IFI44 Orchestrates an IL-10-Driven M2 Macrophage Program in Breast Cancer: An Immune Prognostic Signature.

Introduction
BC poses a complex health challenge, marked by its prevalence and high fatality rates, to women’s wellbeing across the world,1 and is predicted to occur in 1 in 8 women, the leading cause of mortality among females.2 Conventional therapeutic method comprising operations; chemical treatments; radiation; hormone treatment; and precision medicine, although there have been notable improvements in BC therapy, these treatments are often linked to severe side effects, and death rate is still high.3,4 In this situation, immunotherapies and targeted treatments with greater effectiveness and precision have emerged as promising strategies to fight BC. Some new immune checkpoints such as LAG-3, TIM-3, TIGIT, as well as signaling pathways such as JAK/STAT, MAPK, PI3K/AKT/MTOR,5 have solved the problem of low response rate and drug resistance, but the specific mechanism of action is still unclear. Therefore, in order to find more effective immunotherapies, it is crucial to uncover novel and robust predictive biomarkers and explore the underlying mechanisms.
Interferon-induced protein 44 (IFI44) activated by type I interferons such as IFNα and IFNβ,6 was firstly identified as a hepatitis C virus-related microtubule-clustering protein extracted from the liver cells of infected chimpanzees7 and thought to be included in the immune response and to function upstream or downstream of the bacterial presence. In addition, the OS of patients with high levels of IFI44 was significantly lower than in those with low levels of IFI44 in HNSC.8 Studies conducted have revealed that IFI44 is an important factor in immune diseases.9–11 IFI44 contributes to immune cell recruitment, highlighting its potential as a therapeutic target. Yet, there is a lack of knowledge concerning on IFI44’s role in invasion of immune cells in BC.
Macrophages are key components of the immune microenvironment in BC and significantly contribute to tumor progression by promoting aggressive cell behaviors in different cancers.12,13 In solid tumor studies, tumor associated macrophages (TAMS) support the release of CSF-1 from tumors and Epidermal Growth Factor (EGF) from macrophages via paracrine signaling involving invading tumor cells.14–16 The functions of macrophages are flexible, allowing them to transition from an M1 to an M2 polarization state to meet different physiological needs. M2 macrophages secrete type II cytokines, promoting the development of inflammation and tumorigenesis.17,18 There is a need to enhance the indicators that facilitate identification of M2 macrophages phenotypes and to predict the prognosis in BC.
In this study, we used bioinformatics tools including a variety of databases to explore the prognostic significance of IFI44, to study the association of IFI44 with immune cells and to construct the significance of IMS. The ability of IFI44 to proliferate, migrate and invade was evaluated using cell viability, colony-formation assays and transwell tests in vitro, which provided evidence that IFI44 could enhance the expression of IL-10 and thereby enhance the infiltration of M2 macrophages. Survival curve, Receiver Operational Characteristic (ROC) and Pan Cancer Survival Curve were employed to verify the effectiveness of IMS in evaluating prognosis and achieving a good response in multiple validation datasets. Therefore, we demonstrated that IFI44 may be a potential immunotherapy target and that the IMS may help forecast chemotherapy response and outcomes in BC.

Materials and Methods

Acquisition and Processing of Data
Expression profiles containing 337 samples from GSE47994,100 samples from GSE36772 and 86 samples from GSE15852 were downloaded. Clinical profiling and corresponding characteristics for 2509 BC instances (METABRIC) were retrieved from cBioPortal. RNA expression profiling data for clinical and related genes for pan-cancer was obtained using the “Bioconductor” package in The Cancer Genome Atlas database (TCGA) biolinks (RNA SeqV2).19 IFI44 expression in normal and breast cancer tissues was analyzed using immunohistochemistry data from the HPA database. The expression of IFI44 in multiple cell lines was obtained from the HPA database. Correlation between the IFI44 and IL-10 was explored by TIMER and GEPIA2.

Assessment of Immune Infiltration
Targeted therapy in cancer is closely related to the immune microenvironment.20,21 The ESTIMATE algorithm (estimation) was employed to score the purity of the matrix, immunity, and tumor.22
CIBERSORT was applied to estimate the relative fractions of immune cell types in GSE47994 and GSE36772.23 The association between IFI44 levels and the infiltration of key immune cell types in GSE47994 and GSE36772 was evaluated using Spearman correlation.

Correlations Between IFI44 and M2 Marker Genes
Genes with connection to M2 macrophages infiltration were dug by the R packages “limma” and “tidyverse” with a significance level of P < 0.01 and an absolute R > 0.2 value above the defined threshold to look for marker genes.24 The threshold of absolute R > 0.2 was chosen to include genes showing at least a weak-to-moderate association with M2 macrophage infiltration. The more stringent P-value threshold of <0.01 was applied to increase confidence in the selected markers. Then, Spearman’s coefficient was employed to conduct expression correlation assays between IFI44 and M2 marker genes in the TCGA cohort.

Cell Culture and Transfections
To select appropriate cellular models for functional validation, we firstly examined the basal expression of IFI44 across breast cancer cell lines using the HPA database. This led us to choose HCC70 (high endogenous expression) and MDA-MB-231 (low endogenous expression) for subsequent loss- and gain-of-function studies, respectively. The HCC70 BC cells and MDA-MB-231, which obtained from Chinese Academy of Sciences, were cultured in the RPMI-1640 and DMEM medium, respectively. Short hairpin RNA (shRNA) of IFI44 and over-expression of IFI44 (OE) plasmids were obtained from GenePharma, and the plasmids were separately transfected into HCC70 cells and MDA-MB-231 cellls. The shRNA had the following sequences: shIFI44-1:5′-GCGAAGATTCACTGGATGAAA-3′, and shIFI44-2: 5′- CCCTGTAAAGGATGTTCTAAT-3′. Following the manufacturer’s guidelines, shRNA transfection was carried out.25 MDA-MB-231 was transfected with IFI44 plasmid that overexpresses genes.

Real-Time Quantitative
To verify the transfection efficiency of our constructed cell lines, RT-qPCR was used. Trizol (TaKaRa) was used to extract RNA from the MDA-MB-231 and HCC70 cells. The extracted RNA was transcribed into cDNA via the PrimeScript™ II Reverse Transcriptase (Servicebio). PCR amplification of the IFI44 strictly following the manufacturer’s instructions (Servicebio). Primers were in the following manners: ACTIN, forward (F): 5′-CACCCAGCACAATGAAGATCAAGAT-3′, reverse (R): 5′-CCAGTTTTTAAATCCTGAGTCAAGC-3′; IFI44, forward (F): 5′-TCCAAGGGCATGTAACGCAT-3′, reverse (R): 5′-CCTCCCTTAGATTCCCTATTTGCT-3′. The levels of IFI44 mRNA were quantified relative to the ACTIN mRNA using the 2−ΔΔCq method.

Cell Viability and Colony Formation Assays
To detect cell viability, the Cell Counting Kit-8 (CCK-8) from Beyotime (#C0037) was employed based on the instructions given by manufacturer. After 12 h transfection, HCC70 and MDA-MB-231cells were put into 96 well plates at 2000 cells density. A microplate reader was used to record absorption at 450 nm. Over a period of four days, each day’s absorbance data is used to plot cell proliferation curves.26 In the colony formation assay, 500 cells were seeded and cultured in 6-well plates until colonies became visible. 4% formalin fixation and 1% crystal violet were then used to stain and count the colonies.

Transwell Assays
Migration assays were performed with chamber (5-mm pore size, Corning) and invasion assays were supplemented with Matrigel (BD Biosciences). Twelve hours post-transfection, 50,000 cells were placed into the upper chambers using 200μL of medium without serum, and lower chamber only added 600μL of 10% FBS medium. The upper insert cells were transplanted to the lower surface, subsequently treated with 4% formalin and colored using 1% crystal violet, following a 24-hour culture period at 37 °C. The average number of tumor cells was counted in randomly selected microscope fields.

M2 Macrophage Infiltration Assay
The innate immune system plays a fundamental role in tumor immune surveillance and response to immunotherapy, with macrophage polarization directly influencing the immune balance within the tumor microenvironment.27 THP-1, obtained from the Chinese Academy of Sciences, was seeded in 6-well plates using RPMI-1640 medium with 10% FBS. 100 ng/mL PMA was added for 12h to stimulate THP-1 differentiation to M0 macrophages when the density reached 5×105. M0 macrophages were polarized to M2 with 20ng/mL IL-4 (MedChemExpress) and 20ng/mL IL-13 (MedChemExpress) for 48h at 37°C.28 2.5×105 M2 macrophages that had been serum-free for a full 12 hours were seeded in chambers, and 2.5×105 MDA-MB-231 and HCC70 cells were planted in the lower chamber with 10% fetal bovine serum.29 After incubation for totally 24 hours, 5% CO2 concentration at 37°C, cells in the lower surface were fixed, stained and quantified.

Flow Cytometry
Cells collected from lower chamber of the Transwell assay are rinsed and incubated in the dark at 4°C for half an hour. The following anti-human antibodies were used: CD163-PE-Cy7(BD Biosciences); CD206-APC (BD Biosciences).30 The results of positive CD163/CD206 were analyzed using the FlowJo 10.10 software program.

Elisa
IL-10 is a marker for M2 macrophages, and we wonder whether the expression of IFI44 would synchronously affect the expression of IL-10. In this study, IL-10 was quantitatively detected by double-antibody sandwich ELISA (Abcam, ab185986) based on the instructions given by manufacturer.

Patient Tissue Collection
Breast cancer tissues were obtained from 10 individuals, who suffered surgical treatment at the Department of Breast Disease Center, The First Affiliated Hospital of Nanchang. This study’s clinical research protocol was supported from the Ethics Committee at Nanchang University, and the patients signed the informed consent. Tissues were swiftly frozen in liquid nitrogen and maintained at −80°C for later experiments.

Immunohistochemistry
IHC staining was used to assess the different levels of expression for IL-10 and IFI44 in clinical samples (n=10). Tissue sections were stained with 3, 3′-diaminobenzidine for signal detection. anti-IFI44 and anti-IL-10 antibodies were sourced from Proteintech.31 Two seasoned pathologists conducted an independent examination and quantification of the image of sections.

Establishment and Confirming a Predictive Signature in BC
BC patients were split into training (TCGA) and testing (METABRIC) cohorts. The TCGA group was employed to identify IFI44-related M2 marker genes with prognostic significance by univariate and multivariate cox survival analysis. Using LASSO Cox regression by R package “glmnet”, the final four genes were identified through multivariable cox analysis. The risk score=β1X1 + β2X2 +… + βiXi. The formula employs Xi to denote the gene expression value from the predictive model, utilizing Bi as its respective coefficient.32 Kaplan–Meier survival analysis categorized samples into different risk groups by median score. The prognostic model’s predictive efficacy was confirmed via the ROC curve analysis, executed with the R “Time Roc” package. Moreover, for evaluation of the risk-model, the METABRIC cohort evaluated signature prognostic performance with this formula. Therefore, we used forest plots to comprehensively evaluate the risk model’s precision by incorporating the nine clinical information: tumor size; lymph node states; age; menopausal state; ER; HER-2; Nottingham Prognostic Index (NPI); pathological grade and IMS score.

Susceptibility Analysis
Drug sensitivity testing utilized Cancer Drug Susceptibility Genomics 2 database (GDSC2)33 and the relationship between IMS and drug susceptibility was explored using the “onco Predict” R software package.

Statistical Analysis
Data are presented as mean ± standard deviation (SD) from at least three independent experiments, each performed in triplicate (technical replicates). Student’s t-tests were compared to groups with normal distribution, while the Wilcoxon rank-sum test assessed non-normal distributions, and visualize statistical results with GraphPad Prism Version 10. Via the application of cox regression, we meticulously examined the prognostic factors. Subsequently, leveraging the “survival” and “survminer” R packages, we crafted a robust prognostic model. The Kaplan-Meier curve was used for the analysis of survival data, and the association of clinical and pathological parameters with varying risk groups in BC patients was analyzed via the Chi-Square test. If the outcomes indicated that P was less than 0.05, it was considered statistically significant. *p < 0.05; **p < 0.01; ***p < 0.001 by Student’s t test.

Results

IFI44 Demonstrated Increased Expression in BC and Related with Poor Prognosis
Evidence indicated elevated expression of IFI44 in HNSC tumors,8 however, the expression of IFI44 in BC is unexploited. We examined the expression of IFI44 mRNA levels between tumor and normal samples in the TCGA and discovered that IFI44 was notably increased in several cancer types, including BRCA (Figure 1A). Then, our findings found that IFI44 demonstrated increased expression in paired BC tissues from TCGA and GSE15852 databases (Figure 1B and C). Immunohistochemistry from HPA showed that IFI44 was higher than that of normal tissues in BC tissues (Figure 1D). Thus, this study delved into the link between IFI44 expression and prognosis through the K-M plotter database, and we found that the prognosis of patients with increased expression of IFI44 was poor (Figure 1E).

IFI44 Promoted Proliferation, Migration, and Invasion of BC Cells
The role of IFI44 in breast cancer is unknown. In U87 and U251 cells, increased expression of IFI44 facilitated cell proliferation, movement, and invasiveness, whereas knockdown suppressed these phenotype.34 We firstly assessed IFI44 mRNA levels by The Human Protein Atlas in multiple cell lines and found that IFI44 was highly expressed in HCC70 cell line, but low in MDA-MB-231 cell line (Figure 2A). By employing two specific shRNAs, we decreased IFI44 levels in HCC70 BC cells, transfected with IFI44 gene overexpression plasmid in MDA-MB-231 and validated by qRT-PCR (Figure 2B). Based on CCK-8 (Figure 2C), colony formation (Figure 2D) and transwell assays (Figure 2E), it had been presented to inhibit the proliferation and invasion of BC cells in HCC70 cells. However, overexpression of IFI44 in MDA-MB-231 promoted the ability of BC cells in proliferation, invasion, and metastasis (Figure 2F). Together, we have demonstrated that IFI44 facilitated BC cells to proliferate, migrate, and invade.

The Expression of IFI44 Was Strongly Associated with Immune Infiltrating Cells
Since IFI44 induces antiviral responses that is widely reported to be predominantly expressed in virus, there is clinical interest in investigating the function of IFI44 in immune microenvironment. We performed immune infiltration analysis using CIBERSORT algorithm, thus exploring the potential association of IFI44 with immune cells in the GSE47994 and GSE36772 cohorts. In the GSE47994 cohort, we found that IFI44 was positively correlated with T cells CD8, Plasma cells, M2 Macrophage and M1 Macrophage, but negatively with T cells CD4 naive, NK cells activated, and monocytes (Figure 3A). The association between the IFI44 and immune cells was further evaluated in the GSE36772, and we found that the immune scores for M2 Macrophage and M1 Macrophage were increased in the group with elevated IFI44 than in the group with reduced IFI44 (Figure 3B). The trend of the results obtained by Spearman analysis is consistent (Supplementary Figure 1). Notably, M1 Macrophage and M2 Macrophage were discovered to be considerably more prevalent in the IFI44-high group in the GSE47994 and GSE36772 (Figure 3C). Thus, there may be interactions between the immune cell types and IFI44 that deserve future in-depth study.

High Expression of IFI44 in BC Promoted Infiltration of M2 Macrophages
Galectin-9 is highly expressed in the tumor microenvironment and binds to Tim-3 on the surface of macrophages, activating the PI3K/AKT signaling pathway, resulting in macrophage polarization to M2 type, which promotes tumor cell proliferation, movement, and invasiveness, thereby driving tumor development.35 In order to explore whether IFI44 exerts pro-tumor function by regulating the tumor immune microenvironment, we first assessed the relationship between its expression and M2 infiltration. TIMER2 and GEPIA2 database correlation studies confirmed a sustained positive link between IFI44 and four M2 markers (IL10, CD163, TGFB1, and CSF1R), among which IL-10 was the most relevant (Figure 4A and B). Subsequently, IHC staining was used to evaluate the different levels of expression for IL-10 and IFI44 in clinical samples. Results indicated the expression trends of IFI44 and IL-10 were consistent with bio-information in BC (Figure 4C). To verify whether IFI44 can directly regulate M2 macrophage recruitment, we conducted in vitro co-culture experiments. Knocking down IFI44 in HCC70 cells resulted in less infiltration of M2 macrophages, but transwell experiments showed that the IFI44-OE group contained a significantly greater number of cells penetrating chambers in MDA-MB-231 and HCC70 cells (Figure 4D–F). Co-culture with IFI44-overexpressing tumor cells significantly increased the percentage of M2 macrophages (CD206⁺, CD163⁺). In addition, silencing IFI44 impaired the tumor cells’ ability to promote M2 migration (Figure 4G). Elisa’s results indicated that reduction of IFI44 expression in HCC70 cells attenuated the expression of IL-10. However, overexpression of IFI44 in 231 cells increased IL-10 expression (Figure 4H). Early studies had demonstrated that IL-10 signaling promotes the motility of M2 macrophages.36 We speculated that IL-10 might contribute to the movement of M2 macrophages facilitated by IFI44. To verify the hypothesis, the number of M2 macrophages was examined in cells where IFI44 was silenced in HCC70 cell line and IFI44-OE in MDA-MB-231 cell line. As expected, low expression of IFI44 decreased the number the migration of M2 cells, but exogenous IL-10 application to the knockdown cells restored the diminished levels. Over-expression of IFI44 in MDA-MB-231 cells facilitated the movement of M2 macrophages; however, the addition of IL-10Ab inhibited the migration process (Figure 4I). Mechanistically, IFI44 orchestrates an IL-10–driven immunosuppressive microenvironment by enhancing M2 macrophage infiltration.

Construction of the IMS in the TCGA Set and Validation in the METABRIC Cohort
Considering the significant role of IFI44 and M2 macrophage in BC, it is important to formulate a model related to them to estimate the patients’ prognosis with BC. Co-expression analysis identified 210 M2 marker genes, following filter conditions with an absolute R above 0.2 and a p-value below 0.05 selected genes from the former 210 M2 marker genes, and identified the final 155 marker genes with a Pearson correlation coefficient’s absolute value exceeds 0.2, with a p-value under 0.05. Then univariate cox regression analysis was used to identify 42 maker genes about OS, which were shown in the Heat Map (Figure 5A). Finally, the last 7 candidate genes were found using multivariate Cox regression model analysis, and subsequently applied to generate a forecast signature (Figure 5B). The LASSO regression analysis involved 7 genes to further suggest that 4 genes (GPR171, KIR2DS4, NPAS1 and CD79A) were identified when the cross-validation error reached its minimum (lambda. min=0.0079) (Figure 5C and D). The prognostic risk score based on IFI44 M2 signatures (IMS) was derived via the application of the given formula:

This algorithm was used to calculate the risk in each BC patients and stratify the patients into various groups according to the median IMS risk score. This data demonstrated that there was a significantly higher mortality rate in high-risk patients compared to those at low risk (Figure 5E and F). The AUC of the ROC model for predicting the 1-year, 3-year, and 5-year survival of BC patients was 0.705, 0.698, and 0.614, respectively (Figure 5G). To test the validity of the model, we used a METEBIRC cohort to test its predictive performance, using a similar pattern to allocate patients to high- and low-risk cohorts. Consistent with the training cohort, patients in the high-risk group exhibited poorer OS than those in the low-risk group (Figure 5H). Thus, the outcomes of our research suggested that IMS risk score was highly specific and sensitive for predicting OS in BC patients.

The Clinical Value of IMS in BC
The IMS’s status as a standalone predictive indicator was evaluated via a multivariate cox regression analysis, with covariates including tumor size, lymph node states, pausal state, ER-IHC, HER2-SNP6, NPI, tumor grade and IMS score. The analysis revealed that IMS had statistical significance in BC (p ≤ 0.001) (Figure 6A), which further implied the IMS score model could act as an independent prognostic indicator for BC. Moreover, the value of IMS in the therapeutic effects of immunotherapy and chemotherapy in BC was explored. The ESTIMATE algorithm was utilized to derive the ESTIMATE score for every patient, indicating the comprehensive level of immune infiltration. In TCGA cohorts, low-IMS patients had elevated Stromal, Immune, and ESTIMATE scores relative to high-IMS patients (P<0.001, Figure 6B). Moreover, higher ESTIMATE scores might be associated with improved prognostic outcomes. We explored the association of IMS and the IC50 of certain anticancer drugs. As shown in Figure 6C, Cisplatin and Paclitaxel, which are frequently used anti-tumor drugs for TNBC in clinical settings, showed reduced effectiveness in high-IMS groups.

The Diagnostic Utility of This IMS with Pan-Cancer Analysis
To assess the value of IMS in Pan-cancer, we downed TCGA database’s transcriptome expression profiles paired with clinical information for multiple cancer categories and calculated each sample’s risk using a similar formula. To explore the prediction precision of the IMS in other cancers, a survival model was applied to investigate survival patterns across risk groups, with findings indicating notable differences among the stratifications illustrated by the IMS in four types (Bladder Urothelial Carcinoma (BLCA); HNSC; Skin Cutaneous Melanoma (SKCM); and Acute Myeloid Leukemia (LAML)) of cancer (p-value≤0.05) (Figure 7A and B). ESTIMATE’s evaluation of immune infiltration showed a stark disparity can be observed between the high-IMS and low-IMS clusters across all four types of cancer, where the low-IMS clusters exhibit less immune system involvement (Figure 7C). The findings align with BRCA observations. Consequently, the results suggest that IMS could potentially be used for other cancers.

Discussion
IFI44 is activated, and linked to infections by various viruses,6 including papillomavirus and influenza virus.37 Nevertheless, there remains insufficient evidence regarding the role of IFI44 in BC. Here, we found that IFI44 overexpression was associated with unfavorable outcomes, presented potential for modulating the immune system, and the IFI44-based IMS independently predicted prognosis, highlighting its potential as a prognostic indicator. Cellular viability, clonogenic and transwell invasion studies were implemented in vitro to ascertain the function of IFI44, revealing its ability to enhance BC cell proliferation, movement, and invasion. Overall, IFI44 was found to be an elevated biomarker that was closely linked to unfavorable outcomes in BC.
Studies currently emphasize the influence of IFI44 on the INF-α/β mediated anti-viral process,38 pathogenesis of SLE,39 tumor radiation resistance40 and neuroinflammation.41 The possible roles of IFI44 in modulating the immune system during cancer development have been less explored. We investigated the potential association of IFI44 with immune cells by CIBERSORT algorithm, and IFI44 exhibited a positive association with M2 Macrophages in GSE47994 and GSE36772. The tumor microenvironment (TME) significantly influences cancer progression by utilizing tactics such as physical obstructions, cellular fatigue, a suppressed immune milieu and immune cells.42,43 Macrophages fall into two primary types, M1 macrophages and M2 macrophages, playing distinct roles in immune response and monitoring. These cells are not set in stone but can switch between types depending on shifts in their surrounding conditions. M1 macrophages release proinflammatory cytokines that effectively combat pathogens, while M2 macrophages facilitate angiogenesis, tumor progression, and metastasis.44 Correlation analyses revealed a positive connection between IFI44 and M2 macrophage abundance and markers a positive connection between IFI44 and M2 macrophage infiltration and markers, with IL-10 having the most significant positive relationship. Research indicated that IL-10, CD163, TGFB1, CSF1R, parasitic infections, and various stimulations promote macrophage polarization toward the M2 phenotype.45,46 Via IL-10 signaling, IFI44 directed M2 macrophages to move towards BC cells. For instance, in pancreatic ductal carcinoma, inhibiting the expression of key factors of M2 macrophages in the immune microenvironment, such as CD163, will significantly weaken the ability of tumors to invade and spread to distant sites.47 However, increasing the M1/M2 macrophage ratio inhibits tumorigenicity.48 These outcomes collectively demonstrated that IFI44 encouraged the trafficking of macrophages, their M2 polarization, and the motility of M2 macrophages towards BC cells, resulting in an immunosuppressive environment, which contributed significantly to the resistance against immunotherapy in BC.
Furthermore, we established a nomogram incorporating risk scores and clinical variables to predict survival probabilities, which confirmed the risk score as an independent prognostic indicator. To further certify the dependability of the IMS in pan-cancer, we acquired transcriptome expression datasets for various cancer types from the TCGA database and applied our formula to every patient. Several cancers, including BLCA, HNSC, SKCM and LAML, low-risk group members exhibited improved survival rates and greater immune score than those in the high-risk group. M2 macrophage-mediated immunosuppression represents a hallmark of the tumor microenvironment across multiple cancer types.49–52 IFI44, as an interferon-stimulating gene, may be involved in shaping an immunosuppressive environment conducive to M2 polarization in these cancers. IL-10–Driven M2 Macrophage may participate in the immune regulation of these cancers, thus enabling IMS based on this axis to have cross-cancer prediction capabilities.
Current breast cancer immunotherapy faces the challenge of limited response rates and lack of precise biomarkers. The IFI44-IL-10-M2 axis identified in this study reveals a non-classical, intrinsically driven immunosuppressive mechanism by tumor cells, which may partly explain why some tumors are insensitive to existing immune checkpoint inhibitors. Therefore, targeting this axis is expected to provide new combination therapy ideas for overcoming drug resistance.
However, there are additional ideas that could be explored to enhance our research. Initially, mechanism exploration mainly relies on cell line models, which provide strong evidence for causality but may not fully simulate tumor heterogeneity and complex tumor-matrix-immune interaction networks in vivo. Secondly, our constructed IMS signatures and their associations with the immune microenvironment still need to be validated in more prospective clinical cohorts and multi-omics data containing fresh tissue samples. In addition, the limitation of this study is that IFI44 may affect the microenvironment through factors other than IL-10. In the future, through the cytokine profiling analysis of co-culture systems, its immunomodulatory network will be more comprehensively revealed. Overall, our findings not only provide a new mechanistic viewpoint for understanding immune evasion in breast cancer, but also lays a theoretical foundation for the future development of combination immunotherapies based on IFI44 or tumor-associated macrophages.

Conclusion
In summary, our study demonstrates that IFI44 is an oncogenic driver in breast cancer that promotes proliferation, migration, and invasion of tumor cells. Mechanistically, IFI44 orchestrates an IL-10–driven immunosuppressive microenvironment by enhancing M2 macrophage infiltration. The IFI44-based M2 macrophage signature (IMS) we developed serves as a robust and independent prognostic indicator, capable of predicting patient survival and potential chemotherapy response. These findings not only elucidate a novel immune-regulatory role of IFI44 but also propose the IMS as a promising predictive tool for personalized therapy in breast cancer and potentially other malignancies.