Continuous exposure to doxorubicin induces stem cell-like characteristics and plasticity in MDA-MB-231 breast cancer cells identified with the SORE6 reporter

Cell culture and drug

We obtained MDA-MB-231 (HTB-26) and Hs578t (HTB-126) TNBC cells from American Type Culture Collection (ATCC). Only cells below Passage 20 were used for our experiments. MDA-MB-231 cells were cultured routinely in Leibovitz's L-15 with 10% fetal bovine serum (FBS, F3135 Sigma-Aldrich) at 37 °C in a humidified atmosphere without CO₂. Hs578t and HEK293T (CRL-3216) cells were grown routinely in Dulbecco’s Modified Eagle Medium (DMEM), supplemented with 10% FBS, at 37 °C in an atmosphere with 5% CO₂. Doxorubicin (D1515 Sigma-Aldrich) was prepared in Milli Q water and stored at -70 °C with protection from light.

Identification and isolation of GFP + cells (CSC-like)

The identification and quantification of GFP + cell populations (CSC-like) were performed in sublines that were generated by lentiviral transduction of a reporter system that contained 6 concatenated repeats of SOX2/OCT4 response elements, which drive the expression of destabilized green fluorescent protein (SORE6-GFP). A minimal cytomegalovirus promoter, (mCMVp)-GFP, was used to generate control cell lines. Both reporter constructs were kindly provided by Dr. L.M. Wakefield (National Cancer Institute, Bethesda, MD, USA) [22]. Lentiviral particles were generated SORE6-GFP and mCMVp-GFP, as previously described. HEK293T cells were transfected with the psPAX2 packaging plasmid and the pMD2.G envelope plasmid using Lipofectamine 3000 (L3000-001, Invitrogen)[23]. For transduction with lentiviral particles, MDA-MB-231 or Hs578t cells were exposed to lentiviral supernatants diluted 1:1 in fresh medium for 24 h with 8 μg/ml Polybrene (28728–55-4, Santa Cruz). Transduced cells were positively selected through a sequential treatment with puromycin 0.5 μg/mL (P4512, Sigma-Aldrich) for additional 72 h and GFP-based cell sorting (FACS Aria II Cell Sorter).

MDA-MB-231 SORE6 cells were used to isolate GFP + and GFP- populations by flow cytometry. For purification of GFP + and GFP- populations after 19–22 weeks of exposure to doxorubicin, cultured cells were sorted based on the expression of GFP + using gates that were defined by the control mCMVp-GFP. Cell sorting was performed on a FACSAria, and cells were used immediately for the mammospheres assay, CellTrace violet staining or RNA extraction. The purity of cells was > 95% for each population.

MTT cell viability assay

The effects of doxorubicin on viability were determined in cells that were in the exponential growth phase by MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide] assay. Cells were seeded in 96-well plates at a density of 1 × 10⁴ cells/well. Cells were treated with and without doxorubicin as indicated (0–10 μM for MDA-MB-231 or 0-1000 nM for Hs578t). After 72 h, 20 μL MTT solution (5 mg/mL) was added for 3 h at 37 °C. The resulting formazan crystals were dissolved in 100 μL of DMSO. The amount of reduced tetrazolium salt was measured spectrophotometrically at 570 nm (Epoch, Biotek). By nonlinear regression of 4 parameters, we obtained the IC50 and IC75 values (GraphPad Prism v6.0).

Establishment of doxorubicin-resistant MDA-MB-231-SORE6 cells

Resistant cells were established by continuous and gradual exposure at doxorubicin at increasing concentrations (1–10 nM). The concentration range of doxorubicin was determined, according to our dose–response curve (Supplementary Fig. 1). A passage was performed when the culture was 70% to 80% confluent; each passage was considered to be 5–7 days. Acquisition of resistance occurred at approximately 19–22 weeks. The cells were considered to be resistant at peak viability compared with control cells at IC 50 and IC75 (evaluated by MTT cell viability assay). The cultures (control and Dox) were maintained under identical conditions and for the same period to allow for appropriate comparison between the 2 conditions.

Quantification of apoptosis by flow cytometry

To evaluate the intrinsic resistance of MDA-MB-231-SORE6 TNBC cells, we exposed them to doxorubicin (IC50) and evaluated their viability by flow cytometry using Annexin V-Alexa Fluor 647 (A23204, Invitrogen)/7AAD (A9400, Sigma). Briefly, cells were seeded in 24-well plates at a density of 50 × 10⁴ cells/well. Cells were treated with and without doxorubicin (0.43 μM) for 72 h. After incubation, the cells were collected in PBS (pH 7.4) and centrifuged for 5 min at 12,000 rpm. The cells were incubated in 300 μl annexin-binding buffer at room temperature for 5 min and double-stained with Annexin V-Alexa Fluor 647 and 7AAD for 15 min and 20 min, respectively. The stained cells were acquired on a Attune NxT cytometer (Applied Biosystem, Attune NxT), and the data were analyzed in FlowJo, V.10.0.

Mammosphere formation assay

Mammosphere formation assay was performed as reported [24]. Briefly, purified and unpurified cells were plated at low density (100 viable cells per well) on a 96-well ultra-low attachment plate (Corning Costar) with MammoCult medium and growth factors (05620, StemCell Technologies). The number of mammospheres with diameter > 60 μm was quantified on Day 7 by taking micrographs (Eclipse Ti-U microscopy, Nikon) and analyzed in NIS-Elements Imaging Software 4.13. The results are expressed as mammospheres-forming efficiency (MFE), which was calculated with the following equation: $MFE=\frac{N\text{umber of mammospheres per well}}{\text{Number of cells seeded per well}}\times 100$.

CellTrace violet staining

CellTrace Violet (C34571, Molecular Probes) staining was carried out per the manufacturer's instructions. In brief, isolated cells (GFP + or GFP- populations) were stained with 2.5 μM CellTrace Violet in PBS (5 min, room temperature, in the dark) with occasional stirring. Staining was quenched with the addition of 5 volumes of PBS that contained 10% FBS; the cells were then centrifuged (5 min, 1200 rpm), resuspended in complete L-15 medium, and cultured immediately. Data are expressed as Fold dye dilution of CellTrace (FDD) and was calculated with next equation:  $FDD=\frac{\text{median fluorescence intensity}(t0)}{\text{median fluorescence intensity}(t72h)}$.

RNA-seq and analysis of gene expression

Total RNA was extracted from the various cells using the RNeasy Kit (74004 Qiagen). RNA quality was assessed (Agilent Bioanalyzer); all RIN values were greater than 7.0. RNA libraries were generated (KAPA RNA) for RNA-sequencing by 50-bp paired-end sequencing in replicate (Illumina HiSeq 2500). In total, 10⁷ reads were mapped using UCSC hg19, and we followed a workflow to detect differentially expressed (DE) genes and gene ontologies from raw RNA-Seq data using Galaxy. DESeq2 was then used to normalize the read counts and extract DE genes. Heatmap2 and Volcano Plot were used to visualize DE genes in GSEA v2.2.0 using the MSigDB database v5.1. Functional enrichment analysis of the DE genes was performed using g:Profiler version e111_eg58_p18_30541362 (database updated in 2024).

The heat maps and correlations between genes in the same patient cohort were verified and analyzed by data mining in TCGA-BRCA using the UCSC Xena browser; also, survival curves were generated (http://xena.ucsc.edu/). Overall survival (OS) was defined as the time from the first diagnosis of primary breast cancer to death from any cause. All results are displayed as P values by log-rank test.

Statistical analysis

Half-maximal inhibitory concentration (IC₅₀ and IC₇₅) values were calculated by non-linear regression. Statistical significance was determined by one-way ANOVA or Tukey's or student's test. P values ≤ 0.05 are reported. GraphPad Prism (v6.0) was used to perform these analyses.

Doxorubicin-induced enrichment of GFP + cells (CSC-like) in TNBC cells

Cancer stem cells (CSCs) are highly associated with the development of drug-resistant cancer cells, and TNBC is known for its high CSC population [22, 25, 26]. In this study, we used a reporter system (SORE6) [22] that allows the real-time monitoring of such properties as phenotypic plasticity and the response to therapeutic agents in CSC-like (GFP +) cells.

To establish a resistant TNBC cell line, MDA-MB-231-SORE cells were cultured with (Dox) or without doxorubicin (control) under identical conditions and for the same period to allow for an appropriate comparison between the 2 conditions (Fig. 1A). Following treatment with increasing doxorubicin concentrations, the cells became enriched in GFP + populations (CSC-like) compared with control cells (Fig. 1B). We observed enrichment of GFP + cells from the first month of doxorubicin exposure; this enrichment was also seen in Hs578t cells (Supplementary Fig. 1).

Enriched cells develop resistance to doxorubicin and increase their mammosphere-forming efficiency (MFE)

To evaluate resistance in cells that are continuously exposed to increasing concentrations of doxorubicin (1–10 nM), we analyzed the viability of cell culture at IC50 (0.43 μM) or IC75 (1.44 μM) for 72 h (Fig. 2A). The control cells showed 50% cell viability at 72 h, whereas Dox cells showed 90% and 70% viability at 0.43 μM and 1.44 μM doxorubicin, respectively. We confirmed that cells that were exposed to doxorubicin developed resistance, observing viability at concentrations 25-fold greater than the plasma concentrations of doxorubicin in patients (~ 18 nM) [27]. Given that CSCs have been reported as being intrinsically resistant to chemotherapy [22], we analyzed the viability of GFP + (CSC-like) and GFP- populations (non-CSCs) that were exposed to doxorubicin at IC50 in näive cultures. In these cultures, we noted a pronounced cytotoxic effect at the IC50 of doxorubicin, particularly affecting the GFP + populations (Supplementary Fig. 2). Gargini and colleagues also observed that doxorubicin induces cell death and cell cycle arrest in the CD44^high/CD24^low population of MDA-MB-231 cells, demonstrating the susceptibility of CSC-likes [28]. The use of the reporter system (SORE-GFP) allowed us to compare the sensitivity of CSC and non-CSC. Thus, our results confirm that prolonged exposure to a low dose of doxorubicin is necessary to induce the drug-resistant phenotype, as other groups have reported [25, 26, 29, 30].

Next, we performed mammosphere assay to characterize the functional phenotype of doxorubicin-resistant cells. The mammosphere-forming efficiency (MFE) of Dox cells was twice that of control cells. The length of the spheres that were formed by Dox cells was lower than that of control cells (Fig. 2B). Our data show that enrichment of the GFP + population (CSC-like) acquires resistance and MFE due to continuous exposure to low concentrations of doxorubicin.

Doxorubicin induces stemness and plasticity phenotype in GFP- populations

To test whether GFP + cells (CSCs-like) that expand while acquiring doxorubicin resistance have a different stemness phenotype than GFP- cells (non-CSCs), we purified GFP + and GFP- cells (Fig. 3A) from Dox-resistant cultures and performed mammosphere assay (Fig. 3B). GFP + and GFP- cells from Dox-resistant cultures had a similar but higher MFE than purified cells from the control culture. However, the length of mammospheres was similar between the Dox-resistant and control cultures. We and others have reported that continuous stimulation with doxorubicin at a low dose increases the MFE in TNBC cells [25, 26]; however, in the current study, we found that both GFP + and GFP- populations from a culture that has been exposed to doxorubicin have a similar stemness phenotype, resulting in an increased MFE.

It is widely accepted that CSCs can self-renew and give rise to more committed daughter cells, whereas regeneration of CSCs from more differentiated daughter cells is an infrequent phenomenon [31]. To analyze this principle in our enriched, resistant cells with a stemness phenotype, we culture purified cells (GFP + and GFP-) and determined the percentage of GFP + cells (Fig. 4). The GFP + populations regenerated GFP- cells, which increased the percentage slightly with passage in culture. In contrast, GFP- populations from the culture control were largely incapable of regenerating GFP + cells; notably, GFP- populations from the resistant culture rapidly regenerated GFP + cells, which increased the percentage with passage in culture (Fig. 4A). In Passage 3, the purification of GFP + populations resulted in a reduction in the GFP + percentage (89.5% and 84.1% in control and Dox, respectively). This decrease indicates the emergence of a GFP- populations. Conversely, the purification of GFP- populations from the resistant culture effected an increase in the GFP + percentage (23.2%), versus the 2.07% of GFP + cells in the GFP- populations that were purified from the control (Fig. 4B).

The expression of SOX2 and OCT4 was measured by RT-qPCR in GFP + and GFP- populations (Fig. 4C). In cells exposed to Dox, the relative expression of OCT4 is significantly higher in GFP- cells than in GFP + cells. For SOX2, no significant Dox-induced changes were identified. As expected, in control cells, the expression of both SOX2 and OCT4 is higher in the GFP + population than in the GFP − population. This result provides evidence for a positive regulatory role of OCT4 in Dox-induced resurgence in GFP + cells.

To determine whether the increase in GFP + percentage in GFP- populations that were purified from the resistant culture resulted from enhanced proliferative capacity due to doxorubicin, we analyzed cell proliferation. After sorting, purified cells were stained with CellTrace (t0), cultured for 72 h, and analyzed by flow cytometry (Fig. 5A). Purified and unpurified cells from the resistant culture showed lower proliferative capacity compared with control cells (Fig. 5 B). As has been reported, GFP- cells (non-CSCs) can proliferate faster than GFP + cells (CSCs) [32, 33].

mRNA expression profiles of subpopulations purified from resistant culture

Phenotypic plasticity represents a paradigm for understanding cancer initiation, progression, and drug resistance that involves a broad set of phenomena, including the loss of lineage commitment and the acquisition of a stemness phenotype (dedifferentiation). Drug resistance is achieved by promoting the acquisition of a dedifferentiated state by increasing the expression of certain genes [34].

RNA sequencing analysis (RNA-seq) was performed to analyze alterations in genes in the GFP- and GFP + populations from resistant and control cultures (Fig. 6). Principal component analysis (PCA) and subsequent visualization of PC1 and PC2 resulted in the clustering of samples by condition (Fig. 6A). A total of ~ 1000 genes were detected as differentially expressed and heuristically clustered into 4 groups, based on their expression profiles (Fig. 6B). The results from DESeq2, visualized in volcano plots, revealed 150–450 differentially expressed genes, based on predefined thresholds (adjusted p-value < 0.05 and |log2-fold change|), according to the combinations between pairs of doxorubicin and control from GFP + and GFP- populations (Fig. 6C). In the gene ontology analysis of upregulated genes (from control and Dox-resistant GFP- cells), several biological processes were identified, related to cellular response to chemical stimulus, mesenchymal cell differentiation, stem cell differentiation, stem cell population maintenance, and mechanism of DNA repair. Further, positive regulation of wnt signaling pathways and transcription of notch receptor targets was observed. Several identified processes were involved in self-renewal and differentiation of CSCs in several tissues and breast cancer [35]. The main downregulated genes were related to regulation of apoptosis and cellular metabolism (Fig. 6D). The molecular functions (Fig. 6E) were related to ABC-type transporter activity (upregulated genes) and protein-DNA binding and enzymatic activity (downregulated genes). The gene ontology for several distinct clusters of genes is provided in Supplementary Information (Supplementary Fig. 3). We identified 7 genes (ITGB1, SNAI1, NOTCH4, STAT5B, RAPGEF3, LAMA2, and GNAI1) that have been implicated in both stemness and the drug response and validated their relevance through an analysis of data from breast cancer patients in the TCGA database using UCSC Xena (Fig. 6F). Based on our RNA-seq analysis, treatment at low concentrations of doxorubicin induced differential gene expression in GFP- populations, explaining in part the presence of cells with a plastic phenotype, leading to enrichment of resistant cells with a stemness phenotype.

Conflict of interest

The authors declare no competing interests.