Skip to main content
NCBI home page
Cancer Cell International logoLink to Cancer Cell International
. 2025 Jul 28;25:287. doi: 10.1186/s12935-025-03920-w

Construction and validation of acetylation-related gene signatures for immune landscape analysis and prognostication risk prediction in luminal breast cancer

Mengdi Zhu 1,2,#, Jinna Lin 1,2,#, Haohan Liu 1,3,#, Jingru Wang 4, Nianqiu Liu 5, Yudong Li 1,2, Hongna Lai 1,2, Qianfeng Shi 1,2,✉
PMCID: PMC12302894  PMID: 40722090

Abstract

Background

Epigenetic acetylation plays an essential role in the development and drug resistance of luminal breast cancer. However, the acetylation regulatory network in luminal breast cancer remains underexplored.

Methods

We used the TCGA-BRCA database to explore the acetylation regulatory network in luminal breast cancer. Spearman correlation coefficients, Cox proportional hazards, and the STRING database were used to identify genes that were correlated with acetylation regulatory molecules in luminal breast cancer and could predict patient outcomes. An acetylation regulatory risk model was constructed via Consensus Cluster Plus and the LASSO risk model. GSEA, K‒M survival analysis, and receiver operating characteristic (ROC) curve analysis were used to analyze survival and possible regulatory pathways of the risk model. TIDE, Microenvironment Cell Populations-counter, and CIBERSORT algorithms were used to analyze the immune landscape of the risk model population. Patients’ tumor specimens were used to detect the expression of KAT2B and TAF1L. The luminal breast cancer cell lines MCF-7 and T47D were used in cell viability, Transwell, western blotting, and RT‒qPCR experiments to confirm the risk model. Mouse model was constructed for in vivo validation of KAT2B and TAF1L function.

Results

In our study, we utilized the TCGA-BRCA database to conduct a comprehensive analysis of the acetylation regulatory pattern in luminal breast cancer. Using Consensus Cluster Plus and the LASSO risk model, we screened 6 acetylation-related genes (KAT2B, TAF1L, CDC37, CCDC107, C17orf106, and ASPSCR1) and constructed a 6-gene risk model of luminal breast cancer. Based on this model, luminal breast cancer patients were classified into high- and low-risk subgroups. The high-risk subgroup had a poor prognosis. Further analysis revealed that the high-risk subgroup was associated with lower CD8 + T-cell infiltration and greater responsiveness to immune checkpoint inhibitor therapy. In vitro and in vivo experiments revealed that knockdown of KAT2B and TAF1L dramatically inhibited tumor cell proliferation. In vitro experiments also showed knockdown of KAT2B and TAF1L dramatically inhibited tumor cell migration, increased lymphocyte infiltration, and significantly upregulated the expression of CD8 + T-cell-associated chemokines in luminal breast cancer cells.

Conclusions

In this study, we successfully constructed a 6-gene acetylation-associated risk model for luminal breast cancer, providing a new direction and evidence for personalized treatment. Our results also suggested that KAT2B and TAF1L might serve as potential therapeutic targets in luminal breast cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-025-03920-w.

Keywords: Acetylation, Luminal breast cancer, Prognostic signature, CD8 + T cell, Microenvironment.

Background

Female breast cancer is the most prevalent type of cancer worldwide in women and seriously threatens women’s health. According to Global Tumor Statistics, in 2022, there were more than 2.3 million new cases and approximately 0.67 million deaths among breast cancer patients [1]. Breast cancer can be divided into several subtypes according to the expression of the estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor (HER2), and Ki67 [2]. Among these subtypes, the luminal subtype of breast cancer (ER+) accounts for approximately 70%, which can be further divided into the luminal A subtype and luminal B subtype. Different subtypes of breast cancer exhibit very distinctive tumor biology and clinical features. The luminal subtype of breast cancer is estrogen dependent for tumor growth, and its standard treatment regimen is 5–10 years of endocrine therapy. Endocrine therapy is thought to reduce the recurrence and mortality of luminal breast cancer; nevertheless, primary, or secondary drug resistance is still inevitable [3]. How to precisely cluster high-risk patients and how to find new effective therapeutic targets for these patients are key clinical questions to be addressed.

Breast cancer has been established as a highly heterogeneous disease, even within subtypes. Large-scale sequencing-based studies have shown that heterogeneity in breast cancer is also evident at the genetic level [4, 5]. There are several methods available for staging and risk grouping of breast cancer patients based on sequencing data, such as PAM50 staging and the 21-gene recurrence risk score. Furthermore, several molecular tests based on gene expression profiles, such as MammaPrint, Oncotype Dx, and Prosigna, have entered mainstream care for luminal breast cancer patients [6]. While these gene expression analyses allow for effective risk subtyping of luminal breast cancer, no possible therapeutic targets have been validated.

Epigenetic modifications, such as methylation, acetylation, and phosphorylation, regulate the content and function of nucleic acids and proteins. Epigenetic regulation has been reported to participate in many aspects of diseases, especially cancers [7]. Histone and nonhistone lysine acetylation are epigenetic regulatory mechanisms that are thought to be involved in tumor initiation, development, and metastasis. The regulators of lysine acetylation have been reported to be dysregulated in many cancer types [8–10]. Furthermore, lysine acetylation inhibitors have extraordinary therapeutic effects on cancers, such as lymphoma, lung cancer, colon cancer, pancreatic cancer, and breast cancer, which highlights the important role of lysine acetylation regulation in cancers, especially luminal breast cancer [8]. However, not all luminal breast cancers can benefit from epigenetic therapy due to the wide range of acetylation regulation and the heterogeneity of breast cancers. Acetylation-associated analysis and risk subtyping of luminal breast cancers is highly important for these patients.

Lysine acetylation can regulate gene expression in many ways. Histone lysine acetylation can promote the transcription of genes through relaxing chromatin. Nonhistone lysine acetylation can regulate protein expression through degradation and protein stability [11, 12]. Many important genes that play essential roles in breast cancer are regulated by hyperacetylation or hypoacetylation of lysine. For example, SIRT7 has been reported to deacetylate nucleophosmin (NPM), a member of the RAS oncogene family (Ran), and FKBP prolyl isomerase 51 (FKBP51), hence regulating the stability of p53, the nuclear export of NK-κB, and the activation of AKT1 [13–15]. Furthermore, SIRT3 has been reported to regulate metabolism via the deacetylation of lysine residues. Li-Ting Wang reported that KAT2B could acetylate the transcription factor intestine-specific homeobox (ISX) and regulate the expression of epithelial‒mesenchymal transition factors [16]. Based on large-scale transcriptome sequencing data, several studies have investigated the prognostic value associated with acetylation and the role of acetylation signatures in tumors, such as colorectal cancer, ovarian cancer, prostate cancer, and breast cancer [17–20]. However, the global influence of lysine acetylation in luminal breast cancer remains unclear.

The tumor immune microenvironment is thought to play an essential role in tumor development. Tumors tend to evade the immune system to protect themselves. Today, a variety of immune checkpoint inhibitors (ICIs) have been implemented in clinical oncology treatment, but the effectiveness of immunotherapy is less than satisfactory due to the highly heterogeneous nature of tumors. Luminal breast cancer is thought to be a “cold tumor” due to its low infiltration of immune cells in the tumor microenvironment [21]. Determining whether some luminal breast cancer patients can benefit from immune therapy and how to screen for these patients are key and difficult parts of precision therapy.

In this study, we integrated data from 481 luminal breast cancer patients and 65 paired paracancerous tissues from the TCGA database to identify lysine acetylation regulator modification patterns via unsupervised clustering of the gene expression of 37 lysine acetylation regulators. We focused on exploring regulatory molecules of acetylation and conducted correlation analysis to identify related genes. Additionally, we performed Cox proportional hazards survival analysis to determine the significance of acetylation regulators and their associated genes in relation to patient survival. Subsequently, consensus cluster analysis was conducted to identify and discover potential new subclasses in luminal breast cancer via the 30 selected genes. To further screen for molecules significant for the survival and classification of patients with luminal breast cancer, we used a LASSO regression model and validated our findings with internal validation. A total of six molecules (KAT2B, TAF1L, CDC37, CCDC107, C17orf106, and ASPSCR1) were included in the model. Based on this 6-gene model, luminal breast cancer patients were divided into high-risk and low-risk groups. We analyzed the survival curves, ROC curves, and immune landscapes of the high- and low-risk groups. The results indicated that the high-risk group exhibited lower CD8 + T-cell inflammation and was more likely to benefit from immune checkpoint inhibitors. Finally, two of the six genes were selected to validate their molecular functions in vitro in luminal breast cancer cell lines. Our results revealed an acetylation-associated 6-gene risk model in luminal breast cancer that has the potential to be used for clinical risk stratification. Additionally, our study identified KAT2B and TAF1L as potential therapeutic targets in high-risk luminal breast cancer.

Materials and methods

Data acquisition

The RNA-Seq data of 1247 breast samples (65 paired breast epithelial cells samples and 481 luminal breast tumor samples) and corresponding clinical information were obtained from The Cancer Genome Atlas (TCGA) website (https://portal.gdc.cancer.gov/projects/TCGABRCA). Ensemble IDs were converted to official gene symbols using DAVID website, and log2 processing of the data was performed.

Identification of acetylation-related mRNAs

A total of 33 regulators were selected according to the previous study, including the transferase family (KAT2A, KAT2B, KAT3A, KAT3B, KAT5, KAT6A, KAT6B, KAT7, KAT8, TAT1, ESCO1, ESCO2, and KAT1), the deacetylase family (HDAC 1–11, SIRT 1–7, TCF 1, and LEF 1), and acetylation sites readers (BRDT, BRD2-4) [11, 22]. Subsequently, acetylation-related mRNAs were identified using the methos from de Winter JC, et al. (|R2|>0.6 and P < 0.001) [23].

Cox proportional hazards survival analysis

To screen for acetylation regulators and related mRNAs that have prognostic significance in luminal breast cancer, we used Cox regression model [24]. The analysis was performed and plotted at Sanger box (http://www.sangerbox.com/tool) (P < 0.001, KM-P < 0.01).

Construction of the co-expression network

To demonstrate the correlation between the acetylation regulators and their corresponding mRNAs, the acetylation regulators-mRNA co-expression network was constructed and plotted at Cytoscape [25]. Furthermore, we analyzed the protein interaction among the selected molecules and plotted the protein interaction network at the STRING database (https://string-db.org/) [26].

Consensus clustering analysis

For identifying intrinsic groups sharing biological characteristics, intersection prognostic genes were used to find the ‘cleanest’ cluster partition where items nearly always either cluster together giving a high consensus in luminal breast cancer using the “Consensus Cluster Plus” R package [27]. We performed survival analysis and Gene Set Enrichment Analysis (GSEA) of the clusters at Sanger box (http://www.sangerbox.com/tool).

Construction of risk model

We numbered luminal breast cancer cases in the TCGA cohort randomly and divided them into training and testing sets in a 2:1 ratio. Next, we performed the LASSO regression analyses and generate a risk model by control of the first-rank value of Log(λ) at the minimum likelihood of deviance using caret and “glmnet” R packages. The risk score of each luminal breast cancer patient was calculated with the following formula:

graphic file with name d33e415.gif

In the training, testing, and all groups, we visualized risk scores, the status, survival analyses, and receiver operating characteristic (ROC) curves of luminal breast cancer patients using Sanger box (http://www.sangerbox.com/tool). The risk score formula was KAT2B × 0.0581 + TAF1L × 0.3866 + CDC37 × (−0.0476) + CCDC107 × (−0.1235) + C17orf106 × (−0.4843) + ASPSCR1 × (−0.0906).

Drug sensitivity prediction

To predict the response of the high and low-risk luminal breast cancer groups to immune checkpoint inhibitors (ICIs, anti-PD1 and anti-CTLA4 therapy), Tumor Immune Dysfunction and Exclusion (TIDE) algorithm was used (http://tide.dfci.harvard.edu/).

Assessment of immune cell infiltration and immune microenvironment

The difference in immune cell infiltration in the two groups of patients was evaluated using the CIBERSORT algorithm(http://CIBERSORT.stanford.edu/) [28]. Furthermore, we used the R package “MCPcounter” to fit the infiltration degree of 10 immune cells (T cell, CD8 T cell, cytotoxic lymphocyte, B lineage, NK cell, monocytic lineage, myeloid dendritic cell, neutrophil, endothelial cell, and fibroblasts) in the high and low-risk groups based on expression profile data. P < 0.05 was considered significant.

Log-rank survival analysis

The overall survival of luminal breast cancer patients in the SCAN-B (The Sweden Cancerome Analysis Network - Breast) and METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) databases were analyzed and plotted using Breast Cancer Gene-Expression Miner v4.8 (http://bcgenex.ico.unicancer.fr/BC-GEM/GEM-Accueil.php?js=1). The best cutoff value was taken for all survival analyses. The METABRIC database predominantly comprises patient populations from the United Kingdom and Canada, whereas the SCAN-B study focuses on Swedish patient cohorts. Neither of these populations overlaps with the patient cohort included in the TCGA database.

Cell culture

The luminal breast cancer cell lines MCF-7 and T47D were obtained from American Type Culture Collection (ATCC). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco) with 10% fetal bovine serum (FBS, Newzerum). Cells were cultured using the medium described above, and the medium was changed every 2 days. When the cell density reached approximately 70%−80%, cell passaging was performed. The medium was removed, and 1 ml trypsin was added to digest cells for about 30 s–1 min, then 1 ml medium containing serum was added to terminate digestion. Cells were gently blown off the bottom of the culture bottles, then transferred to a 15 ml centrifuge tube, and centrifuged at 1000 rpm for 3 min. The liquid was removed, and cells were resuspended with 1 ml complete medium. Cells were passage into a new culture bottle at 1:3 − 1:4. The number of cell passages in a single experiment should not exceed 15 times.

Si-RNA transfection

For transfection of si-RNAs, cells were seeded at 1 × 105 cells per well in 6-well plates, the cell density was about 60–70%. Then cells were transfected with negative control (NC) or specific siRNAs for KAT2B and TAF1L (100 nM, Gene Pharma, China) respectively using Invitrogen™ Lipofectamine™ RNA iMAX Transfection Reagent. The transfection system was as follows: 250 µl optimal medium with 25 pmol siRNA and 7.5 µl RNA iMAX. The transfection system was mixed well and incubated for 15 min before being added to the cells. After transfection for 24 h, the culture medium of cells was discarded and replaced with new complete medium for another 24 h. The siRNA sequences were listed in Table 1.

Table 1.

SiRNA sequences we used in the in vitro experiments

Gene Forword 5’−3’ Reverse 5’−3’
KAT2B-1 CCGCAUCAACUAUUGGCAUTT AUGCCAAUAGUUGAUGCGGTT
KAT2B-2 GCGACAACUCCUGGAACAATT UUGUUCCAGGAGUUGUCGCTT
TAF1L-1 GCUCCAGUGUAUCUUCAUATT UAUGAAGAUACACUGGAGCTT
TAF1L-2 GACCCAAACAACCCUUCAUTT AUGAAGGGUUGUUUGGGUCTT
NC* UUCUCCGAACGUGUGACGUTT ACGUGACACGUUCGGAGAATT
Open in a new tab

*NC: Negative Control

RNA extraction

Cells were seeded into 6-well plates, and after treatment, total RNA was extracted using the RNA Quick Purification kit (ESscience, China, RN001). Cells were lysis with 500 µl lysis buffer, and then 500 µl ethanol absolute was added and mixed well. The mixture was centrifuged in paired centrifugal columnists at 12,000 rpm for 1 min and then the centrifuged liquid was discarded. The RNA was then washed with 500 µl wash buffer. Finally, 30–50 µl of diethylpyrocarbonate (DEPC) enzyme-free water was used to dissolve the RNA. Total RNA concentration and purity were analyzed in duplicate using a NanoDrop One (Thermo Fisher Scientific, Inc.; cat. no. AZY1705838).

Reverse transcription and quantitative real-time PCR (qRT-PCR)

PrimerScript RT Master Mix (RR036A, Takara) was used to generate cDNA. The reverse transcription reaction system contained 2 µl of 5X prime script RT master mix, 1000 µg of RNA, and DEPC water. The reverse transcription procedure was as follows: 15 min at 37 ℃, 5 s at 85 ℃, and then 4 ℃ for restoration. Real-time quantitative PCR was performed using TB Green Premix Ex Tap II (RR820A, Takara). The RT-qPCR system contained 5 µl SYBR premix ex taqII, 0.4 µl forward primer, 0.4 µl reverse primer, 3.2 µl DEPC water, and 1 µl cDNA product. The reactions were carried out in the LightCycler480 system. The reaction protocol was as follows: 95 °C for 10 min; followed by 40 cycles at 95 °C for 10 s and 60 °C for 30 s. The gene-specific primer sequences were listed in Table 2.

Table 2.

, primer sequences we used in the in vitro experiments

Gene Forword 5’−3’ Reverse 5’−3’
GAPDH ACAACTTTGGTATCGTGGAAGG GCCATCACGCCACAGTTTC
KAT2B CGAATCGCCGTGAAGAAAGC CTTGCAGGCGGAGTACACT
TAF1L TCCGGGCGACTGTTGTTTTA CCGAGTCTGACATGATGGCG
CCL2 GCAATCAATGCCCCAGTCAC GACACTTGCTGCTGGTGATTC
CCL5 CAGTCGTCCACAGGTCAAGG CTTGTTCAGCCGGGAGTCAT
CXCL9 CCAGTAGTGAGAAAGGGTCGC AGGGCTTGGGGCAAATTGTT
CXCL10 GTGGCATTCAAGGAGTACCTC TGATGGCCTTCGATTCTGGATT
CXCL11 GACGCTGTCTTTGCATAGGC GGATTTAGGCATCGTTGTCCTTT
Open in a new tab

Western blotting

Protein was extracted from cells using RIPA lysis buffer which contained protease and phosphatase inhibitors (Thermo Fisher, 78442). The relative protein density was calculated using Pierce BCA Protein Assay Kit (Invitrogen, 23227). Protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred to polyvinylidene difluoride membranes. The membranes were incubated with specific primary antibodies (KAT2B, TAF1L, and GAPDH) at 4 ℃ overnight, followed by anti-mouse or anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibodies (CST, 7076/7074, dilution 1:1000) at room temperature for 1 h. Afterward, the protein–antibody complex was visualized by enhanced chemiluminescence assay (34095, Pierce). Primary antibodies against KAT2B (PTM Bio, PTM-5711, dilution 1:1000), TAF1L (Proteintech, 55170-1-AP-50UL, dilution 1:1000), and HRP-conjugated glyceraldehyde 3-phosphate dehydrogenase antibody (Proteintech, HRP-60004, dilution 1:5000) were used.

Cell viability assays

After transfection with indicated siRNAs, cells were then seeded in 96-well-plates at 1500 per well for cell viability detection. Relative cell viability was detected using MTT (Biofrox, 3580MG250). Briefly, MTT was configured into a 5 mg/ml solution in sterile phosphate-buffered saline (PBS) and added into cell culture media with a ratio of 1:10 at 0–5 days (identify 6 h after planted as 0 day when the cell is attached) respectively. After incubation at 37 ℃ for 4 h, the culture medium was aspirated and the precipitate was dissolved in DMSO, then absorbance at a wavelength of 490 nm was detected by a microplate spectrophotometer. We compared the 1–5 days absorbance with the day 0 absorbance in each group and plot the fold change of relative cell viability using the GraphPad Prism 8.0 software.

Colony formation assay

For colony assay, after being transfected with indicated siRNA for 48 h, cells were plated at 1000 cells per well in 6-well plates and cultured with the complete culture medium for 14–21 days (MCF-7 cells for 14 days and T47D cells for 21 days). Then the culture medium was removed, and cells were fixed with 4% paraformaldehyde for 15 min and stained with 0.1% crystal violet for 20 min. The colony formation was shoot and automatic counting using the fluorescent enzyme-linked immune-spot analyzer (AID vSpot Spectrum). We compared between groups using counted cell colony counts.

Primary human immune cells

Primary peripheral blood mononuclear cells were obtained and isolated from peripheral blood of healthy donors from the Guangzhou Blood Center. The blood samples were negative for antibodies against hepatitis C virus, hepatitis B virus, HIV, and syphilis. All related procedures were performed with the approval of the Internal Review and Ethics Boards of the Sun Yat-Sen Memorial Hospital.

Peripheral blood lymphocytes isolation

Peripheral blood monocytes were isolated by Ficoll density gradient centrifugation. CD8 + T cells were isolated by magnetic-activated cell sorting using direct CD8 Isolation Kit (Cat# 130-094-156, Miltenyi Biotec) according to the manufacturer’s instructions. CD8 + T cells were cultured in RPMI 1640 medium (GIBCO) supplemented with 25mmol/L HEPES, 4mmol/L L-glutamine, 25 µmol/L 2-mercaptoethanol, and 10% FBS at 37 °C. Media were replaced every 2 days.

Trans-well assay

To detect the cell migration ability, transwell assay was performed. After transfection with indicated siRNAs, cells were seeded in the upper 8 μm transwell chambers (Corning, 353497) at 1 × 105 cells with serum-free culture medium. Then 500 µl culture medium with 20% serum was added to the lower transwell chambers. The chambers were incubated in 37 ℃ incubators for 48 h and cells were then fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. After stained, the cells on the upper chambers were wiped off gently, and the cells that migrated to the bottom of the chambers were observed and shot using an inverted light microscope. The number of migration cells was relatively quantitated using ImageJ.

For the T-cell chemotaxis assays, 4 μm transwell chambers were used (Corning, 3460-48EA). The T-cells isolated from human blood samples were seeded in the upper 4 μm transwell chambers at 1 × 105 cells with serum-free 1640 culture medium. And the culture medium which had incubated with cells for 24 h after transfection was added to the bottom of the chambers. The transwell chambers were incubated for 6 h, and the T-cells that migrated to the bottom of the chambers were counted under the microscope using cell counting plates.

Animal tumor model

We used 3-week-old Bal/bc female nude mice for animal model construction. Estrogen was embedded subcutaneously in the back of the mice and used as a tumor estrogen growth stimulus. MCF-7, MCF-7-sh-KAT2B cells were pre-prepared. About 5 days after burying the estrogen tablets the mice were randomly divided into 3 groups of 5 mice each, and each group was planted with 1*107 MCF-7, MCF-7SH cells at the subcutaneous mammary pads, respectively. The tumor volume was measured every 3 days after about 1 week when the tumor had grown to a palpable size. 30 days later, the mice were euthanized, and the tumors were dissected and collected.

Statistical methods

The Wilcox test was used to compare the proportion of tumor-infiltrating immune cells. Spearman correlation analysis was used to analyze the correlation between acetylation regulators and mRNA. Differences in the proportions of clinical characteristics were analyzed by the chi-squared test. Cox univariate regression analysis and multivariate Cox regression analysis were implemented to define the independent prognostic factor for overall survival (OS). The predictive accuracy of the prognostic model for OS was evaluated by performing time-dependent ROC curve analysis. P < 0.05 was considered statistically significant. * P < 0.05, ** P < 0.01, *** P < 0.001.

Results

Identification of acetylation-related genes in luminal breast cancer

We searched for 37 acetylation regulatory molecules through a literature review, including acetyltransferases (KAT2A, KAT2B, KAT3A, KAT3B, KAT5, KAT6A, KAT6B, KAT7, KAT8, TAT1, ESCO1, ESCO2, and KAT1), deacetylases (HDACs 1–11, SIRT 1–7, TCF1, and LEF1), and ε-N-lysine acetylation motif readers BRDT and BRD2-4 [11, 22]. To identify potential genes regulated by acetylation, Spearman correlation coefficients was performed among the acetylation regulators and the whole-genome sequencing results in the TCGA database. A total of 570 genes whose correlation coefficients with acetylation regulators ∣R ∣≥ 0.6 and P < 0.05 were selected. To further screen genes that are significant for survival in patients with luminal breast cancer, Cox proportional hazards survival analysis was performed. A total of 30 acetylation genes whose hazard ratios (HRs) were P < 0.001 and whose KM values were P < 0.01 were selected. The forest plot of the 30 genes revealed that EXOC6B, TAF1L, APOOL, and KAT2B were risk factors for luminal breast cancer, and the remaining 26 genes were protective factors (Fig. 1A). Next, to obtain a comprehensive picture of the correlations of the 30 genes with the 37 acetylation regulatory genes, we plotted a correlation heatmap. The overall correlation heatmap is displayed in Supplementary Fig. 1 A, and the parts with p values less than 0.05 are displayed in Fig. 1B. According to the correlation heatmap, EP300, KAT2B, EXOC6B, TAF1L, and APOOL were mostly negatively correlated with the other genes (Fig. 1B). Further coexpression analysis revealed that most of the molecules were coexpressed with SIRT6. EP300 was coexpressed with EXOC6B, TAF1L and APOOL. SIRT7 was coexpressed with GADD45GIP1 and C17orf106. KAT2B was not coexpressed with any other molecules (Fig. 1C).

Fig. 1.

Fig. 1

Open in a new tab

Identification of 30 acetylation-related genes in luminal breast cancer. (A) The forest figure of 30 genes we selected which were both correlated with acetylation regulatory factors and significantly associated with the prognosis of luminal breast cancer. (B) The heatmap showed the correlation index of the 30 acetylation related genes and 37 acetylation regulatory factors whose p value <0.05. (C) The co-expression network revealed the co-express relationship of 30 acetylation related genes and 37 acetylation regulatory factors. The blue points represented acetylation related genes, and the red points represented acetylation regulatory molecules. (D) The protein interaction network of 30 acetylation related genes and 37 acetylation regulatory factors was plotted at The String database

A protein correlation network was subsequently plotted using the STRING website, and except for CCDC12, GIPC1, TNE1, CCDC107, LENG1, EXOC6B and C14orf80, the remaining molecules all had predictable or proven interactions with one or more molecules at the protein level (Fig. 1D). Overall, we identified 30 molecules associated with acetylation regulatory genes that were significantly correlated with the outcomes of patients with luminal-type breast cancer.

Constructed acetylation subclusters of luminal breast cancer

As acetylation-associated genes might contribute to the development of luminal breast cancer, to further characterize the parable pattern of acetylation regulatory subtypes, cluster models were constructed. Based on the expression of the 30 acetylation-related genes, Consensus Cluster Plus was utilized to identify the different subtypes (K = 2–9) in the risk model. According to the cumulative distribution function (CDF) curves (Supplementary Fig. 2A), delta area plots (Supplementary Fig. 2B), tracking plots (Supplementary Fig. 2C), and CM heatmaps (Fig. 2A), when k = 2, the sample cluster was stable and robust. Furthermore, when the luminal breast cancer cluster was divided into two subclusters, significant survival differences between subclusters were detected (Fig. 2B). The overall survival (OS) time of patients in subcluster 2 was significantly lower than that of patients in subcluster 1 (HR = 0.23, 0.11–0.50; P = 0.000062) (Fig. 2B). Then, the expression and clinical feature (TNM status, stage, HER2 status, and age) heatmaps were compared (Fig. 2C). The results revealed a greater proportion of patients with stage III–IV tumors (26% vs. 22%) and higher T (37% vs. 25%) and N ratios (56% vs. 50%) in Cluster 1, with poorer prognoses. To further understand the possible functions of the clusters, we performed GSEA of Cluster 1 and Cluster 2 (Fig. 2D and E). The results revealed that hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, the phosphatidylinositol signaling system, inositol phosphate metabolism, and N-glycan biosynthesis were enriched in Cluster 1 (Fig. 2E). Oxidative phosphorylation, Parkinson’s disease, ribosome, proteasome, and base excision repair were enriched in Cluster 2 (Fig. 2E). GSEA revealed possible changes in cell functions and pathways in Clusters 1 and 2, providing possible directions for further research in these clusters.

Fig. 2.

Fig. 2

Open in a new tab

Construction of acetylation-related sub-clusters of luminal breast cancer. (A) Consensus Cluster Plus results showed two different subtypes among the risk model. (B) Survival analysis showed that two clusters exhibited significantly different prognosis (HR = 0.23, 95CI% 0.11–0.50, p = 0.000062). (C) The clinical features of cluster 1 and cluster 2 (age, stage, TNM stage, and HER2 status) were showed. D-E. GSEA analysis of cluster 1 and cluster 2

Risk model based on the six subcluster genes

To explore the prognostic value of these 30 acetylation genes, samples from the TCGA-BRCA luminal breast cancer cohort were randomly divided into training (n = 292) and validation (n = 146) groups at a ratio of 2:1.

We built a LASSO risk model of six genes when the first-rank value of Log(λ) was at the minimum likelihood from the 30 acetylation genes in the training group. The risk formula was as follows: risk score = KAT2B × 0.0581 + TAF1L × 0.3866 + CDC37 × (−0.0476) + CCDC107 × (−0.1235) + C17orf106 × (−0.4843) + ASPSCR1 × (−0.0906). The cvfit and lambda figures are shown in Fig. 3A and B. The formula was subsequently validated in the validation group and total patient group. A risk score based on this formula was calculated for the training, validation, and total groups. The cutoff risk score was found to be 0.5, and based on it, the patients were subdivided into low- and high-risk groups (Fig. 3C and F, and 3I). The bottom graphs show the OS time and final follow-up status (dead or alive) for each patient in the scatter plots.

Fig. 3.

Fig. 3

Open in a new tab

LASSO risk model of acetylation related genes in luminal breast cancer. A-B. The cvfit (A) and lambda (B) parameter of the luminal breast cancer LASSO risk model. (C) Risk score of the training group, the cut-off risk score was 0.5. Patients were divided into high-risk group and low-risk group according to the risk score. (D) Survival analysis of the high-risk and low-risk groups in the training group. (E) ROC curve of high-risk and low-risk groups in the training group. (F) Risk score of the validation group. (G) Survival analysis of high-risk and low-risk groups in the validation group. (H) ROC curve of high-risk and low-risk groups in the validation group. (I) Risk score of the validation group. (J) Survival analysis of high-risk and low-risk groups in all patients group. (K) Risk score of all patients group

K‒M survival analysis and ROC curve analysis at 1 year, 3 years and 5 years were performed to explore the independent outcome predictive potential of our signature. The K‒M survival curve revealed that the survival time of the high-risk group was significantly shorter than that of the low-risk group, whether in the training group, validation group, or all patients group (Fig. 3D and G, and 3J). The HR of the training group was 2.72, 95% CI 1.74–4.25, P < 0.0001 (Fig. 3D). The hazard ratio (HR) of the validation group was 2.72 (95% CI 1.69–4.38, P = 0.00033) (Fig. 3G). The hazard ratio (HR) of all patients was 4.57, 95% CI 2.56–8.16, P < 0.0001 (Fig. 3J). Furthermore, the areas under the curve (AUCs) for 3- and 5-year OS were all > 0.700 in the training, testing and all patient groups (Fig. 3E and H, and 3K).

Immune landscape and ICI response of the risk model

Immune checkpoint inhibitors (ICIs) are promising drugs for treating breast cancer. However, the effect of ICIs on luminal breast cancer is not satisfactory. To predict the possible response to ICIs (anti-PD1 and anti-CTLA4 therapy), the TIDE database was used. The results revealed that the possible responder patient ratio was significantly greater in the high-risk group than in the low-risk group (Fig. 4A), indicating that high-risk patients might benefit from ICI treatment. Next, to predict the difference in immune cell infiltration between the high- and low-risk groups, we used the Microenvironment Cell Populations-counter and CIBERSORT algorithms to assess immune infiltration in the TCGA-BRCA sequencing data. The Microenvironment Cell Populations-counter (MCP) algorithm can predict 10 immune cells, including T cells, CD8 + T cells, cytotoxic lymphocytes, B lineages, NK cells, monocytic lineages, myeloid dendritic cells, neutrophils, endothelial cells, and fibroblasts. The MCP results revealed a greater percentage of infiltrating CD8 + T cells in the low-risk group (P < 0.01) and a greater percentage of infiltrating neutrophils in the high-risk group (P < 0.05) (Fig. 4B). Moreover, CIBERSORT revealed that CD8 + T cells, regulatory T cells (Tregs), and M0 macrophages had greater infiltration percentages in the low-risk group (P < 0.001) than in the high-risk group, and CD4 + memory T cells and resting mast cells had greater infiltration percentages in the high-risk group (P < 0.001) (Fig. 4C). Both algorithms revealed that the number of CD8 + T cells was greater in the low-risk group than in the high-risk group, indicating the potential role of CD8 + T cells in our risk models.

Fig. 4.

Fig. 4

Open in a new tab

Immune signature and immune check point inhibitors (ICI) response of the LASSO risk model. (A) TIDE database results showed the predicted response value of ICI in high risk and low risk groups. The blue data represented possible non-responders; the red data represented possible responders. (B) Microenvironment Cell Populations-counter database reveled the 10 immune cells (T cell, CD8 T cell, cytotoxic lymphocyte, B lineage, NK cell, monocytic lineage, myeloid dendritic cell, neutrophil, endothelial cell, and fibroblasts) algorithm result of high and low risk groups. (C) The CIBERSORT algorithms results showed the 22 immune cells fit of high and low risk groups. (D) The expression of 79 critical immune checkpoint genes which were significantly different in high and low groups. * p < 0.05, ** p < 0.01, *** p < 0.001

We also compared the expression of critical immune checkpoint genes in the high- and low-risk groups, and the immune checkpoint genes were extracted from an article by Fei-Fei Hu [29]. A total of 79 genes were included (Supplementary Fig. 2D), 23 of which were significantly different (P < 0.05) between the high- and low-risk groups (Fig. 4D). Most of the immune checkpoint genes were upregulated in the low-risk group, whereas CD274 (PDL1), BTN2A1, CD160, and CD80 were downregulated in the low-risk group. Our results revealed significantly different immune infiltration characteristics between the high- and low-risk groups, which might explain the different predicted responses to ICI treatment in the two groups.

KAT2B and TAF1L affected luminal breast cancer malignancy in vitro

To further validate and explore the role of the six genes we included in the risk model, K‒M survival analysis and receiver operating characteristic (ROC) curves of each gene in the TCGA-BRCA luminal breast cancer database were analyzed and plotted. High expression of KAT2B and TAF1L was related to poor clinical outcomes (Figs. 5A and 6A), whereas high expression of ASPSCR, C17orf106, CCDC107, and CDC37 was related to better clinical outcomes (Supplementary Fig. 3A-D). The ROC curve revealed that the AUCs of KAT2B and TAF1L for predicting 1-, 3-, and 5-year survival were > 0.6 (Figs. 5B and 6B). The ROC curves of the remaining genes are displayed in Supplementary Fig. 3A-D. The prognostic role of KAT2B and TAF1L indicated their possible tumor-promoting roles. To validate the survival role of KAT2B and TAF1L, we analyzed survival in the SCAN-B and METABRIC luminal breast cancer databases. The results showed that in the SCAN-B database, higher expression of KAT2B was associated with a shorter OS time (HR = 1.40, 1.06–1.85; P = 0.0196) (Fig. 5C). According to the METABRIC database, higher expression of TAF1L was associated with a shorter OS time (HR = 1.39, 1.12–1.73; P = 0.0032) (Fig. 6C). The above analysis showed that high expression of genes KAT2B and TAF1L in luminal breast cancer predicted a poor prognosis. This suggested that inhibitors targeting KAT2B and TAF1L might be effective in suppressing luminal breast cancer growth. Meanwhile, we collected 7 pairs of luminal breast cancer and paracancer tissues in the clinic and used immunohistochemistry experiments to detect the expression levels of KAT2B and TAF1L in cancer and paracancer. The results showed that KAT2B and TAF1L were highly expressed in tumor tissues (Supplementary Fig. 4 A). Therefore, we further validated the function of these two molecules in vitro and in vivo.

Fig. 5.

Fig. 5

Open in a new tab

Knock down of KAT2B inhibited the proliferation and invasion of luminal breast cancer cells. (A) Survival analysis of KAT2B in luminal breast cancer in TCGA database. HR = 1.82, 95%CI 1.28–2.6, p < 0.0001. (B) ROC curve of KAT2B in luminal breast cancer. (C) Survival analysis of KAT2B in luminal breast cancer in the SCAN-B database. (D) Cell viability assay and proliferation curve showed knock down of KAT2B could inhibit the proliferation of MCF-7 cell. (E) Colony formation assay showed knock down of KAT2B could decrease the colony formation ability of MCF-7 cell. The statistic graph was shown in the right. (F) Cell viability assay and proliferation curve showed knock down of KAT2B could inhibit the proliferation of T47D cell. (G) Colony formation assay showed knock down of KAT2B could decrease the colony formation ability of T47D cell. The statistic graph was shown at right. H-I. Trans-well assay exhibited that knock down of KAT2B could inhibit the invasion ability of MCF7 (H) and T47D (I) cells. The statistics graph was shown at right. *** p < 0.001

Fig. 6.

Fig. 6

Open in a new tab

Knock down of TAF1L inhibited the proliferation and invasion of luminal breast cancer cells. (A) Survival analysis of TAF1L in luminal breast cancer in TCGA database. HR = 1.82, 95%CI 1.28–2.6, p < 0.0001. (B) ROC curve of TAF1L in luminal breast cancer. (C) Survival analysis of TAF1L in luminal breast cancer in the METABRIC database. (D) Cell viability assay and proliferation curve showed knock down of TAF1L could inhibit the proliferation of MCF-7 cell. (E) Colony formation assay showed knock down of TAF1L could decrease the colony formation ability of MCF-7 cell. The statistic graph was shown in the right. (F) Cell viability assay and proliferation curve showed knock down of TAF1L could inhibit the proliferation of T47D cell. (G) Colony formation assay showed knock down of TAF1L could decrease the colony formation ability of T47D cell. The statistic graph was shown at right. H-I. Trans-well assay exhibited that knock down of TAF1L could inhibit the invasion ability of MCF7 (H) and T47D (I) cells. The statistics graph was shown at right. *** p < 0.001

The human luminal breast cancer cell lines MCF-7 and T47D were used to perform the in vitro experiments. RT‒qPCR and western blot assays revealed the knockdown efficiency of si-KAT2B and si-TAF1L in MCF-7 and T47D cells (Supplementary Fig. 4B‒E). A cell viability assay revealed that knocking down KAT2B and TAF1L inhibited the cell proliferation rate (Figs. 5F and 6F). Moreover, the colony formation assay results revealed that knockdown of KAT2B and TAF1L significantly reduced their colony formation ability (Figs. 5G and 6G). Since KAT2B and TAF1L are reportedly related to tumor metastasis, we used a Transwell assay to detect changes in cell migration ability after knockdown of KAT2B and TAF1L. The results showed that knocking down KAT2B and TAF1L significantly inhibited the migration of MCF-7 and T47D cells (Figs. 5H-I and 6H-I).

Next, we constructed overexpressing plasmid of KAT2B and TAF1L in MCF-7 and T47D cells and detected their proliferative functions (Supplementary Fig. 5A-B). Cell proliferation assay and colony assay showed that overexpression of KAT2B and TAF1L significantly promoted the proliferation of luminal breast cancer cells (Supplementary Fig. 5C-E, G-I). Meanwhile, Transwell assay showed that overexpression of KAT2B and TAF1L could promote the invasive ability of luminal breast cancer cells (Supplementary Fig. 5F and J).

These results suggest that KAT2B and TAF1L can promote the proliferation and metastasis of luminal breast cancer cells, further confirming their cancer-promoting effects. Our results also indicated that KAT2B and TAF1L might be promising therapeutic targets in luminal breast cancer.

Silencing KAT2B and TAF1L increased lymphocyte infiltration in vitro

Our previous results revealed that the risk model we constructed was related to CD8 + T-cell infiltration (Fig. 4B-C). To explore the correlation between CD8 + T-cell infiltration and KAT2B and TAF1L expression, we used qPCR to examine the changes in the expression of chemokines associated with CD8 + T-cell chemotaxis after KAT2B and TAF1L knockdown (Fig. 7A-D). The results showed that after silencing KAT2B, chemotaxis factors CCL5, CXCL10, and CXCL11 were upregulated in both the MCF-7 and T47D cell lines (Fig. 7A-B). Moreover, CCL5, CXCL10, and CXCL11 were also upregulated after TAF1L was silenced in the MCF-7 and T47D cell lines (Fig. 7C-D). The upregulation of chemotaxis factors suggested that KAT2B and TAF1L might be negatively correlated with CD8 + T-cell infiltration; hence, a CD8 + T-cell chemotaxis assay was performed (Fig. 7E) and we counted the lymphocytes that crossed the Transwell membranes. The results showed that silencing KAT2B and TAF1L increased the percentage of CD8 + T cells that migrated through the membrane into the bottom Transwell chamber (Fig. 7F-G). Our results suggest that knockdown of KAT2B and TAF1L upregulates the expression of the chemokines CCL5, CXCL10, and CXCL11. Functional experiments revealed that knockdown of KAT2B and TAF1L induced an increase in lymphocyte chemotaxis, suggesting that the expression of KAT2B and TAF1L negatively correlates with the chemotaxis of CD8 + T cells.

Fig. 7.

Fig. 7

Open in a new tab

Knock down of KAT2B and TAF1L increased lymphocyte infiltration in vitro. A-B. RT-qPCR showed the expression of CD8 + T cell related chemokines in MCF-7 (A) and T47D (B) cells after silencing of KAT2B. C-D. RT-qPCR showed the expression of CD8 + T cell related chemokines in MCF-7 (C) and T47D (D) cells after silencing of TAF1L. (E) The schematic of the lymphocyte chemotaxis experiment. F-G. The relative lymphocyte chemotaxis rate after knocking down of KAT2B (F) and TAF1L (G) in MCF-7 and T47D cells. *** p < 0.001

Silencing KAT2B and TAF1L inhibited luminal breast cancer proliferation in vivo

Further, to validate our results in in vivo experiments, we constructed luminal breast cancer tumor models in Bal/bc nude mice using the MCF-7 cell line. We constructed MCF-7, MCF-7-shKAT2B, and MCF-7-sh-TAF1L mice tumor models. When the tumor was palpable, measure the volume of the tumors every three days until 30 days after tumor formation. Mouse tumors were collected, weighed, and photographed to assess the effect of knockdown of KAT2B and TAF1L on luminal breast cancer proliferation. Our results showed that knock down of KAT2B and TAF1L could effectively inhibited the proliferation rate (Fig. 8A-B) and tumor weight (Fig. 8C). Meanwhile, knockdown of KAT2B and TAF1L did not significantly reduce the body weight of mice (Figure 8D). This suggests that inhibitors targeting KAT2B and TAF1L can significantly inhibit luminal breast cancer proliferation with little side effects, which is of great significance for clinical patients.

Fig. 8.

Fig. 8

Open in a new tab

Knock down of KAT2B and TAF1L inhibited luminal breast cancer proliferation in vivo. MCF-7, MCF-7-shKAT2B, and MCF-7-shTAF1L were inoculated into the mammary fat pads of Bal/bc nude mice. The tumor picture (A), tumor growth curve (B), and tumor wight (C) showed the effect of sh-KAT2B/TAF1L in vivo. The mouse wight figure (D) showed that shKAT2B/TAF1L had no significant effect on the overall state of the mice

Discussion and conclusion

Luminal breast cancer is considered the most common subtype of breast cancer with the best prognosis. However, a portion of luminal breast cancer patients still experience treatment resistance and rapid progression, and identifying and categorizing these high-risk patients is important for precision therapy. Recently, acetylation modifications were suggested to play essential roles in the development of cancers, including luminal breast cancer. Bioinformatics analysis of the role of acetylation modifications in tumors has been performed in many studies. However, studies of acetylation modifications in luminal breast cancer based on large-scale sequencing are still limited. Elucidating the role of acetylation modification in luminal breast cancer is essential for therapeutic guidance.

In our study, we used the TCGA-BRCA database to identify 30 key acetylation-related genes. These genes were correlated with 37 acetylation regulators and independently predicted the outcomes of patients with luminal breast cancer. We subsequently established a correlation network among the 30 key acetylation-related genes and the 37 acetylation regulators. In this network, most of the genes were correlated with the acetylase human lysine acetyltransferase 2B (KAT2B), indicating the important role of KAT2B in the acetylation regulation network of luminal breast cancer. KAT2B has been reported to participate in diverse biological processes, such as transcriptional regulation, epithelial cell mesenchymal transition (EMT), centrosome amplification, angiogenesis, macrophage M2 polarization, and cell proliferation [30–35]. Furthermore, KAT2B has also been reported to be associated with prognosis in several cancers, including pancreatic cancer, lung adenocarcinoma, renal cell carcinoma, and cervical carcinoma; however, its role in luminal breast cancer has not been elucidated [36–41]. Our analysis revealed that KAT2B was upregulated in the high-risk group of luminal breast cancer patients. Furthermore, in vitro experiments demonstrated that silencing KAT2B in luminal breast cancer cell lines resulted in decreased proliferation, reduced metastatic capacity, and increased lymphocyte infiltration. Our results indicated that KAT2B plays a critical role in the acetylation regulatory network of luminal breast cancer.

Although breast cancer can be divided into several subtypes according to surface molecular markers, research has shown that even within these subtypes, the cancers are highly heterogeneous. Therefore, further subgrouping and risk scoring of breast cancer is necessary for further precise treatment. Bioinformatics analysis of molecular expression via large-scale sequencing has shown that tumor heterogeneity is reflected at the molecular level. Based on this idea, several sequencing-based risk models, such as the PAM50 subtype, 21 gene recurrence score, and breast cancer index, have been established. In addition, large-scale sequencing data have shown that epigenetic regulation, especially acetylation regulation, is essential for cancer development. There are also many studies constructing breast cancer risk model predictions based on large-scale sequencing for effective risk stratification of breast cancer [42–45]. However, risk models based on acetylation-related genes in luminal breast cancer have not been developed. In our study, undifferentiated clustering analysis was used to demonstrate that the 30 acetylation-related genes we identified can be used to distinguish molecular subtypes of luminal breast cancer. We then built a risk score model via LASSO regression with six genes (KAT2B, TAF1L, CDC37, CCDC107, C17orf106, and ASPSCR1) and validated the model with an internal validation dataset.

The application of our risk model divided luminal breast cancer patients into two subgroups with completely different prognoses. Further immune infiltration-related characterization of the subtypes also revealed significant differences in CD8 + T-cell infiltration that reflected the efficacy of ICIs in these two subpopulations. Our study effectively utilized acetylation-associated molecules for further subgroup risk stratification of luminal-type breast cancers, with the potential to guide further precision therapy. More large-scale database validation is still needed for this study, and we will also conduct large-scale sequencing of our center’s patient samples in a follow-up study to further validate our findings. Additionally, in subsequent studies, we plan to collect clinical specimens and employ flow cytometry to analyze the infiltration of CD8 + T cells in luminal breast cancer and adjacent tissues in real-world settings, thereby further validating our findings.

To further identify possible therapeutic targets in our model, we performed survival analysis of six model molecules and found that KAT2B and TAF1L were highly expressed in the high-risk subgroup. Functional validation of the in vitro models also confirmed that knockdown of KAT2B and TAF1L significantly inhibited luminal breast cancer cell growth and metastasis and that knockdown of KAT2B and TAF1L significantly increased CD8 + T-cell infiltration. Our results indicate that KAT2B and TAF1L are potential therapeutic targets for high-risk luminal breast cancer and provide new therapeutic regimens and preclinical evidence for treating patients with high-risk luminal breast cancer. In future work, we will collect clinical luminal-positive breast cancer samples and detect the expression of KAT2B/TAF1L to validate the prognostic differences between patients with high and low KAT2B/TAF1L expression through survival analysis. Furthermore, we will collect clinical specimens to establish organoid models and humanized mouse models, while designing specific inhibitors targeting KAT2B and TAF1L to further investigate and validate the therapeutic potential of inhibiting these two targets. We will also explore and validate the modulation of the immune microenvironment and sensitivity to ICI drugs by KAT2B and TAF1L in animal models in future experiments.

In conclusion, our study constructed an acetylation-associated risk model through in-depth analysis of acetylation-related genes in luminal breast cancer and identified two subtypes of luminal breast cancer with completely different clinical outcomes and immune infiltration levels. Further exploration of therapeutic targets revealed KAT2B and TAF1L as potential therapeutic targets in the high-risk group of luminal breast cancer patients. Our study provides a risk model and therapeutic targets for further subgrouping of luminal breast cancer patients and offers a new direction for precision therapy in high-risk luminal breast cancer patients.

Supplementary Information

Supplementary Material 1 (11.9KB, docx)
Supplementary Material 2 (2.4MB, docx)

Author contributions

Q. Shi, M. Zhu, and H. Liu designed the experiments, analyzed data, prepared figures, and wrote the manuscript; J. Lin and Y. Li performed the knock down, proliferation and colony formation experiments; J. Wang, N. Liu, and H. Lai performed the western blot, RT-qPCR and trans-well experiments.

Funding

This project is funded by Natural Science Foundation of China grants (82203087, 82403734), Yat-sen Research and Development Project (YXQH202528), Technology Basic and Applied Basic Research Topic (2024A04J4778), and China Postdoctoral Science Foundation (2023M744042, 2024T171073).

Data availability

The RNA-Seq data of 1247 breast samples (65 paired breast epithelial cells samples and 481 luminal breast tumor samples) and corresponding clinical information were obtained from The Cancer Genome Atlas (TCGA) website (https://portal.gdc.cancer.gov/projects/TCGABRCA).

Declarations

Ethics approval and consent to participate

The data for this study were obtained from online public databases.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

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

Mengdi Zhu, Jinna Lin and Haohan Liu contributed equally.

References

  • 1.Bray F, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. May 2024;74(3):229–63. 10.3322/caac.21834. [DOI] [PubMed]
  • 2.Harbeck N, Gnant M. Breast cancer. Lancet. 2017;389(10074):1134–50. 10.1016/S0140-6736(16)31891-8. [DOI] [PubMed]
  • 3.Ignatiadis M, Sotiriou C. Luminal breast cancer: from biology to treatment, Nat Rev Clin Oncol, 2013;vol. 10, no. 9, pp. 494–506, 10.1038/nrclinonc.2013.124 [DOI] [PubMed]
  • 4.Canino F et al. Jun., Role of Intrinsic Subtype Analysis with PAM50 in Hormone Receptors Positive HER2 Negative Metastatic Breast Cancer: A Systematic Review, IJMS, vol. 23, no. 13, p. 7079, 2022, 10.3390/ijms23137079 [DOI] [PMC free article] [PubMed]
  • 5.Swarbrick A, Fernandez-Martinez A, Perou CM. Gene-Expression profiling to Decipher breast cancer Inter- and intratumor heterogeneity. Cold Spring Harb Perspect Med. Jan. 2024;14(1):a041320. 10.1101/cshperspect.a041320. [DOI] [PMC free article] [PubMed]
  • 6.Munkácsy G, Santarpia L, Győrffy B. Gene expression profiling in early breast Cancer—Patient stratification based on molecular and tumor microenvironment features. Biomedicines. Jan. 2022;10(2):248. 10.3390/biomedicines10020248. [DOI] [PMC free article] [PubMed]
  • 7.Sun L, Zhang H, Gao P. Metabolic reprogramming and epigenetic modifications on the path to cancer, Protein Cell, vol. 13, no. 12, pp. 877–919, Dec. 2022, 10.1007/s13238-021-00846-7 [DOI] [PMC free article] [PubMed]
  • 8.Li Y, Seto E. HDACs and HDAC inhibitors in cancer development and therapy. Cold Spring Harb Perspect Med. Oct. 2016;6(10):a026831. 10.1101/cshperspect.a026831. [DOI] [PMC free article] [PubMed]
  • 9.Mamdani H, Jalal SI. Histone deacetylase Inhibition in Non-small cell lung cancer: hype or hope?? Front Cell Dev Biol. Oct. 2020;8:582370. 10.3389/fcell.2020.582370. [DOI] [PMC free article] [PubMed]
  • 10.Wei H et al. Sep., KDELR2 promotes breast cancer proliferation via HDAC3-mediated cell cycle progression, Cancer Communications, vol. 41, no. 9, pp. 904–920, 2021, 10.1002/cac2.12180 [DOI] [PMC free article] [PubMed]
  • 11.Narita T, Weinert BT, Choudhary C. Functions and mechanisms of non-histone protein acetylation. Nat Rev Mol Cell Biol. Mar. 2019;20(3):156–74. 10.1038/s41580-018-0081-3. [DOI] [PubMed]
  • 12.Shvedunova M, Akhtar A. Modulation of cellular processes by histone and non-histone protein acetylation. Nat Rev Mol Cell Biol. May 2022;23(5):329–49. 10.1038/s41580-021-00441-y. [DOI] [PubMed]
  • 13.Yu J, et al. Regulation of Serine-Threonine kinase Akt activation by NAD + -Dependent deacetylase SIRT7. Cell Rep. Jan. 2017;18(5):1229–40. 10.1016/j.celrep.2017.01.009. [DOI] [PMC free article] [PubMed]
  • 14.Sobuz SU et al. Sep., SIRT7 regulates the nuclear export of NF-κB p65 by deacetylating Ran, Biochimica et Biophysica Acta (BBA) - Molecular Cell Research, vol. 1866, no. 9, pp. 1355–1367, 2019, 10.1016/j.bbamcr.2019.05.001 [DOI] [PubMed]
  • 15.Ianni A et al. Feb., SIRT7-dependent deacetylation of NPM promotes p53 stabilization following UV-induced genotoxic stress, Proc. Natl. Acad. Sci. U.S.A., vol. 118, no. 5, p. e2015339118, 2021, 10.1073/pnas.2015339118 [DOI] [PMC free article] [PubMed]
  • 16.Wang L, et al. PCAF -mediated acetylation of ISX recruits BRD 4 to promote epithelial‐mesenchymal transition. EMBO Rep. Feb. 2020;21:e48795. 10.15252/embr.201948795. no. 2. [DOI] [PMC free article] [PubMed]
  • 17.Bianco-Miotto T et al. Oct., Global Levels of Specific Histone Modifications and an Epigenetic Gene Signature Predict Prostate Cancer Progression and Development, Cancer Epidemiology, Biomarkers & Prevention, vol. 19, no. 10, pp. 2611–2622, 2010, 10.1158/1055-9965.EPI-10-0555 [DOI] [PubMed]
  • 18.Long M, Hou W, Liu Y, Hu T. A Histone Acetylation Modulator Gene Signature for Classification and Prognosis of Breast Cancer, Current Oncology, vol. 28, no. 1, pp. 928–939, Feb. 2021, 10.3390/curroncol28010091 [DOI] [PMC free article] [PubMed]
  • 19.Fang Y, et al. Establishment, immunological analysis, and drug prediction of a prognostic signature of ovarian cancer related to histone acetylation. Front Pharmacol. Sep. 2022;13:947252. 10.3389/fphar.2022.947252. [DOI] [PMC free article] [PubMed]
  • 20.Jing Z et al. Jul., Novel acetylation-related gene signatures for predicting the prognosis of patients with colorectal cancer, Human Cell, vol. 35, no. 4, pp. 1159–1173, 2022, 10.1007/s13577-022-00720-6 [DOI] [PubMed]
  • 21.Ciarka A, Piątek M, Pęksa R, Kunc M, Senkus E. Tumor-Infiltrating lymphocytes (TILs) in breast cancer: prognostic and predictive significance across molecular subtypes. Biomedicines. Mar. 2024;12(4):763. 10.3390/biomedicines12040763. [DOI] [PMC free article] [PubMed]
  • 22.Marmorstein R, Zhou M-M. Writers and Readers of Histone Acetylation: Structure, Mechanism, and Inhibition, Cold Spring Harbor Perspectives in Biology, vol. 6, no. 7, pp. a018762–a018762, Jul. 2014, 10.1101/cshperspect.a018762 [DOI] [PMC free article] [PubMed]
  • 23.De Winter JCF, Gosling SD, Potter J. Comparing the Pearson and Spearman correlation coefficients across distributions and sample sizes: A tutorial using simulations and empirical data., Psychological Methods, vol. 21, no. 3, pp. 273–290, Sep. 2016, 10.1037/met0000079 [DOI] [PubMed]
  • 24.George B, Seals S, Aban I. Survival analysis and regression models. J Nuclear Cardiol. Aug. 2014;21(4):686–94. 10.1007/s12350-014-9908-2. [DOI] [PMC free article] [PubMed]
  • 25.Shannon P et al. Nov., Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks, Genome Res., vol. 13, no. 11, pp. 2498–2504, 2003, 10.1101/gr.1239303 [DOI] [PMC free article] [PubMed]
  • 26.Szklarczyk D et al. Jan., The STRING database in 2021: customizable protein–protein networks, and functional characterization of user-uploaded gene/measurement sets, Nucleic Acids Research, vol. 49, no. D1, pp. D605–D612, 2021, 10.1093/nar/gkaa1074 [DOI] [PMC free article] [PubMed]
  • 27.Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking, Bioinformatics, vol. 26, no. 12, pp. 1572–1573, Jun. 2010, 10.1093/bioinformatics/btq170 [DOI] [PMC free article] [PubMed]
  • 28.Chen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA. Profiling tumor infiltrating immune cells with CIBERSORT. In: Von Stechow L, editor. in Cancer systems biology. Methods in Molecular Biology. Volume 1711. vol. 1711., New York, NY: Springer New York; 2018. pp. 243–59. 10.1007/978-1-4939-7493-1_12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Hu F-F, Liu C-J, Liu L-L, Zhang Q, Guo A-Y. Expression profile of immune checkpoint genes and their roles in predicting immunotherapy response. Brief Bioinform. May 2021;22(3):bbaa176. 10.1093/bib/bbaa176. [DOI] [PubMed]
  • 30.Malatesta M, Steinhauer C, Mohammad F, Pandey DP, Squatrito M, Helin K. Histone Acetyltransferase PCAF Is Required for Hedgehog–Gli-Dependent Transcription and Cancer Cell Proliferation, Cancer Research, vol. 73, no. 20, pp. 6323–6333, Oct. 2013, 10.1158/0008-5472.CAN-12-4660 [DOI] [PubMed]
  • 31.Bharathy N et al. Nov., P/ CAF mediates PAX3–FOXO1 -dependent oncogenesis in alveolar rhabdomyosarcoma, The Journal of Pathology, vol. 240, no. 3, pp. 269–281, 2016, 10.1002/path.4773 [DOI] [PubMed]
  • 32.Fournier M, et al. KAT2A/KAT2B-targeted acetylome reveals a role for PLK4 acetylation in preventing centrosome amplification. Nat Commun. Oct. 2016;7(1):13227. 10.1038/ncomms13227. [DOI] [PMC free article] [PubMed]
  • 33.Fernandez HR, Lindén SK. The aspirin metabolite salicylate inhibits lysine acetyltransferases and MUC1 induced epithelial to mesenchymal transition. Sci Rep. Jul. 2017;7(1):5626. 10.1038/s41598-017-06149-4. [DOI] [PMC free article] [PubMed]
  • 34.Zhang G, et al. MicroRNA-200c and microRNA- 141 are regulated by a FOXP3-KAT2B axis and associated with tumor metastasis in breast cancer. Breast Cancer Res. Dec. 2017;19(1):73. 10.1186/s13058-017-0858-x. [DOI] [PMC free article] [PubMed]
  • 35.Han H, Zhu W, Lin T, Liu C, Zhai H. N4BP3 promotes angiogenesis in hepatocellular carcinoma by binding with KAT2B, Cancer Science, vol. 113, no. 10, pp. 3390–3404, Oct. 2022, 10.1111/cas.15498 [DOI] [PMC free article] [PubMed]
  • 36.Ansari D, et al. Protein deep sequencing applied to biobank samples from patients with pancreatic cancer. J Cancer Res Clin Oncol. Feb. 2015;141(2):369–80. 10.1007/s00432-014-1817-x. [DOI] [PMC free article] [PubMed]
  • 37.Feng C, Xiong Z, Jiang H, Ding Q, Fang Z, Hui W. Genetic alteration in Notch pathway is associated with better prognosis in renal cell carcinoma. BioFactors. Jan. 2016;42(1):41–8. 10.1002/biof.1250. [DOI] [PubMed]
  • 38.Li L, Zhang J, Cao S. Lysine Acetyltransferase 2B predicts favorable prognosis and functions as anti-oncogene in cervical carcinoma, Bioengineered, vol. 12, no. 1, pp. 2563–2575, Jan. 2021, 10.1080/21655979.2021.1935525 [DOI] [PMC free article] [PubMed]
  • 39.Fraunhoffer NA, et al. Multi-omics data integration and modeling unravels new mechanisms for pancreatic cancer and improves prognostic prediction. Npj Precis Onc. Aug. 2022;6(1):57. 10.1038/s41698-022-00299-z. [DOI] [PMC free article] [PubMed]
  • 40.Yang S, Zhao H, Xiao W, Shao L, Zhao C, Sun P. Extracellular vesicle-packaged miR‐181c‐5p from epithelial ovarian cancer cells promotes M2 polarization of tumor‐associated macrophages via the KAT2B/HOXA10 axis. J Gene Med. Oct. 2022;24(10):e3446. 10.1002/jgm.3446. [DOI] [PubMed]
  • 41.Zhou X, Wang N, Zhang Y, Yu H, Wu Q. KAT2B is an immune infiltration-associated biomarker predicting prognosis and response to immunotherapy in non-small cell lung cancer, Invest New Drugs, vol. 40, no. 1, pp. 43–57, Feb. 2022, 10.1007/s10637-021-01159-6 [DOI] [PubMed]
  • 42.Wang Y, et al. Tubulin alpha-1b chain was identified as a prognosis and immune biomarker in pan-cancer combing with experimental validation in breast cancer. Sci Rep. Apr. 2024;14(1):8201. 10.1038/s41598-024-58982-z. [DOI] [PMC free article] [PubMed]
  • 43.Luo Z, Wang Y, Bi X, Ismtula D, Wang H, Guo C. Cytokine-induced apoptosis inhibitor 1: a comprehensive analysis of potential diagnostic, prognosis, and immune biomarkers in invasive breast cancer, Transl Cancer Res, vol. 12, no. 7, pp. 1765–1786, Jul. 2023, 10.21037/tcr-23-34 [DOI] [PMC free article] [PubMed]
  • 44.Huang X, et al. Association of DBNDD1 with prognostic and immune biomarkers in invasive breast cancer. Discov Onc. Feb. 2025;16(1):218. 10.1007/s12672-025-01990-w. [DOI] [PMC free article] [PubMed]
  • 45.Jing Y, et al. Diagnostic value of 5 MiRNAs combined detection for breast cancer. Front Genet. Nov. 2024;15:1482927. 10.3389/fgene.2024.1482927. [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (11.9KB, docx)
Supplementary Material 2 (2.4MB, docx)

Data Availability Statement

The RNA-Seq data of 1247 breast samples (65 paired breast epithelial cells samples and 481 luminal breast tumor samples) and corresponding clinical information were obtained from The Cancer Genome Atlas (TCGA) website (https://portal.gdc.cancer.gov/projects/TCGABRCA).

Articles from Cancer Cell International are provided here courtesy of BMC

RESOURCES