAQP9 + TANs group, injecting the mixture of shAQP9 cells and TANs. The ratio of TANs to BC cells was set at 1:10. Commencing on day 14 postinoculation, TANs were administered via local and intravenous injection every 2 weeks. Tumor progression was monitored by measuring the maximum diameter (L) and minimum diameter (W) of a tumor twice a week. Tumor volume was calculated using the formula 1/2LW [ 2 ]. After 6 weeks, the mice were euthanized and the tumors were excised and weighed, followed by the dissection of the lungs and liver from each mouse. All experimental data were analyzed using GraphPad Prism 8.0 software and SPSS 25.0 software. Differences between groups were assessed using t‐tests or one‐way analysis of variance. The correlation between AQP9 and CD66b expression and the clinical pathological features of patients with BC was examined using chi‐square tests. Kaplan–Meier analysis and ROC analysis were used to assess patient prognosis and risk. Cell experiments were independently replicated three times. Data are presented as mean ± SD. A P value lower than 0.05 was considered statistically significant, with * indicating p  < 0.05, ** indicating p  < 0.01, *** indicating p  < 0.001, and “ns” denoting no statistical significance.

Figure S1. The morphology of neutrophils changed following stimulation with MDA‐MB‐231 cell conditioned medium. Figure S2. (A) BC‐GenExMiner database showed that AQP9 expression was most pronounced in triple‐negative BC. (B,C) The levels of AQP9 were higher in MDA‐MB‐231 and SK‐BR‐3 cells than in MCF‐7 cells. (D,E) Protein efficiency validation of AQP9 overexpression in MCF‐7 cells and subsequent knockdown of AQP9. (F,G) Protein efficiency validation of AQP9 knockdown in SK‐BR‐3 and MDA‐MB‐231 cells. Figure S3. (A,B) Wound closure assay showed that TANCM stimulated MCF‐7 cells migration following AQP9 overexpression. (C,D) Wound closure assay showed that TANCM inhibited MDA‐MB‐231 cell migration after AQP9 knockdown. Table S1. Summary of clinical characteristics of BC patients in IHC study. Table S2. Sequence of the AQP9 plasmid. Table S3. Primers used for RT‐qPCR.