CD19⁺Ki67⁺B cells regulated by NAMPT as key modulators in triple-negative breast cancer with brain metastasis

Abstract

Background

Among breast cancer subtypes, triple-negative breast cancer (TNBC) stands out for its aggressiveness and high frequency of brain metastases. However, the mechanisms driving BrM remain poorly understood.

Methods

We performed integrated single-cell RNA sequencing (scRNA-seq) analysis of TNBC among 15 patients (8 with metastases, 7 without) and combined these data with transcriptomic profiles of BrM from public datasets. B cell heterogeneity was characterized, and the prognostic value of cycling B cells (CD19⁺Ki67⁺) was validated in two independent RNA-seq cohorts (TCGA, GSE65194) and a Xiangya real-world cohort. Functional assays were performed using TNBC-derived organoids co-cultured with CD19⁺Ki67^+/− B cells, and multiplex immunofluorescence was used to evaluate activation of signaling pathways.

Results

scRNA-seq revealed significant enrichment of cycling B cells in metastatic TNBC. High abundance of CD19⁺Ki67⁺ B cells correlated with poor overall survival across cohorts. Functional experiments demonstrated that CD19⁺Ki67⁺B cells enhanced TNBC organoid proliferation, invasion, and metastatic potential compared to CD19⁺Ki67⁻ B cells. Cell-cell communication analysis revealed that activation of the NAMPT/ITGA5/ITGB1 signaling pathway served as a critical mechanism by which B cells regulated crosstalk with cancer cells, which was further validated by multiplex immunofluorescence and a cohort of 74 patient samples.

Conclusions

CD19⁺Ki67⁺ B cells drive TNBC progression and brain metastasis by activating the NAMPT/ITGA5/ITGB1 pathway. These findings provide mechanistic insights into the immune regulation of TNBC BrM and identify potential therapeutic targets to improve clinical outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13058-026-02251-6.

Introduction

Representing approximately 15–20% of breast cancers, TNBC is recognized as the most aggressive form of the disease. It is characterized by cancer cells that exhibit reduced expression of both estrogen and progesterone receptors and produce minimal HER2 protein [1, 2]. Despite substantial advancements in breast cancer diagnostics and therapeutic strategies, the prevalence of metastatic disease—particularly brain metastases (BrM)—remains consistently high, posing a persistent challenge to clinical outcomes [3]. At initial diagnosis, approximately 5% of TNBC cases exhibit BrM. As the disease progresses, up to 30% of these patients develop BrM during their clinical course. Patients diagnosed with TNBC usually experience unfavorable outcomes, as the median overall survival (OS) often falls below six months [4–6]. However, in the absence of precise prognostic assessments and clear therapeutic targets, there remains no effective treatment for brain metastases in TNBC.

Recently, the success of immunotherapy has spurred a growing research focus on the tumor microenvironment (TME) [7]. TME is strongly linked to key biological processes, including tumor heterogeneity, therapeutic resistance, and the tendency toward recurrence and metastasis [8, 9]. The development of BrM is closely associated with a suppressive TME, yet this association warrants further exploration [10, 11]. As a key component of TME, B cells have been largely understudied, with their compositional and functional heterogeneity remaining poorly characterized.

The roles of B cells within TME are inherently complex and multifaceted. Accumulating evidence indicates that tumor-infiltrating B cells (TIBs) activate immunosuppressive programs via the “glutamine metabolism–epigenetic” axis, weakening anti-tumor T cell function, driving tumor progression and immunotherapy resistance [12, 13]. Another study examining B cells across 19 major cancer types reported that TIBs exhibit substantial heterogeneity in abundance and subtype composition. Among these, tumor-associated atypical B cells (TAABs)—characterized by high clonal expansion levels and proliferative capacity—are significantly associated with poor clinical outcomes [14]. Thus, deeper characterization of B cell heterogeneous phenotypes in the TME may reveal new pathways for modulating anti-tumor immunity and enabling targeted cancer therapies.

Here, we conducted single-cell RNA sequencing (scRNA-seq) on tumor samples derived from 15 individuals with TNBC (including 8 with metastatic disease and 7 without metastasis). B cells were found to be significantly enriched in metastatic TNBC samples among the five major immune cell clusters. By integrating this scRNA-seq data with brain metastasis transcriptomic data, we identified cycling B cells (CD19⁺Ki67⁺ B cells) as a subset linked to TNBC brain metastasis and poor prognosis. These findings were subsequently validated by Xiangya clinical cohort. Then intracranial tumor-derived organoids and CD19⁺Ki67^+/− B cells co-cultured experiments were conducted. Results of organoid functional experiments demonstrated that CD19⁺Ki67⁺ B cells significantly enhanced the proliferation, invasiveness and further promoted metastatic potential. Mechanistically, cell-cell communication analysis and multiplex immunofluorescence confirmed CD19⁺Ki67⁺ B cells drive TNBC progression via the NAMPT/ITGA5/ITGB1 pathway. Collectively, our study provided systematic evidence for the immune regulatory mechanisms underlying TNBC metastasis and identified potential therapeutic targets for this aggressive disease subtype.

Methods

Single cell RNA sequencing (scRNA-seq) data processing

Three TNBC scRNA-seq datasets GSE169246, GSE143423 and GSE254379 were collected from the Gene Expression Omnibus (GEO). Patients who had not received preoperative chemotherapy or radiotherapy were enrolled in this study. Single-cell sequencing data were analyzed using the Seurat toolkit (v.4.3.0) for normalization, dimensionality reduction, and clustering, with batch effects removed using Harmony, and differentially expressed genes (DEGs) identified based on the following criteria: log2FC > 0.25, p.adj < 0.05, and an expression ratio greater than 0.1. Cell types in each subpopulation were identified based on their known lineage markers22. Missing annotations were completed using a combination of manual annotation based on Cellmarker2 and automatic annotation with GPTCelltype, while DEGs were identified through FindAllMarkers. Cancer cells from all samples were selected for copy number variation (CNV) analysis, with tumor cells identified as aneuploid and normal hepatocytes as diploid using the R package copykat.

Clinical data collection

TNBC samples were retrospectively collected in Xiangya hospital. Ethical approval for this study was obtained from the Ethics Committees of Xiangya Hospital (approved number: 2025091543).

B cell assessment in cohorts

For the TCGA and GEO (GSE65194) cohorts with RNA-seq data, cycling B cell signatures were estimated via ssGSEA. In the Xiangya cohort, Ki67 and CD19 immunofluorescence was performed on tissue sections, and Ki67⁺ B cells were counted via ImageJ. Patients in both cohorts were stratified by the median value of the signature (TCGA/GSE65194) or cell counts (Xiangya). Kaplan–Meier curves and log-rank tests evaluated survival differences between groups.

Pseudotime trajectory and intercellular communication analysis

Trajectory analysis was performed using the R package Monocle2, with dimensionality reduction via DDRTree, cell progression organized by the order Cells function, and gene expression dynamics visualized through heatmaps generated by the plot_pseudotime_heatmap function. Cell-to-cell communication networks between cycling B cells and cancer cells were constructed and visualized using the CellChat R package based on scRNA-seq data.

Flow cytometry assays

Tumors, spleens, tibias, and femurs of mice were dissociated and ground to get single - cell suspensions. 3 × 10⁶ cells were washed in staining buffer and incubated with anti - mouse CD16/CD32 (BiLegend, 1:200, #156603) for 30 min to block Fc receptors. Then, the cells were incubated with ki67- PE (BiLegend, 1:200, #151209) and CD19 - FITC (BiLegend, 1:200, #152403) at 4 °C for 30 min. Subsequently, After two washes with staining buffer, the cells were preserved at 4 °C until further use. Flow cytometric analysis was carried out on a BD LSRFortessa instrument (BD Biosciences), and the resulting data were processed with FlowJo version 6.4.7.

Orthotopic xenograft experiments in vivo

All procedures strictly adhered to the NIH Guide for the Care and Use of Laboratory Animals. Animal studies were reviewed and authorized by the Institutional Animal Ethics Committee. For the orthotopic implantation model, C57BL/6 mice were anesthetized with 1% pentobarbital sodium and secured in a stereotaxic frame (RWD Life Science). Subsequently, 4 µL of 4T1 cells (5 × 10^5) were stereotactically implanted into the brain parenchyma (coordinates: 2.0 mm lateral, 2.0 mm posterior to bregma, 3.0 mm depth).

Organiod construction

Brain tumors were minced into 1–3 mm² fragments, digested in 5 ml buffer at 37 °C for 1 h, then filtered through a 100 μm cell strainer (NEST, 258365). Red blood cells were lysed, and samples were purified by centrifugation in cold PBS. Collected cells were mixed with Matrigel (Corning, 356234), dispensed as droplets of 40 µL were dispensed into prewarmed 24-well plates (Corning, 3524), and incubated inverted for 20 min before adding organoid culture medium. Organoids were passaged every 2 weeks at a 1:2 − 1:3 split ratio.

EDU

The established breast cancer organoids were incubated in a breast cancer organoid medium supplemented with 20µM EdU for 3 h. The organoids were then fixed in 4% paraformaldehyde. Next, incubation with Apollo dye solution (Cat. No. C10310-1, RiboBIO, China) for 30 min enabled labeling of newly synthesized DNA. The samples were washed for 10 min in 0.5% Triton X-100 to achieve membrane permeabilization. Finally, nuclear counterstaining was performed with DAPI. Fluorescence signals were captured with an N2-DMi8 microscope (Leica, Wetzlar, Germany).

Invasion assays

Matrigel and culture medium were mixed at a 1:1 ratio, plated into wells, and allowed to solidify for 1 h at 37 °C in an incubator. Subsequently, organoids were dissociated and resuspended in culture medium containing 3% Matrigel. The organoid-Matrigel mixture was gently added along the wall of the 24 - well plate. After 3 days of culture, the invasive capacity of the organoids were observed.

Multiplex immunofluorescence staining

Organoids were fixed with 4% paraformaldehyde (PFA; CAS#50-00-0, Sigma-Aldrich) for 15 min at room temperature. Multiplex immunohistochemical staining of tissue sections was performed with the mIHC Fluorescence Kit (RC0086-34RM, RecordBio, Shanghai, China), according to the manufacturer’s protocol. The procedure involved xylene deparaffinization, graded ethanol hydration, antigen retrieval in sodium citrate buffer (0.01 M, pH 6.0), quenching of endogenous peroxidase with H₂O₂, blocking in 10% goat serum, and sequential staining with primary antibodies followed by TSA-RM detection. Each staining cycle commenced with antigen retrieval and concluded with TSA labeling. The antibodies used were: MMP-2 (1:200, 10373-2-AP, Proteintech, China) with TYR-570, E-cad (1:400, 20874-1-AP, Proteintech, China) with TYR-520, NAMPT (1:200, ab236874, abcam, USA) with TYR-690, ITGA5 (1:200, 84468-4-RR, Proteintech, China) with TYR-570, INSR (1:200, ab69508, abcam, USA) with TYR-520, CK5/6 (1:200, 68295-1-Ig, Proteintech, China) with TYR-690. DAPI was applied for nuclear counterstaining. Whole-slide imaging and analysis were carried out with a pathology imaging system (Pannoramic MIDI II, 3D HISTECH, Hungary).

Statistical analysis

Data preprocessing, statistical evaluations, and graphical visualizations were conducted on Ubuntu using R software (version 4.3.3). Results are presented as mean ± SEM, and statistical analyses were additionally performed with GraphPad Prism 7.0. Logistic regression models were used to calculate odds ratios (ORs) with corresponding 95% confidence intervals (CIs). For pairwise comparisons, Student’s t-test was applied, whereas one-way ANOVA was performed when more than two groups were assessed. A p-value threshold of < 0.05 (p < 0.05, *p < 0.01, **p < 0.001) was defined as statistically significant.

Results

Single-cell transcriptomics landscape

ScRNAseq data GSE169246 was frist analyzed. To characterize the immune landscape of triple-negative breast cancer (TNBC) with divergent metastatic statuses, we enrolled 8 TNBC patients with metastatic disease and 7 TNBC patients without metastasis into the analysis. Ultimately, 80,461 high-quality immune cell transcriptomes were obtained at the single-cell level from the patient cohort.

As depicted in Fig. 1A-C, five major immune cell clusters—CD4⁺ T cells, CD8⁺ T cells, B cells, NK&T cells, and myeloid cells—were identified by projecting cells onto a 2D UMAP plot and assessing canonical marker gene expression. Through unsupervised clustering, these primary populations were further resolved into 19 transcriptionally distinct immune cell subpopulations: 5 CD4⁺ T cell clusters, 3 CD8⁺ T cell clusters, 5 B cell clusters, 2 NK&T cell clusters, and 4 myeloid cell clusters (Fig. 1D). Heatmap of marker gene expression revealed distinct transcriptional profiles across subclusters. (Fig. 1E-I).

Fig. 1.

Immune cell landscape in metastatic and non-metastatic TNBC patients. A-B UMAP showed 5 major immune cell clusters in metastatic and non-metastatic samples. C The expression of canonical marker genes which was used to determine major clusters. D UMAP of the 19 detailed immune cell subclusters. E-I Heatmap showing marker gene expression among immune cell subclusters

Enrichment of B cells in metastatic TNBC

To further investigate the heterogeneity of the immune cell between non-metastatic and metastatic tumors, we explored the cell proportions and compositions of distinct tumor cell populations. As shown in Fig. 2A, B cells were markedly enriched in metastatic samples, with an average proportion of 32%. Subsequently, we profiled the cellular composition of 80,461 single cells across metastasis and non-metastasis groups. B cells exhibited a significantly higher proportion in metastatic samples relative to non-metastatic counterparts. Among the detailed subtypes, cycling B cells exhibited a relatively high proportion of the total cell composition across both groups (Fig. 2B). B cell clusters were then identified into 5 subpopulations based on the expression of canonical phenotypical markers (Fig. 2C-F). Naive/resting B cells exhibit high expression of “naive” marker genes, including FCER2 and YBX3. Activated B cells are characterized by high expression of TNFRSF13B, DUSP4, and EGR1. Plasma cells show high expression of IGHG1 and SDC1. Germinal center B cells are marked by high expression of antibody secretion-related genes such as MEF2B and RAPGEF5. Additionally, cycling B cells exhibit high expression of proliferation-associated genes, including PKMYT1, PLK1, and MKI67.

Fig. 2.

Heterogeneity of B cell subsets in metastatic and non-metastatic TNBC patients. A The proportions of B cell clusters in different samples. B The cell type composition of B cell clusters in different samples. C UMAP showed 5 major B cell clusters in metastatic and non-metastatic samples. D The canonical marker genes of B cell clusters. E Violin plot showed the expression level of selected marker genes in each B cell cluster. F Heatmap of DEGs across B cell clusters

Tumor heterogeneity of TNBC brain metastasis

Tumor heterogeneity is intricately tied to tumor metastasis, stemness, and treatment resistance [15, 16]. To further explore the cells that play a key role in triple-negative breast cancer (TNBC) brain metastasis, we integrated transcriptome sequencing data of brain metastases with single-cell data for analysis [17]. First, we performed differential expressed gene (DEG) analysis on the transcriptome data of TNBC brain metastases: the top 50 DEGs were identified by comparing brain metastatic tissues with non-metastatic tissues (Fig. 3A). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that the lipid and atherosclerosis pathway and fluid shear stress and atherosclerosis pathway were notably activated. Additionally, Gene Set Enrichment Analysis (GSEA) showed that tumor necrosis factor α (TNFA) signaling via nuclear factor κB (NFKB) and KRAS signaling up were significantly enriched (Fig. 3B-C).

Fig. 3.

Multi-omics analysis of TNBC brain metastasis. A Volcano plot of differentially expressed genes between brain metastatic group and non-metastatic group. B The pathway enrichment analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG). C The pathway enrichment analysis of Gene Set Enrichment Analysis (GSEA).D The expression of canonical marker genes which was used to determine major B cell clusters. E Survival curves of OS with varying levels of cycling B cells infiltration. F Heatmap showing gene expression dynamics along pseudotime. G Pseudotime trajectory of B cells. H The differentiation state changes of B cells along the pseudotime trajectory. I Distribution of B cell subsets in the trajectory graph. J CD19 gene expression along the pseudotime trajectory

CD19⁺ki67⁺ B cells is closely related with poor prognosis in TNBC patients with brain metastasis

Considering the primary marker gene of cycling B cells is Ki67 (Figure S2), and based on the human tumor B-cell atlas published on Cell, B cells in the proliferative state (Ki67⁺FCRL4⁺ B cells) can contribute to an immunosuppressive tumor microenvironment and poorer prognosis [14]. Therefore, we chose CD19⁺Ki67⁺ B cells as the main focus of our study. To further investigate the presence of cycling B cells and their clinical relevance, we performed multiplex immunostaining on formalin-fixed paraffin-embedded (FFPE) sections from 17 triple-negative breast cancer (TNBC) samples (including 9 metastatic ones) in our Xiangya cohort (Fig. 4A). As shown in Fig. 4B, distinct levels of CD19⁺Ki67⁺ B cell infiltration were presented in TNBC patient samples from metastatic versus non-metastatic cases. We observed a significant enrichment of CD19⁺Ki67⁺ B cells in TNBC samples from patients with brain metastases compared to non-metastatic samples, which is consistent with the findings from our previous single-cell analysis. Moreover, elevated abundance of CD19⁺Ki67⁺ B cells showed potential association with larger tumor size (Fig. 4C). The heat map of patients characteristics was shown in Fig. 4D.

Fig. 4.

Enrichment of CD19⁺Ki67⁺B cells in TNBC and their clinical relevance. A Immunofluorescence imaging showing the overall tumor and high-resolution regions of interest, segmented into fine grids for precise spatial mapping and quantitative spot analysis. Scale bars: top, 625 μm; bottom, 50 μm.B Representative immunofluorescence staining of clinical TNBC samples from metastasis and non-metastasis groups. White arrows highlighted CD19⁺Ki67⁺B cells. Scale bars, 50 μm. C Quantification of CD19⁺Ki67⁺B cells as a proportion of total cells and their association with metastatic status and pathological T stage.D Heatmap of TNBC patients along with key clinical characteristics

CD19⁺ki67⁺ B cells are associated with proliferation, invasion and migration in TNBC organoid

To validate the results of multi-omic analysis, we performed B cell co-culture assays in this study. First, we sacrificed C57BL/6 bearing orthotopic 4T1 cell line TNBC xenografts and isolated tumor-derived organoids from the intracranial tumors. Concurrently, we isolated splenic, bone marrow, and lymph node tissues from these mice for flow cytometry, sorting CD19⁺Ki67⁺ B cells and CD19⁺Ki67⁻ B cells. These sorted B cell subsets were then co-cultured with the established organoids (Fig. 5A-B).

Fig. 5.

The tumor-promoting activity of CD19⁺Ki67⁺ B cells in the TNBC organoids. A Workflow of TNBC organoids co-cultured with Ki67^+/− B cells. B Sorting of CD19⁺Ki67^+/− B cells via flow cytometry. C Proliferation test by EDU assays after co-cultured with Ki67⁺or Ki67⁻B cells; D Inverse circularity score of invasion ability via Ki67^+/− B cells co-culture; E Multiplex immunofluorescence staining images of metastasis-related molecules MMP-2 and E-cad. Scale bars = 50 μm

Subsequently, in vitro assays were performed to investigate the functional role of CD19⁺Ki67⁺ B cells. EdU assays demonstrated that exposure to Ki67⁺ B cells enhanced the proliferative activity of TNBC organoids. The proportions of EdU⁺ cells were increased from 11.2% (Ki67⁻ B cells) to 30.4% (Ki67⁺ B cells) (p < 0.01, Fig. 5C). Then the organoids were observed under brightfield microscopy. After one week, organoids co-cultured with Ki67⁺ B cells exhibited accelerated growth. Similarly, organoid invasiveness was quantified using the inverse circularity score based on previous studies [18]. The mean inverse circularity score of the Ki67⁺ B cell exposed group was notably higher than that of the Ki67⁻ B cell group (3.3 vs. 1.783, p < 0.05, Fig. 5D). Classical markers associated with cellular metastatic capacity were also evaluated (Fig. 5E). Ki67⁺ B cells showed the enrichment of specific markers, including MMP2 and E-cadherin (E-cad). In conclusion, CD19⁺Ki67⁺ B cells display Breg-like properties and potentially promote proliferation, invasion, and migration in TNBC organoids.

CD19⁺Ki67⁺ B Cells promote TNBC progression via the NAMPT/ITGA5/ITGB1 pathway

To further discover the potential mechanism of Ki67⁺ B cells promotion, tumor cells from two datasets (GSE143423 and GSE254379) were isolated for further analysis (Fig. 6A). GSVA and copy number variation (CNV) analyses were performed to characterize pathway activity and genomic alterations in tumor cells, with the results presented in Figure S3.

Fig. 6.

Cell-cell communication analysis between Cycling B cells and cancer cells in TNBC. A UMAP visualization of cancer cells and normal cells derived from integrated single-cell RNA-seq datasets. B Cell-cell communication network illustrating the number of ligand-receptor interactions between cancer cells and distinct B-cell subsets. C Cell-cell communication network showing the relative interaction weights between cancer cells and different B-cell subsets. D Heatmap of outgoing and incoming signaling pathways across different cell populations. D Violin plots presenting the expression levels of NAMPT/ITGA5/ITGB1 across multiple cell populations. F Ligand-receptor communication analysis identifying the NAMPT/ITGA5/ITGB1 pathway between cycling B cells and cancer cells

To explore intercellular communication, we further analyzed cell-cell interactions between cycling B cells and cancer cells. Comparative assessment of the number and strength of predicted ligand-receptor interactions showed that cycling B cells exhibited enhanced communication with tumor cells relative to other B-cell subsets (Fig. 6B-C).

Subsequent pathway-level integration of these interactions revealed significant enrichment of NAMPT-related signaling in cycling B cell-tumor cell communication (Fig. 6D). At the ligand-receptor level, interaction inference indicated that cycling B cells may influence cancer cell behavior through the NAMPT/ITGA5/ITGB1 axis (Fig. 6E). Consistently, analysis of signaling probability identified the NAMPT-ITGA5/ITGB1 pathway as a dominant communication route from cycling B cells to cancer cells (Fig. 6F). Collectively, these findings suggested that cycling B cells may promote tumor cell progression via NAMPT-mediated engagement of ITGA5/ITGB1 signaling.

Next, we validated the above conclusion by organoids in vitro o-cultured with B cells and real-world Xiangya patient cohort. Multiplex immunofluorescence results demonstrated that exposure to CD19⁺Ki67⁺ B cells promoted activation of the NAMPT/ITGA5/ITGB1 pathway in TNBC organoids. The relative fluorescence intensity of Ki67⁺ B cells was 2.329-fold that of Ki67⁻ B cells (p < 0.05, Fig. 7A-D). Subsequently, we further expanded the sample size in the previously established Xiangya cohort to 74 cases and validated the association between NAMPT expression and metastatic status in real-world patient cohort. The baseline characteristics of triple-negative breast cancer (TNBC) patients are presented in Fig. 7E. Consistently, NAMPT expression was significantly higher in TNBC patients with metastasis than in those without metastasis. Survival analysis demonstrated that the cohort with higher NAMPT expression levels had significantly shorter survival time (Fig. 7F).

Fig. 7.

Validation of NAMPT/ITGA5/ITGB1 signaling in TNBC.

A-D Multiplex immunofluorescence staining images of TNBC organoids co-cultured with CD19⁺Ki67^+/− B cells. Markers include NAMPT (pink), ITGA5 (red), and INSR (green). Scale bars: 100 μm. E Heatmap of TNBC patients along with key clinical characteristics. F OS curves of patients with different NAMPT expression in xiangya cohort

Discussion

TNBC remains a major therapeutic challenge due to its high metastatic potential and poor prognosis [19, 20]. Comprehensive analysis of TME in metastatic TNBC is crucial for exploring its tumor heterogeneity and developing effective therapies [21–24]. However, the specific roles of B cells and its functional mechanisms within the TNBC TME remain insufficiently explored and require further investigation. In this study, we focused on B cells from metastatic and non-metastatic TNBC specimens for scRNA-seq. Integrating brain metastasis transcriptomes identified CD19⁺Ki67⁺ B cells as associated with TNBC brain metastasis and poor prognosis. Additionally, Ki67⁺ B cells exerted suppressive regulatory effects, affecting the proliferation, invasiveness and metastatic potential of TNBC brain metastasis organoids. Mechanistically, CD19⁺Ki67⁺ B cells drive TNBC progression by activating the NAMPT/ITGA5/ITGB1 pathway.

Cycling B cell exhibited high expression of PKMYT1, PLK1, SERPINA9, HJURP and Ki67. Based on previous studies, PKMYT1 negatively regulated the key cell cycle kinase CDK1 by inducing its inhibitory phosphorylation. The Inhibition of PKMYT1 caused aberrant CDK1 activation, which suppressed tumor growth [25]. Furthermore, existing studies in TNBC have demonstrated that PKMYT1 promotes the formation of an immunosuppressive TME [26]. As for PLK1, expression levels of PLK1 not only facilitated the identification of patients with metastatic breast cancer but also suggested that it may be implicated in mediating resistance to CDK4/6 inhibitors [27]. SERPINA9 exhibited B-cell-specific expression, and its expression significantly increased when B cells proliferated and differentiated into antibody-producing plasma cells [28]. Serving as another oncogene, HJURP can contribute to tumorigenesis and such dysfunction may further induce the characteristic suppressive immune infiltration [29]. All these results suggested that cycling B cells were associated with an immature and proliferative state, which is also consistent with the subsequent pseudotime analysis results. These characteristics provided a basis for the analysis of the function and activity of cycling B cells subset.

Subsequently, by integrating multi-omics data with the Xiangya patient cohort, we identified that cycling B cells were involved in mediating the process of brain metastasis and were closely associated with poor overall survival in TNBC. Notably, in the diagnosis and treatment of lymphoma, the ki67 index is widely recognized as a marker tightly linked to the invasiveness and malignancy of tumors. Under normal circumstances, B cells typically initiate rapid proliferation upon activation by multiple signals during the immune response. Specifically, during B cell proliferation, events such as gene recombination errors, metabolic stress, and signaling pathway abnormalities are not only etiological factors for the development of hematopoietic cancer but also correlate with the progression of other solid tumors [30–33]. Zhang Zemin et al. demonstrated that cycling B cells in a proliferative state exert regulatory B cell (Breg)-like properties: they were highly expressed in tumor tissues and contributed to the maintenance of an immunosuppressive TME [14]. Consistent with our findings, pseudotime analysis revealed that cycling B cells were localized at the early stage of B cell differentiation and exhibited a proliferative phenotype—with Ki67 being one of the key markers. Therefore, in subsequent experiments, we selected CD19⁺Ki67⁺ B cells as our primary research subject. Herein, our in vitro co-culture experiments of TNBC brain metastasis mouse-derived organoids (BrM MDOs) with targeted B cells suggest that, compared with Ki67⁻ B cells, the presence of Ki67⁺ B cells may significantly enhance the proliferation, invasiveness and metastatic potential of TNBC BrM MDOs [34].

Existing studies suggested that intracranial B cells can support antitumor immunity when they co-localize with TFH cells in well-organized TLS, where local B-cell maturation and T-cell activation were sustained [34, 35]. In brain metastases, however, TLS formation is heterogeneous across tumor types, and TLS-like signatures appear less frequent in breast cancer lesions than in lung cancer or melanoma, implying fewer TLS-supported TFH cell and B cell interactions [36, 37]. Meanwhile, the brain metastatic niche is shaped by astrocytes and resident myeloid cells and is often immunosuppressive, which may favor regulatory B-cell programs rather than TLS-linked antitumor activity [38, 39]. Together, these context differences may help explain why TNBC brain metastasis can exhibit a more pro-metastatic B-cell state, consistent with the NAMPT-ITGA5/ITGB1 communication axis observed in our data.

In summary, our study demonstrated that CD19⁺Ki67⁺ B cells in TNBC are associated with the progression of brain metastasis and poor prognosis, and may serve as an independent prognostic factor. Furthermore, in-depth investigations into Ki67⁺ B cell-targeted inhibitors are expected to improve the therapeutic efficacy of TNBC treatment.

Conclusion

CD19⁺Ki67⁺B cells correlate with brain metastasis progression and poor patient prognosis, and may act as an independent prognostic indicator in TNBC, which offer a potential therapeutic target for tumor treatment.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

QYZ and LYZ performed and analyzed experiments. QYZ and ZMZ performed statistical descriptions and bioinformatic analyses. YXZ and DJ interpreted and analysed sthe data. LYZ provided financial supports. QYZ wrote the first draft of the manuscript. All authors have read and approved the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (82573449, 82171828), Suzhou Multidisciplinary Clinical Research Project for Major Diseases (DZXYJ202304), Suzhou Health Talent Program Research Project (GSWS2022028), Internal Collaboration Project of the State Key Laboratory of Radiation Medicine and Radiation Protection (GZN1202302), Talent Project of The Second Affiliated Hospital of Soochow University (XKTJ-RC202401).

Data availability

All data in this study are available upon request.

Declarations

Ethics approval and consent to participate

Ethical approval for this study was obtained from the Ethics Committees of Xiangya Hospital (approved number: 2025091543).

Consent for publication

All authors consent to publication.

Competing interests

The authors declare no competing interests.