Impact of Combination Drugs on the Transcriptome Profiles of Triple-Negative Breast Cancer Cells

1. Introduction

Triple-negative breast cancer (TNBC) is one of the most aggressive cancer types among women, representing over 10-20% of breast cancers globally [1]. In 2022, approximately 2.3 million new breast cancer cases were reported globally; TNBC accounts for a disproportionate mortality due to its aggressive nature and the limited availability of advanced treatments in low- and middle-income countries [2]. Conventional chemotherapy continues to be the main treatment for most patients; however, approximately 50% of tumors develop resistance after an initial response to the drugs [3]. TNBC has a notably higher incidence among African American (AA) women in the United States, with an estimated rate of 25.2 per 100,000, compared to 12.9 per 100,000 among non-Hispanic white (NHW) women [4].

Recent studies have shown that new treatments, such as Poly(ADP-ribose) polymerase inhibitors (PARPi) and immune checkpoint inhibitors (ICI), improve outcomes for TNBC patients [5,6]. Specifically, olaparib (Ola) has demonstrated improved progression-free survival in germline BRCA-mutant metastatic breast cancer [7], while atezolizumab (Atz) has produced durable clinical responses in PD-L1-positive TNBC [8]. Mechanistic evidence indicates that the synergetic effects of PARP inhibition will be further enhanced by the ICIs [9].

In this study, we investigated the transcriptional effects of Ola (a PARPi) and Atz (an anti-PD-L1 antibody), alone and in combination, among two TNBC cell lines with different ancestries: MDA-MB-157 (AA descent) and MDA-MB-231 (Caucasian descent). These differences in transcriptomic signatures between AA and Caucasian descent cells at the molecular level will allow us to understand the mechanistic rationale for using PARPis and ICIs in TNBC therapy.

2. Materials and Methods

2.1. Cell Culture

Human breast cancer cell lines MDA-MB-157 and MDA-MB-231 were purchased from the American Type Culture Collection. The human breast cancer cell lines were cultured with RPMI-1640 liquid media, and 10% inactivated fetal bovine serum (FBS) (Cytiva, USA). Ola and Atz were obtained from Selleck Biochem, USA. Cell lines were cultured on tissue culture plastic using RPMI-1640 medium, supplemented with 10% (v/v) FBS, and kept in a humidified incubator at 37 °C with 5% carbon dioxide [10,11](Eppendorf, USA).

2.2. Treatment and Groups

Cells were seeded at a density of 0.5 million and allowed to attach in triplicate. Once they reached 70-80% confluence, treatment was applied. Cells were then treated with the following drug concentrations: Ola: 5 nM, Atz: 0.4 nM, Ola+Atz: 5.4 nM, and the DMSO-treated group served as the control for 24 hours in all subsequent experiments.

2.3. Cell Viability Assay

Upon achieving 70-80% confluency in each plate, cells were treated with drugs at the concentrations mentioned above for 24 hours. After this period, cells were harvested and counted using a 0.4% trypan blue method with an automated cell counter (Thermo Scientific, USA). Cell counts were compared between the DMSO-treated (control) group and the drug-treated groups, including Ola, Atz, and Ola+Atz. For each group, equal volumes of cell suspension and trypan blue were mixed and loaded onto a cell counting plate, which was then analyzed using the automated cell counter. Data from the cell viability assay are represented as the mean ± standard error (SE) from three independent experiments. Statistical comparisons were performed among the control and treatment groups using one-way ANOVA in GraphPad Prism. Statistical significance is indicated as ***p < 0.001, **p < 0.01 and *p < 0.05.

2.4. RNA Sequencing Experiments

Cells from the two cell lines were seeded at a density of 0.5 million in triplicate for all treatments (DMSO, Ola, Atz, Ola+Atz). They were allowed to attach and grow. When the cells reached 70-80% confluency, the drug treatment was applied. After 24 hours of drug exposure, the cells were harvested into 700 µL of QiaZol and sent to LC Sciences in Houston for sequencing. At LC Sciences, briefly, a Poly(A) RNA sequencing library was prepared using Illumina’s TruSeq-stranded mRNA sample preparation protocol. RNA integrity was assessed using an Agilent Technologies 2100 Bioanalyzer High Sensitivity DNA Chip. Poly(A)-containing mRNA was purified using oligo (dT) magnetic beads in two rounds. The purified poly(A) RNA was fragmented at elevated temperature in the presence of divalent cations. Quality control and quantification of the sequencing library were performed with an Agilent Technologies 2100 Bioanalyzer High Sensitivity DNA Chip. Paired-end sequencing was conducted on Illumina’s NovaSeq 6000 sequencing system.

2.5. Bioinformatics Analysis

Firstly, we used Cutadapt and Perl scripts from LC Sciences in-house to remove reads containing adaptor contamination, low-quality bases, and undetermined bases. Then, sequence quality was verified using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). We employed HISAT2 [12] to map reads to the genome available at ftp://ftp.ensembl.org/pub/release-107/fasta/homo_sapiens/dna/. The mapped reads for each sample were assembled using StringTie [3]. Subsequently, all transcriptomes were merged to reconstruct a comprehensive transcriptome using Perl scripts and gffcompare. After generating the final transcriptome, StringTie [13] and Ballgown (http://www.bioconductor.org/packages/release/bioc/html/ballgown.html) were used to estimate the expression levels of all transcripts. mRNAs differential expression analysis was performed using the R package DESeq2 [14] between two different groups, and using the R package edgeR [15] between two samples. The mRNAs with a false discovery rate (FDR) below 0.05 and an absolute fold change of at least 2 were considered differentially expressed mRNAs. We used MATS (http://rnaseq-mats.sourceforge.net/) computational tool for detecting the differential alternative splicing events from RNA-Seq data. rmats2sashimiplot (https://github.com/Xinglab/rmats2sashimiplot) was used to visualize rMATS events. The top 20 rMATS events with the smallest FDR were selected for visual display. Volcano plots of DEGs were generated using the ggplot2 package in R to visualize upregulated and downregulated genes, highlighting those with significant expression changes.

2.6. Gene Ontology and KEGG Enrichment Analysis

In order to investigate the influence of various chemotherapy drug treatments and different combinations on the key process, which DEGs in treated and untreated cells were involved, the molecular cues they exerted, and the essential pivotal signaling pathways they were involved in, we performed Kyoto Encyclopedia Gene and Genome (KEGG) and Gene Ontology (GO) term enrichment classification, and enrichment analysis. Gene Ontology was performed using Gene Ontology web server (https://www.geneontology.org/), and KEGG analysis was conducted using KEGG (Kyoto Encyclopedia of Genes and Genomes) database (https://www.kegg.jp/).

3. Results

3.1. Ola and Atz as Mono and Combination Therapy in AA and Caucasian Cells

To identify transcriptomic differences at the PARPi and ICI levels in AA and Caucasian individuals, we used two TNBC cell lines: MD-157 (AA) and MD-231 (Caucasian). The cell lines were treated with Ola and Atz. We have categorized our treatments into the following six groups based on the treatment with drugs as a single agent or combination therapy: i) control (untreated) versus Ola, ii) control versus Atz, iii) control versus the combination of Ola and Atz, iv) Atz versus Ola, v) Ola versus the combination of Ola and Atz, and vi) Atz versus the combination of Ola and Atz for each of the cell lines.

3.2. Ola and Atz Combination Effects on the Cell Viability of AA and Caucasian Cells

The cell viability analysis indicates the effects of Ola, Atz and their combination on the various cell lines Figure 1A,B. In both the AA and white cell lines, exposure to 5nM of Ola resulted in a statistically significant reduction in cell viability at 24 h after treatment (all p < 0.0001; Figure 1A,B). A similar pattern was observed for Atz in both the AA and white cell lines; treatment with 0.4 nM Atz led to a statistically significant decrease in cell viability 24 h post-treatment (all p < 0.05; Figure 1A,B). In the case of exposure to 5.4 nM of a combination of Ola and Atz, resulted in a significantly higher reduction in cell viability was observed among the white cell lines (all p < 0.0001; Figure 1B) compared to the AA cell lines (all p < 0.001; Figure 1A) at 24 h after treatment. These findings indicate that the combination of Ola and Atz exerts a significant inhibitory effect on the growth of MD-MB-157 and MD-MB-231 cells.

3.3. Analysis of the Alternative Splicing Events Among AA and Caucasian Cells

Alternative splicing (AS) is a key process that generates transcriptomic diversity and plays a crucial role in cellular differentiation, tissue-specificity, and responses to environmental stimuli [16]. We examined AS events, such as 3’ splice site (A3SS), alternative 5’ splice site (A5’SS), mutually exclusive exon (MXE), retained intron (RI), and skipped exon (SE), across all our groups.

We identified a set of differentially regulated alternatively spliced regions in all the groups (Figure S1–S12). Our results showed that Ola treatment leads to significant alterations in alternative splicing compared with control, with exon skipping (SE) as the predominant event type in both AA (Figure S7) and Caucasian descent (Figure S1) cell lines, indicating that PARP inhibition through Ola significantly impacts RNA processing, likely through DNA damage-associated transcriptional stress and disruption of splicing factor function. A similar pattern was observed in cells treated with Atz compared with controls in both AA (Figure S8) and Caucasian descent (Figure S2) cell lines, indicating that immune checkpoint blockade primarily affects co-transcriptional splicing dynamics, likely through IFN-γ–driven transcriptional reprogramming and inflammatory signaling cascades. Overall, the AS events were significantly higher in the Ola+Atz versus Atz group among the AA cell lines (Figure S12), indicating a synergistic disruption of RNA processing, likely driven by the combined effects of Ola–induced genomic stress and Atz–mediated transcriptional reprogramming, resulting in enhanced transcriptomic plasticity in AA TNBC cell lines.

3.4. Identification of Differentially Expressed Genes (DEG) Among the AA and Caucasian Cells

Among the differentially expressed genes, we have selected the top 99 genes for our study. DEG analysis showed 11 genes downregulated and 88 upregulated in the Ola versus control group among AA descendant cell lines (Table S1), whereas it identified 29 downregulated and 70 upregulated genes among Caucasian cell lines (Table S2). DEG analysis comparing Atz to the control group revealed 81 genes downregulated and 18 upregulated in AA descendant cell lines (Table S3). In contrast, among Caucasian cell lines, 36 genes were downregulated and 63 upregulated (Table S4). For the other groups, the following number of DEGs were identified: i) In the Atz versus Ola comparison, 67 genes were downregulated and 34 upregulated in the AA cell lines (Table S5), while 61 genes were downregulated and 38 upregulated in the Caucasian cell lines (Table S6); ii) In the Ola+Atz versus control comparison, 82 genes were downregulated and 17 upregulated in the AA cell lines (Table S7), while 88 genes were downregulated and 12 upregulated in the Caucasian cell lines (Table S8); iii) In the Ola+Atz versus Ola comparison, 5 genes were downregulated and 94 upregulated in the AA cell lines (Table S9), while 92 genes were downregulated and 7 upregulated in the Caucasian cell lines (Table S10); iv) In the Ola+Atz versus Atz comparison, 80 genes were downregulated and 19 upregulated in the AA cell lines (Table S11), while 94 genes were downregulated and 5 upregulated in the Caucasian cell lines (Table S12). Figure 2 and Figure 3 illustrate the heatmap of selected DEGs in AA and Caucasian descent cells. Results from the heatmap clustering analysis of Ola versus control showed that the genes RPSAP70, MIR4280HG, HRH1, HDAC9, USP32P2, UPK3BL1, ENSG00000289697, NXF2, ENSG00000289865, ENSG00000289100, and PTH1R among the AA cells (Figure 2A), while the genes CORO7–PAM16, ENSG00000286235, ENSG00000256349, XIST, ENSG00000261732, ENSG00000237870, ENSG00000287484, LCT, HSPE1–MOB4, ENSG00000288053, ENSG00000285565, ENSG00000255931, LINC01602, PSG11, LINC01173, DYNRLB2–AS1, TNFSF12–TNFSF13, LINC01638, ENSG00000287824, ENSG00000284776, ENSG00000241280, MATN3, SNAI2, SRGN, CHML, ENSG00000287396, ZNF860, ENSG00000255639, and HAS2 among the Caucasian cell lines, were top clustered and significantly downregulated in Ola compared to the control samples (Figure 3A). In the case of Atz versus control, results from the heatmap clustering analysis showed that the genes VASN, TRIB3, NCF2, SERPINE1, MIR23AHG, NUPR1, CHAC1, SLC6A9, SLC38A2, PSAT1, SLC38A1, ASNS, GDF15, AMIGO2, ICAM1, CLDN1, TNFSF15, and WARS1 among the AA cells (Figure 2B), while the genes MALAT1, ENSG00000267288, SMG1P7, MIR503HG, and LINC01089 among the Caucasian cell lines, were top clustered and significantly upregulated in Atz compared to the control samples (Figure 3B). In Atz versus Ola group, results from the heatmap clustering analysis showed that the genes ENSG00000268400, ENSG00000280241, ANKRD55, NXF2, ALOX5AP, XAF1, KIAA0408, ENSG00000251320, USP32P2, SGIP1, HEY1, CYP1A2, ENSG00000230479, KLHDC1, ABI3BP, CCDC146, PICART1, ENSG00000271992, ECM2, GDPD1, RBM20, ENSG00000261770, GYPC, ALPP, SLC6A9, IL1RL1, WFDC21P, CCDC144NL-AS1, TNXB, SEL1L3, ASNS, SUN3, ENSG00000276850, NEDD9, MIR503, HDAC9, CTH, COL4A4, PSAT1, VWA5A, MKX, PTPRC, LUM, BIRC3, HCN4, SPINK1, IGDCC4, CLDN1, AMIGO2, LINC02154, PAPPA2, SELP, CHAC1, TNFSF15, KCTD4, BACH1, WNK4, TMEM147-AS1, WARS1, WHRN, CXCL8, ST6GAL1, LINC00520, PDE2A, COL4A3, VLDLR-AS1, and LINC02725 among the AA cells (Figure 2C), while the genes TNFSF12–TNFSF13, LBHD1, ENSG00000272516, ENSG00000288591, ENSG00000273306, ENSG00000285043, LINC03002, RAB6C, UPK3BL2, ENSG00000230551, COL1A2, IGSF9B, ENSG00000287824, LIX1–AS1, ENSG00000260400, HAS2, FAT3, TNC, CHML, LINC01173, ENSG00000287175, MEI4, PEDS1–UBE2V1, ENSG00000269693, ENSG00000262160, ENSG00000287484, RASA1, SLC6A2, CORO7–PAM16, VSIG10L2, ENSG00000288725, ENSG00000285238, ENSG00000286235, ENSG00000286112, LINC01602, ENSG00000243635, LGALS4, and SUMO2P17 among the Caucasian cell lines were top clustered and significantly upregulated in Ola compared to the Atz samples (Figure 3C). In the Ola+Atz group versus the control group, results from the heatmap clustering analysis showed that the genes ELF3, ST14, EPN3, CDH1, RAB25, S100A14, AKR1C2, ALDH3B2, CDC42BPG, PDZK1IP1, TMPRSS13, KRT17, LINC01133, TJP3, FA2H, TPD52L1, and TSPAN13 are among the AA cells (Figure 2D), while the genes MMP3, ENSG00000276851, RHBDL1, BIAP3, OASL, ENSG00000275216, TMEM150A, H2AC6, GMPR, H2BC5, and KBTBD11–OT1 are among the Caucasian cell lines and were top clustered and significantly upregulated in Ola+Atz compared to the control samples (Figure 3D). In the Ola+Atz group versus the Ola group, results from the heatmap clustering analysis showed that the genes SLIT2, FZD8, FAM167A, ENSG00000261915, and ADAMTS1 are among the AA cells (Figure 2E), while the genes SPC25, KIF2C, KIF20A, HJURP, NCAPG, ZWINT, PTN, FAM83D, NUF2, SKA1, PRC1, RRM2, MKI67, NDC80, NEURL1B, CDCA8, SPAG5, TPX2, BRIP1, FBXO43, SHCBP1, ESPL1, NCAPD2, RACGAP1, FAM72C, BUB1B, CIP2A, TTK, IQGAP3, MELK, DEPDC1, FAM72B, MAD2L1, GTSE1, NCAPH, SGO1, FBXO5, EDNRB, CEP55, CCNA2, CDK1, NUSAP1, FAM72A, ARHGAP19, DLGAP5, KIF23, KIF11, FAM72D, PBK, KIF4A, CDC25C, FAM111A, CENPF, KIF18B, STBD1, CDCA2, HMMR, E2F8, RTKN2, TMEFF1, ERCC6L, DEPDC1B, GAS2L3, TOP2A, KNL1, KIF20B, PRR11, SMC4, DIAPH3, BUB1, KNTC1, KIF15, DHFR, ARHGAP11A, CSPG4, INCENP, CKAP2L, KIF14, NCAPG2, RAD51AP1, CENPK, PLK4, KIF24, ASPM, NEIL3, FANCD2, NEK2, SGO2, POGLUT3, CIT, CENPE, ANLN, and RFC3 are among the Caucasian cell lines and were top clustered and significantly downregulated in Ola+Atz compared to the Ola samples (Figure 3E). In the Ola+Atz group versus the Atz group, results from the heatmap clustering analysis showed that the genes KISS1, L1CAM, LOXL4, VEGFC, CLIC6, TIMP2, MMP14, PLAT, HOXB9, NCF2, TIMP3, AXL, HSD17B11, FOXA2, ANKRD1, PSG5, ROBO4, FADS2, TMEM255B, ARMCX1, TGM2, PTX3, ENSG00000275216, ANGPTL2, CCN3, CALB2, ZDHHC2, FMNL1, SLC14A1, MXRA8, GAS6, EMP3, FST, LPAR1, SLAMF7, IGFBP7, EDIL3, AMIGO2, FAM107B, TCEAL9, HMGA2, COL5A1, FADS1, APPB1IP, FSTL3, NEXN, IFI16, SLC37A2, C4BPB, MXRA7, RHOJ, SMURF2, CTNNAL1, AJUBA, BCAT1, HHIP, CDH11, FRMD6, IL6ST, ADAM8, FSTL1, PID1, ADAMTS1, PLAC8, COL8A1, BEX3, NUAK1, MIR100HG, NMNAT2, COLGALT2, CXCL1, CCND2, FAM171A1, SCN5A, SLC2A3, BDNF, RHOF, FBN1, RAB32, NOG, and TOR4A are downregulated among the AA cells (Figure 2F), while the genes ENSG00000258377, IL6, H2BC5, TMEM150A, H2AC6 and CYP1A1 are among the Caucasian cell lines and were top clustered and significantly upregulated in Ola+Atz compared to the Atz samples (Figure 3F).

3.5. Gene Ontology Enrichment Analysis of Differentially Expressed Genes Among AA and Caucasian Cells

Results from the GO enrichment analysis of Ola versus control in AA cells showed that Ola-regulated DEGs are strongly enriched for transcription-related pathways, including regulation of transcription (DNA-templated and RNA polymerase II–mediated), signal transduction, cell proliferation, and protein phosphorylation in the biological process (BP) category, nucleus, cytoplasm, membrane, and cytosol, nucleoplasm, and chromatin in the cellular component (CC) category and protein binding, DNA binding, metal ion binding, and RNA polymerase II–related transcription factor activity in the molecular function (MF) category (Figure 4A, Supplementary Figure S13A). Similarly, GO analysis of Ola versus control in Caucasian cells showed enrichment for signal transduction, inflammatory and immune responses, DNA repair, cell proliferation, IL-12 production, and regulation of immune tolerance in the BP category, membrane components, nucleus, cytoplasm, nucleoplasm, mitochondria, ribosomes, chromatin, and extracellular regions in the CC category and protein binding, ATP binding, nucleotide binding, DNA binding, metal ion binding, hydrolase activity, transcription factor binding, and oxidoreductase activity in the MF category (Figure 5A, Supplementary Figure S14A). GO enrichment analysis of Atz versus control in AA cells showed that Atz-regulated DEGs are strongly enriched for transcriptional regulation, signal transduction, cell cycle regulation, apoptosis, and immune-related processes, including regulation of immune system development and cellular responses to external stimuli in the BP category, strong enrichment for genes localized to the membrane, cytoplasm, and nucleus, along with representation in cytosol, plasma membrane, nucleoplasm, mitochondria, ribosomes, chromatin, and extracellular regions in the CC category and enrichment of protein binding, ATP binding, metal ion binding, DNA binding, nucleotide binding, and transcription factor–related activities in the MF category (Figure 4B, Supplementary Figure S13B). Similarly, GO analysis of Atz versus control in Caucasian cells showed enrichment for chromatin organization, histone lysine methylation, regulation of transcription, kinase and phosphatase activity–related processes, and cellular responses to damage or external stimuli in the BP category, nucleus, cytoplasm, cytoskeleton, plasma membrane, and intracellular organelle compartments in the CC category and protein binding, transferase and kinase activity, ATP binding, and regulatory binding interactions in the MF category (Figure 5B, Supplementary Figure S14B). Results from the GO enrichment analysis of Atz versus Ola in AA cells showed that Atz-vs-Ola DEGs are strongly enriched for signal transduction, transcriptional regulation (including RNA polymerase II–mediated transcription), cell adhesion, cell differentiation, and immune system–related processes in the BP category, strong enrichment for genes localized to membrane, cytoplasm, and nucleus, with additional representation in plasma membrane, cytosol, mitochondria, extracellular regions, and cytoskeletal structures in the CC category and enrichment of protein binding, ATP binding, metal ion binding, DNA binding, and transcription factor–related activities in the MF category (Figure 4C, Supplementary Figure S13C). Meanwhile for GO analysis of the Atz versus control group in the Caucasian cells showed enrichment for signal transduction, chromatin remodeling, cell–cell adhesion, immune response–related processes, and regulation of transcription in the BP category, membrane, nucleus, cytoplasm, and extracellular or plasma membrane–associated regions in the CC category and protein binding, nucleotide and ATP binding, metal ion binding and DNA binding in the MF category (Figure 5C, Supplementary Figure S14C). Results from the GO enrichment analysis of Ola+Atz versus control in AA cells showed that Ola+Atz-regulated DEGs are strongly enriched for transcriptional regulation (including RNA polymerase II–mediated transcription), signal transduction, cell proliferation and differentiation, apoptosis, immune system processes, and G-protein–coupled receptor signaling in the BP category, membrane, cytoplasm, and nucleus, plasma membrane, cytosol, nucleoplasm, mitochondria, extracellular regions, Golgi apparatus, and cytoskeletal compartments in the CC category and protein binding, metal ion binding, DNA binding, ATP binding, transferase activity, hydrolase, and RNA polymerase II cis-regulatory region sequence-specific DNA binding in the MF category (Figure 4D, Supplementary Figure S13D). Similarly, GO analysis of Ola+Atz versus control in Caucasian cells showed enrichment for cell cycle regulation, cell division, signal transduction, transcriptional regulation (including RNA polymerase II–mediated transcription), and cellular responses to DNA damage in the BP category, cytoplasm, nucleus, membrane, cytosol, plasma membrane, nucleoplasm, extracellular regions, endoplasmic reticulum, Golgi apparatus, and centrosome in the CC category, protein binding, metal ion binding, nucleotide and ATP binding, DNA binding, transferase, hydrolase, and transcription-related binding functions in the MF category (Figure 5D, Supplementary Figure S14D). Results from the GO enrichment analysis of Ola+Atz versus Ola in AA cells showed that DEGs are strongly enriched for transcriptional regulation, signal transduction, immune and inflammatory responses, cell differentiation, and apoptosis-related processes in the BP category, membrane, cytoplasm, and nucleus, plasma membrane, cytosol, extracellular regions, endoplasmic reticulum, Golgi apparatus, and cytoskeletal compartments in the CC category, and protein binding, ATP binding, metal ion binding, DNA binding, transferase, hydrolase, and kinase functions in the MF category (Figure 4E, Supplementary Figure S13E). Similarly, GO analysis of Ola+Atz versus Ola in Caucasian cells showed enrichment for cell cycle regulation, cell division, cellular responses to DNA damage, signal transduction, transcriptional regulation (including RNA polymerase II–mediated transcription), and immune-related processes in the BP category, nucleus, cytoplasm, membrane, and cytosol, plasma membrane, chromosomal and spindle-associated structures, centrosomes, extracellular regions, and endomembrane compartments in the CC category, protein binding, nucleotide and ATP binding, metal ion binding, DNA binding, helicase, transferase, hydrolase, and ATPase activity in the MF category (Figure 5E, Supplementary Figure S14E). Results from the GO enrichment analysis of Ola+Atz versus Atz in AA cells showed that DEGs are strongly enriched for cell cycle regulation, cell division, DNA damage response, transcriptional regulation (including RNA polymerase II–mediated transcription), and signal transduction in the BP category, nucleus, cytoplasm, membrane, cytosol in the CC category, and protein binding, ATP and nucleotide binding, DNA binding, metal ion binding in the MF category (Figure 4F, Supplementary Figure S13F). Similarly, GO analysis of Ola+Atz versus Atz in Caucasian cells showed enrichment for cell cycle progression, cell division, DNA damage response, signal transduction, and transcriptional regulation in the BP category, cytoplasm, nucleus, membrane, and cytosol in the CC category, protein binding, DNA binding, ATP and nucleotide binding, and metal ion binding in the MF category (Figure 5F, Supplementary Figure S14F).

3.6. Pathway Enrichment Analysis of Differentially Expressed Genes Among AA and Caucasian Cells

Results from the KEGG pathway enrichment analysis of Ola versus the control group showed that the pathways, transcriptional misregulation in cancer, viral carcinogenesis, systemic lupus erythematosus, MAPK signaling, TGF-β signaling, phospholipase D signaling, ferroptosis, Th1 and Th2 cell differentiation, were significantly enriched in AA cells (Figure 6A), whereas Caucasian cells showed enrichment of tryptophan metabolism, biosynthesis of cofactors, metabolic pathways (Figure 7A). Results from the KEGG pathway enrichment analysis of Atz versus the control group showed that the pathways, DNA replication, cell cycle, pathways in cancer, rheumatoid arthritis, cytokine–cytokine receptor interaction, Herpes simplex virus 1 infection, Human T-cell leukemia virus 1 infection, HIF-1 signaling, FoxO signaling, p53 signaling, cellular senescence, ECM–receptor interaction, and biosynthesis of amino acids, were significantly enriched in AA cells (Figure 6B), whereas Caucasian cells showed enrichment of lipid metabolism and metabolic pathways (Figure 7B). Results from the KEGG pathway enrichment analysis of Atz versus the Ola group showed that the pathways in cancer, systemic lupus erythematosus, alcoholism, viral carcinogenesis pathways, HIF-1 signaling, Hippo signaling, PI3K–Akt signaling,p53 signaling, JAK–STAT signaling, cell adhesion molecules, tight junction, breast cancer, bladder cancer, gastric cancer, small cell lung cancer, glycine, serine and threonine metabolism and central carbon metabolism in cancer, were significantly enriched in AA cells (Figure 6C), whereas Caucasian cells showed enrichment of ECM–receptor interaction, focal adhesion, proteoglycans in cancer, platelet activation, PI3K–Akt signaling, AGE–RAGE signaling, and relaxin signaling (Figure 7C). Results from the KEGG pathway enrichment analysis of Ola+Atz versus the control group showed that the pathways in cancer, PI3K–Akt signaling, Rap1 signaling, cell adhesion molecules, ECM–receptor interaction, focal adhesion, proteoglycans in cancer, MAPK signaling, Ras signaling, FoxO signaling and AGE–RAGE signaling were significantly enriched in AA cells (Figure 6D), whereas Caucasian cells showed enrichment of cell cycle, homologous recombination, Fanconi anemia pathway, DNA replication, oocyte meiosis, progesterone-mediated oocyte maturation, purine metabolism, pyrimidine metabolism and tryptophan metabolism (Figure 7D). Results from the KEGG pathway enrichment analysis of Ola+Atz versus the Ola group showed that the pathways in complement and coagulation cascades, pathways in cancer, regulation of the actin cytoskeleton, cell adhesion molecules, signaling pathways regulating pluripotency of stem cells, breast cancer, gastric cancer, non-small cell lung cancer, hepatocellular carcinoma, basal cell carcinoma, glycine, serine and threonine metabolism, starch and sucrose metabolism, biosynthesis of unsaturated fatty acids were significantly enriched in AA cells (Figure 6E), whereas Caucasian cells showed enrichment of cell cycle, homologous recombination, Fanconi anemia pathway, DNA replication, p53 signaling, oocyte meiosis, progesterone-mediated oocyte maturation, purine metabolism, pyrimidine metabolism, one-carbon pool by folate, cellular senescence, mismatch repair and alcoholism (Figure 7E).

4. Discussion

TNBC is linked to an aggressive tumor biology, early metastasis, and a poorer prognosis compared to other subtypes of breast cancer [17]. Despite several advances in TNBC therapy, patients still have a significantly shorter survival time and often develop distant metastases, emphasizing the urgent need for developing new therapeutic strategies [18,19,20]. Recently, immune checkpoint inhibitors (ICIs), especially PD-1/PD-L1 inhibitors such as Pembrolizumab and Atz, have shown significant clinical benefit in PD-L1-positive patients [21,22,23] and PARPi such as Ola have been approved for the treatment of TNBC patients with BRCA1/2 mutations [23]. Further, TNBC is known to disproportionately affect population groups, with individuals of African ancestry being more likely to be diagnosed compared to European American women [24,25,26].

Recent studies showed that combination therapies, such as integrating with chemotherapy and antibody-drug conjugates, are able to overcome resistance towards ICI [27] and PARPi [28]. In the present study, our data support the idea that combination therapy enhances efficacy by targeting multiple pathways or mechanisms, whereas monotherapy leads to incomplete pathway suppression and adaptive resistance. To predict the impact of the combination therapy of PARPi Ola and ICI Atz at the transcriptomic level in cells derived from two disproportionate TNBC burden population groups [29,30,31], such as AA (MDA-MB-157) and Caucasian (MDA-MB-231), we first performed cell viability assays (Figure 1A,B). We found significant differential effects of monotherapy and combination treatment on cellular survival in both AA (Figure 1A) and Caucasian cells (Figure 1B), with the lowest cell viability being observed in the combination (Ola + Atz) group compared to the control, indicating a strong cytotoxic or growth-suppressive effect.

Differential gene expression analysis, typically performed with RNA-seq, is commonly used to identify genes that differ between tumor and normal tissues, as well as between drug-treated and untreated, and sensitive and resistant, phenotypes in TNBC patients [32,33,34,35,36]. To this end, we performed DEG analysis for the six groups: Ola versus Control, Atz versus Control, Atz versus Ola, Ola+Atz versus Control, Ola+Atz versus Ola, and Ola+Atz versus Atz in both AA and Caucasian descent cells. We initially conducted heatmap-based profiling to identify distinct, treatment-specific gene expression signatures, emphasizing both drug-specific and race-related molecular responses (Figure 2 and Figure 3). In AA-derived cells (Figure 2A,B), both Ola and Atz showed significant DEGs related to control conditions. Ola treatment showed a significant change in gene expression patterns involving the general transcription (HDAC9), Gα signaling (HRH1, PTH1R), GPCR downstream signaling (HRH1), and RNA metabolism (NXF2) pathways, consistent with PARP inhibition (Figure 2A). In contrast, Atz-treated cells showed a transcriptional profile enriched for immune-related gene expression changes, including NCF2, SERPINE1, ICAM1, TNFSF15, and WARS1 (Figure 2B). On the other hand, the Ola+Atz group exhibited widespread upregulation and downregulation across multiple gene clusters, indicating synergistic transcriptional reprogramming (Figure 2D). Specifically, the Ola+Atz versus Atz group (Figure 2F) demonstrated a significant change in gene expression compared to Ola+Atz versus Ola group (Figure 2E), indicating that the addition of Ola induces significant transcriptional changes beyond those mediated by Atz alone. These findings are consistent with the hypothesis that PARP inhibition can enhance tumor immunogenicity, potentially by increasing DNA damage, genomic instability, and downstream immune activation signals, thereby potentiating the effects of immune checkpoint blockade. In Caucasian-derived cells (Figure 3A,B), both Ola and Atz exhibited notable DEGs compared to control conditions. Similar to AA-derived cells, Ola treatment showed significant expression of genes XIST, LCT, PSG11, MATN3, SNAI2, SRGN, CHML, and HAS2, which are involved in PARP inhibition (Figure 3A), whereas Atz-treated cells showed a transcriptional profile enriched for immune-related gene expression changes, such as MALAT1 and SMG1P7, compared to the control conditions (Figure 3B). Overall, the different transcriptional responses observed in AA (Figure 2) and Caucasian-derived cells (Figure 3) emphasize the need to include diverse biological systems in drug testing. These results indicate that molecular features linked to race may impact responses to targeted treatments and immunotherapy, which is directly relevant to TNBC health disparities.

Gene Ontology enrichment analysis plays a crucial role in drug treatment research, enabling systematic interpretation of transcriptomic changes into biologically meaningful pathways and helping us uncover drug mechanisms, pinpoint therapeutic targets, and understand differences in treatment responses [37,38]. Our GO analysis of AA-descent cells showed that GO terms were predominantly associated with processes related to DNA damage response, DNA repair, and cell cycle regulation in the Ola versus control group (Figure 4A), consistent with the known mechanism of PARP inhibition. In contrast, the Atz group, compared to the control group (Figure 4B), showed enrichment of GO terms related to immune response, cytokine signaling, and antigen presentation pathways, which aligns with previous studies where Atz, as an immune checkpoint inhibitor targeting PD-L1, enhances immune-related signaling and inflammatory pathways [39]. In the combination treatment group (Ola+Atz versus control, Figure 4D), there was a concurrent enrichment of GO terms related to DNA repair, apoptosis, and immune pathways, suggesting that combination therapy activates both genomic stress and immune responses. This may enhance therapeutic effectiveness through complementary mechanisms. Similarly, in comparisons of Ola+Atz versus Olaparib (Figure 4E) and Ola+Atz versus Atz (Figure 4F), the combination treatment shows further enrichment of GO terms related to cell death, stress response, and immune modulation, suggesting a synergistic or additive effect beyond the effects of the individual treatments. These findings, showing more enriched GO terms and gene counts in important pathways, support the idea that combination therapy amplifies biological disturbances. A similar pattern was observed in Caucasian-descent cells also where Ola versus the control group was enriched with DNA repair, replication stress, and cell cycle regulation (Figure 5A); Atz versus the control group was enriched with cytokine signaling, antigen presentation, and inflammatory responses (Figure 5B); Atz versus Ola group was enriched with genomic stability, DNA repair pathways and immune signaling pathways (Figure 5C); combination therapy groups Ola+Atz versus control (Figure 5D), Ola+Atz versus Ola (Figure 5E) and Ola+Atz versus Atz (Figure 5F) were enriched with DNA damage response, apoptosis, and immune-related pathways. Overall, the groups within the AA (Figure 4) and Caucasian cells (Figure 5) exhibited clear functional patterns, emphasizing both treatment-specific and common molecular mechanisms.

At the pathway level, KEGG pathway enrichment analysis (Figure 6 and Figure 7) indicates that the transcriptomic response to Ola, Atz, and their combination is significantly influenced by genetic ancestry. AA descent cells (Figure 6) showed a significant enrichment of signaling cascades (PI3K–Akt, MAPK, Ras, Rap1, HIF-1, JAK–STAT, TGF-β, FoxO), immune/inflammatory pathways (Systemic lupus erythematosus, Rheumatoid arthritis, Complement and coagulation cascades), and cytoskeletal/adhesion pathways (Axon guidance, Focal adhesion, ECM–receptor interaction, Cell adhesion molecules). In contrast, Caucasian descent cells (Figure 7) under combination treatment are dominated by DNA replication and repair pathways (Cell cycle, Fanconi anemia pathway, Homologous recombination, DNA replication, p53 signaling, Oocyte meiosis, Pyrimidine/Purine metabolism). These results indicate that the combination of Ola and Atz causes synthetic lethality and replication stress mainly in Caucasian cells, whereas it activates more extensive oncogenic signaling and immune pathway changes in AA cells. Furthermore, KEGG pathway enrichment analysis identified pathways common to both populations, particularly under combination treatment. Some of these common pathways include cancer, Rheumatoid arthritis, Systemic lupus erythematosus, Cell adhesion molecules, ECM–receptor interaction, PI3K–Akt signaling, and p53 signaling, suggesting common oncogenic mechanisms that transcend ancestry. Homologous recombination and DNA replication are observed in both (Figure 6B, Figures 7D–F), but they are significantly more prominent in Caucasian cells. Overall, KEGG pathway enrichment analysis showed that AA-derived cell lines (Figure 6) have a wider and more significant transcriptional and pathway response to both single and combination therapies than Caucasian cells (Figure 7), especially under Ola+Atz treatment.

Limitations of our study: Our study has a few limitations, including the following. Our RNA-seq data might not fully capture the complexity of tumors in vivo or the heterogeneity among patients, especially with respect to treatment response. Our observed transcriptomic differences might not entirely represent population-level biology; they could instead result from cell-line selection rather than genuine ancestry-driven mechanisms. Our study results may not capture the intra-group heterogeneity. Our study uses only one cell line per racial group, so the response might depend on specific mutations rather than ancestry. A key limitation is including a BRCA wild-type cell line to evaluate PARPi efficacy, which relies on synthetic lethality only present in BRCA-mutant cells. Additionally, testing Atz response in a cell-only culture without T cells does not fully reflect the patient’s situation or the drug’s mechanism. While some response may be observed, it lacks the complexity of actual patient biological interactions.

5. Conclusions

In conclusion, these findings show that genetic ancestry plays a significant role in shaping the transcriptomic response of TNBC to drugs, with important implications for understanding racial disparities in treatment outcomes. The fact that AA and Caucasian cells activate distinct molecular programs in response to the same therapies underscores the vital need to include diverse biological systems in preclinical drug research. Although our study has limitations common to cell line models, such as incomplete representation of tumor diversity in vivo and potential cell-line-specific artifacts, the results offer a valuable foundation for future research focused on developing ancestry-informed, personalized treatment strategies for TNBC.