Abstract

   The aim of this study was to identify biomarkers associated with breast
   cancer prognosis and to explore the underlying pathogenic mechanisms.
   Interferon alpha-inducible protein 6 (IFI6), known as a proliferative
   and anti-apoptotic factor, has been implicated in various malignant
   diseases. However, its biological roles in breast cancer remain poorly
   understood. To address this, we employed bioinformatics analyses to
   investigate the expression and prognostic significance of IFI6 in
   breast cancer. Our findings revealed that IFI6 was upregulated in
   breast cancer and was associated with histological subtypes and lymph
   node metastasis status. Kaplan-Meier plotter analysis demonstrated that
   high IFI6 expression correlated with poor prognosis in breast cancer
   patients with ER-positive, PR-positive, HER2-positive, and lymph
   node-positive subtypes. To further enhance clinical applicability, we
   constructed a prognostic nomogram incorporating IFI6 expression and
   clinicopathological factors, which showed favorable predictive
   performance for overall survival. Additionally, IFI6 expression showed
   significant correlations with infiltrating immune cells, including
   regulatory T cells (Tregs), M1 macrophages, naïve B cells, and plasma
   cells. Single-cell RNA sequencing analysis revealed that IFI6 was
   predominantly expressed in epithelial tumor cells and was associated
   with altered immune cell composition, suggesting the potential role in
   shaping the immune microenvironment. Moreover, IFI6 expression was
   closely associated with several immunomodulators. In conclusion, IFI6
   serves as a potential biomarker for immune infiltration and poor
   prognosis in breast cancer and may offer novel insights into risk
   stratification and immunotherapeutic strategies.

Supplementary Information

   The online version contains supplementary material available at
   10.1038/s41598-025-20489-6.

   Keywords: Bioinformatics analysis, IFI6, Immune infiltration, Breast
   cancer, Prognosis

   Subject terms: Cancer, Immunology

Introduction

   Recent years, the incidence of breast cancer (BC) has surpassed that of
   lung cancer, making it the most prevalent malignant tumor worldwide.
   Additionally, it has become the leading cause of cancer-related deaths
   among women worldwide^[30]1. Advances in medical care, such as surgery,
   conventional chemotherapy, precision radiotherapy, endocrine therapy,
   and monoclonal antibody treatments, have significantly improved the
   prognosis of breast cancer patients^[31]2. However, the treatment
   outcomes for advanced breast cancer remain poor, with many patients
   continuing to suffer from the threat of recurrence and mortality^[32]3.
   Studies have demonstrated that the prognosis and therapeutic responses
   in breast cancer were significantly correlated with clinical,
   pathological, and molecular characteristics^[33]4. Although numerous
   molecules have been identified as biomarkers for the early diagnosis
   and prognostic assessment of breast cancer, these biomarkers still
   exhibit certain limitations, highlighting the urgent need to identify
   novel and reliable prognostic biomarkers for breast cancer^[34]5.

   Interferon alpha-inducible protein 6 (IFI6), a member of the FAM14
   protein family, is primarily localized to the inner mitochondrial
   membrane and is one of the interferon-stimulated genes^[35]6. IFI6 has
   been implicated in several malignancies, including breast and gastric
   cancers^[36]7. Previous studies have shown that IFI6 is a target of
   estrogen signaling and is associated with resistance to tamoxifen
   treatment^[37]8. Furthermore, IFI6 has been identified as an
   anti-apoptotic protein in multiple myeloma and breast cancer^[38]9.
   Additionally, Liu et al. reported that higher expression of IFI6 was
   linked to an immunosuppressive microenvironment in esophageal squamous
   cell carcinoma^[39]10. However, its comprehensive role in the tumor
   microenvironment of breast cancer remains unclear, and few studies have
   systematically explored its value in predicting immune infiltration and
   clinical prognosis using integrative computational approaches.

   In recent years, machine learning and optimization algorithms have been
   increasingly applied to improve disease management, early screening,
   and outcome prediction across various medical domains^[40]11–[41]14.
   Among them, prognostic modeling based on molecular and clinical
   features has become an important strategy in cancer research^[42]15.
   Among various statistical approaches, Cox proportional hazards
   regression remains one of the most widely used and robust methods for
   evaluating the association between gene expression and patient survival
   outcomes^[43]16. Multivariate Cox analysis allows adjustment for
   potential confounding factors and the identification of independent
   prognostic markers with clinical relevance. In parallel, nomograms have
   emerged as effective tools for individualized risk prediction, which
   also facilitates clinical decision-making and personalized treatment
   planning^[44]17. By integrating independent prognostic variables such
   as gene expression levels and clinicopathological parameters, nomograms
   provide visual and quantitative representations of patient-specific
   survival probabilities. Random forest as an ensemble learning method
   also provides robust variable importance rankings and has shown
   excellent performance in classification and risk prediction
   tasks^[45]18.

   While previous studies have associated IFI6 with cancer cell apoptosis,
   estrogen signaling, and therapy resistance, the broader immunological
   role of IFI6 within the breast cancer microenvironment remains poorly
   characterized. Importantly, no studies to date have systematically
   examined the predictive value of IFI6 in relation to immune cell
   infiltration and overall prognosis using integrative computational and
   machine learning approaches. In this study, we comprehensively analyzed
   the expression and prognostic significance of IFI6 in breast cancer,
   and developed a nomogram incorporating IFI6 and clinicopathological
   features for individualized survival prediction. Moreover, we explored
   the association between IFI6 expression and the immune landscape using
   CIBERSORT and validated immune associations using multiple algorithms.
   These findings expand the understanding of IFI6 beyond its previously
   known roles and underscore its potential as a novel immune-related
   biomarker for prognostication in breast cancer.

Results

Elevated expression of IFI6 in breast cancer

   The expression of IFI6 in various human cancers was assessed using the
   TIMER 2.0 database. As depicted in Fig. [46]1A, IFI6 expression levels
   were significantly elevated in several cancer types, including bladder
   urothelial carcinoma (BLCA), breast cancer (BRCA), cholangiocarcinoma
   (CHOL), colorectal cancer (COAD), esophageal carcinoma (ESCA), head and
   neck squamous cell carcinoma (HNSC), clear cell carcinoma of kidney
   (KIRC), hepatocellular carcinoma (LIHC), prostate adenocarcinoma
   (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD) and
   uterine corpus endometrial carcinoma (UCEC). In contrast, IFI6
   expression was found to be lower in kidney chromophobe (KICH) and lung
   squamous cell carcinoma (LUSC). Additionally, data from the GEPIA
   database demonstrated that the mRNA expression levels of IFI6 in BC
   were significantly higher than those in normal tissues (P < 0.05)
   (Fig. [47]1B). This higher mRNA expression of IFI6 in breast cancer was
   further confirmed using the UALCAN cancer database (P < 0.001)
   (Fig. [48]1C). Furthermore, Western blot (WB) analysis was performed to
   examine IFI6 expression at the protein level in breast cancer cell
   lines. As shown in Fig. [49]1D, IFI6 protein levels were notably higher
   in HCC1937, BT549, MCF-7 and MDA-MB-231, further substantiating its
   dysregulation in breast cancer.

Fig. 1.

   [50]Fig. 1
   [51]Open in a new tab

   IFI6 was evaluated in breast cancer. (A) Differential expression of
   IFI6 between tumor and adjacent normal tissues across various cancer
   types (TIMER 2.0). (B) IFI6 expression in breast cancer and normal
   breast tissue (GEPIA). (C) IFI6 expression in breast cancer and normal
   tissue based on the UALCAN database. (D) WB analysis showed increased
   IFI6 protein expression in breast cancer cell lines. (*P < 0.05,
   **P < 0.01, ***P < 0.001)

Clinicopathological significance of IFI6 expression in breast cancer

   The ULCAN database revealed that IFI6 was associated with clinical
   stages in BC. Specifically, IFI6 expression was significantly
   upregulated across stages 1 to 4 in BC patients (Fig. [52]2A). In terms
   of BC subtypes, IFI6 expression was notably elevated in patients with
   luminal, HER2-positive, and triple-negative BC (TNBC) subtypes
   (Fig. [53]2B). Additionally, IFI6 expression progressively increased
   according to nodal metastasis status, with higher expression observed
   in BC patients classified as N0, N1, N2, and N3 (Fig. [54]2C). Further
   analysis using the BC-GEM online tool revealed differences in IFI6
   expression across various clinical parameters. The results showed that
   IFI6 expression was significantly associated with age, with higher
   expression levels in younger BC patients (≤ 51 years, Fig. [55]2D).
   Furthermore, IFI6 expression varied across histological types, being
   higher in micropapillary carcinoma and lower in invasive lobular
   carcinoma (ILC) (Fig. [56]2E). BC patients with higher
   Scarff-Bloom-Richardson (SBR) grades and advanced Nottingham Prognostic
   Index (NPI) values exhibited significantly elevated IFI6 mRNA levels
   (Fig. [57]2F and G). Moreover, IFI6 expression was positively
   correlated with progesterone receptor (PR) and HER2 status (Fig. [58]2I
   and J). However, no significant association was found between IFI6
   expression and estrogen receptor (ER) status or TNBC subtypes
   (Fig. [59]2H and K).

Fig. 2.

   [60]Fig. 2
   [61]Open in a new tab

   Association of IFI6 expression with clinicopathological features of BC
   patients. Clinicopathologic features included (A) clinical stage; (B)
   molecular subtypes; (C) lymph node metastasis status; (D) age; (E)
   histological subtype; (F) SBR grading; (G) NPI; (H) ER status; (I) PR
   status; (J) HER-2 status; (K) TNBC status. (*P < 0.05, **P < 0.01,
   ***P < 0.001)

Prognostic value of IFI6 expression in breast cancer

   The association between IFI6 expression and prognosis in BC patients
   was evaluated using the BC-GEM database. Elevated expression of IFI6
   was significantly associated with poor overall survival (OS,
   Fig. [62]3A) and recurrence-free survival (RFS, Fig. [63]3B). ROC curve
   was conducted to assess the diagnostic performance of IFI6, yielding an
   area under the curve (AUC) value of 0.870 (95% CI: 0.837–0.903;
   Fig. [64]3C). Kaplan-Meier plotter analysis further confirmed that
   higher IFI6 expression correlated with significantly worse OS
   (Fig. [65]3D) and RFS (Fig. [66]3E). Additionally, increased IFI6
   expression was linked to poor OS in BC patients with ER-positive
   (Fig. [67]3F), PR-positive (Fig. [68]3G), HER-2-positive (Fig. [69]3H)
   and lymph node-positive subtypes (Fig. [70]3I), as well as poor RFS in
   patients with ER-positive (Fig. [71]3J), PR-positive (Fig. [72]3K), and
   lymph node-positive subtypes (Fig. [73]3M). However, no significant
   association was observed between IFI6 expression and RFS in
   HER-2-positive patients (Fig. [74]3L). These findings indicated that
   IFI6 expression significantly impacted the prognosis of BC patients
   across various clinicopathological factors.

Fig. 3.

   [75]Fig. 3
   [76]Open in a new tab

   Prognostic significance of IFI6 in BC patients. (A,B) The OS (A) and
   RFS (B) in BC cohorts obtained through BC-GEM. (C) ROC curve. (D,E) The
   OS (D) and RFS (E) in BC cohorts obtained from the Kaplan-Meier plotter
   database. (F–I) OS curves based on different molecular subtypes:
   ER-positive (F), PR-positive (G), HER-2 positive (H) and lymph
   node-positive subtypes (I). (J–M) RFS curves based on different
   molecular subtypes: ER-positive (J), PR-positive (K), HER-2 positive
   (L) and lymph node-positive subtypes (M).

Construction and validation of the IFI6-based prognostic model

   We further constructed a prognostic model for BC patients via several
   datasets. The METABRIC cohort was utilized as the training set. Forest
   plots of multivariate Cox regression revealed that IFI6 (HR: 1.082, 95%
   CI: 1.020–1.148, P = 0.009) was a prognostic predictor independent of
   BC (Fig. [77]4A). We also generated a nomogram that combines the
   clinical factors and IFI6 (Fig. [78]4B). The calibration curves of
   internal validation showed good agreement between the
   nomogram-predicted and observed OS at 1-, 3-, and 5-year time points
   (Fig. [79]4C), indicating favorable calibration performance of the
   model. The AUC values for 1-, 3-, and 5-year OS prediction were 0.714,
   0.622, and 0.626 (Fig. [80]4D), reflecting a moderate discriminatory
   ability. The decision curve analysis (DCA) curve also indicated that
   the nomogram provided a certain degree of clinical net benefit
   (Fig. [81]4E). In the external validation cohorts ([82]GSE7390 and
   [83]GSE42568), the nomogram also showed favorable predictive
   performance. Calibration curves indicated good consistency between the
   predicted and observed OS at 1, 3, and 5 years (Fig. [84]4F). The AUCs
   for predicting 1, 3, and 5-year OS were 0.654, 0.607, and 0.631
   (Fig. [85]4G), indicating moderate discrimination ability. Furthermore,
   the DCA curve demonstrated that the nomogram provided clinical net
   benefit across a reasonable range of threshold probabilities,
   confirming its potential utility in clinical decision-making
   (Fig. [86]4H).

Fig. 4.

   [87]Fig. 4
   [88]Open in a new tab

   Construction and verification of the predictive model. (A) Forest plots
   of multivariate Cox regression. (B) The nomogram predicts prognosis for
   BC patients. (C) The calibration curve evaluates the precision of the
   nomogram model in forecasting 1-, 3-, and 5-year survival rates. (D)
   Receiver operating characteristic (ROC) curves show the predictive
   accuracy of the nomogram. (E) DCA illustrates the overall improvement
   of the predictive risk model. (F–H) Validation of the prognostic
   nomogram in external cohorts: calibration curves (F), ROC curves (G),
   and DCA curve (H).

Gene set enrichment analysis (GSEA)

   We further performed GSEA analysis to identify key biological processes
   involved in IFI6-mediated mechanisms in breast cancer. A detailed
   enrichment analysis of hallmark pathways was presented in Supplementary
   Table [89]S1. The crucial pathways included the interferon gamma
   response (Fig. [90]5A), the interferon alpha response (Fig. [91]5B),
   the IL-6-JAK-STAT3 signaling pathway (Fig. [92]5C), the inflammatory
   response (Fig. [93]5D), the complement (Fig. [94]5E), and the DNA
   repair (Fig. [95]5F). Collectively, these results suggested that the
   role of IFI6 in tumorigenesis was likely associated with tumor
   immunology.

Fig. 5.

   [96]Fig. 5
   [97]Open in a new tab

   Gene set enrichment analysis of IFI6. The enriched pathways included
   (A) interferon gamma response, (B) interferon alpha response, (C)
   IL-6-JAK-STAT3 signaling pathway, (D) inflammatory response, (E)
   complement, (F) DNA repair.

Network construction of IFI6 co-expressed genes in breast cancer

   To investigate the mechanism underlying IFI6 in tumorigenesis and
   prognosis, co-expressed genes of IFI6 were also identified from the
   BRCA dataset in TCGA using the LinkedOmics database. The analysis
   revealed 5785 genes positively correlated and 6534 genes negatively
   correlated with IFI6 expression, as illustrated in the volcano plot
   (Fig. [98]6A, Supplementary Table S2). The top 50 significantly
   positively and negatively correlated genes were visualized in heat maps
   (Fig. [99]6B and C). Among the positively correlated genes, the top
   three were ISG15 (r = 0.898, P < 0.05), IFIT1 (r = 0.893, P < 0.05),
   and OAS1 (r = 0.850, P < 0.05) (Fig. [100]6D and E, and F). Conversely,
   the top three negatively correlated genes were FAM172A (r=-0.442,
   P < 0.05), TMEM168 (r=-0.396, P < 0.05), and FAM168B (r=-0.404,
   P < 0.05) (Fig. [101]6G and H, and I).

Fig. 6.

   [102]Fig. 6
   [103]Open in a new tab

   Co-expression network of IFI6. (A) Volcano plot of IFI6-associated
   genes (LinkedOmics). (B,C) Heatmaps of the top 50 positively (B) and
   negatively (C) genes correlated with IFI6. (D-F) Spearman correlations
   between IFI6 and the top three positively correlated genes: IFIT1,
   ISG15, and OAS1. (G-I) Spearman correlations between IFI6 and the top
   three negatively correlated genes: FAM172A, TMEM168, and FAM168B. All
   P-values were adjusted using the Bonferroni correction and
   Benjamini-Hochberg method.

Enrichment analysis of IFI6 and co-expressed genes

   We performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathway enrichment analyses on the top 100 genes most
   strongly associated with IFI6. The pathways were ranked by P-value.
   Detailed summaries of the enrichment terms for Biological Process (BP),
   Cellular Component (CC), Molecular Function (MF), and KEGG pathways are
   provided in Supplementary Table S3. In the BP analysis, we observed
   significant enrichment in pathways related to type I interferon
   signaling, cellular response to type I interferon, and response to type
   I interferon (Fig. [104]7A). For CC, the enrichment was mainly observed
   in the postsynaptic endocytic zone, dendritic spine head, and
   phagocytic vesicle membrane (Fig. [105]7B). In the MF category, the
   most enriched terms included double-stranded RNA binding,
   single-stranded RNA binding, and RNA helicase activity (Fig. [106]7C).
   KEGG pathway analysis revealed significant enrichment of IFI6
   co-expressed genes in pathways related to Hepatitis C, Influenza A,
   Measles, and Epstein-Barr virus infection (Fig. [107]7D).

Fig. 7.

   [108]Fig. 7
   [109]Open in a new tab

   Enrichment analysis of GO and KEGG pathway enrichment of IFI6
   co-expressed genes. (A–C) GO Enrichment analysis: Biological Process
   (A), Cellular Component (B), Molecular Function (C); (D) KEGG pathway
   enrichment analysis. Although FDR values were computed, only P-values
   are shown in the figure for visualization purposes.

IFI6 is relevant to immune infiltration

   Based on the network of co-expressed genes and enrichment analysis, we
   hypothesized that IFI6 played a significant role in tumor immunology.
   For further investigation, the relationship between IFI6 expression and
   immune infiltration was analyzed via the ESTIMATE algorithm. As shown
   in Fig. [110]8A, IFI6 expression was positively correlated with immune
   score (r = 0.26, P < 0.05) and ESTIMATE score (r = 0.13, P < 0.05), but
   not with stromal score (r = 0.0017, P > 0.05), indicating its relevance
   to immune infiltration rather than stromal content. Moreover, IFI6
   showed a negative correlation with tumor purity (r=-0.14, P < 0.05).
   Additionally, we estimated the proportions of 22 immune cell types in
   each sample (Fig. [111]8B). Among these 22 immune cell types, 20 showed
   significantly different abundances between high and low IFI6 expression
   groups (Fig. [112]8C). Furthermore, correlation analyses of
   infiltrating immune cells were performed, with scores representing the
   degree of correlation. Moreover, the resulting correlation heatmap
   indicated that regulatory T cells and M1 macrophages were strongly and
   positively correlated with IFI6 expression. In contrast, naïve B cells
   and plasma cells were negatively correlated with IFI6 expression
   (Fig. [113]8D and Supplementary Table S4).

Fig. 8.

   [114]Fig. 8
   [115]Open in a new tab

   Enrichment analysis of GO and KEGG pathway enrichment of IFI6
   co-expressed genes. (A) The correlation between IFI6 and tumor
   microenvironment. (B) Heatmap of the proportions of 22 immune cell
   types in the low and high IFI6 expression groups of cancer samples.
   (C) The box diagram of the proportion of immune cells based on IFI6
   expression level. (D) The correlation matrix of immune cell and IFI6.
   All P-values were adjusted using the Benjamini-Hochberg method.

Single-cell RNA sequencing (scRNA-seq) reveals IFI6-associated immune
landscape in BC

   The Seurat package was used to normalize the scRNA-seq data. As shown
   in the UMAP in Fig. [116]9A, a total of 794,443 cells were grouped into
   39 clusters based on the top 2000 highly variable genes and the top 30
   principal components. According to known marker genes, the 39 clusters
   were annotated into nine major cell types, including T cells,
   epithelial cells, myeloid cells, fibroblasts, pericytes, B cells,
   endothelial cells, plasma cells, and basal cells (Fig. [117]9B). The
   dot plot (Fig. [118]9C) visualizes the expression levels and
   proportions of canonical marker genes across the annotated cell types,
   further validating the classification.

Fig. 9.

   [119]Fig. 9
   [120]Open in a new tab

   Identification of IFI6-associated immune landscape via scRNA-seq
   analysis. (A) A total of 39 clusters were identified in the dataset.
   (B) UMAP plot displaying the distribution of 9 distinct cell types. (C)
   Heatmap depicting the expression levels of the marker genes specific to
   each indicated cell type. (D,E) UMAP plot illustrating IFI6
   heterogeneously expressed among different cell clusters. (F) Bar chart
   highlighting the different proportions of each cell type based IFI6
   expression level.

   To explore the immune-related features of IFI6, we mapped its
   expression across the cell populations. As shown in Fig. [121]9D, IFI6
   was heterogeneously expressed among different cell clusters, with
   noticeable expression in specific regions. Cells were then dichotomized
   into IFI6-positive and IFI6-negative groups based on expression levels
   (Fig. [122]9E). As visualized in the stacked bar chart (Fig. [123]9F),
   IFI6-positive cells were enriched in epithelial cells, while other cell
   types such as T cells and myeloid cells showed predominantly
   IFI6-negative status, suggesting high IFI6 expression may be associated
   with a “cold” immune microenvironment.

The expression of IFI6 is associated with immunomodulators

   To further elucidate the novel function of IFI6, we utilized the TISIDB
   database to assess the relationship between IFI6 expression and the
   abundance of immunomodulators. The results indicated that IFI6 was
   positively correlated with a variety of immunostimulatory molecules
   (Fig. [124]10A). The six immune-stimulators with the strongest
   correlations were CD80 (r = 0.438, P < 0.05), CD86 (r = 0.349,
   P < 0.05), TNFSF13B (r = 0.344, P < 0.05), ICOS (r = 0.336, P < 0.05),
   LTA (r = 0.290, P < 0.05), and CXCL12 (r=-0.143, P < 0.05)
   (Fig. [125]10B). Additionally, IFI6 expression was associated with a
   variety of immune-suppressive agents (Fig. [126]10C), with the six
   strongest associations being LGALS9 (r = 0.553, P < 0.05), LAG3
   (r = 0.494, P < 0.05), IDO1 (r = 0.354, P < 0.05), CTLA4 (r = 0.328,
   P < 0.05), TGBR1 (r=-0.224, P < 0.05), and KDR (r=-0.155, P < 0.05)
   (Fig. [127]10D). Furthermore, IFI6 was positively correlated with
   several major histocompatibility complex (MHC) molecules
   (Fig. [128]10E), particularly TAP1 (r = 0.611, P < 0.05), HLA-B
   (r = 0.590, P < 0.05), B2M (r = 0.572, P < 0.05), HLA-A (r = 0.560,
   P < 0.05), TAP2 (r = 0.520, P < 0.05), and HLA-C (r = 0.519, P < 0.05)
   (Fig. [129]10F).

Fig. 10.

   [130]Fig. 10
   [131]Open in a new tab

   Relationship between IFI6 expression and the abundance of
   immunomodulators in breast cancer using TISIDB database. (A,B)
   Correlation between IFI6 expression and the abundance of
   immune-stimulators; (C,D) Correlation between IFI6 expression and the
   abundance of immunosuppressants; (E,F) Correlation between IFI6
   expression and the abundance of MHC molecules.

Discussion

   IFI6, also referred to as GIP3, belongs to the FAM14 protein
   family^[132]19. Due to its role in stabilizing mitochondrial function,
   IFI6 has been identified as an anti-apoptotic factor in various
   cancers^[133]20. Cheriyath et al. demonstrated that ectopic expression
   of IFI6 could induce tamoxifen resistance in the ER+ breast cancer cell
   line MCF-7, while silencing IFI6 triggered apoptosis in BT-549 cells
   and inhibited the growth of MCF-7 cells^[134]21. However, previous
   research has primarily concentrated on basic studies involving a
   limited number of breast cancer cell lines and has not integrated
   clinical data. Therefore, it is essential to conduct more comprehensive
   studies that explore the potential role of IFI6 in breast cancer. This
   study aims to enhance the understanding of the underlying mechanisms of
   IFI6 in tumor immunology and position IFI6 as a diagnostic and
   therapeutic target for personalized breast cancer treatment.

   In the present study, we found that IFI6 expression was significantly
   upregulated in pan-cancer including breast cancer compared with
   adjacent normal tissues, which is consistent with the recent
   studies^[135]22. To validate this observation, we performed western
   blotting analysis and confirmed the overexpression of IFI6 in breast
   cancer cell lines compared to normal breast epithelial cells. Moreover,
   the expression of IFI6 was closely associated with advanced
   clinicopathological features including higher tumor stage, lymph node
   metastasis, SBR grade, NPI score, and hormone receptor status. Patients
   with higher IFI6 expression exhibited worse OS and RFS, which suggested
   that IFI6 may serve as a prognostic marker. Furthermore, we performed
   subtype-stratified survival analyses based on ER, PR, HER2 status, and
   lymph node involvement. Within most clinical subgroups, patients with
   high IFI6 expression consistently exhibited poorer OS and RFS,
   reinforcing the prognostic relevance of IFI6 across diverse breast
   cancer subtypes and clinical backgrounds. To enhance the clinical
   utility of these findings, we constructed a nomogram model integrating
   IFI6 expression with established clinicopathologic factors to predict
   individualized OS in breast cancer patients. The nomogram demonstrated
   favorable predictive accuracy and may serve as a practical
   decision-support tool for risk stratification in clinical
   settings^[136]23. While ER, PR, HER2, and Ki-67 are fundamental in
   classifying breast cancer into distinct subtypes, they do not fully
   account for prognostic heterogeneity within each subtype^[137]24. In
   the present study, high IFI6 expression remained significantly
   associated with worse survival outcomes even among ER-positive,
   PR-positive and HER2-positive patients, which indicates that IFI6 may
   reflect additional biological features not captured by existing
   markers. Incorporating IFI6 into multivariate models improved
   prognostic accuracy and supports its potential as a complementary
   marker to refine risk stratification.

   The potential mechanism by which IFI6 impacts cancer development and
   progression is complex and not fully understood^[138]25. Co-expression
   analysis is a widely used approach for inferring putative gene
   functions and elucidating the roles of genes in phenotypic
   variations^[139]26. In this study, IFI6 exhibited a strong positive
   correlation with IFIT1, ISG15, and OAS1. IFIT1, an interferon-induced
   protein, has been implicated in promoting cancer development^[140]27.
   ISG15 mediates the ISGylation of KPNA2, thereby maintaining cancer stem
   cell-like characteristics in anaplastic thyroid carcinoma^[141]28.
   Furthermore, OAS1 has been identified as a prognostic biomarker in
   pan-cancer, potentially contributing to cytotoxic T lymphocyte (CTL)
   dysfunction and macrophage M2 polarization^[142]29. GSEA further
   revealed that IFI6 was associated with key biological processes,
   including oxidative phosphorylation, proteasome function, and antigen
   processing and presentation, all of which are critical pathways linked
   to the development and progression of breast cancer^[143]30,[144]31.

   Based on the co-expression analysis, we hypothesized that IFI6 may be
   associated with tumor immunity. Although the present study identified a
   positive correlation between IFI6 expression and immune score, the
   potential confounding factors such as tumor purity and stromal
   contamination may influence the estimation of immune cell
   infiltration^[145]32. Further analysis revealed a negative correlation
   between IFI6 expression and tumor purity (r=-0.14, P < 0.05), which
   suggests that patients with high IFI6 expression tend to have a lower
   proportion of tumor cells and a higher level of immune infiltration.
   Additionally, no significant association was found between IFI6
   expression and stromal score. These findings alleviate concerns that
   tumor purity or stromal contamination may bias immune infiltration
   estimates, supporting a potential role of IFI6 in modulating the tumor
   immune microenvironment. Moreover, scRNA-seq revealed that IFI6 was
   predominantly expressed in epithelial cells and largely absent in
   immune populations such as T cells and myeloid cells, suggesting a
   tumor-intrinsic expression pattern^[146]33. Notably, IFI6-positive
   cells were enriched in epithelial compartments, whereas IFI6-negative
   status dominated immune cell populations. These findings are consistent
   with the negative correlation observed between IFI6 expression and
   tumor purity, as well as the lack of association with stromal score (P
   > 0.05), which indicates that immune relevance of IFI6 likely stems
   from the modulation of immune infiltration rather than stromal or
   purity-related artifacts. Our study also noticed that IFI6 expression
   was positively correlated with Tregs. As acknowledged, Tregs
   infiltrated heavily into tumor tissue, which was usually associated
   with a poor prognosis in cancer patients^[147]34. This may represent a
   potential mechanism underlying the poor prognosis observed in patients
   with elevated IFI6 expression. M1 macrophages are classically activated
   macrophages that play a critical role in anti-tumor and anti-infection
   immune responses. The polarization of macrophages toward the M1
   phenotype can be induced by interferons, particularly interferon-γ,
   which in turn activates interferon-stimulated genes (ISGs) such as
   IFI6, OAS1, ISG15^[148]35,[149]36. This aligns with the positive
   correlation between IFI6 expression and M1 macrophages identified in
   the present study. We also found that IFI6 expression was negatively
   associated with B cell naïve and plasma cells, which indicated that
   higher IFI6 expression was associated with reduced recruitment or
   activity of B cell subtypes within the tumor microenvironment,
   potentially contributing to an immunosuppressive state that favors
   tumor progression^[150]37. Our findings are in line with prior studies
   that characterized immune cell infiltration patterns in breast cancer
   using deconvolution algorithms such as CIBERSORT and TIMER^[151]38.
   However, compared to earlier studies that often focused on a single
   algorithm or dataset, our analysis employed a multi-platform approach
   that integrates TIMER 2.0, CIBERSORT, GEPIA, TISIDB, and single-cell
   RNA-seq data. In addition, tumor purity and stromal content were
   adjusted in correlation analyses to reduce potential confounding. This
   comprehensive strategy provides a more nuanced view of the tumor immune
   microenvironment and enhances the robustness of our conclusions.

   Although existing evidence suggests that IFI6 contributes to the
   formation of a tumor-suppressive immune microenvironment, its precise
   role in shaping the tumor immune landscape remains to be fully
   elucidated. IFI6 is a well-characterized ISG with known mitochondrial
   localization. Emerging studies indicate its involvement in regulating
   mitochondrial reactive oxygen species (mtROS)^[152]39, potentially
   linking interferon signaling, mitochondrial metabolism, and immune
   modulation within the tumor microenvironment^[153]40. Elevated mtROS
   levels have been shown to promote the differentiation of Tregs and
   suppress the activity of effector T cells^[154]40,[155]41.
   Particularly, moderate mtROS levels facilitate Foxp3 expression and
   IL-10 production in Tregs, which are crucial for the maintenance of
   local immune suppression^[156]42. In this context, IFI6 may enhance the
   expression of immunosuppressive cytokines such as IL-10, or chemokines
   that recruit Tregs to contribute to the immunosuppressive tumor
   microenvironment. Moreover, IFI6 is frequently co-expressed with other
   ISGs including ISG15 and OAS1, which suggests a role in sustaining a
   chronic interferon-driven immune state^[157]43. While acute interferon
   signaling is typically protective, prolonged activation has been
   associated with immune exhaustion and tumor immune evasion^[158]44.
   These findings raise the possibility that IFI6 may paradoxically
   promote tumor immune escape under chronic inflammatory conditions.
   These results support a mechanistic hypothesis in which IFI6 serves as
   a key mediator connecting interferon responses, mtROS production, and
   immune suppression. Further experimental validation is warranted to
   determine whether targeting IFI6 could reshape the immunosuppressive
   tumor microenvironment and improve anti-tumor immunity. To further
   elucidate the role of IFI6 in the tumor microenvironment, the
   correlation between IFI6 immunomodulators was analyzed. IFI6 expression
   was significantly positively correlated with immune inhibitors, immune
   stimulators and MHC molecules. The association of IFI6 with both
   immunostimulatory and immunosuppressive molecules highlights its
   complex and dualistic role in the tumor microenvironment. On one hand,
   IFI6 could enhance the anti-tumor immunity response through its
   correlation with immunostimulatory molecules. On the other hand, its
   association with immunosuppressive molecules may reduce immune
   efficacy, ultimately facilitating immune escape and tumor progression.
   The strong positive correlations identified in this study suggest that
   IFI6 acts as a significant immunomodulator and a potential target for
   future immunotherapy strategies. Targeting IFI6 in combination
   therapies, particularly in pathways involving specific
   immunosuppressants and MHC molecules, could provide a novel approach to
   enhance anti-tumor immunity. Further experimental studies are necessary
   to clarify the precise mechanisms by which IFI6 influences these immune
   molecules and their associated signaling pathways.

   Despite the comprehensive bioinformatics analyses performed in this
   study, several limitations should be acknowledged. First, although we
   validated IFI6 expression patterns using independent datasets such as
   METABRIC and incorporated multiple immune deconvolution algorithms
   including TIMER and CIBERSORT. We did not employ additional frameworks
   such as xCell or MCP-counter, which use distinct gene signatures and
   statistical models to estimate immune cell populations^[159]45. The use
   of multiple complementary algorithms would enhance the robustness and
   reproducibility of immune infiltration findings. Our findings remain
   based solely on computational analyses and lack experimental
   validation. Future studies incorporating in vitro and in vivo
   functional assays, such as IFI6 knockdown or overexpression in breast
   cancer cell lines and subsequent immunological assays, as well as
   further validations including qPCR and immunohistochemistry (IHC) in
   clinical samples, will be crucial for confirming the mechanistic roles
   proposed. Second, this study is exploratory, and the associations
   observed between IFI6 expression and clinical or immune parameters
   should not be interpreted as causal. While approaches such as Mendelian
   Randomization (MR)-Egger regression could help assess potential
   pleiotropy and causality^[160]46, these were beyond the scope of the
   present work. We suggest that future studies incorporate such
   frameworks to elucidate the mechanistic role of IFI6 better. Third,
   while single-cell RNA sequencing data revealed that IFI6 is
   predominantly expressed in epithelial cells, suggesting a
   tumor-intrinsic expression pattern, the spatial heterogeneity of immune
   infiltration and the potential interaction between IFI6 and specific
   immune subsets remain to be clarified. Spatial transcriptomic profiling
   and multiplex immunohistochemistry could further unravel the contextual
   cellular crosstalk in the tumor microenvironment^[161]47. Fourth,
   although we stratified IFI6 expression across breast cancer molecular
   subtypes, the functional significance of IFI6 may differ across these
   subgroups. Future work should explore whether IFI6 acts as a prognostic
   or predictive biomarker preferentially in specific subtypes such as
   triple-negative or HER2-positive breast cancers. Additionally, the
   ancestry bias inherent in public datasets such as TCGA^[162]48, which
   underrepresent non-European populations, may limit the generalizability
   of our findings.

   Our findings indicate that IFI6 is closely linked to immune
   infiltration patterns and patient prognosis in breast cancer,
   suggesting its potential as a useful clinical biomarker. Future
   research should focus on evaluating whether IFI6 adds diagnostic and
   prognostic value beyond well-established markers like ER, PR, and HER2.
   Large-scale prospective studies are needed to determine if
   incorporating IFI6 expression into existing molecular subtyping or risk
   stratification models can improve the accuracy of predicting patient
   outcomes and treatment responses. To facilitate clinical application,
   it is also important to explore how IFI6 testing could be integrated
   into routine diagnostic workflows, such as standardized IHC or
   RNA-based assays. Given the association between IFI6 and immune cell
   infiltration, we propose that IFI6 could serve as a stratification
   marker for immunotherapy to identify patients more likely to benefit
   from immune checkpoint inhibitors. Further preclinical studies should
   investigate whether targeting IFI6 can modulate the tumor immune
   microenvironment and enhance immunotherapy efficacy. Specifically, IFI6
   inhibition may reduce immunosuppressive signaling and reshape
   macrophage or T cell populations by affecting mtROS and interferon
   signaling pathways. This raises the possibility that combining IFI6
   inhibition with PD-1/PD-L1 blockade could have synergistic effects in
   breast cancer treatment. Validation of these hypotheses through in vivo
   models and combination therapy studies will be critical for advancing
   IFI6 as both a predictive and therapeutic biomarker in clinical
   practice.

Methods

Data collection and IFI6 gene expression analysis

   To analyze IFI6 expression across various cancer types, we utilized the
   TIMER 2.0 platform ([163]http://timer.cistrome.org/), which processes
   RNA-seq data using standardized pipelines and applies TPM normalization
   to estimate gene expression levels. In parallel, we used the GEPIA web
   server ([164]http://gepia.cancer-pku.cn/), which integrates data from
   TCGA and GTEx, normalizes expression data using log2(TPM + 1)
   transformation, and performs differential expression analyses with
   pre-applied statistical filters, including Benjamini-Hochberg
   correction for multiple testing. Analyses were conducted using the gene
   symbol ‘IFI6’, selecting the ANOVA module to compare tumor versus
   normal expression in breast cancer. Furthermore, the UALCAN database
   ([165]https://ualcan.path.uab.edu/analysis.html/) was used to validate
   IFI6 expression levels in breast cancer, with gene expression values
   log2-transformed prior to comparison. Differentially expressed genes
   (DEGs) were identified using thresholds of |log2 fold change|>1 and
   false discovery rate (FDR) < 0.05. For the underlying gene expression
   analyses based on TCGA data, raw RNA-seq count data were downloaded and
   normalized using the Trimmed Mean of M-values (TMM) method via the
   edgeR R package to account for library size differences. Batch effects
   across sequencing runs were corrected using the ComBat algorithm
   implemented in the sva package, ensuring consistency for cross-dataset
   comparisons and reducing technical variability.

Cell lines and cell culture

   The human breast epithelial cell line MCF-10A and breast cancer cell
   lines including MCF-7, BT-549, MDA-MB-231, and HCC1937 were used in
   this study. All cell lines were obtained from the American Type Culture
   Collection (ATCC) and cultured according to ATCC protocols. MCF-10A
   cells were maintained in DMEM/F12 medium supplemented with 5% fetal
   bovine serum (FBS), 20 ng/mL EGF, 0.5 µg/mL hydrocortisone, 100 ng/mL
   cholera toxin, and 10 µg/mL insulin. MCF-7, BT-549, MDA-MB-231, and
   HCC1937 cells were cultured in RPMI-1640 or DMEM supplemented with 10%
   FBS and 1% penicillin-streptomycin. All cells were maintained at 37 °C
   in a humidified incubator with 5% CO₂.

Western blot

   Total protein was extracted from cultured cells using RIPA lysis buffer
   (Beyotime, Shanghai, China) supplemented with protease and phosphatase
   inhibitors. Protein concentrations were quantified using a BCA Protein
   Assay Kit (Thermo Fisher Scientific, USA) according to the
   manufacturer’s protocol. Equal amounts of protein (25 µg) were
   separated on a 15% SDS-PAGE gel and transferred onto PVDF membranes
   (Millipore, USA). After blocking with 5% non-fat milk in TBST
   (Tris-buffered saline with 0.1% Tween-20) for 1.5 h at room
   temperature, membranes were incubated overnight at 4 °C with primary
   antibodies against: IFI6 (Abclonal, Cat# A6157, dilution 1:1000), actin
   (Proteintech, Cat# 20536-1-AP, dilution 1:4000). After washing,
   membranes were incubated with appropriate HRP-conjugated secondary
   antibodies for 1 h at room temperature. Protein bands were visualized
   using an ECL detection reagent (Millipore) and imaged with a Clinx
   imaging system (Qinxiang, Shanghai, China). Densitometric analysis was
   performed using Image J (Version 1.53t, National Institutes of Health,
   Bethesda, MD, USA; [166]https://imagej.net/ij/), with IFI6 expression
   normalized to actin.

Clinicopathological correlation analysis of IFI6

   Gene expression data from TCGA were accessed via the UALCAN database to
   analyze the correlation between IFI6 expression and clinical stage,
   molecular subtypes, and lymph node metastasis in breast cancer.
   Analysis of variance (ANOVA) was used to perform difference tests on
   multiple sample groups. Additionally, the Breast Cancer Gene-Expression
   Miner (BC-GEM) ([167]http://bcgenex.centregauducheau.fr/BC-GEM/), a
   comprehensive data mining tool integrating 36 published and annotated
   genomic datasets, was used to further investigate the relationship
   between IFI6 expression and various clinicopathological features. The
   data was last updated in December 2024.

Prognostic significance of IFI6

   The prognostic value of IFI6 was assessed using the BC-GEM. The
   diagnostic accuracy of gene signatures is typically evaluated using the
   AUC value, which represents the area under the ROC curve. We downloaded
   and collated RNA-seq data from the STAR pipeline of the TCGA-BRCA
   (breast invasive carcinoma) project from the TCGA database
   ([168]https://portal.gdc.cancer.gov), extracted data in TPM format,
   removed duplicate data, and processed the data with log2(value + 1)
   transformation. Subsequently, using R version 4.2.1, we performed ROC
   analysis on the data with the pROC package [1.18.0], and visualized the
   results with ggplot2 [3.4.4]. The Kaplan-Meier Plotter
   ([169]http://kmplot.com/analysis/), an online database that integrates
   gene expression and clinical data, was used to evaluate the prognostic
   significance of IFI6 expression based on OS and RFS in breast cancer
   patients. Patients were dichotomized into high and low expression
   groups according to the median IFI6 expression. Log-rank p-values,
   hazard ratios (HR), and 95% confidence intervals (CIs) were calculated
   automatically by the platform and reported to evaluate statistical
   significance.

The construction and validation of IFI6-based nomogram

   To construct and validate an IFI6-based prognostic nomogram, we used
   the METABRIC cohort as the training set, incorporating clinical
   variables such as age, tumor stage, lymph node status, hormone receptor
   status (ER, PR), HER2 status, and IFI6 mRNA expression. Univariate Cox
   regression identified variables significantly associated with OS, which
   were then included in a multivariate Cox model to determine independent
   prognostic factors. Based on the multivariate model, we constructed a
   nomogram to estimate individualized 1-, 3-, and 5-year OS probabilities
   using the ‘rms’ R package (version 8.0.0) and ‘regplot’ R package
   (version 1.1). Model calibration was assessed using calibration curves
   generated with 1,000 bootstrap resamples. Discrimination ability was
   evaluated via time-dependent ROC curves, and the AUC at 1, 3, and 5
   years was calculated using the ‘timeROC’ R package (version 0.4). To
   visualize survival curves and perform log-rank tests, we used the
   ‘survival’ R package (version 3.8-3) and ‘survminer’ R package (version
   0.5.0).

   In addition, we implemented DCA curve to evaluate the clinical net
   benefit of the nomogram using the ‘ggDCA’ R package (version 1.2).
   External validation was conducted on two independent datasets
   [170]GSE7390 and [171]GSE42568, applying the same statistical
   procedures and evaluation metrics to assess the model’s
   generalizability. P < 0.05 was considered significant difference.

Gene set enrichment analysis

   Patients with invasive breast cancer from the TCGA BRCA dataset were
   classified into two groups based on the median IFI6 mRNA level. GSEA
   analysis was conducted using GSEA software (version 4.1.0) to identify
   key biological pathways associated with IFI6. P < 0.05, FDR < 0.25, and
   normalized enrichment scores (|NES|) > 1 were considered significant.
   FDR were calculated using the Benjamini-Hochberg method to account for
   multiple testing.

Identification and enrichment analysis of IFI6 co-expressed genes

   LinkedOmics ([172]http://www.linkedomics.org) was utilized to analyze
   genes co-expressed with IFI6 in the TCGA BRCA cohort^[173]49. The
   Pearson correlation coefficient was employed to assess statistical
   correlations between IFI6 expression and other genes. Volcano plots
   were generated using the “ggplot2” R package (version 3.4.4), and
   heatmaps of the top 50 positively and negatively correlated genes were
   constructed using “pheatmap” (version 1.0.12). GO functional annotation
   and KEGG pathway enrichment analyses were performed using the “cluster
   Profiler” R package (version 4.8.1). To control for multiple hypothesis
   testing, both Bonferroni correction and Benjamini-Hochberg FDR
   adjustments were applied, and terms with P < 0.05 were considered
   statistically significant.

Immune infiltration analysis

   The ESTIMATE algorithm (R package estimate) was used, alongside Pearson
   correlation analysis, to calculate the correlation between IFI6
   expression and the immune-related score including the immune score,
   stromal score, ESTIMATE score, and tumor purity in breast cancer. The
   CIBERSORT algorithm was applied to determine the infiltration levels of
   22 immune cell types from TCGA expression profiles. Pearson correlation
   analysis was employed to evaluate the relationship between immune cell
   proportions and IFI6 expression levels^[174]50. All correlation
   analyses were adjusted for multiple comparisons using the
   Benjamini-Hochberg method, and FDR < 0.05 was considered statistically
   significant. The significance of differences between two groups was
   determined using the Dunn test, while comparisons among multiple groups
   were evaluated with the Kruskal–Wallis test. Furthermore, the
   association between IFI6 expression and immunomodulators was analyzed
   using TISIDB database
   ([175]http://cis.hku.hk/TISIDB/index.php)^[176]51. Spearman correlation
   coefficients and FDR-adjusted p-values were reported to identify
   significant immunoregulatory interactions.

Single-cell RNA sequencing data processing and analysis

   scRNA-seq data were obtained from [177]GSE176078 and processed using
   the ‘Seurat’ R package (version 5.3.0). Initial quality control steps
   involved filtering cells with fewer than 200 detected genes and genes
   expressed in fewer than 3 cells. Cells with high mitochondrial gene
   content (> 10%) were excluded to remove low-quality or dying cells. The
   filtered gene expression matrix was normalized using the “LogNormalize”
   method with a scale factor of 10,000. Highly variable genes were
   identified for downstream analyses. Principal component analysis (PCA)
   was performed based on the variable genes, and significant principal
   components were selected for clustering using the Louvain algorithm.
   Cell clusters were visualized via Uniform Manifold Approximation and
   Projection (UMAP). Cluster annotation was conducted by comparing
   canonical marker gene expression to known cell type signatures. To
   assess IFI6 expression patterns, cells were dichotomized into
   IFI6-positive and IFI6-negative groups based on normalized expression
   thresholds. The distribution of IFI6 expression among cell clusters was
   visualized using feature plots and stacked bar charts. Differential
   expression analyses between IFI6-positive and negative cells were
   performed where relevant. All analyses were conducted in R (version
   4.4.3).

Supplementary Information

   Below is the link to the electronic supplementary material.
   [178]Supplementary Material 1^ (1.2MB, pdf)

Abbreviations

   IFI6
          Interferon alpha-inducible protein 6

   BC
          Breast cancer

   BLCA
          Bladder urothelial carcinoma

   BRCA
          Breast cancer

   CHOL
          Cholangiocarcinoma

   COAD
          Colorectal cancer

   ESCA
          Esophageal carcinoma

   HNSC
          Head and neck squamous cell carcinoma

   KIRC
          Clear cell carcinoma of kidney

   LIHC
          Liver hepatocellular carcinoma

   PRAD
          Prostate adenocarcinoma

   READ
          Rectum adenocarcinoma

   STAD
          Stomach adenocarcinoma

   UCEC
          Uterine corpus endometrial carcinoma

   KICH
          Kidney chromophobe

   LUSC
          Lung squamous cell carcinoma

   TNBC
          Triple-negative BC

   ILC
          Invasive lobular carcinoma

   SBR
          Scarff-Bloom-Richardson

   NPI
          Nottingham prognostic index

   PR
          Progesterone receptor

   ER
          Estrogen receptor

   AUC
          Area under the curve

   BP
          Biological process

   CC
          Cellular component

   MF
          Molecular function

   Tregs
          Regulatory T cells

   MHC
          Major histocompatibility complex

   mtROS
          Mitochondrial reactive oxygen species

   CTL
          Cytotoxic T lymphocyte

   TCGA
          The Cancer Genome Atlas

   BC-GEM
          Breast cancer gene-expression miner

   ROC
          Receiver operating characteristic

   OS
          Overall survival

   RFS
          Recurrence-free survival

   HR
          Hazard ratios

   GSEA
          Gene set enrichment analysis

   FDR
          False discovery rate

   NES
          Normalized enrichment scores

   GO
          Gene Ontology

   KEGG
          Kyoto Encyclopedia of Genes and Genomes

   ATCC
          American type culture collection

   HR
          Hazard ratios

   CIs
          Confidence intervals

   PCA
          Principal component analysis

   UMAP
          Uniform manifold approximation and projection

   ANOVA
          Analysis of variance

Author contributions

   M.L, and Z.Z. designed research and supervised the experimental work;
   L.J. and C.X. performed all experiments; L.J. and C.X. analyzed the
   data and interpreted the results; L.J. and C.X. wrote the manuscript;
   L.J., M.L, and Z.Z. revised the manuscript. M.L, and Z.Z. funding. All
   authors reviewed and approved the manuscript before submission.

Funding

   This work was supported by the National Natural Science Foundation of
   China (grant number 82274296 to ZZ, grant number 82473449 to ML). This
   work was also supported by the Dalian Medical University
   Interdisciplinary Research Cooperation Project Team Funding (grant
   number JCHZ2023009 to ML).

Data availability

   All data generated or analyzed during this study are included in this
   published article [and its supplementary information files].

Declarations

Competing interests

   The authors declare no competing interests.

Consent for publication

   Not applicable.

Footnotes

   Publisher’s note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   Lili Jiang and Chan Xing contributed equally to this work.

Contributor Information

   Man Li, Email: man_li@dmu.edu.cn.

   Zuowei Zhao, Email: dmuzhaozuowei@163.com.

References