Polyamine Anabolism Promotes Chemotherapy-Induced Breast Cancer Stem Cell Enrichment

Abstract

Breast cancer patients may initially benefit from cytotoxic chemotherapy but experience treatment resistance and relapse. Chemoresistant breast cancer stem cells (BCSCs) play a pivotal role in cancer recurrence and metastasis, however, identification and eradication of BCSC population in patients are challenging. Here, an mRNA-based BCSC signature is developed using machine learning strategy to evaluate cancer stemness in primary breast cancer patient samples. Using the BCSC signature, a critical role of polyamine anabolism in the regulation of chemotherapy-induced BCSC enrichment, is elucidated. Mechanistically, two key polyamine anabolic enzymes, ODC1 and SRM, are directly activated by transcription factor HIF-1 in response to chemotherapy. Genetic inhibition of HIF-1-controlled polyamine anabolism blocks chemotherapy-induced BCSC enrichment in vitro and in xenograft mice. A novel specific HIF-1 inhibitor britannin is identified through a natural compound library screening, and demonstrate that coadministration of britannin efficiently inhibits chemotherapy-induced HIF-1 transcriptional activity, ODC1 and SRM expression, polyamine levels, and BCSC enrichment in vitro and in xenograft and autochthonous mouse models. The findings demonstrate the key role of polyamine anabolism in BCSC regulation and provide a new strategy for breast cancer treatment.

Keywords: HIF‐1 inhibitor; breast cancer stem cell; britannin; chemotherapy; hypoxia‐inducible factor 1; polyamine anabolism.

Figure 1

An mRNA expression‐based BCSC signature demonstrates chemotherapy‐induced BCSC enrichment. A) Development of BCSC signature. B,C) GSEA and Kaplan–Meier analysis of relapse‐free survival was performed based on BCSC P‐Sig and N‐Sig expression in patient samples from TCGA BRCA. D) BCSC P‐Sig and N‐Sig expression in primary tumor tissue from metastatic (Yes) versus non‐metastatic (No) cancer within 5 years in METABRIC was compared. E) Adherent or sphere breast cancer cells were cultured and qPCR assay was performed for the mRNA expression of top 10 BCSC P‐Sig and N‐Sig genes. F‐I) SCID mice were injected with MDA‐MB‐231 cells and treated with vehicle (V) or paclitaxel (P). ALDH (F, G), mammosphere (H), and RNA sequencing (I) assays were performed. J) Breast cancer cell lines were treated with vehicle, paclitaxel (P), gemcitabine (G), or carboplatin (C) for 3 days and qPCR assay was performed. K) BCSC P‐Sig and N‐Sig expression in primary tumor tissue from patients who received neoadjuvant chemotherapy (Yes) and who did not receive chemotherapy (No) in METABRIC was compared. ^* p < 0.05, ^*** p < 0.001.

Figure 2

Chemotherapy induces BCSC enrichment through activation of polyamine anabolism. A,B) Metabolites information from 23 breast cancer patients in TCGA BRCA was compared between BCSC^high and BCSC^low groups (A), and metabolite ontology enrichment analysis was performed (B). C) Scheme of polyamine anabolic pathway. D) Adherent or sphere breast cancer cells were cultured and qPCR assay was performed. E) MDA‐MB‐231 cells were treated with vehicle (V) or 10 nM paclitaxel (P) for 3 days and polyamine metabolites were measured (mean ± SEM; n = 3). F) Breast cancer cells were treated with V or P for 3 days and qPCR assay was performed. G) Polyamine metabolites from primary breast cancer patient samples collected at Qilu Hospital of Shandong University were measured and compared between patients who received neoadjuvant chemotherapy (Yes) and patients who did not (No) (mean ± SEM; n = 17 in each group). H) Immunoblot assay was performed in MDA‐MB‐231 subclones for ODC1/SRM knockdown. I,J) MDA‐MB‐231 subclones were treated with V or P for 3 days, and ALDH assays were performed (mean ± SEM; n = 3). K) SCID mice were injected with indicated numbers of MDA‐MB‐231 NTC or ODC1/SRM knockdown subclone cells and tumor‐initiating cell frequency (TCF) with 95% confidence intervals (CI) was calculated 70 days after tumor injection. L‐N) 2 × 10⁶ MDA‐MB‐231 subclone cells were implanted into SCID mice. When tumor volume reached 200 mm³ (day 0), mice were treated with V or P (10 mg kg⁻¹, days 0, 5, and 10), and tumor volumes were measured every 2–3 days (L). Tumors were harvested on day 13, and the percentage of ALDH⁺ cells (M) and polyamine levels in tumor tissue (N) were measured (mean ± SEM; n = 5). O) 2 × 10⁶ MDA‐MB‐231 subclone cells were implanted into SCID mice. When tumors were palpable, mice were treated with 10 mg kg⁻¹ paclitaxel every 5 days until tumors were no longer palpable. Kaplan–Meier survival curves of tumor‐free (left), tumor‐bearing (center), and recurrence‐free (right) were plotted (n = 8). P,Q) mRNA expression of ODC1 and SRM in primary tumor tissue from patients who received neoadjuvant chemotherapy and who did not receive chemotherapy (P), and from patients who had metastasis within 5 years and who did not have metastasis within 5 years (Q), in METABRIC, was compared. R) Kaplan–Meier analysis of relapse‐free survival was performed based on ODC1 or SRM mRNA levels in patient that received chemotherapy (n = 1372). ^* p < 0.05, ^*** p < 0.001; ns, not significant.

Figure 3

HIF‐1 promotes polyamine anabolism and BCSC enrichment in response to chemotherapy. A‐C) SCID mice injected with MDA‐MB‐231 NTC or HIF‐1α knockdown subclone cells were treated with paclitaxel, and tumors were harvested for ALDH (A; mean ± SEM; n = 3), mammosphere (B; mean ± SEM; n = 3), and RNA sequencing (C) assays. D) Scheme of HIF‐1 expression fluorescent tracking system. E‐G) SCID mice were injected with MDA‐MB‐231 cells transfected with HIF‐1 expression fluorescent tracking system and treated with paclitaxel. Tumor tissues were collected and sorted into DsRed⁺/GFP⁻ or GFP⁺ populations (E), and qPCR (F) and secondary limiting dilution transplantation (G) assay were performed. H,I) MDA‐MB‐231 subclones were treated with V or P for 3 days, and protein expression (H), and intracellular ornithine and polyamine levels (I; mean ± SEM; n = 3) was measured. J,K) MDA‐MB‐231 cells were treated with V or P and chromatin immunoprecipitation (ChIP) was performed with control IgG or antibody (Ab) against HIF‐1α or HIF‐1β (K; mean ± SEM; n = 3) and qPCR primers flanking the candidate HIF‐1 binding sites in the ODC1 or SRM gene (J). L,M) MDA‐MB‐231 subclones were treated with V or P for 3 days in the absence or presence of spermidine, and ALDH assay was performed (mean ± SEM; n = 3). N) SCID mice were injected with indicated numbers of MDA‐MB‐231 subclone cells, and tumor‐initiating cell frequency (TCF) with 95% confidence intervals (CI) was calculated 70 days after tumor injection. O‐Q) 2 × 10⁶ MDA‐MB‐231 subclone cells were implanted in SCID mice. When tumor volume reached 200 mm³ (day 0), mice were treated with V or P (10 mg kg⁻¹, days 0, 5, and 10), and tumor volumes were measured every 2–3 days (O). Tumors were harvested on day 13, and the percentage of ALDH⁺ cells (P) and polyamine levels in tumor tissues (Q) were measured (mean ± SEM; n = 5). ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001; ns, not significant.

Figure 4

Britannin inhibits HIF‐1‐regulated polyamine anabolism and eradicates BCSCs. A) Scheme of HIF‐1 dual luciferase reporter system. B) Chemical structure of Britannin. C) SUM159 cells were transfected with HIF‐1 dual luciferase reporter system, treated as indicated for 3 days, and the FLuc/RLuc ratio was determined (mean ± SEM; n = 3). D‐G) MDA‐MB‐231 cells were treated with paclitaxel, alone or in combination with indicated doses of britannin, for 3 days. mRNA expression of HIF‐1‐target genes (D), levels of ornithine and polyamine (E; mean ± SEM; n = 3), the percentage of ALDH⁺ cells (F; mean ± SEM; n = 3), and mRNA expression of top 10 BCSC P‐Sig and N‐Sig genes (G) was determined. H) MDA‐MB‐231 cells were treated as indicated for 3 days and immunoblot assay was performed. I) Scheme of HIF‐1α TAD reporter system. J) MDA‐MB‐231 cells were transfected with HIF‐1α TAD function reporter system, treated as indicated for 3 days, and the Fluc/RLuc ratio was determined (mean ± SEM; n = 3). K) MDA‐MB‐231 cells were treated as indicated and ChIP assays were performed (mean ± SEM; n = 3). L‐P) 2 × 10⁶ MDA‐MB‐231 cells were implanted in SCID mice. When tumor volume reached 200 mm³ (day 0), mice were treated with vehicle, paclitaxel (P; 10 mg kg⁻¹, days 0, 5, and 10), britannin (B; 5 mg kg⁻¹, days 1–13), or the combination of paclitaxel and britannin (P + B). Tumor volumes (L) and mouse body weight (M) were measured every 2–3 days. Tumors were harvested on day 13, and the percentage of ALDH⁺ cells (N), mRNA expression of ODC1 and SRM (O), and levels of ornithine and polyamine in tumor tissues (P) were determined (mean ± SEM; n = 5). Q) 2 × 10⁶ MDA‐MB‐231 cells were implanted in SCID mice. When tumors were palpable, mice were treated with P or P + B, until tumors were no longer palpable. Kaplan–Meier survival curves of tumor‐bearing and recurrence‐free were plotted (n = 10). R‐U) MMTV‐PyMT mice were treated with V, P (5 mg kg⁻¹, days 0, 5, and 10), B (5 mg kg⁻¹, days 1–13), or P + B, when accumulated tumor volume of each mouse reached 150 mm³. Tumors were harvested on day 13, and the accumulated tumor weight (R), polyamine levels in tumor tissues (S), and the percentage of ALDH⁺ cells (T) were determined (mean ± SEM; n = 5). Tumor tissues were digested to single cells and secondary limiting dilution transplantation assay was performed in SCID mice (U). V) A proposed model of HIF‐1 in the regulation of polyamine anabolism and BCSC enrichment in response to chemotherapy. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001; ns, not significant.