The Cancer Genome Atlas provided the breast cancer patient data (TCGA, https://portal.gdc.cancer.gov/) and Genotype-Tissue Expression (GTEx) (https://www.gtexportal.org/home/) database [9]. GSE109169 datasets were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database and the Human Protein Atlas (HPA, http://www.proteinatlas.org/) database was used for validation [10]. Data were preprocessed and log-transformed using R packages. The MRPL18 RNA expression was investigated using the ConsensusPathDB (https://cpdb.molgen.mpg.de/) and GTEx datasets, while the FANTOM5 dataset (https://fantom.gsc.riken.jp/) was employed to analyze MRPL18 protein expression levels. 60 pairs of breast tissues, comprising normal and cancerous samples, were gathered from The First People’s Hospital of Changde City (Changde, China). The Ethics Committees of The First People’s Hospital of Changde City approved all sample collection procedures (ethics approval: 2024-147-01).

The analysis of all protein-coding genes was conducted using single-cell RNA sequencing (scRNA-seq), resulting in their classification into 15 distinct cell groups (https://www.proteinatlas.org/MRPL18/single+cell+type (accessed on 02 June 2025)) [11].

Analyze the specificity of different types of immune cells by referring to the immune cell section in the Human Protein Atlas (https://www.proteinatlas.org/immune+cell (accessed on 02 June 2025)). This study examined various immune cell lineages and total peripheral blood mononuclear cells (PBMCs). The datasets from Monaco and Schmiedel were used to analyze MRPL18 expression across different blood cell lineages [12].

Using a microtome (Leica RM2235), the paraffin blocks were sliced into Section 3 μm thick. To remove paraffin, the sections were subjected to a series of xylene washes (2 × 10 min) followed by rehydration in graded ethanol (100%, 90%, 70%; 2 × 5 min each) and finally rinsed in distilled water. Antigen retrieval was conducted. Sections were then incubated for an hour at room temperature using a blocking buffer that included 5% bovine serum albumin to decrease nonspecific binding. Afterward, the sections were incubated using the MRPL18 primary antibody (Atlas Antibodies, HPA028775, 1:200, Solna, Sweden) overnight. The sections were washed 5 times with PBS (pH 7.3, ×1) after incubation with secondary antibodies (Thermo Fisher Scientific, A11034, 1:1000, Kitchener, ON, Canada). Positive staining was quantified in five randomly selected microscopic fields. The immunohistochemistry results were evaluated by at least three experienced pathologists [13].

Using logistic regression analysis, the connection between MRPL18 expression levels and clinicopathological features was evaluated. To determine the diagnostic relevance of MRPL18, the ROC curve was employed with the help of the ‘pROC’ package (v1.18.0) and visualized through ‘ggplot2’ (v3.4.2). For Kaplan-Meier survival analyses, the ‘survival’ (v3.5–5) R packages were employed, using the ‘surv_cutpoint’ function to establish cutoffs.

To identify independent predictors of overall survival (OS) in patients, both univariate and multivariate Cox regression analyses were carried out. The analyzed variables included histological type, T, N, and M stages, PR and ER statuses, PAM50 subtype, menopause status, anatomic neoplasm subdivisions, and MRPL18 expression.

The GeneMANIA website (https://genemania.org/) was utilized for analyzing the PPI network [14]. The ‘clusterProfiler’ package (v4.4.4) was utilized for Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analyses, as previously described [15].

The MDA-MB-231 (HTB-26) and BT-549 (HTB-122) cell lines, known for their high invasiveness as triple-negative breast cancer models, were sourced from the specialized company (ATCC, Manassas, VA, USA). Additionally, routine checks for Mycoplasma contamination and STR identification were performed before starting any experimental activities. All cell lines were cultured in a humidified incubator by Thermo Fisher Scientific at 37°C with 5% CO₂.

RiboBio Co., Ltd. (Guangzhou, China) designed and synthesized the siRNA specifically aimed at MRPL18. Qingke Biotechnology Co., Ltd. (Beijing, China) developed the MRPL18 overexpression vector by incorporating the complete MRPL18 coding sequence into the pcDNA3.1 plasmid. Lipofectamine™ 2000 (11668030, Thermo Fisher Scientific, Waltham, MA, USA) was used for transfection with either 50 nm siRNA or a 2 μg MRPL18 overexpression vector. The culture medium was refreshed eight hours post-transfection. The primer sequences were synthesized by Qingke Biotechnology Co., Ltd. (Beijing, China) and were listed in the Table S1.

The protein sample, weighing 35 μg, was processed through SDS-PAGE and subsequently transferred to a Polyvinylidene fluoride (PVDF) membrane (MilliporeSigma, 05317, Louis, MO, USA). The membrane was followed by an overnight incubation with primary antibodies: MRPL18 (1:500, ZEN BIO, Chengdu, China), Phospho-PI3 Kinase (1:1000, CST, Boston, MA, USA, 4228), Phospho-AKT (1:1000, CST, MA, USA, 9271), and GAPDH (1:1000, Proteintech, Wuhan, China). The membrane was incubated with HRP-conjugated goat anti-rabbit IgG secondary antibody (ZEN BIO, Chengdu, China; 1:3000) for 45 min. The protein bands were visualized using the Tanon 5200 chemiluminescence imaging system (Tanon Science and Technology Co., Ltd., Shanghai, China) after adding the chemiluminescent substrate (Biosharp, BL520A, Hefei, China). Protein expression levels were determined by quantifying band intensities with the help of ImageJ software (version 1.52a, NIH, Bethesda, MD, USA).

Cell Counting Kit-8 (CCK-8) assay: A total of 1000 cells (MDA-MB-231 and BT-549) were seeded into each well of a 96-well plate. After cell attachment, 10 μL of CCK-8 reagent (CA1210-100T, Solarbio, Beijing, China) was added to each well. Colony formation assay: The experiment involved seeding 1000 cells (MDA-MB-231 and BT-549) in a 35 cm dish, with the growth medium changed every three days. After 14 days of continuous culture, fix the cells with methanol (CH3OH ≥ 99%) for 15 min, followed by staining with 1% crystal violet solution (C8470, Solarbio, Beijing, China) for 10 min. Rinse thoroughly with running water, air dry the plates, and perform statistical analysis. Transwell assay: the MDA-MB-231 and BT-549 cells (2.5 × 10⁴) in serum-free medium were seeded into the upper chambers of transwell inserts (8 μm pore size; Corning, NY, USA). After passing through the membrane, cells were fixed with methanol (CH3OH ≥ 99%), stained with 1% crystal violet, and examined using an inverted microscope (Nikon Instruments Inc., Eclipse Ti2, Melville, NY, USA). Wound healing assay: The cells, MDA-MB-231 and BT-549, were introduced into 6-well or 24-well plates and cultured until they formed a monolayer that was 90%–100% confluent. Using a sterile 200 μL pipette tip (or an appropriately sized one), a straight scratch (wound) was made on the cell monolayer. PBS (pH: 7.2–7.4, ×1) was used to gently rinse the wells one to two times, removing any detached cells and debris from the wound area. Immediately after scratching (0 h), and then again at 24 h, images were captured using a Leica DMI40008B microscope (Leica, Wetzlar, Germany). Cell migration rate (%) was calculated using the following equation: Migration rate (%) = [(scratch width at 0 h − scratch width at 24 h)/scratch width at 0 h] × 100% [15,16].

Sixteen female BALB/c nude mice, aged 3–4 weeks, were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd., (Beijing, China). Under controlled conditions, the mice experienced a 12-h light/dark cycle, temperatures of 22°C–25°C, and humidity of 50%–70%.

MDA-MB-231 cells were seeded (2 × 10⁵ cells per well) in 10 cm dishes and transfected with either 200 nM siRNA targeting MRPL18 or 25 μg of the MRPL18 overexpression plasmid. For the control groups, 200 nM of negative control siRNA or 25 μg of an empty vector was used for transfection. After incubating for 24 h, the cells were harvested. The cells were subsequently injected into the dorsal area under the skin of nude mice. Tumor volume was assessed weekly and calculated using the equation: V = 0.5 × length × width². Following a period of 28 days, the mice were anesthetized and executed by a 1% pentobarbital sodium overdose, and the tumor tissues were harvested for later examination. Approval for all experimental protocols was granted by the Ethics Committee at Central South University (CSU-2024-0193).

Harvest the tissue and promptly immerse it in 4% paraformaldehyde in PBS (pH: 7.2–7.4, ×1) for 4 h at room temperature or overnight. Following fixation, rinse the tissue 5 times in PBS (pH: 7.2–7.4, ×1) for 4 min per wash. Embed the fixed tissue in a paraffin compound. Sectioning the tissue at a thickness of 3 µm using a microtome. Tissue samples underwent antigen retrieval, were blocked, and incubated overnight at 4°C with the Ki67 primary antibody (27309-AP, Proteintech, China; 1:50). Afterward, the cells were treated with a fluorescent secondary antibody (AS007, ABclonal, Wuhan, China; 1:100) for 45 min, and then stained with 4^′,6-Diamidino-2-Phenylindole (DAPI) solution (10 µg/mL, C0065, Solarbio) for nuclear staining. Fluorescence imaging was performed using a Zeiss microscope (Axio Imager 2, Jena, Germany).

Tissue cells were subjected to TUNEL staining using a TUNEL staining kit (C1086, Beyotime, Shanghai, China). The sections underwent treatment with 20 μg/mL of DNase-free Proteinase K (Beyotime, ST532, 20 mg/mL) for 15 min, followed by three 5-min washes with PBS (pH: 7.2–7.4, ×1). For 60 min, the sections were treated with 50 μL of TUNEL reaction solution at 37°C in the absence of light. Post-incubation, the sections underwent three 5-min rinses with PBS (pH: 7.2–7.4, ×1).

Cells (MDA-MB-231 and BT-549) apoptosis was assessed using an Annexin V-FITC/propidium iodide double staining kit (Vazyme, KGA106, Nanjing, China). After staining with Annexin V-FITC (1–5 µg/mL) and propidium iodide (1–10 µg/mL), the apoptotic rate was measured using a flow cytometer (Cytek Biosciences, Fremont, CA, USA) as previously described [17].

Version 4.4.2 of R was used for immune analysis. Using the ssGSEA approach, the infiltration levels of 24 immune cell types were measured in breast cancer samples. Using the ‘ggplot2’ (version 3.4.2) and ‘pheatmap’ (version 1.0.12) packages, the link between MRPL18 expression and immune checkpoint genes was analyzed.

Statistical analyses were performed using GraphPad Prism 9.3 (GraphPad Software, Inc., San Diego, CA, USA). A Student’s t-test determined group significance, and Spearman’s correlation coefficient assessed correlations. Results are expressed as mean ± SD, with significance indicated by *p < 0.05, **p < 0.01, and ***p < 0.001. All experiments were repeated three times.

Firstly, the results revealed that MRPL18 was expressed in many human tissues and organs, including muscle tissues, the brain, and the proximal digestive tract (Fig. 1a). The RNA expression of MRPL18 was validated using the ConsensusPathDB. According to the ConsensusPathDB dataset, MRPL18 expression levels were found to be highest in skeletal muscle, the tongue, and the cerebral cortex (Fig. 1b). In FANTOM5 dataset, the findings indicated that the higher expression of MRPL18 were observed in the vagina, skeletal muscle, and liver among all the tissues and organs assessed (Fig. 1c).

Subsequently, the scRNAseq data derived from breast tissues were examined (Fig. 1d). Immune cell type characteristics in hematopoietic tissues were verified using the Monaco and Schmiedel datasets. The Monaco dataset reveals that intermediate monocytes and myeloid dendritic cells have notably increased MRPL18 expression, as shown in Fig. 1e. The schmiedel dataset, as depicted in Fig. 1f, shows that naive CD8 T-cells, naive CD4 T-cells, and non-classical monocytes had the highest MRPL18 expression levels among hematopoietic immune cells.

The disparity in MRPL18 expression between cancerous and healthy tissues is shown in Fig. 2a,b. In addition, by examining data from 1087 patients in TCGA (Table 1), MRPL18 expression was much less in healthy tissues when compared to breast cancer tissues (p < 0.001) (Fig. 2c). Besides, MRPL18 was notably elevated in tumor tissue compared to matched normal breast tissue in 112 pairs of samples (Fig. 2d). To strengthen our conclusions, we referenced the GEO database dataset GSE109169, which provided additional support for our main findings (Fig. 2e). To confirm the protein level of MRPL18 in breast cancer using the HPA database, followed by validation of protein level of MRPL18 through immunohistochemistry experiments. The results showed a notable rise in MRPL18 levels in breast cancer tissue (Fig. 2f–h). In conclusion, our study provides strong evidence that MRPL18 is significantly overexpressed in breast cancer tissues.

Subsequently, the clinical results showed that MRPL18 was significantly correlated with T stage, PMA 50 subtype, PR status, ER status, and histological type (all p < 0.05), as illustrated in Fig. 3a–e. Notably, MRPL18 expression demonstrated a strong ability to distinguish between cancerous and normal tissues, with an AUC value of 0.762 (Fig. 3f). Furthermore, Kaplan-Meier survival curves indicated that patients with high MRPL18 expression had significantly shorter OS, disease-specific survival (DSS), and progression-free interval (PFI) compared to those with low MRPL18 expression (p < 0.05) (Fig. 3g–i). Additionally, higher MRPL18 expression indicates a shorter OS in different subgroups of breast cancer (Fig. 3j–l).

The independent predictive capability of MRPL18 expression was evaluated using univariate and multivariate cox regression models. Analysis indicated that MRPL18, along with N stage, M stage, and age, were significant factors in predicting overall survival in breast cancer (all p < 0.05) (Table 2). Subsequently, a nomogram incorporating MRPL18, N stage, M stage, and age was developed. Specifically, nomograms predicting 1-, 3-, and 5-year OS were generated, as shown in Fig. 4a. As the scores for the four prognostic factors increased, patient prognosis progressively worsened. Furthermore, the calibration curve, as shown in Fig. 4b, confirmed the nomogram’s precise prediction of OS. Time-dependent ROC curves confirmed the prognostic significance of MRPL18 in forecasting survival rates (Fig. 4c).

A PPI network was initially constructed to identify 20 hub genes related to MRPL18, including MRPL19, as showed in Fig. 5a. A positive correlation between MRPL18 and MRPL19 was found in the analysis (R = 0.198, p < 0.001) (Fig. 5b). Differential analysis was performed to generate volcano plots comparing samples with high and low MRPL18 expression, as shown in Fig. 5c. The GO and KEGG revealed that MRPL18 is associated with cell-substrate adhesion, adherens junctions, cell adhesion mediator activity, the PI3K-AKT signaling pathway, proliferation, metastasis, and other functions, as depicted in Fig. 5d,e.

To investigate the effects of MRPL18 on cellular functions, a variety of in vitro experiments were carried out to assess its influence on cell proliferation, migration, and apoptosis. After transfecting cells with the MRPL18 overexpression plasmid, the results showed a significant upregulation of MRPL18 expression. Conversely, MRPL18 expression was markedly reduced after being transfected with three specific siRNA sequences (Fig. S1). Moreover, CCK-8 and colony assays showed that overexpression of MRPL18 was found to promote cell proliferation and colony formation ability, while MRPL18 knockdown significantly decreased these abilities (Fig. 6a–g).

Traswell and wound healing assays showed that MRPL18 promotes cell migration ability, while knockdown of MRPL18 was found to suppress the migratory ability of breast cancer cells (Figs. 6h–k and S2). Flow cytometry results showed that si-MRPL18 promoted cell apoptosis, while overexpression of MRPL18 inhibited apoptosis (Fig. S3). Additionally, the western blotting results showed that reducing MRPL18 levels significantly inhibited the PI3K-AKT pathway, while increasing MRPL18 levels enhanced the pathway’s activation (Fig. 6l,m).

The xenograft tumor model using nude mice to explore MRPL18’s impact on in vivo tumor growth. (Fig. 7a,b). The results showed that subcutaneous tumors had notably reduced volume and weight in the si-MRPL18 group compared to the negative control group, while overexpression of MRPL18 tumor growth (Fig. 7c–f). Immunofluorescence experiments revealed a significant suppression of Ki67 expression in the si-MRPL18 compared to the negative control group, while Ki67 expression was significantly upregulated in the MRPL18-OE group compared to the vector control group (Fig. 7g–i). TUNEL experiments revealed that MRPL18 knockdown promotes cell apoptosis, while overexpression of MRPL18 suppresses cell apoptosis (Fig. 7j–l). Additionally, The MRPL18 protein level significantly decreased in the si-MRPL18 group compared to the negative control group, while it significantly increased in the MRPL18-OE group (Fig. 7m–o).

Single-sample GSEA was conducted to assess how MRPL18 expression influences immune infiltration within the tumor microenvironment. Spearman’s correlation analysis showed a positive relationship between MRPL18 expression and the infiltration levels of different immune cells, including Th2 cells, but it also had a negative correlation with the infiltration levels of certain immune cells, like central memory T cells (Tcm) cells (p < 0.05) (Fig. 8a,b). Furthermore, MRPL18 expression was positively correlated with tumor mutation burden (p < 0.05) (Fig. 8c). Finally, we identified a negative correlation with genes such as NRP1, CD28, and CD200 (p < 0.05) (Fig. 8d). MRPL18 may promote an immunosuppressive tumor microenvironment, potentially reducing the efficacy of immunotherapy.