Tumor-Associated CD19⁺CD39⁻ B Regulatory Cells Deregulate Class-Switch Recombination to Suppress Antibody Responses

In the immune system, B cells are an essential component of humoral immunity. They are classically positive modulators that regulate inflammation and immune responses. Solid tumors often contain B-cell populations, suggesting B cells have a role in influencing the tumor microenvironment alone or in cooperation with other resident cells (1, 2). The role of B cells in cancer is equivocal. On one hand, B cells are thought to be the immune system's defensive powerhouse as they secrete antibodies and build up a memory B cell–mediated immune response against tumors (3). On the other hand, tumors show an enhancement in metastasis following CD20 depletion from the total B-cell pool (4, 5). Further research led to the identification of a new protumor B-cell branch with a suppressive function that exists alongside the well-known antitumoral B-cell subsets like memory and plasma B cells. The newly observed immunosuppressive B cells were later designated as B regulatory (Breg) cells (6, 7).

Despite reaching a partial consensus on the IL10-mediated effector function of Breg cells, the field has yet to produce a unified view on the phenotypic status of Breg cells (8, 9). Several IL10-producing Breg-cell subsets with different surface markers have been described so far. Among them, CD19⁺CD24^hiCD38^hi B cells have a phenotype associated with transitional B cells (10), and CD19⁺CD24^hiCD27⁺, a phenotype associated with memory B cells (11), make up the majority of IL10-secreting Breg cells. It is now widely accepted that Breg cells perform immunoregulatory functions primarily through the secretion of cytokines such as IL10, IL35, and TGFβ as well as other proteins such as granzyme-B and programmed death–ligand 1 (PD-L1; refs. 7, 12, 13). Although the presence of suppressive B cells in various autoimmune disorders has been known for more than 30 years (14), very little is known about their presence and function in the field of tumor biology. Breg cells appear to suppress T helper cells, induce FOXP3⁺ regulatory T (Treg) cells, and target other tumor-infiltrating lymphocytes such as myeloid-derived suppressor cells, macrophages, and natural killer cells to thwart antitumor immunity (15, 16). Various studies have also shown that Breg cells are closely associated with many clinicopathologic factors in patients with solid tumors and may be potential biomarkers for predicting patient survival (14). A study on gastric cancer found that Breg cells suppress the immune system by downregulating Th1 cells and increasing Treg-cell generation (17, 18). These findings suggest the potential for therapeutic targeting of Breg cells in patients with cancer. Breg cells are widely known to be immunosuppressive B-cell subsets. However, other than intracellular IL10, no phenotypic surface signature has yet been assigned to this B-cell subset in the tumor microenvironment.

In this study, using high-dimensional flow-cytometric and transcriptomic analysis, we identified a phenotypic signature for tumor-associated IL10-producing Breg cells, which can be used to live-sort these cells. Our ex vivo findings indicated that, in addition to their T cell–mediated immunoregulatory function, Breg cells play an important role in limiting the production of germline transcripts (GLT), activation-induced cytidine deaminase (AID) expression involved in class-switching recombination (CSR), B-cell differentiation, and antibody production in the tumor microenvironment.

Patients with breast cancer (n = 45; females, age 18–64 years), classified as invasive carcinoma of no special type (ductal), were enrolled in the study. Fifteen age-/sex-matched healthy individuals (aged 22–62 years), 8 patients with rheumatoid arthritis (RA), and 6 Hepatitis B virus (HBV)–infected patients were studied in parallel as controls. Blood and tissue samples were obtained between August 2017 and September 2022. Tissues from primary breast tumor lesions were collected from patients undergoing surgical procedures. In all cases, research personnel obtained and processed tissue within 1 hour of surgery. All the freshly collected tissues and bloods were processed immediately. In an unbiased way, patients with RA who had a moderate to high disease activity index and did not have any other systemic diseases were included in the study. The ethical approval for the collection of postoperative breast tumor tissue samples and peripheral blood from patients with breast cancer, patients with RA, and healthy individuals and subsequent experiments with them for this study was sanctioned under the provisions of the ethics committee, ESI Post-Graduate Institute of Medical Science and Research, Kolkata, India (Approval no: ESI-PGIMSR/MKT/IEC/13/2017), the Institute of Post-Graduate Medical Education and Research Oversight Committee (Approval no. IPGME&R/IEC/2018/643) and the Human Ethics Committee, Bose Institute (Approval no: BIHEC/2017–18/7). Informed written consent was obtained from all patients enrolled in the study in compliance with the Declaration of Helsinki. Patient characteristics are in Supplementary Tables S1–S3. Healthy individuals’ data are also added in Supplementary Table S4.

Collected peripheral blood was layered over a lymphocyte separation medium (HiSep, Himedia, catalog no. LS001) in a 1:1 ratio. Then a density gradient centrifugation was performed at 1,000 rpm for 40 minutes at room temperature. The buffy coat was then collected, and the cells washed with PBS and resuspended to obtain total leukocytes. Naïve T cells were isolated using a naïve T-cell isolation kit (BioLegend, catalog no. 480041). After obtaining leukocytes, 10 μL of the biotin-antibody cocktail was added to 10⁷ cells. After thorough mixing, the cells were incubated on ice for 15 minutes. Then, 10 μL of streptavidin nanobeads were added, mixed well, and incubated on ice for 15 minutes. The cells were then treated with 2.5 mL of PBS before being transferred to a 5 mL (12 × 75 mm) polypropylene tube and placed in the BD cell separation magnet (BD Biosciences, catalog no. 552311) for 5 minutes. The unbound portion of the sample containing the desired naïve CD4⁺ T cells was then collected. Flow cytometry was used to determine the purity of the sample. CD19⁺ B cells were isolated using a B-cell enrichment cocktail (BD Biosciences, catalog no. 558007). Five-microliter biotinylated human B lymphocyte enrichment cocktail was added for every 1×10⁶ cells, mixed well by pipetting and incubated for 15 minutes at room temperature. After that, the labelled cells were washed with 10× excess volume of 1× BD IMag buffer (BD Biosciences, catalog no. 552362), centrifuged at 300 × g for 7 minutes and the supernatant aspirated. The cells were resuspended in 100 μL of BD IMag buffer before we added 5 μL of BD IMag streptavidin particle, and incubated for 30 minutes at room temperature. Next, we transferred the sample to a 12×75 mm round-bottom test tube and placed the tube on the BD cell separation magnet for 8 minutes. After that, we collected the unbound fraction of the sample containing the desired CD19⁺ B cells. The purity of enriched cells was determined by flow cytometry and was consistently > 90%. Some of the CD19⁺ B cells were further separated on the basis of CD39 expression using biotin mouse anti-human CD39 (BioLegend, catalog no. 328204), yielding, CD19⁺CD39^‒ Breg cells and CD19⁺CD39^‒ Breg-depleted peripheral blood mononuclear cells (PBMC).

Isolated cells were cultured in RPMI1640 containing l-glutamine and NAHCO3 (Sigma-Aldrich, catalog no. R8758) supplemented with 25 mmol/L HEPES (Sigma-Aldrich, catalog no. H4034) and 10% FBS (Thermo Fisher Scientific, catalog no. 16000069), and penicillin–streptomycin (Thermo Fisher Scientific, catalog no. 15140122). Isolated naïve T cells were activated with anti-CD3/anti-CD28 (BioLegend, catalog nos. 317302 and 302902, respectively). First, we prepared a 10 μg/mL solution of anti-CD3 and anti-CD28 in sterile PBS. Then, 50μL of the anti-CD3 solution was applied to each microwell of the 96-well test plate and incubated at 37°C for two hours. After washing, we added 200 μL of naïve T-cell suspension (RPMI medium with 10% FBS; 2×10⁵ cells) and 50 μL of anti-CD28 to each well and incubated for 72 hours at 37°C. Recombinant human IL6 (PeproTech, catalog no. 200–06; 25 ng/mL) and IL21 (PeproTech, catalog no. 200–21; 10 ng/mL) were added in the naïve T-cell cultures (2×10⁵ cells) for 72 hours in the presence with anti-CD3/anti-CD28 to polarize T follicular helper (Tfh) cell. CD19⁺ B cells (5×10⁵ cells) were activated for 72 hours with 1 μg/mL recombinant CD40LG (Sino Biological, catalog no. 10239-HO8E-250). For the induction of the memory/plasma phenotype and germinal transcript detection, cells were stimulated for 7 days with CD40LG+IL2 (PeproTech, catalog no. 200–02) +IL4 (PeproTech, catalog no. 200–04) to develop a memory phenotype, CD40LG+IL2+IL10 (IL10, PeproTech, catalog no. 200–10) to develop plasma cells, and CD40LG+IL4+IL10 for GLT detection. Concentrations used for recombinant proteins were CD40LG (1 μg/mL), IL4 (40 ng/mL), IL10 (50 ng/mL), and IL2 (25 ng/mL). To check Breg-cell stability, sorted and unsorted Breg cells from tumor PBMCs were activated with CD40LG (1 μg/mL) in presence of lipopolysaccharide (LPS) (20 μg /mL; Sigma-Aldrich, L2630) for 7 days. For adenosine-mediated AID production, 5×10⁵ CD19⁺ B cells from the tumor patient's blood were cultured with CD40LG (1 μg/mL) and adenosine (20 μmol/L; Himedia, catalog no. TC083–5G) for 7 days.

Flow cytometry was performed using fluorochrome-conjugated antibodies specific for the following human markers: CD19-BB515/-FITC/-APC/Percp-Cy5.5, CD24-PE-Cy7, CD38-APC, IgM-V500, IgD-APC-Cy7, PD1-Percp-Cy5.5, CD27-PE, IL10-PE/-BV421, CD39-BV421, TGFβ-BV421, CD138-Percp-Cy5.5, IFNγ-PE, Ki67-BV421, IL21-BV421/PE, CXCR5-PE-Cy7/APC, CD4-APC-H7/FITC, CD25-PE-Cy7, FOXP3-APC, Annexin-V-FITC/APC, 7AAD, and CD45-PE. All antibody details can be found in Supplementary Table S5. For multicolor flow-cytometric surface staining, cells were stained at 4°C for 30 minutes. For analysis of intracellular FOXP3 and Ki67, cells were fixed and permeabilized with transcription factor buffer (BD Biosciences, catalog no. 562574). For detection of the intracellular cytokines IL10 and IFNγ and IL21, cells were stimulated with a cell activation cocktail (phorbol 12-myristate 13-acetate and ionomycin) in the presence of brefeldin-A (BioLegend, catalog no. 423304) for the last 5 hours at 37°C. Then the cells were washed, fixed, permeabilized using cytofix and cytoperm buffer set (BD Biosciences, catalog no. 554714), and stained with anti–IFNγ-PE, anti–IL21-BV421 and anti–IL10-PE/-BV421 (see Supplementary Table S5 for details) for 30 minutes using the manufacturer's protocol. Appropriate isotype controls and fluorescence minus one (FMO) were used to set the gates for cytokine detection. CD4⁺ T-cell apoptosis was evaluated by flow cytometry using 7AAD (BD Biosciences, catalog no. 559925). BD FACS Verse was used to acquire the data and BD FACS Verse Suite software (BD Biosciences), and FlowJo software (BD Biosciences, version 10.0) were used to analyze all flow-cytometry data. For quantifying stained cells in contour plots, dot plots, and quadrants, lines were drawn on the basis of the signals using FMO and unstained controls. For high-dimensional data analysis, t-distributed stochastic neighbor embedding (t-SNE), FlowSOM, and Uniform Manifold Approximation and Projection (UMAP) plugins are used using FlowJo v.10 online guidelines.

For confocal microscopy, isolated CD19⁺ B cells were treated and stimulated with a cell activation cocktail with brefeldin-A (BioLegend, catalog no. 423304) for 5 hours at 37°C. Then 200 μL of cell suspension (4×10⁴ cells) was applied to the poly-L-lysine-coated coverslip and allowed to adhere for 40 to 60 minutes at 37°C. Fixation was performed using 4% paraformaldehyde for 20 minutes at room temperature, followed by permeabilization with cold-methanol for 8 minutes at −20°C. After blocking with 4% BSA (30 minutes), the cells were incubated with anti–human-CD19 (Biotenylated-Ab, BioLegend, catalog no. 302204), anti–IL10 (rabbit-mAb, Cell Signaling Technology, catalog no. 12163S), followed by fluorescence-tagged streptavidin particle (FITC-Streptavidin, BioLegend, catalog no. 405202) and secondary antibodies (BioLegend, catalog no. 406418 and Invitrogen, catalog nos. A11034, A11010, and A11003) for 45 minutes, and DAPI (Sigma, catalog no. 28718–90–3) was added for the last 15 minutes. The cells were examined under a Leica confocal microscope at 63× magnification using DPX2 mounting medium. Images were analyzed by ImageJ software (version 1.53k).

Cell supernatants from 72-hour culture of Breg and non-Breg cells were harvested and analyzed for IL10 (Ray Biotech, catalog no. ELH-IL10). Cell supernatants from differentiated naïve T cells to Tfh cells were harvested at different time points and monitored by ELISA for IL21 levels (Invitrogen, catalog no. BMS2043). Both ELISAs were run in a Thermo Labsystems Multiskan instrument, EX at 450 nm according to the manufacturer's instructions (RayBiotech & Thermo Fisher Scientific). For serum isolation from human blood (healthy donor and tumor patient blood), we collected the blood in a tube that contained no anticoagulant and allowed the blood to clot for 30 minutes at room temperature. Then we centrifuged at 1,000 × g for 10 minutes in a refrigerated centrifuge machine. The serum was separated from the clotted blood and analyzed for different immunoglobulins (IgG, IgM, IgA) using a nephelometer instrument (BN ProSpec System, Siemens Healthineers) and a system kit specific to BN ProSpec. The IgG subsets from cell culture (Tumor PBMC and Breg depleted tumor PBMC) supernatants were analyzed using the IgG subclass human ELISA kit (Thermo Fisher Scientific, catalog no. 99–1000). Kit based standard IL10, IL21, and IgGs were used to develop a standard curve.

Magnetic bead–sorted CD19⁺CD39^‒ Breg cells (1×10⁵ cells) from breast cancer patients’ blood were cocultured with purified autologous naïve T cells (2×10⁵ cells) activated with anti-CD3/anti-CD28 for 72 hours. From each culture experiment, flow cytometry was used to look for IFNγ, FOXP3 expression, and KI67, and annexin-V positivity in T cells. The change in percentage of IFNγ⁺, FOXP3⁺, KI67⁺ and annexin-V⁺ CD4⁺ cells compared with CD4⁺ cells that were cultured alone was calculated. Similarly, CD4⁺ T-cell proliferation was monitored by the CFSE (BioLegend, catalog no. 423801) dilution assay. From a prepared 5-μmol/L working solution of CFSE, 100 μL of CFSE solution was dissolved in 900 μL of PBS containing 5×10⁵ to 10×10⁵ naïve T cells. This was incubated for 20 minutes at 37°C in the dark. Staining was quenched by adding 5 times the initial staining volume of 10% FBS cell culture media. The cells were then pelleted and resuspended in a cell culture pre-warmed media. After data acquisition on a flow cytometer (BD FACS Verse), the cell-proliferation index was calculated using FlowJo software.

RNA-array data containing three B-cell populations CD19⁺CD24^hiCD38^hi, CD19⁺CD24⁺CD38^‒, and CD19⁺CD24^intCD38^int cells from 5 healthy donors with the identifier GSE76272 was downloaded from the Gene Expression Omnibus (GEO) database. The normalized gene expression data was downloaded in Excel format. A log2 fold change and a Student t test were carried out in Excel to identify the significant differentially expressed genes. At first, only significant genes were chosen using P < 0.05 as a cut-off value, followed by selecting the top 50 up- and downregulated genes. Surface-expressing genes were selected by using the Ingenuity Pathway Analysis (IPA) programme (QIAGEN; ref. 19). Heat map analysis for transcriptomic data was performed using Multiple Experiment Viewer (MeV_4_8) software (20).

Semi-qPCR was conducted for the detection of isotype class-specific GLTs and AID. According to the manufacturer's protocols, RNA was isolated from the CD40LG+IL4+IL10—treated PBMCs (germline transcription detection, already mentioned Cell culture) cells using the TRIzol reagent (Thermo Fisher Scientific, catalog no. 15596026). cDNA synthesis was performed with 500 ng of RNA using the cDNA Synthesis Kit (Clontech TaKaRa, catalog no. RR037A). Amplification of 1,000 ng of the cDNA was conducted using specific primers and PCR cycles (Supplementary Table S6). The mean intensity of each band was evaluated using Image J (version 1.53k). Then, using Excel, the fold change was calculated. qPCR analysis of Bcl6 was performed using FastStart Essential DNA Green Master (Roche, catalog no. 06402712001) and run on a LightCycler 96 (Roche). RNA and cDNA synthesis for qPCR analysis was performed as described for PCR. The melting curve was used to assess amplification products using SYBR-green detection with the LC96 SW1.1 software. The data were analyzed using 2^–ΔΔCt method, with β-actin serving as housekeeping controls.

CD19⁺B, CD19⁺CD39^‒ Breg and CD19⁺CD39⁺ non-Breg cell subsets were purified using magnetic beads as described in Immune subset isolation. For adenosine detection, 10,000 cells from each group were cultured in 200 μL PBS in 96-well plates for 40 minutes in the presence of 20 μmol/L ATP (Himedia, catalog no. TC085–5G). One B-cell sample was also treated with a CD73 inhibitor [Adenosine 5′-(α,β-methylene)diphosphate; Sigma-Aldrich, catalog no. M3763] at a concentration of 50 μmol/L. Cell supernatants were collected, centrifuged at 6,000 × g for 2 minutes, boiled for 2 minutes to inactivate ADO-degrading enzymes, and kept at −80°C for further investigation. Adenosine levels were determined using an electrospray-mass spectrometry (ESI–MS) direct infusion (Waters Xevo G2-S QTof) by measuring the intensity of the Adenosine Mass (268). For tuning and calibration for ESI, MS-QTof (Machine name: Water xevo g2 xs-QTof), leucine enkephalin (molar mass 555.62 g/mol) was used. The detection mode was +ve ion mode and the collision energy was 6. The source capillary (kV) was 3.00, the sampling concentration was 40 and the source offset was 80. The temperature of the source was 100°C and desolvation was 250°C. The concentration of the gas (L/h) for gas flow was 50 L/h and the desolvation of gas (L/h) was 400 L/h. The sample injection volume was 5 μL, and the acquisition running time was a minute. The data was analyzed by Masslynx-v.4.1 software.

Using an anti-human Th1/Th2/Th17 CBA array kit (BD Biosciences, catalog no. 560484) according to the manufacturer's instructions, the concentrations of IL2, IL4, IL6, IL10, IL17, TNFα, and IFNγ in healthy donor and tumor patient–derived serum, as well as tumor patient–derived non-Breg and Breg cell–derived cultured supernatant, were evaluated simultaneously. The serum isolation procedure was as described above (see ELISA assay). The BD FACS Verse machine was used to gather the data, and FCAP Array Software v3.0 from BD Biosciences was used to analyze it.

Data analysis was performed using GraphPad Prism 7.0 software. All data was performed in triplicate. Mean ± SD is calculated using GraphPad Prism 7.0 software. The unpaired and paired Student t tests, ANOVA and χ² tests were used to evaluate the significance of differences observed between groups, accepting the value of P < 0.05 as statistically significant.

The microarray data analyzed in this study were obtained from GEO at GSE76272. Other data generated in this study are available within the article and its Supplementary Data files or from the corresponding author upon reasonable request.

Tumor-infiltrating lymphocytes are important modulators of tumor progression (21, 22). While T cells have been extensively studied, the role of B cells in regulating antitumor immunity is less well understood (23). Such a limited understanding of B cells' role in tumor immunology prompted us to explore the atlas of B-cell subsets in the breast tumor microenvironment.

We investigated B-cell subset marker expression, cytokine production, and antibody response in patients with breast cancer and compared them with age and gender-matched healthy people (Supplementary Fig. S1A). On the basis of our concatenated flow-cytometry data (Fig. 1A) of lymphocytes obtained from healthy individuals and breast cancer patients’ blood, we found a higher percentage of B cells in blood from patients with cancer. As B cells are primarily responsible for antibody production, we analyzed serum levels of immunoglobulins, observing that the level of serum-IgG was significantly lower and the level of IgM was significantly higher in cancer patients’ serum than in serum from healthy controls, while the level of IgA remained unchanged (Fig. 1B).

We next conducted a thorough investigation of B-cell subsets and their roles in breast cancer dynamics using multiparametric flow cytometry. A detailed t-SNE analysis of CD19⁺-gated B cells (Fig. 1C) enabled the identification and differential expression of four B-cell subpopulations in the peripheral blood of patients with breast cancer: plasma B cells/plasmablasts (CD19⁺CD38^hiCD27^hi), mature B cells (CD19⁺CD24^intCD38^int), memory B cells (CD19⁺CD24^hiCD27⁺), and transitional B cells (CD19⁺CD24^hiCD38^hi).

The number of transitional B cells was significantly greater in cancer patients’ blood than in blood from healthy donors, while the percentage of CD27⁺ memory B cells and their eventual CD27^hiCD38^hi plasma B cells was lower (Fig. 1C). This finding was validated in a cohort of 10 healthy females and 15 patients with breast cancer (Fig. 1D). Apart from the predicted classical Th2-cytokine bias response (IFNγ skewed to IL4), cytometric bead array results revealed a significantly higher level of IL10 in breast cancer patients' serum (Fig. 1E; Supplementary Fig. S1B).

We investigated the intracellular level of IL10 in the tumor-associated transitional B-cell subset because of the high frequency of transitional B cells in the peripheral blood, increased IL10 level in the serum of patients with breast cancer, and previously reported IL10-positivity as a widely used marker for Breg cells identification (24). From the tumor patient's blood, intracellular IL10 labelling revealed that a considerably larger percentage of CD19⁺CD24^hiCD38^hi transitional B cells were IL10⁺ than mature or memory B cells (Figs. 1F; Supplementary Fig. S1C). Aside from IL10, transitional B cells secrete significantly less IFNγ, IL6, and IL17 than the other B-cell subsets (Figs. 1G; Supplementary Fig. S1D). All of this evidence confirmed that transitional CD19⁺CD24^hiCD38^hi B cells were more prevalent in patients with cancer and contributed to a high amount of IL10 secretion. In the tumor microenvironment, these IL10-producing transitional B cells are Breg cells.

The fact that we found that patients with breast cancer have a higher frequency of IL10-secreting transitional Breg cells encouraged us to investigate the expression pattern of IL10⁺ Breg cells with tumor advancement. Therefore, we used flow cytometry to investigate the frequency distribution of these cells in various pathophysiological conditions. Figure 2A shows that the percentage of CD19⁺CD24^hiCD38^hi Breg cells in the peripheral circulation (10%–12%) and tissues (22%) of the patients with breast cancer were higher than in healthy donor PBMCs. Blood from patients with pathologic disorders such as autoimmune inflammatory disease (RA), on the other hand, showed lower levels of Breg cells than healthy individuals. Breg-cell percentage has also been reported to be low in autoimmune disorders (25). We also checked the percentage of CD19⁺CD24^hiCD38^hi Breg cells in HBV-infected patients’ blood and found that they had a similar expression level as healthy individuals. In Graves' disease and systemic lupus erythematosus patients, CD19⁺CD24^hiCD27⁺ B cells have been defined as Breg cells (11, 26). So, we investigated their status and observed that the frequency of this cell type was lower in breast cancer patients' blood than in healthy donor's blood (Supplementary Fig. S2A), implying that CD19⁺CD24^hiCD38^hi transitional B cells are the only subset that increased in breast tumors.

The interaction between tumor and IL10-producing Breg cells was investigated further using flow cytometry. When comparing cancer patients’ blood and tumor tissue with blood from their healthy counterparts, the percentage of IL10⁺ B cells increased, indicating a positive association between Breg cells and tumors (Fig. 2B). We also used confocal microscopy to validate our findings (Fig. 2C). Supplementary Figure S2B depicts all the controls. After identifying an accumulation of Breg cells in breast tumors, we assessed the percentage of CD19⁺CD24^hiCD38^hi Breg cells in the blood of patients with different stages of disease. Flow cytometry was used to determine the percentage of Breg cells in the patient's blood (Supplementary Fig. S2C), and the multiple data points were then represented in a bar graph (Fig. 2D) revealing that the percentage of Breg was increased as the tumor stage advanced. Our findings collectively imply that CD19⁺CD24^hiCD38^hi Breg cells increased in number during tumor development.

Intracellular IL10-positivity is the hallmark of Breg-cell identification (27). Being an intracellular protein, IL10 can be a signature molecule for Breg-cell identification but cannot be an isolation marker. However, because they have three highly expressed cell surface markers, identified IL10-producing CD19⁺CD24^hiCD38^hi Breg cells have limitations in purifying functional B cells. These are the major reasons behind many functional aspects of the Breg cells being still ill-defined.

We performed a transcriptome analysis using data from a GEO database (GSE76272) to identify distinct Breg-cell surface characteristic markers (10). At first, we shortlisted only significant (P < 0.05) top 50 upregulated and downregulated genes in CD24^hiCD38^hi Breg cells and compared them with CD24^intCD38^int mature and CD24⁺CD38^‒ memory B-cell subsets (Fig. 3A, top). From the IPA, only surface-expressing genes were evaluated (Fig. 3A, bottom). CD39 (ENTPD1), an ATP-hydrolase that converts ATP to ADP/AMP, is one of the most promising surface markers, with minimal surface expression in CD19⁺CD24^hiCD38^hi Breg cells and high expression in both mature and memory B-cell subsets. Sean C. Bendall's group also noticed a similar pattern of CD39 expression in their single-cell multi-omics analysis (28).

Our multicolor flow-cytometric dot-contour overlay analysis from the PBMCs of healthy individuals and tumor patients (Fig. 3B) confirmed that CD39^‒ B cells are CD24^hiCD38^hi. CD19⁺CD39^‒ cells, highlighted as red (red box), and the rest of the grey dots are CD19⁺CD39⁺. In the same CD38 versus CD24 plot, we overlaid this compartment of cells. The vast majority of CD19⁺CD39^‒ cells (red) were CD19⁺CD24^hiCD38^hi. This finding was confirmed in a cohort of breast cancer patients' samples, where there was no discernible difference in the frequency of CD19⁺CD39^‒, CD19⁺CD24^hiCD38^hi, and CD19⁺CD24^hiCD38^hiCD39^‒ B cells (Supplementary Fig. S3A). Clustering (t-SNE) and visualization (FlowSOM) plugins were used to analyze the high-dimensional flow-cytometric data. Within the entire breast cancer patients’ B-cell pool, CD39^‒ B cells were limited in the same place as CD24^hiCD38^hi cells, and their numbers were nearly comparable (Fig. 3C; Supplementary Fig. S3B). The FlowSOM analysis revealed a B-cell subset with negligible CD39 expression but high CD24 and CD38 expression (Fig. 3D).

It is worth noting in this context that previously reported CD19⁺CD39⁺ B cells are immunosuppressive in nature (26, 29). But these cells are not altered in tumor condition. Only CD19⁺CD39^‒ B cells, which are also CD19⁺CD24^hiCD38^hi are increased in tumor conditions (Supplementary Fig. S3C and S3D). We examined numerous features of the CD19⁺CD39^‒ Breg-cell subset to better characterize it. We checked various B-cell developmental markers, including PD-L1, IgM, IgD, and CD73 (Supplementary Figs. S3E and S3F) and found that CD39-positivity was strongly associated with three of the B-cell subsets in breast cancers, including mature and, most notably, memory B cells and plasma B cells, whereas CD39^‒ B cells were not associated with any of the aforementioned subsets except for transitional B cells (Fig. 3E; Supplementary Fig. S3G).

CD19⁺CD39^‒ was found to be an alternate Breg-cell isolation marker to CD24^hiCD38^hi based on transcriptome and flow-cytometric data. We evaluated IL10 status to ensure that CD19⁺CD39^‒ cells were Breg cells. We used a customized magnetic bead–sorting method to separate CD19⁺CD39⁺ and CD19⁺CD39^‒ B cells from breast cancer patients’ blood and activated them for 72 hours in the presence of CD40LG. CD39^‒ B cells made up a higher percentage and secreted more IL10 than CD39⁺ B cells, according to intracellular labelling from flow cytometry (Fig. 3F and G) and cultural supernatants data from ELISA (Fig. 3H). When CD40LG (CD40’s counterpart) was involved, IL10 was elevated in the CD19⁺CD39^‒ Breg-cell compartment. Apart from IL10, our cytometric bead array data from the cell supernatant of CD19⁺CD39⁺ non-Breg and CD19⁺CD39^‒ Breg cells cultured for 72 hours in the presence of CD40LG revealed that CD39^‒ B cells produced significantly lower levels of pro-inflammatory cytokines like IL6, IFNγ, and IL17 than CD39⁺ B cells (Supplementary Fig. S3H). When isolated total B cells and Breg cells were activated and cultured in the presence of LPS, the CD39^‒ Breg percentage was maintained in both cultures even after 7 days (Supplementary Fig. S4A). We also found that Breg cells never redifferentiated from mature B cells in post-Breg depleted samples (Supplementary Fig. S4B) under exogenous stimulus. (CD40LG with IL2 and IL10). These data indicate that the CD39^‒ Breg-cell population is a stable, self-contained population and does not re-differentiate.

We also tested the class-switching properties of Breg cells and observed that the CD39^‒ B-cell subset was class-switch impaired because it could not synthesize adenosine (Fig. 3I and J; Supplementary Fig. S4C) and exogenous adenosine can salvage tumor derived B cells CSR ability (Supplementary Fig. S4D). This class-switch deficient feature of CD39^‒ Breg cells makes it a different B-cell subset from other conventional B-cell subsets.

The identification of suitable surface markers for Breg cells, as well as their prevalence in the breast tumor microenvironment, prompted us to investigate the functional attributes of these CD19⁺CD39^‒ Breg cells. To assess their immunoregulatory function, we isolated CD19⁺CD39^‒ cells from PBMCs from patients with breast cancer (Supplementary Fig. S5A) and cocultured them for 72 hours with purified autologous naïve CD4⁺ T cells (Supplementary Fig. S5B and S5C). The coculture experiment is depicted schematically in Supplementary Fig. S5D. Anti-CD3 and anti-CD28 were used to stimulate naïve CD4⁺ T cells, and the frequencies of CD4⁺IFNγ⁺, CD4⁺FOXP3⁺, and CD4⁺Ki67⁺ cells were assessed using flow cytometry.

When CD4⁺ T cells were cocultured with CD19⁺CD39^‒ Breg cells, it was observed that Breg cells had an inhibitory effect on IFNγ production (Fig. 4A). The inhibitory effect was proportional to the cell ratio (Supplementary Fig. S5E). The Ki67-positivity assay demonstrated that CD19⁺CD39^‒ Breg cells suppressed the proliferation of activated CD4⁺ T cells (Fig. 4B). The Breg cells also induced effector T cell death when they were cultured together (Supplementary Fig. S5F). In addition, the immunosuppressive CD19⁺CD39^‒ Breg-cell subset promoted the generation of suppressive CD4⁺FOXP3⁺ Treg cells from cocultured autologous activated T cells (Fig. 4C). Finally, coculture with Breg cells resulted in reduced cell proliferation of activated CD4⁺ T cells, as measured by the proliferation index level (CFSE dilution; Fig. 4D). All of this evidence suggests the CD19⁺CD39^‒ Breg cells generated in the tumor being immunosuppressive and maintaining immunogenic tolerance in patients with breast cancer.

The adaptive immune response is characterized by somatic DNA recombination such as the CSR required to alter the antibody's effector function. H-chain class switching occurs in three stages: germline gene transcription (GLT), DNA recombination, and B-cell differentiation into memory B cells, followed by Ig-secreting plasma cells (Fig. 5A; ref. 30). Figure 1 shows that Breg-cell frequency was negatively related to memory and plasma B cells. In blood from a cohort of 10 healthy individuals and 10 patients with breast cancer, we detected that the frequency of Breg cells versus memory B cells and plasma B cells has an inverse correlation in patients with breast cancer (Supplementary Fig. S6A). From this observation, we hypothesized that Breg cells may be interfering with the establishment of memory B cells. Memory B cells are classified into two groups based on their IgG and IgM status. Non–class-switched memory B cells (CD19⁺CD27⁺IgD⁺) do not undergo class-switching and generate IgM, whereas class-switched memory B cells (CD19⁺CD27⁺IgD^‒) undergo CSR to produce IgG (Fig. 5B; Supplementary Fig. S6B) (31).

We depleted Breg cells from breast cancer patients' PBMCs and cultured them ex vivo for 7 days in media containing either CD40LG+IL2+IL4 to generate memory phenotype B cells, CD40LG+IL2+IL10 to generate plasma cells, or CD40LG+IL4+IL10 to promote germline-γ1/-γ2 and AID transcription to answer the enigma of Breg cells’ impact on germline gene transcription (GLT), and B-cell differentiation (Fig. 5C). Flow cytometry was used to analyze the differentiation of memory B cells (CD19⁺CD27⁺), more precisely, class-switched (CD19⁺CD27⁺IgD^‒)/non–class-switched (CD19⁺CD27⁺IgD⁺) memory B cells, and plasma B cells (CD38^hiCD138⁺), and PCR was used to analyze CSR of germline-γ1/-γ2 and AID. Our flow-cytometry analysis demonstrated a rise in memory B cell percentages, specifically class-switched memory (Fig. 5D) and plasma B cells (Fig. 5E) in Breg-depleted conditions The class-switched transcripts germline-γ1/-γ2 and AID were also amplified in such conditions (Fig. 5F and G) In this regard, we investigated the role of mature B cells in raising AID levels in Breg-depleted samples. We demonstrated that while a small increase in the fraction of mature B cells could contribute to an increase in AID expression under Breg-depleted conditions, it could not be the sole cause of a large change in AID expression (Supplementary Fig. S6C). In the Breg-depleted condition, the elevated AID level eventually leads to greater surface level IgG (Fig. 5H) and secreted antibodies (IgG1 and IgG2; Fig. 5I) from B cells.

Next, we generated plasma cells ex vivo by incubating Breg-depleted patients' PBMCs with plasma B cell–specific stimuli in combination with CD40LG+IL2+IL10 (Fig. 6A) to investigate the role of Breg cells in the differentiation of Ig-secreting plasma B cells. When Breg cells were eliminated from the cancer patient lymphocyte pool, the frequency of CD38^hiCD138⁺ plasma B cells increased significantly (from 2% to 45%; Fig. 6B, top; Fig. 6C), as did the frequency of class-switched memory B cells (from 7% to 37%; Fig. 6B, bottom; Fig. 6C) after 14 days of culture. We also observed that Breg-depletion alone could cause memory B-cell generation even in the absence of any memory B cell–specific activation (Fig. 6A–C). Our UMAP plots also demonstrate that memory and plasma B-cell cluster frequencies increased in the Breg-depleted condition (Fig. 6D).

All of this evidence supports the notion that CD19⁺CD39^‒ Breg cells have a negative impact on CSR, particularly at the level of GLTs and AID expression. As a result, the generation of class-switched memory B cells is halted, and the number of antibody-producing plasma B cells is reduced resulting in a lower IgG level, which provides an explanation as to why we detected less IgG in breast cancer patients' serum (Fig. 1B).

Previous investigations on B-cell biology have demonstrated that Tfh cells are one of the important immune subsets that trigger class-switching by producing IL21 (Fig. 5A; ref. 32). As our data from Figs. 5 and 6 indicate that Breg cells prevent CSR, we investigated whether Breg cells could disrupt CSR by modulating Tfh-cell responses. To do this, purified naïve T cells (Supplementary Fig. S5B) were cocultured alone or with Breg cells (Supplementary Fig. S5A) and total conventional B cells except Breg (additional control) under Tfh cell–polarizing conditions.

During the process of Tfh-cell differentiation, we noticed that Breg cells had a considerable suppressive effect on CXCR5⁺ programmed cell death protein 1 (PD-1)^hi Tfh-cell development (Fig. 7A), as well as the synthesis of the transcription factor BCL6 (Fig. 7B). Tfh-cell generation was unaffected by the addition of conventional B cells to the naïve T-cell culture. These data altogether suggest that downregulation of Tfh-cell responses were only halted in the presence of Breg cells. Tfh cells are a key source of IL21 (Supplementary Fig. S7A; ref. 33), and we found the percentage of CD4⁺IL21⁺ T cells was considerably lowered in the presence of Breg cells (Fig. 7C; Supplementary Fig. S7B). Following that, we looked at whether, apart from Tfh-cell differentiation and IL21 production, Breg cells had any negative impacts on IL21 levels from already generated Tfh cell. As a result, we ran another flow-cytometry experiment with the same number of Tfh cells from Breg culture, total B-cell culture (except Breg) or grown alone. This time, we observed that Breg cells could also inhibit IL21 production by already differentiated Tfh cells (Fig. 7D).

Next, we attempted to assess plasma-cell generation in relation to the Tfh cell–polarizing conditions. We designed an experiment in which we grew ongoing Tfh differentiated cells alone or with Breg cells and conventional B cells without Breg for 72 hours and collected cell supernatant, followed by a further 5 days of culture with autologous unstimulated B cells (excluding Breg) (Fig. 7E) to assess plasma B-cell generation.

Because recombinant IL21 was added as part of the Tfh-cell differentiation protocol, it was not clear whether the plasma-cell generation was mediated by exogenous IL21 or Tfh-cell secreted IL21. To answer this question, we performed a time-dependent ELISA of IL21 in the conditioned media and noticed that after 24 hours, the majority of the exogenously added IL21 bound to naïve T cells (Tfh cell–polarizing conditions). After 48 hours, the level of IL21 gradually increased in the culture supernatant and peaked at 72 hours (Fig. 7F). Although some background IL21 (recombinant IL21) existed in the cell supernatant after 72 hours, this basal IL21 level was the same for all culture conditions (T-cell alone, with Breg cells, and with B cells).

From the supernatants collected after 72 hours of Tfh-cell generation in various culture conditions, we observed that Breg cells suppressed the IL21 release from differentiating Tfh cells (Fig. 7G). As IL21 getting lowered after Breg engagement, a significant reduction in plasma B-cell generation was observed (Fig. 7H). All of these results suggested that Breg cells inhibit antibody responses by suppressing Tfh-cell differentiation and Tfh-cell IL21 production (Supplementary Fig. S7C).

Antibody responses generated by B cells have the potential to destroy tumor cells, and IgG plays an important role in this process (34, 35). Numerous B-cell biology investigations have found that IgG levels in patients with metastatic breast cancer are much lower and IgM levels are greater than in persons without metastasis (35–38). Low IgG levels have been linked to a poor prognosis (35).

To investigate the reasons for this, we looked at the B-cell atlas and noticed that in the blood of patients with breast cancer there was an increase in the IL10-secreting transitional CD19⁺CD24^hiCD38^hi Breg-cell population, which have immunosuppressive properties and showed an inverse relationship with switched memory and antibody-producing plasma B cells. The inverse correlation between memory and Breg cells has also been reported in pediatric immune thrombocytopenia (39). This information suggested the possibility that transitional Breg cells have a negative impact on B-cell differentiation and antibody-mediated immunity. To test this hypothesis, we needed to isolate viable CD19⁺CD24^hiCD38^hi Breg cells. However, these markers have limitations for isolating functional Breg cells because they comprise three highly expressed cell surface markers, leaving flow sorting as the only option. Unfortunately, as the cell numbers are lower and they have to pass through the laser, it is extremely difficult to acquire sufficient unstressed live Breg cells.

High-dimensional flow cytometry and microarray analysis helped us to find CD19⁺CD39^‒ as an alternate marker for CD19⁺CD24^hiCD38^hi B cells, allowing us to overcome this limitation. CD39 expression on the cell surface is strongly associated with mature B cells, memory B cells, and plasma B cells. CD39-negativity, on the other hand, is not linked to any of the above subsets, except for transitional B cells that are CD19⁺CD24^hiCD38^hi. CD39 is an E-NTPDase that hydrolyzes ATP to ADP/AMP and finally to adenosine by CD73 from activated B cells (40). The released adenosine binds to the B cells and helps in differentiation into class-switched B cells (41). So, it can be assumed that in the absence of any of these cell-surface molecules, CD39 and CD73 or if the pathway is blocked, adenosine synthesis may be hindered, resulting in impaired CSR. Consistent with the fact that our identified Breg cells lack CD39, we observed that these cells produce less adenosine than CD39⁺ B cells and that addition of exogenous adenosine could restore the CSR events in B cells. As a result of a greater number of CD39^‒ Breg cells failing to produce adenosine in tumor, cannot assist themselves or other B cells in class switching.

We determined that the IL10-producing CD19⁺CD39^‒ tumor-associated Breg cells we identified are a stable, self-contained population and do not re-differentiate. These Breg cells perform a variety of immunoregulatory functions including inhibiting T helper–cell proliferation, type-1 cytokine production, and T effector–cell survival, and inducing the generation of CD4⁺FOXP3⁺ Treg cells. Studies indicate that IL10 is a characteristic molecule of transitional Breg cells that binds to T cells and promotes immunosuppression (13). TGFβ, another immunosuppressive molecule, has little to no effect on the suppressive action of Breg cells because its expression by Breg cells is similar to that of non-Breg populations. Several studies have revealed that interactions of CD80/CD86 on Breg cells with CD28/CTLA4 on CD4⁺ T cells work synergistically to release IL10 from B cells, suppressing T-cell pro-inflammatory cytokine production (13, 42). PD-L1 expressed by Breg cells, which can bind to PD-1 on T cells, also limits T helper–cell responses (43, 44). All of these immunosuppressive processes in patients with cancer convert the environment from immunogenic to tolerogenic, which helps the tumor to grow more aggressively.

To test the possibility that Breg cells have a negative impact on antibody responses, we depleted Breg cells from the PBMCs of patients with breast cancer. Breg depletion resulted in substantial formation of germinal transcripts and expression of AID, which is involved in H-CSR, and differentiation of B cells into class-switched memory and antibody-producing plasma B cells. It is well established that Tfh cells play an important role in the establishment of germinal centers (32, 33). Breg cells were found to have a negative impact on Tfh-cell generation and IL21 production. As Breg cells prevent CSR by inhibiting the Tfh/IL21 axis, long-lived plasma cell generation is halted, resulting in lower IgG and a higher fraction of B cells unable to undergo CSR events, resulting in higher IgM. Furthermore, CD39^‒ transitional B cells (abundant in patients with cancer) produce high-IgM/low-IgG because they do not participate in CSR due to their lower adenosine-generating capacity. The combined effects of Breg cells on Tfh cells and on adenosine levels might be the cause of the cancer patient's high-IgM/low-IgG event. In this regard, it is worth noting that the molecule(s) responsible for Breg cell–mediated suppression is still under investigation. PD-L1 could be the molecule expressed by Breg cells that prevents class-switching. PD-L1 may bind to PD-1 on Tfh cell, inhibiting IL21 production and impairing both memory and plasma B cell development (44). In-depth studies using single-cell genomics and conditional-knockout mouse models may provide more insight into the impact of Breg cells in B cell–mediated immune regulation during tumor progression.

Our current work suggests that CD19⁺CD39^‒ is a key signature surface-marker for tumor-associated IL10⁺ Breg cells, and their augmentation in patients with breast cancer restricts Tfh-cell expansion and IL21 secretion, hence inhibiting CSR during antibody responses (Supplementary Fig. S8). Breg-cell depletion could, therefore, be a promising future strategy for enhancing the magnitude of antibody responses and the dynamics of memory and plasma B cell–mediated cancer immunotherapy.

G. Sa reports grants from Department of Biotechnology; and grants from Council of Scientific and Industrial Research during the conduct of the study; in addition, G. Sa has a patent for Breg cell sorting pending. No disclosures were reported by the other authors.

S. Pati: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft. S. Mukherjee: Visualization, methodology. S. Dutta: Methodology. A. Guin: Methodology, writing–review and editing. D. Roy: Methodology. S. Bose: Methodology. S. Paul: Writing–review and editing. S. Saha: Software, visualization. S. Bhattacharyya: Conceptualization. P. Datta: Clinical sample providers. J. Chakraborty: Clinical sample providers. D.K. Sarkar: Clinical sample providers. G. Sa: Conceptualization, resources, formal analysis, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.

The authors acknowledge R.K. Dutta and G. Banik for technical help with flow-cytometry studies, A.K. Poddar and S.G. Chakraborty for assistance with confocal microscopy, and S. Roy for his help in mass spectrometry. The authors also thank A. Ghosh, Department of Rheumatology, IPGMER-SSKM Hospital, Kolkata, for his assistance in collecting Nephelometer data and S. Moulik for providing HBV blood. G. Sa is the recipient of NASI Platinum Jubilee Senior Scientist Fellowship. The study was funded by grants from the Department of Biotechnology and Council of Scientific & Industrial Research, Indian council of Medical Research, Government of India.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.