Triple negative MDA-MB-231 breast carcinoma cells were purchased from ATCC and were routinely cultured in RPMI 1640 (Invitrogen, 11875-093) medium supplemented with 10% FBS (Gemini Bio Products, GBP-100106), penicillin-streptomycin (Gibco, catalog number #10378016/15070-063) and L-glutamine (Catalog number # A2916801 complete growth medium (Thermo Fisher Scientific, USA) at 37°C and 5% CO₂ as reported previously (Kaur et al., 2016).

HEK 293 cells were cultured using DMEM complete media with 10% FBS, Penicillin-Streptomycin, and L-glutamine at 37° with 5% CO₂. SIRPα-Fc plasmid DNA ((Miller et al., 2019) was transfected in to HEK293 cells using DMEM serum free media with 0.01%BSA. After 2 days, the media were collected and concentrated using Millipore Sigma Amicon Ultra-15 Centrifugal Filter Units (Fisher Scientific, USA) with a 10 kDa cutoff. Concentrated condition media were purified via Protein A IgG Purification Kit (Thermo Fisher Scientific, USA). Purity of the SIRPα-Fc was confirmed using gel electrophoresis, and concentration was determined using the BCA assay (Thermo Fisher Scientific). For some experiments SIRPα-Fc was purchased from R&D systems. Monovalent SIRPα-Avi-Biotin (SIRPmv) was used as reported previously (Miller et al., 2019).

MDA-MB-231 cells were pretreated with the Celgene antibody for times indicated in the Figure Legends. MDA-MB-231 Cells were then stained with PE anti-human CD47, APC/Cy7 anti-human CD24, FITC anti-human CD44 along with isotype control antibodies; FITC Mouse IgG1, PE Mouse IgG1, APC/Cy7 IgG2a (Biolegend, USA). MDA-MB-231 cells were washed three times and resuspended in Hanks’ balanced salt solution at 1 × 10⁶ cells in 500 μL. Samples then were acquired on a LSRII (BD Biosciences). Flow cytometry data were analyzed using FlowJo software (TreeStar).

bCSC were generated using sorted CD44⁺/CD24^- (Supplementary Figure S1F) cells from the MD-MB-231 cell line. The suspension cells were cultured using SmartDish™ with MammoCult™ Human Medium Kit from Stem Cell Technologies. The suspension cells were treated with CC90002 (1 μg/mL) or SIRPα-Fc (1 μg/mL) for 72 h, and analyzed using APC/Cy7 anti-human CD24, FITC anti-human CD44 along with FITC Mouse IgG1, κ isotype Ctrl, κ Isotype Ctrl (FC), APC/Cy7 IgG2a, κ isotype Control flow antibodies. For mRNA expression, suspension cells were generated from MDA-MB-231 by using AggreWell™400 with AggreWell™ EB Formation Medium from Stem Cell Technologies and treated with CC90002 (1 μg/mL) and SIRPα-Fc (1 μg/mL) for 72 h as described above.

MDA-MB-231 cells were washed with PBS, dissociated with Gentle cell dissociation Reagent (STEM Cell Technologies) and centrifuged for 5 min at 1,200 rpm. The cells were re-suspended into cell isolation buffer containing phosphate-buffered saline (PBS), pH 7.2, 0.5% bovine serum albumin (BSA), and 2 mM EDTA. MDA-MB-231 cells were incubated with biotin anti-human CD24 (Biolegend) for 10 min on ice. Mojosort™ Streptavidin Nanobeads were added to the respective tubes containing cells and incubated for another 10 min on ice. Enriched CSCs cells were separated using MiniMac Separation columns, type MS (Miltenyi Biotech Inc.) according to the manufacturer’s instructions. The CD24^high (CD24⁺) cells became attached to beads, while the CD24^Low cells (CD24⁻) flowed through the columns and were collected into new tubes as shown in Supplementary Figure S1G via flow cytometry analysis. The sorted CD24⁺ and CD24⁻ cells were further treated with divalent or monovalent SIRPα-Fc for 24 h, and total RNA was extracted for Real Time PCR analysis.

Total RNA was isolated using either TRIzol Reagent (Thermo Fisher Scientific) or TriPure™ Isolation Reagent from Roche (Millipore Sigma). cDNA was synthesized using the Maxima First Strand cDNA Synthesis Kit for RT-qPCR, with dsDNase (Thermo Fisher Scientific) Quantitative real-time PCR was performed using KLF4, OCT4, SOX2 using 18S or B2M, SNAIL, SLUG, ZEB1, ZO-1, ALDH1A1, CDH1, CTNNB1 and CSNK1A1 primers using SYBR Green (Thermo Fisher Scientific) on an MJ Research Opticon I instrument (Bio-Rad) as control as described previously (kaur et al., 2019, scientific reports). EMT and cancer stem cell primers (Table 1) were purchased from Integrated DNA Technologies (IDT, USA). The results were quantified as Ct values and expressed as fold gene expression (the ratio of treated/control).

MDA-MB-231 cells were cultured in AggreWell™ Embryoid Body (EB) Formation Medium using SmartDish™ (STEMCELL Technologies) and treated with CC90002 (1 μg/mL) and SIRPα-Fc (1 μg/mL) alone for 72 h at 37°. The loosely aggregated spheroids were suspended into single cells and subjected to BODIPY-aminoacetaldehyde (BAAA) aldehyde dehydrogenase substrate and/or DEAB as negative control staining using ALDEFLUOR™ kit from STEMCELL Technologies. The protocol and flowcytometry analysis were followed according to manufacturer’s instructions. % ALDH^br cells from three independent experiments were calculated using background subtraction of negative control.

IncuCyte NucLight for cell nuclear labeling (ESS4717, Essen BioScience) was used for living cells. MDA-MB-231 (∼1,000 cells/50 µL) were plated in a 96-well plate. CC90002 antibody (1 μg/mL) and SIRPα-Fc (1 μg/mL) were added alone or in combination. Diluted IncuCyte Nuclight Rapid live cell labeling reagent (50 µL) added to each well according to manufactures instructions (Essen BioScience, Santorious). The cells were analyzed every 4 h for 7 days using Basic Analyzer analysis (IncuCyte system software).

MDA-231 cells were grown in culture, (RPMI media w/1% Pen-Strep, 1% L-glutamine, and 10% fetal bovine serum at 37° with 5% CO₂. Approximately 1000 MDA-MB-231 cells were transferred to a small petri dish along with AggreWell™ Embryoid Body (EB) Formation Medium or MammoCult™ Human Medium Kit and treated with CC90002 (1 μg/mL) or SIRPα-Fc (1 μg/mL) alone for 10 days at 37°. The total number of mammospheres were counted using a light microscope, and number of spheroids were plotted using Prism software. For live cell mammospheres formation, MDA-MB-231 triple-negative breast carcinoma cells were labeled with PKH26 dye (Sigma) at day 0 according to the manufacturer’s instructions. Approximately 1,000 cells were transferred into each well of a 96-well plate using mamo cult media. Images were captured every 4 h using Spheroid analysis with the IncuCyte system software.

MDA-MB-231 cells were trypsinzed, and expression of CD47 and SIRPα were measured using flow cytometry (Supplementary Figure S1A, B). To investigate effects of CC90002 Ab on SIRPα-CD47 binding, cells were incubated with monovalent SIRP-α-biotin (Miller et al., 2019) on ice for 30 min in the presence or absence of CD47 antibodies. The expression of SIRPα was analyzed using streptavidin AF-488 via flow cytometry analysis.

MDA-MB-231 cells were treated CC90002 (1 μg/mL) and SIRPα-Fc (1 μg/mL) alone for 72 h at 37° along with untreated and human isotype control (Celgene) using AggreWell media (Stem Cell Technologies) according to manufacturer’s instructions. Total RNA was extracted using ISOLATE II RNA Mini Kit from BIOLINE, and RNA integrity and quantification was measured using RNA-Bioanalyzer.

All mRNA-seq samples were pooled and sequenced on NextSeq using Illumina TruSeq Stranded mRNA Library Prep and paired-end sequencing (GEO Accession Number: GSE247052). The samples had 20 to 33 million pass filter reads with a base call quality of above 83% of bases with Q30 and above. Reads of the samples were trimmed for adapters and low-quality bases using Trimmomatic software before alignment with the reference genome (Human - hg38) and the annotated transcripts using STAR. The average mapping rate of all samples was 95%. Unique alignment is above 57%. There were 1.81%–38.56% unmapped reads. The mapping statistics were calculated using Picard software. The samples had 1.04% ribosomal bases. Percent coding bases are between 44% and 60%. Percent UTR bases are 31%–42%, and mRNA bases were between 79% and 92% for all the samples. Library complexity was measured in terms of unique fragments in the mapped reads using Picard’s Mark Duplicate utility. The samples had 59%–91% non-duplicate reads. In addition, the gene expression quantification analysis was performed for all samples using STAR/RSEM tools.

The RNA-Seek pipeline (https://github.com/CCBR/RNA-seek) was used to process reads. Expected counts from RSEM for both genes were imported into the NIH Integrated Data Analysis Platform (Palantir Technologies) for downstream analysis as described earlier (Nath et al., 2022). Briefly, low count genes were filtered prior to CPM and quantile normalization using Limma voom (Smyth, 2004; Law et al., 2014), followed by differential expression of genes analysis. Pre ranked gene set enrichment analysis (GSEA) using molecular signatures database v6.2 (Subramanian et al., 2005; Liberzon et al., 2011). Batch correction was performed using the ComBat function of the sva (Johnson et al., 2007).

All experiments were replicated at least three times on different groups of cells. All data are expressed as mean ± SD. The differences were considered significant at p values <0.05 as indicated in the figure legends.

Based on our previous studies demonstrating effects of the CD47 antibody B6H12 on stem cell markers in MDA-MB-231 cells (Kaur et al., 2016), which have a high number of bCSC as compared to MCF7 or T47D1 cell lines (Honeth et al., 2008), we examined effects of SIRPα-Fc and the CC90002 antibody on the expression of CD44 and CD24. Flow cytometry confirmed that MDA-MB-231 cells highly express cell surface CD47, whereas only a minor population express detectable SIRPα (Supplementary Figure S1A, B). Binding of SIRPα-Fc to cell surface CD47 was inhibited in the presence of CC90002 (Supplementary Figure S1C). However, treatment of MDA-MB-231 cells with SIRPα-Fc (SIRPFc) or CC90002 (CG) for 72 h did not significantly alter cell surface expression of the stem cell markers CD44 or CD24 (Figures 1A, B). Consistent with these results, treatment of MDA-MB-231 cells with 1 μg/mL of SIRPα-Fc or the CC90002 antibody for 72 h did not significantly change the mRNA expression of OCT4, SOX2, KLF4, CD44, and CD24 genes. However, SIRPα-Fc treatment significantly downregulated CD44 mRNA (Figures 1C–G). Therefore, the effects of SIRPα-Fc and CC90002 antibody on mRNA expression of pluripotent stem cell markers differ from those of B6H12.

Aldehyde dehydrogenase 1 (ALDHA1) is a CSC marker expressed in basal breast cancers (Tan et al., 2013). Expression of specific isoforms of (ALDHA1) and the corresponding enzymatic activity are useful markers of CSC differentiation and can be quantified by flow cytometry using Aldefuor (Moreb, 2008; Marcato et al., 2011; Tomita et al., 2016). MDA-MB-231 cells cultured on plastic in RPMI medium had undetectable ALDHA1 mRNA (Supplementary Figure S1D). MDA-MB-231 cells cultured in AggreWell medium exhibited expression of ALDHA1, but treatment with CC90002 or SIRPα-Fc did not significantly change expression of ALDHA1 mRNA (Supplementary Figure S1E). However, assessment of ALDH1 activity using the ALDEFLUOR Kit, indicated that, SIRPα-Fc significantly increased the percentage of Aldefluor-positive cells as compared to untreated (Figure 1H, J, K). Treatment with CC90002 resulted in a smaller increase in Aldefluor-positive cells that was not statistically significant (Figure 1I, K). Therefore, SIRPα-Fc may be a selective CD47-dependent inducer of stem cell character in these breast carcinoma cells.

Formation of mammospheres provides a quantitative assessment of stem cell character and aggressiveness in breast carcinoma cells (Shaw et al., 2012; Margaryan et al., 2019). SIRPα-Fc treatment resulted in ∼3-fold increase in the number of mammospheres and their combined area (Figure 2A). The area of mammospheres increased less in cultures treated with CC90002 antibody and did not achieve significance. This data indicates that SIRPα-Fc increases either the proliferation or self-renewal capacity of breast cancer stem cells (Figure 2A). We also assessed Mammosphere size using a spheroid dye dilution assay on the IncuCyte instrument (Figure 2B) along with unlabeled control treatments. Depletion of the PKH dye signal indicated increased cell proliferation in spheroids. Decreased fluorescence in SIRPα-Fc-treated cells indicated strong stimulation of cell proliferation, whereas the CC90002 antibody had minimal activity (Figure 2B). SIRPα-Fc (1 μg/mL) treatment increased mammosphere size over 5 days, whereas CC90002 antibody was less active as shown in Figure 2A.

To assess effects of CD47 ligands more directly on MDA-MB-231 bCSC proliferation, we used the IncuCyte Rapid Red Dye assay (Figure 2C). SIRPα-Fc treatment of MDA-MB-231 cells resulted in the most increase in cell growth. CC90002 modestly increased cell proliferation of MDA-MB-231 cells.

To examine effects of these CD47 ligands on the global transcriptome, MDA-MB-231 cells were grown in AggreWell Embryoid Body (EB) Formation media for 72 h in the presence of SIRPα-Fc (1 μg/mL), CC90002 (1 μg/mL), IgG isotype antibody control, or untreated controls (n = 3). Bulk RNA sequencing analysis was performed and analyzed using the NIDAP platform. The QC with Batch Correction, Differential Expression of Genes and Visualization data indicated consistent high-quality results (Supplementary Figure S2A-C). Differentially expressed genes between SIRPα-Fc and CC90002 antibody treatments compared with untreated or control IgG treatment were determined using a threshold p-value of 0.05 (Supplementary Figure S2D). The differentially expressed genes between SIRPα-Fc versus UT, and SIRPα-Fc versus IgG or CC90002 versus UT and CC90002 versus IgG comparisons are shown in Supplementary Data S1. Gene set enrichment analysis (GSEA) for multiple pathways (Supplementary Figure S2E) and list of significant GSEA datasets is summarized in Supplementary Data S2.

GSEA revealed a negative enrichment between SIRPα-Fc versus untreated, and SIRPα-Fc versus IgG treatments (Figures 3A, B, ES = −0.39, p-value 0.00037 and ES = −0.44, pval 0.00032) for genes that regulate epithelial mesenchymal transition (EMT), whereas CC90002 treatment showed a positive enrichment of the same gene set (ES = 0.34) (Figure 3C), which was significant compared to untreated samples but not significant for Isotype control antibody treated (Data S2).

The stem cell marker CD24 is associated with EMT and cell proliferation (Nakamura et al., 2017) and has additive effects with CD47 to regulate phagocytosis (Ozawa et al., 2021). Therefore, we examined expression of genes in this pathway in CD24⁺ enriched and CD24⁻ MDA-MB-231 cells (Figures 4A–H). COL4A1 and ITGA3 genes were selected from list of GSEA Figures 3A, B to validate the RNA sequencing analysis based on their role in proliferation, EMT and stemness of breast cancer cells (Halsted et al., 2008; Jin et al., 2017; Wang et al., 2020b; Zhang et al., 2020). SIRPα-Fc but not SIRPmv treatment significantly enhanced expression of COL4A1 mRNA in CD24⁺ cells but not in CD24⁻ cells, whereas ITGA3 mRNA expression was significantly upregulated in CD24⁻ cells and decreased in CD24⁺ cells with SIRPmv treatment (Figures 4A, B). Treatment using divalent SIRPα-Fc significantly increased casein kinase 2 (CK2/CSNK2A1) mRNA levels in CD24⁺ cells but only modestly in CD24⁻ cells. In contrast, treatment using monovalent SIRPmv significantly downregulated casein kinase 2 mRNA in both CD24⁺ and CD24⁻ cells (Figure 4C). These divergent responses suggest that dimerization of cell surface CD47 by SIRPα-Fc results in a different signal than monovalent ligation of CD47 by SIRPmv. SIRPα-Fc but not SIRPmv treatment downregulated E-cadherin (CDH1) mRNA expression in CD24⁺ cells, but modestly increased expression in CD24⁻ cells (Figure 4D). Similarly, SNAIL mRNA was modestly upregulated with SIRPα-Fc but not SIRPmv treatment only in CD24⁺ cells (Figure 4E). The EMT markers ZEB1, SLUG and ZO-1 were not significantly changed with SIRPα-Fc or SIRPmv treatments in both CD24⁻ and CD24⁺ cells (Figures 4F–H).

In contrast to SIRPα-Fc. CC90002 antibody treatment had no significant effect on CK2 and CDH1 mRNA expression in CD24⁺ and CD24⁻ cells (Figures 5A, B). However, using unfractionated MDA-MB-231 cells, CC90002 antibody treatment significantly downregulated CK2 and ZEB1 mRNA expression (Figures 5C, D), but no other EMT markers were significantly altered (Supplementary Figure S3A-C). Loss of E-cadherin has been associated with progression and survival in human breast cancer (Lipponen et al., 1994; Singhai et al., 2011). This may indicate that SIRPα-Fc treatment increases cancer cell survival, and CK2 may be a key player for EMT driven proliferation or spheroid formation (Zhang et al., 2014). However, the effects of CC90002 diverged from those of SIRPα-Fc, indicating that effects of therapeutic CD47 binding molecules on breast cancer stem cells are ligand-specific.