High PANX1 Expression Leads To Neutrophil Recruitment And Formation of A High Adenosine Immunosuppressive Tumor Microenvironment In Triple-Negative Breast Cancer

High adenosine levels are an important characteristic of the tumor microenvironment (TME) in triple-negative breast cancer (TNBC). Pannexin 1 (PANX1) can release intracellular ATP to the extracellular space and elevate extracellular ATP (exATP) levels under physiological conditions. We performed public database bioinformatics analysis, surgical specimen histological validation, RNA sequencing, and exATP/extracellular adenosine (exADO) assays in vitro and in vivo to reveal the role of PANX1 in regulating the immune microenvironment of TNBC. Our results revealed that PANX1 was highly expressed and acted as a poor prognostic factor in TNBC. The PANX1 expression level was positively correlated with exATP and exADO levels in the TNBC TME. PANX1 promoted inltration of tumor-associated neutrophils (TANs) through exATP, and TANs highly expressed ENTPD1 (CD39) / NT5E (CD73). This study suggests that high PANX1 expression could promote TAN accumulation and adenosine production in the TNBC TME which have been shown to induce local immunosuppression.


Introduction
Breast cancer is the most common malignant tumor in women, and triple-negative breast cancer (TNBC) molecular subtype, which accounts for 15-20% of all breast cancers, is characterized by high invasiveness and ease of recurrence and metastasis and is associated with a poor prognosis due to lack of speci c treatments (1). Compared with other subtypes, TNBC is characterized by abundant immune cell in ltration (2), and the immune status of the tumor microenvironment (TME) has a signi cant impact on the treatment and prognosis of TNBC (3,4). Therefore, it is of clinical importance to investigate the key regulatory genes related to the immune microenvironment of TNBC.
Pannexin 1 (PANX1), a member of the gap junction protein family, forms hemichannels in cell membranes as hexamers or gap junction channels between cells, which mediates the release of intracellular ATP to the extracellular microenvironment (5,6). Extracellular ATP (exATP) and its metabolite extracellular adenosine (exADO) are important factors that regulate the local immune TME (7). Under physiological conditions, exATP released by PANX1 promotes innate and adaptive immune responses by attracting immune cells (8). In the TME, this process is disrupted, and exATP is rapidly metabolized by the nucleotidases ENTPD1 (CD39) and NT5E (CD73) to generate exADO (9). In the TME, exADO is a key factor that contributes to local immunosuppression (10). Recently, PANX1 expression was found to be important in suppressing airway in ammation in the asthma mouse model, and knockdown of PANX1 resulted in increased airway in ammation (11). In breast cancer, the way in which PANX1 affects tumorin ltrating immune cells (TIICs) and the immune microenvironment by regulating exATP levels has not been reported.
This study revealed that TNBC tissues had high PANX1 expression, which promoted the recruitment of tumor-associated neutrophils (TANs) and the accumulation of exADO, forming an immunosuppressive TME.

Single cell transcription data analysis
The TNBC single-cell dataset (23) was downloaded from the Broad Institute Single Cell Portal (https://singlecell.broadinstitute.org/single_cell/study/SCP1106/). The dataset was analyzed using the Seurat R package. Nonlinear dimensional reduction (tSNE) was used to visualize clustering results.
Epithelial cell clusters with highly variable CNVs were determined to be malignant by the inferCNV R package. The SingleR R package was used for cell type annotation, and the default annotated le (Wu_EMBO_metadata.csv) was used as a reference. The single-cell dataset was divided into PANX1 high and low expression groups according to the relative PANX1 expression level of each tumor sample in the count matrix (PANX1 low expression group: CID4513, CID4515 and CID44971; PANX1 high expression group: CID44041 and CID44991).
Immuno uorescence staining Gene colocalization was validated by monoclonal antibody-based immuno uorescence. FFPE sections were subjected to antigen retrieval by heating the slides in citrate buffer for 2 minutes, after which they were incubated with primary antibodies (anti-human PANX1 and MPO antibodies) at 4°C overnight. Then, the slides were incubated with uorescein-labeled secondary antibodies at room temperature, stained with DAPI and photographed under a laser confocal microscope (OLYMPUS IX83-FV3000-OSR).

Neutrophil isolation
Fresh surgical specimens were cut into small pieces and digested in medium containing 1 mg/mL collagenase IV (#V900893, Sigma-Aldrich) at 37°C in a constant temperature shaker for 2 h. The cell suspension was ltered through 40-µm nylon mesh (BD FALCON, #352340) for subsequent detection or culture. The separation of neutrophils from surgical specimens and peripheral blood was performed with a human neutrophil isolation kit according to the manufacturer's protocol (Miltenyi Biotec, Auburn, CA, USA).

RNA sequencing
Human genomic DNA was isolated using a QIAamp DNA Mini Kit according to the manufacturer's protocol (Qiagen, Valencia, CA, USA). Single-end libraries were subsequently constructed using the standard protocol provided by BGI (BGI-Shenzhen) and were then sequenced on the BGISEQ-500 platform. Key immune-related gene expression was analyzed, and the heatmap was generated by the ComplexHeatmap R package.
PANX1 expression was evaluated in control shRNA-transfected cells and was compared with that in nontransfected cells. Single-cell colonies of PANX shRNA-expressing cells were selected with puromycin and examined for PANX1 knockdown. Stable knockdown samples showed a 70-90% reduction in PANX1 expression. Cells were maintained under puromycin selection pressure and were periodically examined for effective PANX1 knockdown using western blot.

Extracellular adenosine assay
In 24-well plates, 2 × 10 4 cells per well were allowed to adhere overnight after which they were removed by centrifugation at 4°C for 10 minutes; the supernatant was then collected for ATP/ADO detection assays. Fresh surgical TNBC tissues and normal specimens were cut into small pieces and digested in medium containing 1 mg/mL collagenase IV (#V900893, Sigma-Aldrich) at 37°C at a constant temperature. The cell suspension was then ltered through 40-µm nylon mesh (#352340, BD FALCON) for subsequent ATP/ADO detection assays. ATP/ADO levels were measured using the ATP/ADO Assay Kit ( uorometric) (Abcam, Cambridge, MA, USA) according to the manufacturer's protocol.

Statistical analyses
Statistical signi cance was determined using unpaired two-tailed Student's t-tests and analysis of variance (ANOVA) followed by Tukey's test. For all in vitro assays, data are representative of at least three independent experiments, which each included three technical replicates unless otherwise stated. Differences in PANX1 expression in different subtypes were assessed using one-way ANOVA (Tukey's test). Paired differences in ATP/ADO levels and ENTPD1/NT5E expression between different groups were assessed using two-tailed t-tests. A log-rank test was performed using the online tool KM plotter to assess PANX1 as a survival biomarker. Statistical analyses were performed using GraphPad Prism software (version 9.0) and R software (version 4.0.5, R Core Team, http://www.R-project.org/). The results are given as the mean ± S.D. and p values < 0.05 were considered signi cant (unless otherwise stated). The details of the statistical analysis are provided in the results section and gure legends.
1. PANX1 was highly expressed in TNBC and was negatively correlated with TNBC prognosis Compared with normal tissues, we found that PANX1 was highly expressed in breast cancer (BRCA), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and stomach adenocarcinoma (STAD) based on the web tool TIMER (TCGA-BRCA data) (p < 0.001 was signi cant) (Fig. 1A). To further investigate the role of PANX1 expression in breast cancer, we analyzed TCGA-BRCA and METABRIC transcriptome data and found that PANX1 was highly expressed in the basal-like subtype (PAM50 model) ( Fig. 1B-C). The prognostic value of PANX1 in different breast cancer molecular types was analyzed using the KM plotter web tool, and high PANX1 expression suggested poor prognosis in terms of overall survival in TNBC (p = 0.015) (Fig. 1D). In the other three molecular subtypes (Luminal A & B and HER2-enriched), no signi cant correlation was observed between PANX1 expression and overall survival (p > 0.05) (Fig. 1D). The above results suggested that PANX1 was highly expressed and was associated with a poor prognosis in TNBC.
2. PANX1 promoted high ENTPD1/NT5E expression in the TME PANX1 is acknowledged as a dominant regulator of exATP release, while exATP and its metabolite exADO are believed to induce an immune-suppressive TME (9). Analysis of TCGA-BRCA basal-like subtype transcriptome data revealed that the PANX1 high expression (top 50%) group had higher ENTPD1 and NT5E expression levels than the PANX1 low expression (bottom 50%) group (p < 0.001) ( Fig. 2A). Further linear analysis suggested a positive correlation between PANX1 expression and ENTPD1 and NT5E expression in TCGA-BRCA basal-like subtype transcriptome data (p < 0.001) (Fig. 2B). The RNA-seq data of surgical breast cancer specimens (TNBC PANX1 high: n = 3; TNBC PANX1 low: n = 3; Luminal subtype: n = 3) suggested higher expression of ENTPD1 and NT5E in the TNBC PANX1-high group compared to TNBC PANX1-low and Luminal group (p < 0.05) (Fig. 2C). According to immunohistochemistry of TNBC surgical specimens, the expressions of ENTPD1 and NT5E were higher in the PANX1 high group than in the PANX1 low group (Fig. 2D). The above results indicated that high PANX1 expression promoted high ENTPD1/NT5E expression in the TNBC TME.

PANX1 expression was positively correlated with TAN in ltration in TNBC
Using CIBERSORT LM22, we analyzed the effect of PANX1 expression in the TCGA-BRCA basal-like subtype on the tumor immune microenvironment. We found that the abundances of in ltrating neutrophils (p < 0.05), resting memory CD4 + T cells (p < 0.05), follicular helper T cells (p < 0.05), monocytes (p < 0.05), CD8 + T cells (p < 0.05), and activated natural killer (NK) cells (p < 0.05) were signi cantly different between tumors with high PANX1 expression (top 50%) and those with low PANX1 expression (bottom 50%) (Fig. 3A). The top 250 coexpressed PANX1-related genes were obtained through GEPIA2, and the gene coexpression network was constructed by GeneMANIA after intersection with the ImmPort gene set (Fig. 3B). Furthermore, gene ontology (GO) analysis suggested that PANX1 and its coexpressed genes were related to granulocyte chemotaxis (Fig. 3C). The relationship between PANX1 expression and TIICs was assessed using TIMER in the TCGA-BRCA basal-like subtype data, and a positive correlation was found between PANX1 and the in ltration level of TANs (p < 0.05) (Fig. 3D). The above results suggest a positive correlation between PANX1 expression and neutrophil in ltration in TNBC. The coexpression of PANX1 and myeloperoxidase (MPO) in TNBC para n-embedded surgical specimens was observed using immuno uorescence. PANX1 and MPO expression was positive correlated, indicating that high PANX1 expression promoted TAN in ltration in the TNBC TME (Fig. 3E). However, the way in which PANX1 establishes an immunosuppressive microenvironment with TANs should be further explored.

Immunosuppressive TANs demonstrated more in ltration in TNBC with high PANX1 expression
The quanTIseq and TIMER deconvolution methods have high deconvolution performance for RNA-seq data from different tumor types and can be suitable tools for further exploration of tumor-in ltrating neutrophils (24). RNA-seq data from fresh surgical specimens were converted into in ltrating immune cell information by quanTIseq and TIMER. The results revealed that TNBC with high PANX1 expression had more in ltrating TANs than TNBC with low PANX1 expression and the Luminal subtype (p < 0.05 for TIMER; p = 0.106 for quanTIseq) (Fig. 4A-C). The relationship between ENTPD1/NT5E expression and the in ltration level of TANs in the basal-like subtype was evaluated by TIMER, and ENTPD1/NT5E expression was positively correlated with the in ltration level of TANs (p < 0.01) (Fig. 4D). Neutrophils were puri ed from fresh TNBC surgical specimens and paired peripheral blood samples, and the transcriptome analysis of TANs revealed higher expression of the nucleotidases ENTPD1 and NT5E (p < 0.05) and high expression of the immunosuppressive cell recruitment-related cytokines CCL2 and CCL17 (p < 0.05) compared with the peripheral blood (Fig. 4E). The above results suggest that PANX1 promotes the recruitment of TANs in TME, and TANs have the ability to convert exATP to exADO with highly expressing ENTPD1 and NT5E.

High PANX1 expression induced a high exADO immunosuppressive TME in TNBC
To clarify whether high PANX1 expression in TNBC could establish an immunosuppressive TME with local high exADO levels, we measured the exATP and exADO levels in the supernatant of cultured breast cancer cell lines and found that exATP and exADO levels in the MDA-MB-231 and HCC-1937 (basal-like subtype cell lines) cell culture media were signi cantly higher than those in the MCF-7 (Luminal subtype cell line) cell culture media (p < 0.05) (Fig. 5A). PANX1 knock down in the MDA-MB-231 and HCC-1937 cell lines led to a downregulation of the exATP and exADO levels in the cell culture media (p < 0.01) (Fig. 5A).
The exATP and exADO levels in the digested tissue supernatant of TNBC surgical samples were signi cantly higher than those in Luminal A breast cancer surgical samples (n = 9) (p < 0.05) (Fig. 5B). The correlation between PANX1 expression and TIIC in ltration in the TCGA-BRCA basal-like subtype was analyzed by TIMER. The results indicated that PANX1 expression was positively correlated with the in ltration level of regulatory T cells (Tregs), M2-like macrophages and myeloid-derived suppressor cells (MDSCs) (p < 0.05) and was negatively correlated with the in ltration level of CD8 + T cells and NK cells (p < 0.05) (Fig. 4C). The TNBC single-cell transcriptome data were analyzed (23) and it was found that tumor samples with high PANX1 expression had lower in ltration of B cells, CD4 + T cells, CD8 + T cells, myeloid cells, NK cells and NK T cells and higher in ltration of cancer-associated broblasts (CAFs), plasma cells and Tregs (Fig. 4D-E). The above results suggested that PANX1 is a key gene responsible for exADO accumulation and establishment of an immunosuppressive TME in TNBC.

Discussion
This study suggests that PANX1 is highly expressed in TNBC and is a prognostic factor associated with poor prognosis in TNBC. PANX1 promotes TAN in ltration by increasing exATP levels in TNBC, and TANs highly express ENTPD1/NT5E, which synergistically contributes to an immunosuppressive environment with high exADO levels and promotes TNBC progression.
Malignant tumors are known as "never healing wounds", and chronic in ammation is one of the key features of malignancy. Chronic in ammatory processes are involved in tumorigenesis and tumor progression. Purine nucleosides (ATP and ADO) exert a strong immunomodulatory ability in the TME, and exATP/exADO can regulate local immune responses by activating immune cell purinergic P2 receptors. Previous studies have focused on the role of the nucleotidases ENTPD1 and NT5E in exADO production and the establishment of an immunosuppressive microenvironment (25,26). However, the source and regulatory mechanism of exATP in the TME, which is the substrate for exADO, has not been as well studied.
Stewart et al. found that PANX1 expression was required for breast development during lactation and that high PANX1 expression was associated with worse clinical outcomes in breast cancer (27). In other malignant tumors, PANX1 exhibited procarcinogenic effects in hepatocellular carcinoma (28), testicular carcinoma (29) and melanoma (30) and anticarcinogenic effects in glioma (31) and rhabdomyosarcoma (32). The above results suggest that the role of PANX1 in tumorigenesis might be complex and might depend on the tumor type and stage. In this study, we investigated the effect of PANX1 in TNBC primary lesions. We found that PANX1 was highly expressed in TNBC and could increase the exATP level in the TME and that high PANX1 expression was associated with poor prognosis in TNBC. These prognosis results are consistent with those in previous breast cancer studies (29,33). However, previous studies have primarily focused on the effects of PANX1 on breast cancer cells (33), and the role of PANX1 in the formation of the immunosuppressive microenvironment has not been fully explored. In vitro studies had shown that PANX1 promoted the activation of NLRP3 (NOD-, LRR-and pyrin domain-containing protein 3) in ammasome and increased the level of interleukin 1β (IL-1β) in the local microenvironment (34).
In this study, we found that the abundance of in ltrating neutrophils was signi cantly higher in the PANX1 high expression TNBC and that the genes coexpressed with PANX1 were related to granulocyte chemotaxis. In the early stage of tumor development, as the rst immune cells to enter the tumor microenvironment, neutrophils mediate subsequent immune responses and regulatory processes (35,36). The exATP secreted by PANX1 is considered to be an important damage-associated molecular pattern (DAMP) signal (37) that could lead to neutrophil recruitment (8). exATP could exacerbate the local immune response by mediating NLRP3 in ammasome activation and IL-1β secretion via the P2Y7 receptor (P2X7R) on neutrophils (38). exATP could also delay neutrophil apoptosis via the P2Y11 receptor (P2Y11R) (39). This study also demonstrated that TANs expressing high levels of ENTPD1/NT5E could promote the hydrolysis of exATP to exADO to aggravate the immunosuppressive TME. This is consistent with the results of studies of in ammatory states (40,41). Chen et al. demonstrated that neutrophils were chemotactic to exATP and hydrolyzed exATP to exADO by NT5E to promote cell migration (41,42). In addition, exADO inhibits neutrophil adhesion and the release of TNF-α and chemokines from LPSstimulated neutrophils (43). Neutrophil-expressed adenosine A2A receptor (A2AR) could inhibit the neutrophil recruitment cascade (44). Previous studies also revealed that the purinergic receptor P2Y6 receptor (P2Y6R) (45), adenosine A2B receptor (A2BR) (46), and adenosine A3 receptor (A3R) (47) on neutrophils are involved in the regulation of neutrophil extracellular traps (NETs) in in ammatory states. Whether exATP/exADO can regulate NETs or the N1/N2-like subtype transition of TANs through purinergic receptors and thus affect the development and metastasis of breast cancer requires further investigation.
In this study, we demonstrated that ENTPD1 and NT5E expression was higher in the PANX1 high expression TNBC and PANX1 upregulated exADO levels in the TME. PANX1 is an important immunomodulator in the TME. When the results of single-cell transcriptome data analysis were combined with TCGA-BRCA data analysis in TNBC, we found that PANX1 expression was negatively correlated with the in ltration levels of CD8 + T cells and NK cells and was positively correlated with the in ltration levels of Tregs. The levels of CAFs were also higher in PANX1 high expression TNBC. A similar phenomenon has been observed during renal injury. Higher PANX1 expression caused an increase in exATP, which was further catabolized by ENTPD1 and then converted to adenosine by NT5E (48). PANX1 plays an important role in the maintenance of chronic in ammation. Targeting PANX1 could inhibit tumorassociated in ammatory responses and reverse the local immunosuppressive microenvironment. Therefore, in-depth investigations of PANX1 may help elucidate tumor-in ammation interactions.
In summary, the high expression of PANX1 in TNBC was a poor prognostic factor and positively correlated with the exADO level. PANX1 could promote TAN in ltration through exATP secretion, and the high expression of ENTPD1/NT5E in TANs could synergistically establish an immunosuppressive TME with high exADO levels. In this study, the relationship between high exATP/exADO levels and TANs was investigated to elucidate the properties of PANX1 and its ability to reshape the metabolicimmunosuppressive TME and provide new targets and strategies for TNBC treatment.

Availability of Data and Materials
The data from previously reported studies and datasets, which supporting this research, have been cited in the manuscript. All the raw data and results generated during and/or analyzed during the current study have been publicly deposited in the Zenodo repository and can be accessed at https://doi.org/10.5281/zenodo.5168915.