Abstract

   Explore the relationship between breast invasive carcinoma (BRCA) and
   cuproptosis-related genes (CRGs). CRGs related to prognosis were
   calculated using Lasso analysis and multivariate Cox analysis based on
   BRCA data from the TCGA, CRG signatures were then generated to
   categorize patients based on their risk scores into high-risk and
   low-risk categories. The GEO dataset was used as external validation. A
   nomogram was constructed in order to further predict the survival of
   patients. We also examined differences in the infiltrative status of
   immune cell subsets present between the high-risk and low-risk
   categories. The prognostic gene expression were validated utilizing
   real-time quantitative PCR (RT-qPCR). We identified nine CRGs
   associated with survival and built a risk model to separate patients
   into high - and low-risk groups with distinct differences in survival
   time. Risk model performance was confirmed by the ROC curve and
   nomogram. Additionally, we found a significant difference between the
   two patient groups in the extent of immune cell infiltration. qRT-PCR
   analysis revealed differential expression of seven CRGs (AK7, CEL,
   GRIA3, KCNE2, NT5C1A, PGK1, NOS1) across various breast cancer cell
   lines compared to MCF-10 A cells, showing both positive and negative
   regulation in different cell lines. These results may help illuminate
   the functions that CRGs perform in BRCA, which might improve our
   knowledge of cuproptosis and facilitate the implementation of more
   successful immunotherapy techniques.

Supplementary Information

   The online version contains supplementary material available at
   10.1038/s41598-025-14621-9.

   Keywords: Cuproptosis, Breast invasive carcinoma, Gene signature,
   Prognosis, Tumor immune microenvironment

   Subject terms: Breast cancer, Cancer, Computational biology and
   bioinformatics

Introduction

   Breast invasive carcinoma (BRCA) is the most frequent kind of cancer.
   There are very high frequencies of BRCA occurrence throughout the
   world, and in the year 2020, BRCA replaced lung cancer, accounting for
   11.7% of new cancer cases (nearly 2.3 million), and 70.000
   fatalities^[32]1. There is a strong link between the immune milieu and
   the onset and advancement of BRCA^[33]2–[34]4. Previous research has
   also shown that the BRCA patients’ prognoses might vary significantly
   owing to the diverse nature of the tumor cells^[35]5,[36]6. For this
   reason, it is vital and crucial to generate unique prognostic and
   diagnostic markers for the BRCA to guide treatment and enhance overall
   survival (OS) rates.

   Copper is known to act as a cofactor for vital enzymes and, as a
   result, plants, animals, and even humans depend on it for a wide range
   of biological processes. Copper is a trace metal, and illnesses may be
   brought on by either a lack of or an excess of the element^[37]7.
   Tsvetkov’s team discovered a new way of programmed cell death that
   relies on copper but differs from other cell death processes, they
   named this mode of programmed cell death cuproptosis. As a result of
   its ability to form direct bonds with lipid acylated components of the
   tricarboxylic acid cycle (TCA), copper initiates the process of
   cuproptosis, in a mechanism independent of apoptotic signaling,
   proteotoxic stress leads to cell death^[38]8. It has been suggested
   that alterations in serum copper affect the prognosis of endometrial
   cancer^[39]9 and head and neck cancers^[40]10. However, BRCA prognosis
   has not yet been associated with cuproptosis-related genes (CRGs)
   variants.

   In this research, we sought to find biomarkers that are related to CRGs
   and to design risk models for CRGs that may predict the immunological
   milieu, prognosis, and responsiveness to immunotherapy among BRCA
   patients. In the future, it might be helpful in clinical
   decision-making.

Materials and methods

Datasets

   Gene expression data were obtained from The Cancer Genome Atlas (TCGA),
   as well as clinical-pathological and prognostic information
   ([41]https://portal.gdc.cancer.gov/). Normal samples and breast cancer
   patients with a survival time of 0 were removed, resulting in the
   retention of 967 patients. A prognostic validation dataset,
   [42]GSE20685, was procured from Gene Expression Omnibus (GEO)
   ([43]https://www.ncbi.nlm.nih.gov/geo/).

Formulation of the CRGs-related risk signature

   In this investigation, we integrated information on gene expression,
   survival status, and survival time by employing the Lasso-Cox technique
   of regression analysis and the glmnet package included in the R
   programming language^[44]11. In total, 967 patients with BRCA were
   randomly divided into two groups: a training set of 484 patients and a
   test set of 483 patients. And we performed the validation of the
   external dataset, which was constituted by 327 tumor samples from
   [45]GSE20685. In order to find the optimal model, we used a 10-fold
   cross-validation procedure. Using the median risk score, patients in
   each cohort were classified as Low-risk (L group) or High-risk (H
   group). The receive operating characteristic (ROC) analysis for the
   365, 1095, and 1825 time points was carried out with the help of the
   ROC function included in the proc R package^[46]12. Moreover, by
   utilizing the ROC function of proc, the area under the curve (AUC) and
   confidence intervals were analyzed to derive the final AUC values. The
   survival time of patients was further calculated using risk scores and
   established clinical risk factors. Using a consistency index (C-index)
   and calibration curves, we evaluated the nomogram’s predictive power.

Protein-protein interaction (PPI) network

   With the online tool Search Tool for Retrieval of Interacting Genes
   (STRING), we obtained the PPI network ([47]https://string-db.org/). PPI
   network analysis was used to collect and integrate different
   expressions of CRGs and their possible interactions.

GeneMANIA

   The GeneMANIA website ([48]http://www.genemania.org) generates
   hypothesis about genes, analyzes gene lists, analyzes gene functions,
   and identify possible priorities of genes for subsequent functional
   assays. In this study, we used this tool to look at CRGs and their
   associated interacting genes and protein interactions, pathways and
   functions.

Patients and tissue samples

   Human Protein Atlas (HPA) immunohistochemistry data was used to
   evaluate the expression levels of CRGs in breast tissues
   ([49]https://www.proteinatlas.org/).

Functional enrichment analysis

   Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes
   (KEGG)^[50]13,[51]14 investigations were implemented to contrast the
   differential signaling pathway and the biological impact to shed light
   on the possible biological impacts of CRGs. We employed the KEGG test
   ([52]https://www.kegg.jp/kegg/rest/keggapi.html). For gene set
   enrichment analysis, the most recent gene annotation of the KEGG
   pathway was taken and processed through clusterprofiler (version
   3.14.3) using R^[53]15. It was determined that the minimum gene set
   should be 5, whereas the maximum gene set should be 5000. A value of
   < 0.05 and an FDR of < 0.05 were judged to have statistical
   significance. To acquire gene set enrichment findings, the GO
   annotations of genes were obtained from the R package org.Hs.eg.db
   (3.1.0) and mapped into the background genes utilizing the R program
   clusterprofiler (version 3.14.3) as the background set for enrichment
   analysis^[54]15.

Tumor immune estimation resource (TIMER)

   The TIMER dataset ([55]https://cistrome.shinyapps.io/timer/)Information
   on 32 different cancer types was included and included 10,897 samples
   from TCGA. The data from this dataset was used to study the
   relationship between CRG levels and those of six immune cells (CD8 + T
   cells, CD4 + T cells, macrophages, neutrophils, B cells, and dendritic
   cells) in BRCA.

Characterization of immune landscape between two risk subgroups

   IOBR is a computational tool for the study of immuno-oncology
   properties and tumor biology, and here, by utilizing the R package
   IOBR^[56]16, we selected the CIBERSORT and ESTIMATE methods to
   calculate immune infiltrating cell scores for each sample.

Single cell analysis

   The CRGs expression within the tumor microenvironment was examined at
   the single-cell level using the Tumor Immune Single Cell Center (TISCH)
   database ([57]http://tisch.comp-genomics.org/).

Survival analysis

   Kaplan Meier (K-M) analysis was executed with the survminer and the
   survival package in R to make a comparison of OS between the L group
   and the H group^[58]17. Multivariate Cox analysis was implemented to
   assess if the risk score independently acted as a risk factor for OS
   among BRCA individuals.

Cells and treatments

   The breast cancer cell lines MCF-7, SK-BR3, BT-474, and MDA-MB-231,
   along with the normal breast epithelial cell line MCF-10 A, were
   procured from Pricella (China). For MCF-10 A cells, cultivation was
   performed in MCF-10 A specific medium (Cat. CM-0525, Pricella, China).
   For MCF-7 cells, cultivation was performed in Dulbecco’s modified
   Eagle’s medium (DMEM, Cat. C11995500BT, Gibco, USA) enriched with fetal
   bovine serum (FBS, Cat. 10099–141 C, Gibco, USA, 10%), penicillin (100
   U ml − 1), and streptomycin (100 mg ml − 1) (Cat. 15140-122, Gibco,
   USA). For SK-BR3 and BT-474 cells, cultivation was conducted in
   RPMI-1640 (Roswell Park Memorial Institute medium 1640, Cat. 11875093,
   Gibco, USA) enriched with FBS (Cat. 10099–141 C, Gibco, USA, 10%),
   penicillin (100 U ml − 1), and streptomycin (100 mg ml − 1) (Cat.
   15140-122, Gibco, USA). For MDA-MB-231 cells, cultivation was performed
   in DMEM (Cat. C11995500BT, Gibco, USA) enriched with FBS (Cat.
   10099–141 C, Gibco, USA, 10%), penicillin (100 U ml − 1), and
   streptomycin (100 mg ml − 1) (Cat. 15140-122, Gibco, USA). Cell
   cultivation was sustained in a moisture-controlled setting at 37 °C
   with 5% CO2 utilizing a Thermo Scientific HERACELL 240i CO2 Incubator
   (240i, Thermo Scientific, USA). In subsequent experiments, the cells
   were counted using a cell counter to ensure that the number of cells
   was accurate, three technical replicates were performed per sample to
   ensure accuracy.

qRT-PCR

   Real-time PCR(qRT-PCR) assays were performed according to the
   experimental instructions of the 7500 Real-Time PCR system (Takara,
   RR820A China). Total RNA was extracted from cell lines using TRIzol
   reagent (Invitrogen, USA) following the manufacturer’s protocol. cDNA
   was synthesized using the PrimeScript RT Reagent Kit (Takara, China).
   The qRT-PCR reactions were performed using TB Green Premix Ex Taq II
   (Takara, China) with the following cycling conditions: initial
   denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s
   and 60 °C for 34 s.

   The mRNA expression levels of CRGs were assessed by converting the
   threshold cycle (CT) values using the 2^-ΔΔCt formula and normalized to
   β-actin. The following primer sequences were used:

   AK7 forward:5’-CCT ACA GCA GCG GAA ACA TC-3’.

   AK7 reverse: 5’-GCC CAC AAT CTG GAA TGT GC-3’.

   NOS1 forward: 5’-TTCCCTCTCGCCAAAGAGTTT-3’.

   NOS1 reverse: 5’-AAGTGCTAGTGGTGTCGATCT-3’.

   NT5C1A forward: 5’-GCC AAG ATT TTC TAC GAC AAC CT-3’.

   NT5C1A reverse: 5’-AGC GAT GGT GAC TGC ATT CTG-3’.

   GRIA3 forward: 5’-TCC GGG CGG TCT TCT TTT TAG-3’.

   GRIA3 reverse: 5’-TCC ACC TAT GCT GAT GGT GTT-3’.

   KCNE2 forward: 5’-TGT ACC TCA TGG TGA TGA TTG GA-3’.

   KCNE2 reverse: 5’-GGT CAT TGG AGT GTT CCC GTC-3’.

   PGK1 forward: 5’-TGG ACG TTA AAG GGA AGC GG-3’.

   PGK1 reverse: 5’-GCT CAT AAG GAC TAC CGA CTT GG -3’.

   CEL forward: 5’-TGG GTG ACT CTG TGG ACA TCT-3’.

   CEL reverse: 5’-GCA GGC ATC TCT TCT TGA AGT T-3’.

   β-actin forward: 5’-CGTGCGTGACATTAAGGAGAA-3’.

   β-actin reverse: 5’-AGGAAGGAAGGCTGGAAGAG-3’.

   All experiments were performed in triplicate, and relative mRNA
   expression was calculated using MCF-10 A cells as a calibrator.
   Statistical significance was determined using one-way ANOVA followed by
   Dunnett’s multiple comparisons test using GraphPad Prism 8.0 software
   (GraphPad Software Inc., USA). P values less than 0.05 were considered
   statistically significant.

Statistical analyses

   Statistical analysis was executed utilizing R (v4.4.1), incorporating
   KM survival analysis and the log-rank test for survival comparison. A
   significance threshold of P < 0.05 was employed for statistical
   analyses.GraphPad Prism v. 9.01 (GraphPad Software) was employed for
   data analysis. Categorical parameters were assessed through the χ^2
   test or Fisher’s exact test, whereas continuous parameters were
   examined utilizing Student’s t-test for paired samples. The findings
   are denoted as means ± SEM derived from three autonomous experiments,
   each conducted in duplicate. In addition, various statistical tests,
   including the chi-squared test, Student’s t-test, two-way ANOVA,
   Wilcoxon test, and Kruskal–Wallis test, were employed to determine the
   significance of observed differences between distinct groups.

Results

Development and validation of the CRGs risk signature

   Tsvetkov et al. found a total of 2977 CRGs^[59]8, these CRGs are shown
   in Supplementary Table 1. First, we constructed a prognostic model for
   BRCA patients based on 7 CRGs (AK7, CEL, GRIA3, KCNE2, NT5C1A, PGK1,
   NOS1) using LASSO algorithm and multivariate Cox regression
   (Fig. [60]1A-C). Eventually, we identified 7 genes after setting the
   lambda value to 0.0291. The following is a description of the proposed
   model formula:

Fig. 1.

   [61]Fig. 1
   [62]Open in a new tab

   Development and verification of the risk signature of the CRG. (A,B)
   CRGs were screened by the LASSO-Cox regression model; (C) multivariate
   Cox analyses were done to assess the CRGs’ diagnostic and prognostic
   significance.
   graphic file with name d33e441.gif

   We divided the patients into H group and L group according to their
   risk scores. There were more deaths in H group in the training set,
   test set and all samples (Fig. [63]2A–D). Using K-M survival curves to
   assess the effect of the model on OS in BRCA patients, we found that OS
   was worse in the H group (Figure [64]2E-H). We then used ROC curves to
   assess the accuracy of the model, and we found that the AUC values were
   greater than 0.5 in the training set, the test set and all samples
   (Fig. [65]2I–L).

Fig. 2.

   [66]Fig. 2
   [67]Open in a new tab

   Prognosis capability of the model in the three patient sets. (A–D) The
   distribution of risk scores, each patient’s survival status, and
   heatmaps of the predictive 7-genes signature in the TCGA database;
   (E–H) Analyses premised on the Kaplan–Meier technique show that the
   risk model in TCGA has significant prognostic significance; (I–L) ROC
   curves with AUC values.

Effect of CRGs on prognosis and expression level

   Multivariate Cox regression revealed 7 CRGs (AK7, CEL, GRIA3, KCNE2,
   NOS1, NT5C1A, PGK1), of which CEL (P = 0.04), KCNE2 (P = 9.6e-3), PGK1
   (P = 0,01) were independent unfavorable factors for BRCA. Additionally,
   we found that GRIA3, AK7, NT5CA1 and NOS1 showed no significant effect
   on the survival of BRCA patients (P > 0.05) (Fig. [68]3A-G).

Fig. 3.

   [69]Fig. 3
   [70]Open in a new tab

   K-M survival curves for the 7 prognostic CRGs. (A–G) CEL, KCNE2 and
   PGK1 were independent unfavorable factors for BRCA.

   In order to determine the expression level of CRGs in BRCA, we used HPA
   to search. We found that there were different expressions of CRGs in
   immunohistochemistry(Fig. [71]4A-G).

Fig. 4.

   [72]Fig. 4
   [73]Open in a new tab

   Immunohistochemistry images of CRGs expression in BRCA tissues (HPA).

Evaluation of the clinical utility of the CRGs nomogram

   A nomogram was created to predict the overall survival (OS) rates of
   BRCA patients at one, three, and five years based on
   clinicopathological characteristics and risk scores. The prognostic
   model using CRGs showed a negative correlation between risk score and
   prognosis (Fig. [74]5A). Calibration curves were constructed to assess
   the agreement between predicted OS rates and actual observed OS rates,
   which demonstrated consistency for 1-year, 3-year, and 5-year
   predictions (Fig. [75]5B). The multivariate Cox regression analysis
   revealed that the CRGs risk score can be utilized as a potential
   prognostic factor for BRCA patients in an independent manner
   (Fig. [76]5C).

Fig. 5.

   [77]Fig. 5
   [78]Open in a new tab

   Construction and validation of the nomogram. (A) Nomogram constructed
   with risk scores and clinicopathological characteristics; (B) Nomogram
   for the calibration curves; (C) Multivariate Cox regression analysis
   assess the prognostic value of the risk signature in BRCA patients in
   an independent manner (*P < 0.05, **P < 0.01,***P < 0.001, ****P <
   0.0001)

Gene interaction analyses and signal pathways in 7 CRGs

   By utilizing multivariate cox, we pinpointed 7 CRGs in BRCA. We next
   performed protein-protein interaction (PPI) network analysis on the
   STRING database and used GeneMANIA to analyze the associations between
   CRGs (Fig. [79]6A,B). Enrichment of these genes was found in Long-term
   depression, Metabolic pathways, Circadian entrainment, Purine
   metabolism, Cation channel complex, Ion channel complex, Transmembrane
   transporter complex, Transporter complex, and so on (Fig. [80]6C,D).

Fig. 6.

   [81]Fig. 6
   [82]Open in a new tab

   Gene interaction analyses and signal pathways in 7 CRGs. (A) Analysis
   of PPI among CRGs using STRING; (B) interaction of CRGs using
   GeneMANIA; (C) the KEGG signaling pathway enrichment study is shown
   here utilizing a dot plot; (D) a representation of the GO signaling
   pathway enrichment analysis is shown using a dot plot.

Correlation analysis between CRGs expression level and the infiltrating
immune cells

   Seven CRGs were identified and their mRNA expressions were analyzed in
   relation to immune cell infiltration including neutrophils,
   macrophages, B cells, dendritic cells, CD4 + T cells, and CD8 + T cells
   using the TIMER database. We found that AK7 was significantly
   correlated with B cells, CD8 + T cells and macrophages. CEL was
   significantly correlated with neutrophils, macrophages, dendritic
   cells, and CD8 + T cells. GRIA3 and NOS1 were significantly correlated
   with all 5 other immune cells except for B cells. NT5C1A was
   significantly correlated with neutrophils, dendritic cells, CD4 + T
   cells and CD8 + T cells. PGK1 was significantly correlated with all 6
   immune cells, and KCNE2 was not correlated with any of the 6 immune
   cells (Fig. [83]7A–G).

Fig. 7.

   [84]Fig. 7
   [85]Open in a new tab

   Relationship between the CRGs expression levels and the immune cell
   infiltration level using TIMER. (A–G) AK7, CEL, GRIA3, KCNE2, NOS1,
   NT5C1A, PGK1 expression levels and the immune cell infiltration level.

Single cell analysis of CRGs

   In our current research, 7 CRGs identified as the key factors
   significantly linked to the prognosis in BRCA. To delve deeper into the
   interplay between 7 CRGs expression and the tumor immune landscape, we
   utilized the TISCH. We found the expression of 6 of these CRGs (PGK1,
   NOS1, KCNE2, GRIA3, CEL and AK7) from the BRCA_GSE161529 dataset, with
   PGK1 having the highest expression in immune cells in Fig. [86]8A–G.
   These outcomes indicate a profound role for CRGs in the cellular
   composition and immune dynamics of BRCA.

Fig. 8.

   [87]Fig. 8
   [88]Open in a new tab

   Single cell analysis of CRGs in BRCA. (A–G) Cellular-level mapping
   unveiled the allocation of PGK1, NOS1, KCNE2, GRIA3, CEL and AK7 across
   diverse immune cell populations in BRCA_GSE161529.

Tumor microenvironment landscape in different subtypes

   In this work, we compared the two subtypes by analyzing the differences
   in the composition of the tumor microenvironment (TME). We used the
   CIBERSORT technique along with the lm22 feature matrix to explore the
   variations in immunological infiltration level of 22 distinct immune
   cells that were found between the two categories. We found that
   patients with different subtypes had different expressions in B cell
   native, CD8 T cells, resting CD4 T cell memory, activated CD4 T cell
   memory, T cell gamma delta, NK cell resting, monocytes, macrophages M0
   and M2, dendritic cell activated and mast cell resting (Fig. [89]9A).
   In contrast with the H subtype, the L subtype had higher Stromal and
   Estimate scores (Fig. [90]9B). Additionally, significant variations
   occurred in the expression of CD274, CTLA4, TIGIT, PDCD1 and SIGLEC15
   in the two subgroups (Fig. [91]9C), and in human leukocyte antigen
   (HLA), genes we found that there were also substantial variations in
   the expression of HLA-A, HLA-B, HLA-C, HLA-DMA, HLA-DOA, HLA-E, HLA-F,
   HLA-H, HLA-J, HLA-K, HLA-L, HLA-V and HLA-W between H and L subgroups
   (Fig. [92]9D).

Fig. 9.

   [93]Fig. 9
   [94]Open in a new tab

   Immune milieu of H and L groups. (A) The box plot illustrates the
   significant variations in immune cells that exist across various
   subtypes; (B) The Immune, the Stromal, and the Estimate scores were
   contrasted between H and L subtypes; (C,D) The expression of several
   immune checkpoints (C) and HLA genes (D) is shown to be differentially
   expressed across H and L subtypes using violin plots (*P < 0.05, **P <
   0.01,***P < 0.001,****P < 0.0001)

The association of CRGs risk signature with tumor microenvironment

   We finally investigated the link between the TME and the CRGs risk
   score. Patients who had a high-risk score were significantly associated
   with native B cells, resting dendritic cells, activated NK cells,
   resting NK cells, M0 macrophages, M2 macrophages, CD8 cells, helper
   follicular T cells, and gamma delta T cells (Fig. [95]10A–I).

Fig. 10.

   [96]Fig. 10
   [97]Open in a new tab

   The link between the risk signature incorporating CRGs and the tumor
   microenvironment. (A–I) The link between the risk score and the
   infiltration levels of native B cells, resting dendritic cells,
   activated NK cells, resting NK cells, M0 macrophages, M2 macrophages,
   CD8 cells, helper follicular T cells, and gamma delta T cells are shown
   by scatter plots.

In vitro functional investigation of CRGs

   The qRT-PCR analysis revealed significant differences in the expression
   levels of CRGs across various breast cancer cell lines compared to
   normal breast epithelial cells (MCF-10 A). AK7 expression was
   significantly higher in MCF-7 and SK-BR3 cells compared to MCF-10 A
   cells, but no significant differences were observed in BT-474 and
   MDA-MB-231 cells (Fig. [98]11A). CEL expression was significantly
   downregulated in MCF-7 and BT-474 cells, while SK-BR3 and MDA-MB-231
   cells showed no significant difference compared to MCF-10 A cells
   (Fig. [99]11B). GRIA3 expression was significantly higher in BT-474
   cells than in MCF-10 A cells. However, no significant differences were
   observed in MCF-7, SK-BR3, and MDA-MB-231 cells (Fig. [100]11C). KCNE2
   expression was significantly upregulated in MDA-MB-231 cells. MCF-7 and
   SK-BR3 cells showed no significant difference compared to MCF-10 A
   cells, while BT-474 cells exhibited a non-significant increase
   (Fig. [101]11D). NT5C1A expression was significantly higher in MCF-7
   cells. SK-BR3, BT-474, and MDA-MB-231 cells showed no significant
   difference compared to MCF-10 A cells (Fig. [102]11E). PGK1 expression
   was significantly upregulated in MCF-7 and SK-BR3 cells. BT-474 cells
   showed a significant decrease, while MDA-MB-231 cells exhibited no
   significant difference compared to MCF-10 A cells (Fig. [103]11F). NOS1
   expression was significantly higher in MCF-7 and SK-BR3 cells. BT-474
   and MDA-MB-231 cells showed no significant difference compared to
   MCF-10 A cells(Fig. [104]11G).

Fig. 11.

   [105]Fig. 11
   [106]Open in a new tab

   qRT-PCR analysis of CRGs expression in breast cancer cell lines. (A–G)
   Relative mRNA expression levels of seven CRGs in normal breast
   epithelial cells (MCF-10 A) and breast cancer cell lines representing
   different molecular subtypes (MCF-7, SK-BR3, BT-474, and MDA-MB-231).
   (A) AK7, (B) CEL, (C) GRIA3, (D) KCNE2, (E) NT5C1A, (F) PGK1, and (G)
   NOS1. The expression levels were normalized to β-actin. Data are
   presented as mean ± SEM from three independent experiments. Statistical
   significance: *P < 0.05, ***P < 0.001, ****P < 0.0001, ns not
   significant, compared to MCF-10 A cells.

Discussion

   Even though copper is required in appropriate concentrations for many
   metabolic activities, an overabundance of copper ions may cause cell
   death. An extensive literature review indicates that copper homeostasis
   is strongly linked to the advancement of diverse cancers and that
   cytotoxicity generated by its disequilibrium may influence the growth
   and proliferative rate of cancer cells^[107]18. An extensive
   investigation of the mechanisms of enhanced intracellular toxicity
   produced by a disruption in copper homeostasis associated with copper
   ion buildup may aid in the effective and selective killing of cancer
   cells in immunotherapeutic interventions^[108]19. The findings of
   Tsvetkov and his colleagues^[109]8 demonstrate that copper ions mediate
   and modulate cuproptosis, which is an innovative kind of cell death
   that is entirely distinctive from other types of cell death, such as
   necroptosis, ferroptosis, apoptosis, and pyroptosis. Copper ions bind
   directly to lipid-acylated pathway constituents of the tricarboxylic
   acid cycle in the process known as cuproptosis, culminating in an
   aberrant clustering of lipid-acylated proteins, the depletion of
   iron-sulfur cluster proteins, and proteotoxic stress reaction,
   ultimately resulting in the death of the cells. It is of the utmost
   importance to find biological markers that can identify cuproptosis in
   complex human tumor tissues in a reliable and accurate manner. The
   patterns of CRGs in BRCA and the prospective capacity for anticipating
   the prognosis have not been clarified up to this point.

   In this research, we found 25 CRGs related to BRCA through Lasso-cox
   analysis, then the multivariate analysis illustrated that 7 CRGs,
   including AK7, CEL, GRIA3, KCNE2, NOS1, NT5C1A, PGK1. AK7 is a
   cytosolic human AK isoform with a close relationship to ciliary
   homeostasis. Ciliary dysfunction has been linked to the effects of AK7
   on various types of cancer, affecting their initiation, progression,
   and prognosis^[110]20. CEL is predominantly found in pancreatic acinar
   cells and lactating breast tissue, with high expression levels
   associated with poor survival rates in breast cancer patients^[111]21.
   GRIA3, a subunit of the ionotropic glutamate receptor (AMPAR), promotes
   tumor growth in glioma^[112]22,[113]23, non-small cell lung
   cancer^[114]24, and pancreatic cancer^[115]25. KCNE2 expression is
   downregulated in gastric cancer and hepatocellular carcinoma patients;
   low levels are correlated with unfavorable overall survival
   rates^[116]26,[117]27. NOS1 is highly expressed across multiple tumors
   and contributes to tumor progression^[118]28–[119]31. NT5C1A plays a
   role in gemcitabine resistance by reducing intracellular dFdCTP
   amounts; it’s strongly expressed in pancreatic ductal adenocarcinoma
   cells^[120]32. PGK1 is an essential enzyme for aerobic glycolysis
   pathway that can promote tumor cell proliferation when highly expressed
   intracellularly but inhibit them when extracellularly present at high
   levels; it also affects chemoradiotherapy resistance^[121]33.

   In this study, we utilized lasso Cox regression analysis to build a
   model using 25 genes before conducting univariate and multivariate
   analyses to identify the seven most significantly associated genes. We
   then classified patients into two subtypes based on their risk scores:
   high-risk score individuals had an unfavorable prognosis compared to
   those with low-risk scores. STRING analysis revealed that most of these
   seven genes interacted with each other, and then we found that these
   CRGs were enriched in Long-term depression pathways as well as
   metabolic pathways such as circadian entrainment or purine metabolism
   among others like cation channel complex or transporter complex by
   GeneMANIA, GO and KEGG. The tumor microenvironment (TME) encompasses
   the cellular milieu where tumor cells coexist with immune cells,
   stromal cells, and other non - cancerous cells^[122]34. Within the TME,
   the interplay between malignant and non - malignant cells can have an
   impact on the growth and advancement of cancer^[123]35. As the TME is a
   cellular environment composed of diverse cell types, including tumor
   cells, it has garnered increasing research attention^[124]36,[125]37.
   Mounting evidence points to TME’s pivotal role in BRCA pathogenesis.
   TME not only promotes tumor cell proliferation through interaction but
   also impacts treatment^[126]38. Despite significant research efforts
   into BRCA’s complex mechanisms, the current understanding of TME,
   therapeutic targets, and prognostic factors remains
   inadequate^[127]39–[128]41.To understand how these genes relate to the
   immune microenvironment differences between subsets analyzed using
   CIBERSORT and ESTIMATE revealed significant findings that warrant
   further investigation. We found significant differences in multiple
   immune checkpoints and HLA genes between the H and L groups, suggesting
   that different risk scores have a significant impact on the immune
   infiltration of patients.

   Our study identifies a new model that can evaluate BRCA prognosis.
   However, our study has some limitations. We used public databases to
   collect data, we need to use different external datasets for further
   validation. Moreover, larger samples and experimental validation are
   needed to further define the immune mechanism between CRGs and BRCA.

Supplementary Information

   Below is the link to the electronic supplementary material.
   [129]Supplementary Material 1^ (46.6KB, xlsx)

Author contributions

   G.S., Z.X., and J.Z. developed the methodology and wrote the main
   manuscript text. D.P. and X.L. were responsible for the software
   development. X.L. and J.Z. conducted data curation and conceptualized
   the study. J.Z. performed the experiments. All authors reviewed the
   manuscript and approved the final submitted version.

Data availability

   The data examined in this research were derived from openly accessible
   databases: The Cancer Genome Atlas (TCGA-BRCA)
   ([130]https://portal.gdc.cancer.gov/), the Gene Expression Omnibus
   (GEO) ([131]https://www.ncbi.nlm.nih.gov/geo/), and the Kyoto
   Encyclopedia of Genes and Genomes (KEGG) ([132]https://www.kegg.jp/).
   All datasets used in this study are publicly available and can be
   accessed through the provided URLs.

Declarations

Competing interests

   The authors declare no competing interests.

Ethics approval and consent to participate

   This study utilized publicly available data from TCGA, GEO, and KEGG
   databases. Research based on publicly available datasets does not
   require institutional ethics approval or waiver documentation.

Footnotes

   Publisher’s note

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

Contributor Information

   Xinghe Liao, Email: fdzlxinghe@163.com.

   Jiangang Zhao, Email: zhaojiangang78@163.com.

References