Targeting NANOS1 in triple-negative breast cancer: synergistic effects of digoxin and PD-1 inhibitors in modulating the tumor immune microenvironment

Taraxacum officinale (Product Standard No. GHT1091), was purchased from Liaoning Senkangyuan Ecological Agriculture Co. The plant material was refluxed with 75% ethanol for 3 hours at 60°C, followed by filtration and concentration. This process was repeated three times, and the combined extract was purified using a D-101 macroporous resin column, where sugars were removed by water elution. The ethanol leaching fraction was collected, evaporated, and spray-dried to obtain Taraxacum officinale extract (TOE).

PD-1inhibitor (PD-1) and Digoxin (Dig) were purchased from Selleckchem. Algestone acetophenide (AA) was purchased from TOPSCIENCE. All chemicals were dissolved in DMSO for the in vivo and in vitro studies.

Dissolve 10 g of tribromoethanol (Sigma) in 10 mL of tert-amyl alcohol (Sigma) at room temperature to prepare a crystal-free stock solution. Filter the stock solution using a 0.22 μm filter (Millex-GP SLGPR33RB, Millipore), then dilute it with physiological saline (National Drug Approval No. H11021190; 0.9 g/100 mL) to prepare a 19.2 mg/mL (2.5% avertin) working solution. Incubate the working solution at 37°C for 4 hours, mix thoroughly by shaking, and store at 4°C protected from light. To establish a mammary fat pad tumor model in mice, intraperitoneally inject 2.5% avertin at a dose of 160 μL/10 g body weight (307 mg/kg) for anesthesia. Once fully anesthetized, the mice were placed in a supine position on a sterile workbench, and their skin was disinfected with povidone-iodine. Make a small incision (~0.5 cm) above the fourth right mammary gland using ophthalmic scissors. Carefully lift the skin with a cotton swab to expose the mammary fat pad, then inject 40 μL of serum-free suspension containing 5 × 10^5 4T1 cells into the fat pad. Mice in the monotherapy groups received intraperitoneal injections of Dig (5 mg/kg) daily or AA (10 mg/kg) every two days, and the combination group also received a 5 mg/kg PD-1 inhibitor every two days for 14 days. Tumor volume was measured using calipers, and the formula used was: volume (mm^3) = [width^2 (mm^2) × length (mm)]/2. On the 15th day of treatment, the mice were euthanized with 2.5% avertin anesthesia. Freshly isolated tumors were rapidly dissociated to obtain viable single cells. Tumor-infiltrating cells were analyzed by fluorescence-activated cell sorting (FACS) to assess antitumor immune responses.

The 4T1 and MDA-MB-231 cell lines were obtained from the Procell Life Science & Technology Co., Ltd. MDA-MB-231 cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Procell). 4T1 cells were cultured in RPMI-1640 supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Procell). Cells were cultured at 37°C with a 5% CO2 atmosphere.

MDA-MB-231 or 4T1 cells were seeded in a 96‐well plate at 2.0×10^5 cells/mL and cultured in medium at 37°C with a 5% CO2 atmosphere for 24 h. After that, the cells were pretreated with various concentrations of TOE for 24 h. At the end of the stimulation, the medium was removed, and 20 µL of CCK-8 reagent (Elabscience, China) was added to each well, and the incubation was continued for 1 h. The absorbance of each well was measured at 450 nm.

The migration and invasion abilities of cells were assessed using Transwell chambers (Corning). For invasion assays, 30 µg of Matrigel matrix (Corning) was added to the upper chamber and incubated at 37°C for 1 hour. For migration assays, no Matrigel was added. A total of 200 μL of cells (2.5×10^5/mL, treated with TOE, Dig, or AA at IC50 concentration in serum-free medium) was added to the upper chamber, and 750 μL of serum-containing medium was added to the lower chamber. The cells were incubated for 20 hours at 37°C in a 5% CO2 atmosphere. After incubation, the medium was removed, and the chambers were washed twice with PBS. The cells were fixed with 3.7% formaldehyde in PBS for 15 minutes, washed twice with PBS, stained with Giemsa dye for 10 minutes, and the remaining cells on the upper surface of the membrane were gently wiped away using a cotton swab. The number of migrated cells was then counted.

MDA-MB-231 cells were seeded in 10 cm dishes and grown to 90% confluency. 9 mL of complete media or TOE (at a concentration of IC50) were added to the control and treatment groups, respectively. After incubated at 37°C with a 5% CO2 atmosphere for 24 h, the medium was removed and washed twice with PBS. Cells were collected in Trizol (Invitrogen). RNA quality control, library construction, and sequencing were performed by Biomarker Technologies.

MDA-MB-231 cells were seeded into 10 cm culture dishes and incubated until they reached 90% confluence. The knockdown efficiency of siNANOS1#1 was validated by qPCR. The control and treatment groups were transfected with either siNC or siNANOS1#1, respectively, using Lipofectamine 3000 (Thermo Fisher). The cells were incubated at 37°C in a 5% CO2 atmosphere for 48 hours, after which the medium was removed, and the cells were washed twice with PBS. Cells were harvested in TRIzol (Invitrogen). RNA quality control, library construction, and sequencing were performed by GeneWiz Inc.

Gene expression data were analyzed using the bc-GenExMiner v4.8 platform (http://bcgenex.ico.unicancer.fr/↗) (20), which integrates publicly available breast cancer transcriptomic data, including 11,359 DNA microarray samples and 4,421 RNA-seq samples. These datasets include healthy breast tissue, adjacent normal tissue, tumor tissue, and various subtypes of breast cancer. The expression of the NANOS1 gene was compared across these tissue types, including healthy tissue, tumor tissue, and triple-negative breast cancer (TNBC) subtypes. Statistical analysis was performed using t-tests or ANOVA to assess the significance of gene expression differences, with a p-value of less than 0.05 considered statistically significant. Gene expression variations were visualized using box plots to provide a clearer representation of differences between groups. Additionally, the platform enables correlation analysis between gene expression data and clinical characteristics, further exploring potential associations between gene expression and clinical outcomes in breast cancer.

The Human Protein Atlas (THPA, https://www.proteinatlas.org/↗) (21) provided protein expression from the Tissue Atlas and Pathology Atlas. This platform combines gene expression, protein data, and clinical outcomes to investigate the relationship between NANOS1 expression levels and prognosis in breast cancer patients. The relationship between the NANOS1 gene and the prognosis of breast cancer patients was analyzed using the Kaplan-Meier Plotter online analysis tool (https://kmplot.com/analysis/↗) (22) which sources its databases from GEO, EGA, and TCGA. Kaplan-Meier survival analysis was performed to classify patients into high-expression and low-expression groups based on NANOS1 protein levels. Survival curves were generated and compared to evaluate the association between NANOS1 expression and patient survival. The log-rank test was used to assess statistical significance, with a p-value of less than 0.05 considered statistically significant. Through this analysis, The Human Protein Atlas provides valuable insights into the potential of NANOS1 as a prognostic biomarker in breast cancer.

TIMER2.0 (http://timer.cistrome.org/↗) was used to assess immune cell infiltration in various cancer types based on deconvolution algorithms (23). The platform allows for the evaluation of the infiltration levels of different immune cells in the tumor microenvironment and correlates these with gene expression and clinical data. In this study, we analyzed the correlation between the expression of the NANOS1 gene and immune cell infiltration in tumor samples from The Cancer Genome Atlas (TCGA). Additionally, TIMER2.0 was used to explore the relationship between immune infiltration and clinical outcomes, such as patient survival and tumor stage. The tool supports analysis across multiple cancer types, providing insights into immune response patterns and the potential role of NANOS1 in immune evasion and immunotherapy.

The structure of NANOS1 protein was obtained from the PDB database with PDB number 4CQO (24). Hydrogen atoms were added using UCSF Chimera software and protonated states were assigned using the H++3.0 program (25). SiteMap (26) was then used to predict the optimal small molecule binding site.

AutoDock Vina1.2.0 software (27) was used for high-throughput virtual screening, and the active site predicted by SiteMap was set as a docking center with docking center coordinates X,Y and Z of 14.25, -5.6, and 50.11, respectively, and the box size was set as a square with a side length of 22.5 Å. The conformation is sampled and scored using a genetic algorithm, and the best conformation is selected by ranking the conformations according to the docking score.

3177 molecules from the ZINC database were used as screening targets in this study, and they are all drug molecules approved for various applications. Some potential drug molecules that may bind to the NANOS1 protein were obtained after virtual screening by AutoDock Vina1.2.0 software using the above binding pocket as the docking site.

Hematoxylin and eosin (H&E) staining was performed to evaluate tissue morphology. Briefly, tumor tissues were collected and fixed in 4% paraformaldehyde overnight at 4°C. The samples were then dehydrated through a graded series of ethanol (70%, 85%, 95%, and 100%) and embedded in paraffin. Thin tissue sections (5 μm) were cut and mounted onto glass slides. The sections were deparaffinized and rehydrated before staining with hematoxylin for 5 minutes and eosin for 3 minutes. After washing and dehydration, the slides were mounted with neutral balsam and observed under a light microscope.

The tumor dissociation procedure was performed to obtain a single-cell suspension for subsequent flow cytometry analysis. Mouse tumor tissues were first excised and placed into a culture dish containing PBS (Procell). The tissues were minced using scissors and transferred into a new culture dish containing enzymatic digestion solution [Collagenase IV 2 mg/mL, DNase I 0.1 mg/mL, RPMI-1640 medium (Procell)] and incubated at 37°C for 1 hour, with gentle shaking every 15 minutes to facilitate complete digestion. Collagenase IV (Thermo Fisher Scientific) assisted in the breakdown of the extracellular matrix, while DNase I (Thermo Fisher Scientific) aided in the removal of nuclear debris. After digestion, the tissue was filtered through a 70 μm cell strainer to eliminate undigested tissue chunks and impurities. The filtered cell suspension was centrifuged (300×g for 5 minutes) to collect the cell pellet. The pellet was resuspended in red blood cell lysis solution (Biyuntian Biotechnology) and incubated at room temperature for 5 minutes with gentle shaking to ensure uniform lysis. The suspension was washed twice with PBS (Procell) to remove any residual lysis solution and cell debris. Finally, the cell pellet was resuspended in PBS, resulting in a single-cell suspension suitable for flow cytometry analysis.

Small interfering RNA (siRNA) oligonucleotides targeting NANOS1 were provided by Xianghong Biotech Co., Ltd. (Beijing, China) (Supplementary Table S1). Each siRNA was transfected using Lipofectamine 3000 (Invitrogen). The specific sequences of the target gene are provided in the supplementary information.

Total RNA was extracted using Trizol reagent (Invitrogen) and reverse transcribed using the PrimeScript™ RT reagent kit (Takara). Quantitative real-time PCR (qRT-PCR) was performed in triplicate using TB Green Premix Ex Taq (Takara). The primer sequences used for qRT-PCR of the target genes are listed in. 1

Flow cytometry was used to analyze the immune cell populations within the tumor microenvironment. The single-cell suspension was aliquoted into two tubes for different fluorescent antibody labeling. One tube was labeled with F4/80, CD11b, CD45.2 antibodies, while the other tube was labeled with CD3e, CD4, CD8a, CD279 (PD-1), CD223 (LAG-3), and CD366 (TIM3) antibodies, all obtained from Thermo Fisher Scientific. The cells were incubated on ice for 30 minutes, allowing surface antigens on the cell membranes to bind to the respective fluorescently labeled antibodies. Following incubation, the cells were washed twice with PBS (Procell) to remove unbound antibodies. The cells were then fixed with 4% paraformaldehyde for 15 minutes, followed by two PBS washes. After all staining and fixation steps, the cells were washed again and resuspended in PBS for flow cytometry analysis. Antibodies were purchased from Thermo Fisher Scientific, and their specific details are provided in. 1

Flow cytometry data were acquired using a BD FACSCalibur flow cytometer. Initial gating was performed using forward scatter (FSC) and side scatter (SSC) plots. Immune phenotype analysis was conducted using multiple fluorescence channels. The flow cytometry data were analyzed with FlowJo software, and FSC and SSC gating were used to select the appropriate cell populations. Immune phenotype analysis was based on distinct fluorescence markers. The gating strategy is outlined in. 1

Cells were treated with Dig and AA for 48 hours, then lysed on ice using RIPA buffer (Solarbio) containing 1% PMSF. Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Equal amounts of protein (20 µg) were separated by 10% SDS-PAGE (Vazyme), transferred to a PVDF membrane (Millipore), and blocked with 5% nonfat milk at room temperature. The membrane was incubated overnight with primary antibody at 4°C, washed with TBST, and incubated with secondary antibody at room temperature. Protein signals were detected using a high-sensitivity ECL chemiluminescent detection kit (Proteintech) on the BG-gdsAUTO730. Anti-NOS-1 was purchased from abcam (working concentration: 2 µg/mL), Beta Actin Polyclonal antibody from ProteinTech Group, Inc. (1:2000 dilution), and Multi-rAb HRP-Goat Anti-Rabbit Recombinant Secondary Antibody (H+L) from ProteinTech Group, Inc. (1:5000 dilution).

Data were collected from independent experiments and expressed as the mean ± SEM. All graphs and analyses were generated using GraphPad Prism software. The significance among multiple (three or more) groups was compared using one-way ANOVA analysis, and differences between the different groups were analyzed by a Student's t-test.

As shown in Figure 1A, TOE significantly inhibited the growth of MDA-MB-231 and 4T1 cells, with the inhibitory effect increasing in a concentration-dependent manner. Additionally, TOE reduced the migration and invasion abilities of these cells (Figure 1B). MDA-MB-231 cells were treated with an IC50 dose of TOE (635.4 μg/ml), and RNA sequencing (RNA-seq) was performed to assess their transcriptomes. A total of 8451 differentially expressed genes (DEGs) were identified following 24 hours of treatment with TOE, with 2,020 genes upregulated and 2,048 genes downregulated (Figure 1C). The top 600 upregulated and downregulated genes were selected for KEGG pathway enrichment analysis based on the |log2(FC)| values, and the top 20 pathways with the lowest p-values are shown in Figure 1D. The five most significantly enriched pathways included pathways in cancer, the MAPK signaling pathway, transcriptional misregulation in cancer, microRNA in cancer, and the cell cycle.

Kaplan-Meier survival analysis was performed to assess the prognostic significance of the top 600 upregulated and downregulated differentially expressed genes (DEGs). Genes positively correlated with survival were identified, focusing on the low expression of downregulated genes and the high expression of upregulated genes. A total of 132, 269, and 179 genes were found to be positively associated with overall survival (OS), relapse-free survival (RFS), and distant metastasis-free survival (DMFS), respectively. Subsequently, 640 prognostic genes were selected based on TCGA data and antibody-based protein data. Of these, 209 genes were associated with an unfavorable prognosis, while 431 genes were linked to a favorable prognosis for breast cancer at the protein level. Ultimately, 8 genes were identified as being associated with prognosis at both the transcript and protein levels (Figure 2A, Table 1). Among these, NANOS1, a less studied and the only downregulated gene, was chosen for further analysis. The prognostic value of NANOS1 protein expression in breast cancer was evaluated using the online tool available at www.proteinatlas.org↗, and the results demonstrated that lower expression of NANOS1 was associated with better prognosis (p < 0.001) (Figure 2B). At the mRNA level, reduced NANOS1 expression was also positively correlated with OS (p < 0.001), RFS (p < 0.001), and DMFS (p < 0.001) in breast cancer patients (Figures 2C–E).

The correlation between NANOS1 expression and immune cell infiltration levels in BRCA-BASAL was assessed, given the association of immune infiltration with cancer development and treatment outcomes. Based on deconvolutional procedure, NANOS1 expression was positively associated with Macrophage (P=8.20e-04), CD4+ T cell (non-regulatory) (P=2.30e-04) and Endothelial cell (P=3.09e-02), meanwhile, NANOS1 expression was negatively associated with B cell (P=7.14e-03),Myeloid dendritic cell (P=3.72e-02), activated NK cell (P=5.44e-03), and T cell follicular helper (P=2.35e-03) (Figure 2F). Among these immune cells, NK cell infiltration was associated with a reduced risk of TNBC, and survival analysis revealed that high NK cell infiltration correlated with better prognosis in patients with low NANOS1 expression (P=0.0216) (Figure 2G). Furthermore, the level of NK cell infiltration in the somatic copy number amplifications (sCNA-Amplifications) state of NANOS1 was demonstrated by violin plots. As shown in Figure 2H, a significant difference in NK cell infiltration was observed between high amplification samples and normal samples (p=0.032).

Typically, targetable proteins need to have a typical binding pocket, so the first step in this study was to use the SimteMap tool to analyze the targetable binding pocket on the surface of the NANOS1 protein. By calculation, SiteMap found a large potential ligand binding pocket consisting of the following amino acids: ALA2076, PRO2077, ARG2080, CYS2118, ASP2119, VAL2120, ILE2121, PRO2122, PRO2123, ASN2124, ARG2129, GLY2207, ASN2208, ARG2209, TYR2210, ASN2211, LEU2212, GLN2213, ASN2216, GLU2261, LEU2265, ASN2268, ALA2269, ASN2272, ARG2311. This indicates the presence of a targetable binding pocket for NANOS1 (Figures 3A, B). To explore potential interactions between approved drugs and the NANOS1 protein, we specifically selected drugs from the ZINC database for virtual docking simulations. The ZINC database was chosen due to its extensive collection of commercially available, drug-like molecules, as well as its provision of detailed compound information, including structural data and molecular properties, which are essential for accurate molecular docking. This approach allowed us to identify candidate drugs that may bind effectively to NANOS1. Next, the binding affinity of FDA-approved drugs was predicted using AutoDock Vina. A total of 3177 drug molecules were ranked based on their docking scores, from highest to lowest. The results indicated that Dig and AA both achieved the highest scores, with a docking score of -9.9 kcal/mol for each drug (a larger absolute value indicates stronger binding affinity) (Table 2). The binding modes were then visualized. Figures 3C, D show the interaction between Dig and the NANOS1 protein, where Dig forms five hydrogen bonds with the protein. This interaction involves hydroxyl groups as hydrogen bond donors and acceptors, and a hydrogen bond is formed between the carbonyl group of the terminal cyclic ether and Arg2209. No π-stacking or salt bridge interactions were observed. Additionally, Dig exhibits significant hydrophobic interactions with the protein due to its hydrophobic backbone. Figures 3E, F illustrate the interaction between AA and the NANOS1 protein. AA forms two hydrogen bonds, one H-π stacking interaction, and hydrophobic interactions with the protein. The hydrogen bonds occur between the side chain of Arg2080 and the oxygen atom on the furan ring, as well as between the oxygen atom on the carbonyl group and the amide backbone of Asn2124. The π-stacking interaction occurs between the charge center of Asn2208 and the phenyl ring at the terminal end of AA. Additionally, AA interacts with several hydrophobic amino acids such as Pro2122, Pro2123, Leu2265, and Leu2212.

Dig and AA were selected for in vitro validation. CCK-8 assays showed that both drugs significantly reduced cell viability, inhibited proliferation, and slowed growth (Figure 4A). The 24-hour IC50 values for Dig were 0.6806 µM and 1.162 µM for MDA-MB-231 and 4T1 cells, respectively, and for AA, 23.25 µM and 21.63 µM. Transwell assays demonstrated that Dig and AA significantly reduced the migration and invasion of MDA-MB-231 and 4T1 cells (Figures 4B, C). To assess the potential of Dig and AA in inhibiting TNBC tumor growth in vivo, a 4T1 mouse tumor model was established. Mice were randomly divided into six groups, receiving different treatments: PD-1 inhibitor monotherapy (5 mg/kg), Dig (5 mg/kg), AA (10 mg/kg), Dig + PD-1 inhibitor combination, AA + PD-1 inhibitor combination, and a saline control. Both Dig and AA significantly inhibited tumor growth compared to the control group (Figure 4D). Moreover, the combination of Dig and PD-1 inhibitor showed superior tumor suppression compared to PD-1 inhibitor alone, highlighting the potential of this combination as a therapeutic strategy for TNBC (Figure 4E). No significant weight changes were observed in the combination treatment groups (Figure 4F).

Freshly excised tumor tissues were harvested; one portion was used for flow cytometry analysis to assess immune cell populations, while the remaining portion was processed for H&E staining to examine tissue morphology and immune cell infiltration (Figure 5A). To evaluate the potential of combination therapy in promoting tumor lymphocyte infiltration and modulating the immunosuppressive environment within 4T1 tumors, several parameters were assessed. The combination of Dig and PD-1 significantly increased the frequency of tumor-associated macrophages compared to the control group (Figures 5B, G). Evaluation of effector T cells revealed that the combination therapy of Dig and PD-1 resulted in a significant increase in the frequency of effector CD8+ T cells compared to the saline group, and this effect was superior to the PD-1 monotherapy group. AA enhances the proportion of CD8+ T cells. However, although the AA + PD-1 inhibitor combination showed a more pronounced increase in CD8+ T cells, the lack of statistical significance (adjusted p value = 0.2576) indicates that the effect of the AA and PD-1 inhibitor combination is less effective than that of the Dig and PD-1 inhibitor combination (Figures 5C, H). These findings indicate that the combination therapy involving Dig effectively enhances the immune response to immune checkpoint blockade (ICB) therapy.

This study also investigated the effects of Dig and AA, either as monotherapies or in combination with PD-1 inhibitors, on exhausted T cells. The results showed that Dig, either alone or in combination with PD-1 inhibitors, effectively reduced the frequency of LAG-3-expressing exhausted T cells (Figures 5D, I). The combination of Dig and PD-1 exhibited synergistic effects in reducing TIM-3 expression levels (Figures 5E, J). Dig monotherapy also effectively reduced the frequency of PD-1-positive exhausted T cells (Figures 5F, K). The analysis of exhausted T cells revealed a concerning trend in the AA + PD-1 inhibitor group, where the proportions of LAG-3+, TIM-3+, and PD-1+ exhausted T cells were higher than those in the AA alone group. The adjusted p-values of 0.5676, 0.8546, and 0.9632 indicate an increase in T cell exhaustion with the AA + PD-1 combination, which may account for the diminished efficacy observed with this combination. This suggests that while AA alone exhibits significant antitumor activity, its combination with PD-1 inhibitors may induce a state of T cell exhaustion, potentially limiting the overall therapeutic benefit. These findings provide strong experimental evidence for the development of novel tumor immunotherapy strategies. The "drug repurposing" approach has emerged as a new paradigm in antitumor drug discovery, offering the advantage of bypassing established toxicological and pharmacokinetic evaluations. Furthermore, this study demonstrates that modulating the tumor microenvironment, as evidenced by increased immune cell infiltration in the tumor, can enhance the efficacy of existing immunotherapies, significantly reducing development time and research costs.

To investigate the role of NANOS1, we silenced its expression using siNANOS1#1 in MDA-MB-231. The silencing of NANOS1 was confirmed by quantitative PCR analysis. Differential gene expression analyses identified a notable number of genes exhibiting upregulation and downregulation (Figure 6A). Among these, IL6 was identified as the most significantly differential gene, with an adjusted p-value of 4.67E-191. A differential gene expression heatmap was generated to visualize these changes (Figure 6B). These 77 differentially expressed genes were then subjected to GO and KEGG pathway analyses to explore their biological functions and relevant pathways (Figures 6C, D). KEGG biological pathway analysis showed gene enrichment on pathways such as the TNF-signal pathway. In the triple-negative breast cancer cell lines MDA-MB-231 and 4T1, siRNA interference, Dig, and AA treatments effectively downregulated the expression of NANOS1 (Figure 6E). Furthermore, Dig and AA treatments reduced the protein expression of NANOS1 in TNBC cell lines (Figure 6F). We discuss the role of TNF-α in promoting tumor cell vascular adhesion and supporting angiogenesis. Previous studies have shown that TNF-α enhances tumor metastasis by upregulating the expression of adhesion molecules and stimulating the expression of angiogenesis factors (28 –30). Our experimental results indicate that siRNA interference, Dig, and AA treatments all suppress TNF-α gene expression, which is consistent with previous findings. Transwell assays revealed that siRNA-mediated silencing of NANOS1 expression significantly reduced the migration and invasion abilities of MDA-MB-231 and 4T1 cells (Figure 6G).