|Home | About | Journals | Submit | Contact Us | Français|
Stromal cells, infiltrating immune cells, paracrine factors and extracellular matrix have been extensively studied in cancers. However, autocrine factors produced by tumor cells and communications between autocrine factors and intracellular signaling pathways in the development of drug resistance, cancer stem-like cells (CSCs) and tumorigenesis have not been well investigated, and the precise mechanism and tangible approaches remain elusive. Here we reveal a new mechanism by which cytokines produced by breast cancer cells after chemotherapy withdrawal activate both Wnt/β-catenin and NF-κB pathways, which in turn further promote breast cancer cells to produce and secrete cytokines, forming an autocrine inflammatory forward-feedback loop to facilitate the enrichment of drug-resistant breast cancer cells and/or CSCs. Such an unexpected autocrine forward-feedback loop and CSC enrichment can be effectively blocked by inhibition of Wnt/β-catenin and NF-κB signaling. It can also be diminished by IL8-neutralizing antibody or blockade of IL8 receptors CXCR1/2 with reparixin. Administration of reparixin after chemotherapy withdrawal effectively attenuates tumor masses in a human xenograft model and abolishes paclitaxel-enriched CSCs in the secondary transplantation. These results are partially supported by the latest clinical data set. Breast cancer patients treated with chemotherapeutic drugs exhibited poor survival rate (66.7 vs 282.8 months, P=0.00071) and shorter disease-free survival time if their tumor samples expressed high level of IL8, CXCR1, CXCR2 genes and Wnt target genes. Taken together, this study provides new insights into the communication between autocrine niches and signaling pathways in the development of chemotherapy resistance and CSCs; it also offers a tangible approach in breast cancer treatment.
Breast cancer is a leading cause of death in women, with about 1.7 million new cases and more than half a million deaths in the world each year.1 Despite considerable advances, most chemotherapeutic regimens that are administered at intervals to avoid irreparable damage to vital host functions ultimately fail to control disease progression.2 This is largely due to the development of drug resistance and the recovery and repopulation of drug-resistant tumor cells between treatment cycles.3, 4, 5
Although acquired drug resistance is frequently due to the reactivation of signaling pathways suppressed during therapies, treatments that block one pathway are not durable and are less effective when treating cancer recurrence.6, 7, 8, 9, 10 In addition, drug-resistant cells and/or cancer stem-like cells (CSCs) capable of initiating new tumors in vivo have been considered as key cellular compartments in cancer recurrence.11, 12 The driving forces behind drug resistance and CSC development have been closely linked to pathways that mediate communication networks between tumor cells, inflammatory factors, and other microenvironment niches.13, 14
Stromal cells, infiltrating immune cells, paracrine factors and extracellular matrix components contribute to cancer microenvironments that have been extensively studied.15 However, autocrine factors produced by tumor cells and their communication with intracellular signaling pathways in drug resistance, CSC development and tumorigenesis after chemotherapy withdrawal have not been well investigated, and precise mechanistic insight remains lacking. Cytokines (such as IL6, IL8 and CCL2) and their signaling pathways have been demonstrated to have important roles in breast cancer initiation, migration, invasion and disease progression.16, 17, 18 However, it remains unclear whether breast cancer cells are capable of producing large amount of cytokines acting as autocrine factors to self-propel the development of drug resistance and CSCs after chemotherapy withdrawal.
In this study, we show that breast cancer cells increase production and secretion of IL6, IL8, CSF2 and CCL2 cytokines after withdrawal of chemotherapeutic drugs (paclitaxel, 5-fluorouracil or doxorubicin). These cytokines activate both NF-κB and Wnt/β-catenin pathways, which in turn promote the production and secretion of more cytokines, forming an autocrine inflammatory forward-feedback loop to enrich drug-resistant cancer cells and CSCs. Accordingly, inhibition of NF-κB and Wnt/β-catenin pathways, neutralization of IL8 or inhibition of IL8 receptor CXCR1/2 with reparixin significantly diminishes forward-feedback loop and CSC enrichment in vitro. Administration of reparixin after paclitaxel withdrawal not only effectively attenuates tumor masses in a human breast cancer xenograft model, but also significantly inhibits the enrichment of CSCs and the growth of paclitaxel-resistant breast cancer cells in secondary transplantation.
To determine whether autocrine mechanisms are implicated in the tumor microenvironment after intermittent exposure to chemotherapeutic drugs and enhance breast CSC phenotypes, we treated breast cancer cells with vehicle or chemotherapeutic drugs (paclitaxel, 5-fluorouracil or doxorubicin) for 4 days, followed by washes, medium change and continual culture in fresh medium in the absence of any drug for an additional 4 days. At day 8, the drug-free and cell-free supernatants were collected and used for culture of the same line of breast cancer cells. The supernatants derived from cells receiving control vehicle-treatment were denoted as vehicle-derived supernatants and from cells receiving paclitaxel treatment as paclitaxel-derived supernatants (Figure 1a).
After exposure of untreated cells to paclitaxel-derived supernatants for 4 days, both CD44high/CD24-/low and ALDH-positive subpopulations in SUM149 and MDA-MB-231 cells (triple negative breast cancer – TNBC), and SUM190 (inflammatory breast cancer) cells increased significantly by 2-3 fold (Figure 1b; Supplementary Figure 1, flow cytometry). A marked upregulation of ALDH1, CD44 and OCT4 proteins were also observed (Figure 1c, western blot). As CD44high/CD24-/low and/or ALDH1 have been commonly used in the characterization of breast CSCs,16, 19, 20, 21 these results suggest that some factors produced by breast cancer cells themselves (i.e., autocrine factors) after drug withdrawal facilitate CSC enrichment.
As a functional measure of CSC properties, we performed mammosphere assays based on the ability of breast CSCs to generate multicellular spheroids in suspension culture. SUM149, MDA-MB-231and SUM190 cells were treated for 4 days with vehicle- or paclitaxel-derived supernatants, and reseeded in an ultra-low 6-well plate (2000 cells/well) for an additional 8 days. Cells pretreated with paclitaxel-derived supernatants formed ~2–3 folds more mammospheres and also generated larger spheroids than cells cultured in vehicle-derived supernatants (Figure 1d). The increased mammosphere forming efficiency suggests that the autocrine factors in paclitaxel-derived supernatants enhance the self-renewal capacity of CSCs.
To further characterize CSC phenotype, we assessed gene expression profiles. The qPCR results showed that treatment with paclitaxel-derived supernatants resulted in a marked upregulation of stem cells-associated genes ALDH1, SOX2, c-MYC, OCT4 and NANOG (Figure 1e). In addition, epithelial–mesenchymal transition (EMT)-related genes SLUG, MMP9, ZEB1 and SNAIL were also elevated, while E-cadherin, an epithelial cell marker, were reciprocally decreased in all three breast cancer cell lines examined (Figure 1e). As expected, after exposure to different proportions of paclitaxel-derived supernatants for 4 days, breast cancer cells became less susceptible to subsequent paclitaxel killing in a dose-dependent manner (Supplementary Figure 2A). Moreover, pre-exposure to paclitaxel-derived supernatants also led to reduced apoptosis in bulk and CSC populations while increased CSCs (both CD44high/CD24-/low and ALDH+ subpopulations) in response to paclitaxel treatment (Supplementary Figures 2B–D and 3, flow cytometry). Collectively, these data indicate that autocrine factors produced by breast cancer cells themselves after chemotherapy withdrawal lead to the induction of CSC properties and chemoresistance.
Since inflammatory cytokines have been closely associated with cancer progression and CSC development,22 we asked whether paclitaxel-derived supernatants possess high levels of inflammatory cytokines that led to CSC enrichment. We found that, after 4-day paclitaxel withdrawal, the gene expression levels of cytokine/chemokine in SUM190, SUM149 and MDA-MB-231 cells remained extremely high (Figure 2a). In particular, the gene expression levels of IL6, IL8, CSF2 and CCL2 were significantly elevated. Consistently, the protein levels of these cytokines/chemokines in supernatants were also markedly increased (e.g., up to 80-fold greater for IL8) as measured by multiple human cytokines assays (Figure 2b), indicating robust production of cytokine proteins by breast cancer cells themselves after exposure to chemotherapeutics followed by drug withdrawal.
The high levels of cytokine proteins in supernatants produced by breast cancer cells after paclitaxel withdrawal prompted us to examine the status of several inflammation-associated pathways, including signal transducer and activator of transcription 3 (STAT3), nuclear factor-kappa-B (NF-κB) and Wnt/β-catenin signaling pathways. As shown in Figure 2c, exposure to paclitaxel-derived supernatants (autocrine factors) for 4 days resulted in a significant increase in the expression of phosphorylated NF-κB, IκBα and STAT3 proteins in the same line of breast cancer cells. In addition, a significant increase in Wnt reporter activity was observed in 7xTCF-transduced cells after culture with paclitaxel-derived supernatants for 4 days (Figure 2d). The 7xTCF is a real-time fluorescent Wnt reporter construct containing 7 tandem Tcf/Lef consensus binding sites upstream of a minimal promoter driving expression of GFP and an SV40-puromycin selection cassette.23 Consistently, both NF-κB and Wnt/β-catenin target genes were significantly upregulated (Figure 2e). These results suggest that breast cancer cell-produced supernatant-factors (autocrine factors) can activate both NF-κB and Wnt/β-catenin pathways of the same line of breast cancer cells.
After analysis of the METABRIC data set,24 we found that among TNBC patients treated with chemotherapeutic drugs, samples with increased expression of Wnt/β-catenin target genes and NF-kB target genes exhibited poor survival rate and shorter disease-free time as compared with samples that expressed lower levels of these genes (Figure 2f). Together, our in vitro results and the clinical data analysis suggest that chemotherapeutic treatment may lead to enhanced secretion of inflammatory cytokines that can activate inflammatory pathways in the same line of breast cancer cells, which correlates positively with enhanced CSC phenotypes and poor clinical outcomes.
The NF-κB signaling pathway is known to play a critical role in the induction and maintenance of inflammatory cytokines.6, 7, 25, 26 Thus, consideration was given to the possibility that activation of NF-κB initiated by paclitaxel-induced supernatants enabled breast cancer cells to continually produce inflammatory cytokines, thereby forming an autocrine inflammatory forward-feedback loop that further promotes CSC enrichment. In addition, whether or not co-activation of Wnt/β-catenin and NF-κB pathways is necessary in the above process was determined by using a loss-of-function approach.
Significantly, siRNA knockdown of β-catenin and/or NF-κB p65 markedly diminished the upregulation of cytokine genes initiated by the exposure to paclitaxel-derived supernatants produced by the same line of breast cancer cells (Figure 3a). In addition, knockdown of NF-κB and/or Wnt/β-catenin significantly suppressed stemness genes and abolished CD44high/CD24−/low enrichment induced by paclitaxel-derived supernatants (Figures 3b and c). Consistently, the expression levels of CSC marker proteins CD44, OCT4 and c-MYC were also significantly suppressed after siRNA knockdown (Figure 3d, western blot), suggesting that both Wnt/β-catenin and NF-κB pathways are required. The knockdown efficiency had been confirmed by the significant reduction of NF-κB and Wnt target genes (Supplementary Figures 4E and F) and p65 and β-catenin proteins (Figure 3d).
Similar results were obtained by using small molecule inhibitors. Blockade of Wnt/β-catenin with XAV939 and NF-κB with Bay-11-7821 robustly suppressed the expression of inflammatory cytokine genes, CSC-associated genes and proteins (Figures 3e–g), CD44high/CD24−/low CSC subpopulation and cell migration (Supplementary Figure 5). To a certain degree, these data are consistent with those shown in Figures 1 and and22 where similar results were obtained from three different breast cancer cell lines. As suppression of STAT3 signaling pathways with a special small molecule inhibitor did not show significant effects (data not shown), it suggests that co-activation Wnt/β-catenin and NF-κB has an important role in the aforementioned biological processes.
Collectively, these data suggest that tumor cell-produced inflammatory cytokines activate both Wnt/β-catenin and NF-κB pathways of the same line of tumor cells, which in turn enhance inflammatory cytokine production, forming an autocrine forward-feedback loop to promote CSC enrichment after chemotherapy withdrawal. It seems that formation of autocrine inflammatory forward-feedback loop is a conserved response to chemotherapeutic drugs. Similar to paclitaxel-derived supernatants, doxorubicin-derived supernatants or 5-fluorouracil-derived supernatants also increased gene expression of IL8 and IL6, enhanced IκBα phosphorylation and 7xTCF-eGFP reporter activity (indicating activation of NF-kB and Wnt pathways), upregulated the expression of CSC-associated c-MYC protein, and augmented expression of CSC-associated genes ALDH1, OCT4, SOX2 and c-MYC (Supplementary Figures 4A–D).
Given that inhibition of NF-κB and Wnt/β-catenin pathways effectively blocked inflammatory autocrine forward-feedback loop, we next asked whether this forward-feedback loop could also be impeded by blocking key cytokines and/or their receptors (part of the upstream loop). As increases in IL8 protein levels after treatment with paclitaxel-derived supernatants were greatest in all three breast cancer cell lines (Figure 2b), we postulated that blockade of IL8 or its receptor would diminish the autocrine forward-feedback loop and CSC development. This possibility was supported by significant upregulation of IL8 receptors (CXCL1 and CXCL2) in clinical invasive breast cancer samples but not in normal controls (Figure 4a). After further analyzing the recently available clinical data set containing chemotherapeutic information (breast cancer TCGA, cBioPortal, Nature Communications 2016), we found that TNBC patients treated with chemotherapeutic drugs exhibited poor survival rate (66.7 vs 282.8 months, P=0.00071) and shorter disease-free survival time (P=0.00208) if tumor samples expressed higher levels of IL8, CXCR1, CXCR2 and the Wnt target genes than those expressed lower levels (Figure 4b). These clinical data suggest a tangible correlation between IL8/IL8 receptors, Wnt target genes and breast cancer progression after chemotherapy.
To determine whether IL8 has a causative role in the formation of inflammatory autocrine forward-feedback loop, we performed tumorsphere formation and wound scratch assays, two functional assessments commonly used in the study of CSC self-renewal and cancer cell migration/invasion.25, 26, 27 Breast cancer cells were cultured in vehicle- or paclitaxel-derived supernatants in the presence of anti-IgG control antibody, anti-IL8 antibody, vehicle or CXCR1/2 inhibitor for 4 days, followed by assessment of tumorsphere-forming or scratch-migration capacity. As anticipated, blockade of IL8 using a neutralizing antibody significantly decreased tumorsphere formation induced by paclitaxel-derived supernatants (Figure 4c). Pharmacological inhibitor reparixin capable of suppressing IL8 receptor CXCR1 and CXCR2 also markedly reduced tumorsphere formation (Figure 4c). In addition, quantification of wound width at 24-hour post injury revealed that anti-IL8 antibody or reparixin significantly inhibited tumor cell migration induced by paclitaxel-derived supernatants (Figure 4d). These results suggest that, in addition to NF-κB and Wnt/β-catenin pathways, IL8 and IL8 receptors are important components in the formation of autocrine inflammatory forward-feedback loop to promote CSC properties after chemotherapy withdrawal.
As inhibition of IL8 receptor CXCR1/2 with reparixin diminished CSC tumorsphere forming and cancer cell migration induced by autocrine inflammatory forward-feedback loop (Figures 4c and d), we then asked whether reparixin would inhibit CSC enrichment after paclitaxel withdrawal in vitro. We treated breast cancer cells with paclitaxel for 4-day followed by reparixin for an additional 4-day in the absence of paclitaxel (Figure 5a). The CD44high/CD24-/low subpopulation was analyzed by flow cytometry at day 4 immediately after paclitaxel withdrawal (Figure 5b) and at day 8 after treatment with vehicle or reparixin (Figure 5c). We found that CD44high/CD24-/low subpopulation treated with paclitaxel followed by vehicle was increased ~2.5-fold at day 8 (Figure 5c) as compared with that treated with vehicle or reparixin alone, and 1.5-fold as compared with that treated with paclitaxel alone at day 4 (Figure 5b), suggesting continual enrichment of CSC subpopulation after paclitaxel removal. In sharp contrast, breast cancer cells treated with reparixin after paclitaxel withdrawal markedly reduced CSCs subpopulation in comparison to vehicle-treated group at day 8 or paclitaxel alone treated group at day 4 (Figures 5b and c, flow cytometry).
Analysis of gene expression profiling showed that expression of cytokines- and EMT-related genes IL6, IL8, CCL2, CSF2, MMP9 and ABCB1 as well as stem cell-associated genes ALDH1, SOX2, OCT4 and CD44 was continuously increased at day 8 after paclitaxel withdrawal for 4 days, and inhibited by reparixin after removal of paclitaxel (Figures 5d and e, qPCR). Thus, reparixin treatment can hamper autocrine inflammatory forward-feedback loop and CSC enrichment in vitro.
To determine whether blockade of autocrine inflammatory forward-feedback loop after paclitaxel withdrawal exhibits similar effects in vivo, we transplanted SUM149 cells into the mammary fat pads of athymic nude mice. When the tumor reached a mean diameter of 4 mm after 15 days of implantation, mice were randomized into 4 groups and injected intraperitoneally with vehicle, paclitaxel (10mg/kg on days 15, 20 and 25), IL8 receptor CXCR1/2 inhibitor reparixin (25mg/kg once every other day starting from days 26 for 10 days), or paclitaxel following by reparixin (Figure 6a). To determine CSC pool in vivo, we harvested tumors at the end of the treatment (at day 35) and assessed CD44high/CD24-/low subpopulation and ALDH-positive population using flow cytometry. As shown in Figures 6b–d, in comparison to vehicle control, paclitaxel treatment alone increased CSC subpopulation over 2-fold after paclitaxel withdrawal for 10 days (P<0.05), although it impeded tumor growth (P<0.05), consistent with in vitro findings that chemotherapy withdrawal enriched CSCs. While reparixin treatment alone only showed a moderate effect on tumor growth, it slightly suppressed CSC enrichment. Notably, reparixin administration after paclitaxel withdrawal markedly reduced tumor burden compared with all other groups, and also inhibited paclitaxel treatment-induced CSC enrichment. Reparixin administration following conventional chemotherapy, therefore, effectively inhibits both bulk tumor and tumorigenic drug-resistant cancer cells and CSCs.
To determine whether tumors containing drug-resistant breast cancer cells and/or increased CD44high/CD24-/low fractions possess greater tumor-initiating potential, we performed secondary transplantation. We serially diluted tumor cells containing various percentage of CD44high/CD24-/low populations isolated from primary tumors, and transplanted them into athymic nude mice without further treatment. Significantly, tumor cells isolated from paclitaxel-treated mice exhibited a greater tumor-initiating capacity. On the contrary, tumor cells isolated from reparixin administration after paclitaxel withdrawal had markedly reduced tumor-initiating capacity (Figure 6e). Based in vitro and in vivo results, our data indicate that inhibition of autocrine inflammatory forward-feedback loop after chemotherapy withdrawal enhances drug efficacy by reducing tumor burden and enrichment of drug-resistant breast cancer cells and CSCs.
Chemotherapy is a standard of care in clinical oncology today due to its effectiveness in reducing tumor burden and improving overall survival. Most chemotherapeutic drugs are administered at intervals to reduce severe side effects.3, 4, 5 A critical question that remains unanswered is whether drug resistance is associated with intermittent drug administration. If so, what are the underlying mechanisms and effective approaches to prevent it? Our results indicate that after paclitaxel withdrawal, breast cancer cells produce inflammatory cytokines to activate both Wnt/β-catenin and NF-κB signaling pathways which in turn further promote cytokine production from breast cancer cells (Figures 1, ,22 and and3).3). Formation of an autocrine inflammatory forward-feedback loop after termination of chemotherapy leads to repopulation of drug-resistant breast cancer cells/CSCs (Figure 6).
It has been suggested that breast cancer cells produce cytokines and/or growth factors to facilitate their survival and expansion. In the present studies, we demonstrate that secretion of CCL2, IL8, IL6 and CSF2 are markedly increased after paclitaxel withdrawal (Figure 2). After termination of paclitaxel treatment for several days, breast cancer cells continually produce higher levels of cytokines to promote CSC enrichment (Figures 2,,33,,44 and and5).5). An autocrine forward-feedback loop involving the released inflammatory cytokines and intracellular signaling pathways may take place during intermittent administration of chemotherapeutic drugs in typical clinical settings.
Indeed, after removal of paclitaxel, culture supernatants enriched with inflammatory cytokines produced by breast cancer cells markedly enhance phosphorylation of NF-κB (Figure 2). Cytokine signaling is known to activate NF-κB, and NF-κB signaling is known to promote cytokine production.28, 29 However, inhibition of NF-κB signaling alone is insufficient to increase sensitivity to cytotoxic drugs30 and was ineffective at diminishing cytokine production from breast cancer cells (Figure 3). We revealed that activation of Wnt/β-catenin and NF-κB is important for the formation of such an autocrine inflammatory forward-feedback loop. As a result, blockade of Wnt/β-catenin and NF-κB effectively abrogates the cytokine feedback loop, diminishes the expression of stemness genes and the enrichment of CSC subpopulation (Figure 3). Co-activation of NF-κB and Wnt/β-catenin pathways has recently been shown to be essential for the development of breast CSCs in response to toll-like receptor 3 stimulation,25 and also have a cooperative role in conferring CSC properties in an intestinal cancer model.31 In addition, phosphorylated NF-κB has been shown to bind to β-catenin via CREB-binding protein to strengthen β-catenin transcriptional activities, resulting in generation of strong self-renewal signaling in myeloid leukemia stem cells.32 This suggests that inhibition of Wnt/β-catenin and NF-κB pathways after chemotherapy withdrawal will maximize clinical efficacy.
We also identify IL8 as an alternative target for blockade of autocrine inflammatory forward-feedback loop after chemotherapy withdrawal (Figures 2 and and4).4). IL8 signal, via two cell surface G-protein–coupled receptors CXCR1 and CXCR2, has been shown to be upregulated in several types of cancer including breast cancer and associated with increased CSC pool in vitro and poor prognosis in patients.33, 34, 35 Our in vitro data demonstrated that blockade of IL8 with a neutralizing antibody or blockade of IL8 receptors CXCR1 and CXCR2 with reparixin inhibits autocrine inflammatory forward-feedback loop triggered by paclitaxel treatment (Figures 4 and and5).5). Our in vivo data also show that administration of reparixin after paclitaxel withdrawal markedly reduced tumor mass and diminished drug-resistant breast cancer cells/CSC subpopulation, subsequently diminishing tumor-initiating capacity (Figure 6). Consistently, patients with TNBC treated with chemotherapeutic drugs exhibited very poor survival rate and shorter disease-free survival time if their tumor samples expressed high levels of IL8, CXCR1/2 and Wnt signals (Figure 4b).
In our in vitro experiments, inhibition of NF-κB and/or Wnt signaling pathways (with siRNA or small molecules) exhibits better potency than neutralization of IL8 with antibody or inhibition of CXCR1/2 with reparixin in the blockade of stem cell- and EMT-associated gene expression. Further optimization of the inhibitory dosage of IL8 antibody and reparixin seems to be required. It is also possible that other undefined factors may have cooperative roles, warranting further studies.
In conclusion, our studies identify an autocrine inflammatory forward-feedback loop after withdrawal of chemotherapeutic drugs, which leads to the repopulation of drug-resistant breast cancer cells/CSCs and facilitate tumor progression. Our results suggest that inhibition of NF-κB and Wnt/β-catenin pathways or key cytokine signals along autocrine inflammatory forward-feedback loop after chemotherapy withdrawal will reduce tumor mass and secondary tumor initiation. Given that NF-κB and Wnt/β-catenin inhibitors and reparixin have been approved by FDA for clinical trials, administration of these reagents between intermittent cycles of chemotherapy may reduce toxicity, and are worth studying in the context of future clinical trials to reduce disease relapse and improve patient outcomes.
SUM149 and SUM190 breast cancer cells were obtained from Asterand (Detroit, MI, USA) and cultured in Hams F12 media (Mediatech, Manassas, VA, USA) containing 5μg/ml insulin, 1μg/ml hydrocortisone, 10mM HEPES and antibiotics (penicillin/streptomycin). Medium for SUM149 cells was further supplemented with 5% fetal bovine serum (HyClone, Logan, UT, USA), and for SUM190 cells was further supplemented with 5mM ethanolamine, 5μg/ml transferrin, 6.6ng/ml 3,3',5-triiodo-l-thyronine sodium salt, 8.7ng/ml sodium selenite, and 1mg/ml bovine serum albumin. Cells were cultured at 37°C in a 5% CO2 incubator. Breast cancer cell line MDA-MB-231 was purchased from the American Type Culture Collection (Manassas, VA, USA) and maintained in DMEM-F12 (1:1) supplemented with 10% Fetal Bovine Serum. Paclitaxel, insulin, hydrocortisone, HEPES, and bovine serum albumin were purchased from Sigma-Aldrich (St. Louis, MO, USA). Reparixin was purchased from MedChemExpress (MedChemExpress Biotechnology, Monmouth, NJ, USA).
Cancer cells dissociated from the transplanted tumor tissues or from culture plates were counted and re-suspended in 100μl of HBSS (Gibco, Langley, OK, USA) containing 2% heat-inactivated fetal bovine serum (FACS buffer) and 105 cells. Five microliters of mouse IgG solution (1mg/ml) was added and incubated on ice for 10min. According to the manufacturer’s recommendation, appropriate antibodies were added and incubated for 30min on ice. The cells were then washed twice with FACS buffer and re-suspended in 0.2ml of FACS buffer that contained 7-aminoactinomycin D (7-AAD, eBioscience, San Diego, CA, 1μg/ml, final concentration) to exclude dead cells. Antibodies used were anti-CD44 (APC) and anti-CD24 (PE), which were purchased from BD Pharmingen. ALDH activity was examined with the ALDEFLUOR kit (Stem cell Technologies, Vancouver, Canada). Cell apoptosis was assessed using an Annexin V-PE-Cy7 Apoptosis Detection Kit (eBiosciences, Burlington, ON, Canada) for Beckman Cyan-ADP 9 flow cytometer or V450 Annexin V (BD) for BD LSR Fortessa. Flow cytometric data were analyzed with Kaluza software (Beckman Coulter, USA) or FlowJo (FlowJo LLC, Ashland, Oregon, USA).
A soft-agar assay was performed on 12-well plates with a base layer of 0.5% agarose gel containing DMEM. To generate the cell layer, 5 × 103 cells/well were suspended in 0.35% top agarose gel in DMEM/F12 medium containing B27 supplement, 20ng/ml of EGF and 20ng/ml of basic FGF. Plates were incubated at 37°C in 5% CO2 for 17 days to allow colony formation, and cell viability was determined by staining with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT, Sigma-Aldrich, St. Louis, MO, USA, 1mg/ml). Colonies of each cell line were counted (>100μm in diameter). All experiments were performed in triplicate, and data are presented as means±S.D.
For western blot analysis, cells were harvested and prepared using RIPA buffer (Sigma-Aldrich) and sub-cellular fractions prepared using the NE-PER Nuclear Protein Extraction Kit (Thermo Scientific, Nepean, ON, Canada) containing protease inhibitor cocktails (Roche, Mannheim, Germany). Protein concentration was determined using a Bio-Rad DC protein assay kit (Bio-Rad, Hercules, CA, USA). Subsequently, 25–30μg of total protein for each sample was loaded onto an 8-12% SDS-PAGE for electrophoresis and then transferred to a PVDF membrane. Protein was identified by incubating the membrane with primary antibodies, followed by horseradish peroxidase-conjugated secondary antibodies and an enhanced chemiluminescence solution (Pierce, Thermo Scientific, USA). Antibodies used in this study include: anti-c-MYC polyclonal antibody (1:1000, D84C12, Cat. 5605), anti-CD44 (8E2) monoclonal antibody (5640), anti-phospho-Stat3 (Tyr705) monoclonal antibody (1:1000, D3A7, Cat. 9145), anti-Stat3 monoclonal antibody (1:1000,124H6, Cat. 9139)) from Cell Signaling (Danvers, MA, USA); anti-ALDH1A1 antibody (1:1000, ab105920) from ABCAM (Cambridge, UK); anti-β-actin monoclonal antibody (1:5000, AC15) from Sigma-Aldrich (St. Louis, MO, USA); anti-Oct4 antibody (ab137427) from ABCAM (Toronto, ON, Canada); Anti-NF-κB p65 monoclonal antibody (1:1000, 112A1021), anti-phospho-NF-κB p65 pSer 536 monoclonal antibody (1:1000, T.849.2), anti-I-kappa-B-alpha monoclonal antibody (1:1000, T.937.7), and anti-phospho-I-kappa-B-alpha pSer32/36 monoclonal antibody (1:1000, H.709.9) from Thermo scientific (Rockford, USA).
The Alamar Blue assay was used for cell viability or relative cytotoxicity analysis according to manufacturer instructions (Bio-Rad Laboratories GmbH, Munich, Germany). Briefly, Alamar Blue dye (BUF012B, Bio-Rad Laboratories), a cell metabolism indicator, was added to a final concentration of 1% (vol/vol) for 4-6h. Fluorescence was measured using a Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek Instruments, Winooski, VT, USA) in top-reading mode with excitation at 560nm and emission at 590nm. Readings were analyzed using Omega analysis software. Cells grown in media+vehicle served as a vehicle control whereas media alone containing 1% (vol/vol) Alamar Blue served as blank and subtracted from the fluorescence value obtained for each well. Results were expressed as relative fluorescence (%) after normalized by designating vehicle control (100%) according to the manufacturer's protocol.
SUM190, SUM149 and MDA-MB-231 cells (2 × 103/well) were reseeded in Ultra-Low attachment plates (Costar, Corning, NY, USA) after treated with vehicle-derived or paclitaxel-derived supernatants for 4 days. The 6-well plates were incubated at 37°C in 5% CO2 for 8 days to allow mammosphere formation. Colonies (>100μm in diameter) of each group were counted.
Total RNAs were extracted using RNeasy kit (QIAGEN) and real-time qPCR analysis was performed using Bio-Rad MyiQ (Bio-Rad) as previously described.36, 37 The conditions for qPCR reactions are: one cycle at 95°C for 20s, followed by 40 cycles at 95°C for 3seconds and annealing at 60°C for 30s. Results were normalized to the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Relative expression level of genes from different groups were calculated with the 2ΔΔCT method and compared with the expression level of the corresponding gene in control cells, Specific primer sequences for individual genes are listed in Supplementary Table 1.
Cell-free supernatants were generated from SUM190, SUM149 and MDA-MB-231 cells after 8 days culture. The cells were first cultured in the presence of vehicle or paclitaxel (15nM), doxorubicin (0.35μM) or 5-fluorouracil (75μM) for 4 days. After washed with PBS, the cells were continuously cultured in fresh medium without any drug for additional 4 days. At Day 8, the drug- and cell-free supernatants were collected for subsequent experiments and for human cytokine assay. A 11-plex human cytokine profiling kit was used for human cytokine arrays for protein concentrations of cytokines in the supernatants, including IFN-γ, GM-CSF, interleukin 1β (IL1β), interleukin 2, interleukin 4, interleukin 6, interleukin 8, interleukin10, interleukin 12, CCL2/MCP-1 and tumor necrosis factor α (TNF-α). The arrays were performed blindly by Eve Technologies (Eve Technologies, Calgary, Alberta, Canada).
siRNAs for NF-κB p65 and β-catenin, and control scrambled siRNA were purchased from Thermo Scientific (Dharmacon, USA) as SMARTpools. For siRNA transfections, cells were transiently transfected with these oligos using Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer’s instructions. Transfection efficiency was assessed using qPCR and western blot. After treatments, cells were trypsinized and subjected to western blot and flow cytometry assays.
The Wnt reporter breast cancer subline 7TGP-SUM190 were generated by transduction of Wnt reporter construct 7TGP 23 into SUM190 cells by lentiviral transduction. The β-catenin/TCF/LEF-dependent reporter plasmid (7 × Tcf-eGFP//SV40-PuroR, 7TGP) containing seven Tcf/Lef-binding sites and a puromycin resistance gene was provided by Dr. Nusse via Addgene (#24305).23 Lentiviral production was carried out as described previously.25, 38 Briefly, ten 10-cm dishes were seeded with 6 × 106cells per dish overnight before transfection. For two dishes, 8μg of the lentiviral vector, 5.4μg of the psPax2 envelope plasmid, 3.6μg of the packaging plasmid (pMD2.G) were used. The medium was replaced overnight, and lentiviral supernatant was harvested after 48hours, filtered through a 0.45-μm PES filter, and concentrated with Lenti-X concentrator (Clontech, Mountain View, CA, USA). For viral infection, once SUM190 cells in 6-well plate reached 40-50% confluence, 1ml of concentrated lentiviral supernatant and 8μg/ml of polybrene were added for 24h. The infected cells containing the reporter TCF-eGFP cassette were selected with puromycin. The TCF-eGFP expression levels after treatment with vehicle- or paclitaxel-derived supernatants were determined by FACS.
Scratch wound-healing assay was performed as reported previously.27, 39 Briefly, cells were seeded in 24-well plates and grown to confluence. The monolayer was scratched with a 200μl sterile pipette tip, washed twice with PBS to remove detached cells and debris, and then incubated with medium containing different reagents as indicated. Each experimental condition was tested in quadruplicate. The cells were photographed at different time points using a Zeiss Axiovert 40 CFL microscope (Carl Zeiss AG, Feldbach, Switzerland). The open area was quantified on each photo to obtain mean values. Migration was reported as the difference (in mm2) between the scratch dimensions observed at 0 and 24-h.
All mouse experimentation was conducted in accordance with standard operating procedures approved by the Animal Care Committee at the University of Ottawa. Athymic nude mice (6–8 week old, female, 20 to 25g body weight) were obtained from Charles River Laboratories. To establish breast cancer xenografts in nude mice, SUM149 cells were mixed with Matrigel (BD Biosciences, Bedford, MA, USA) and injected under aseptic conditions into mammary fat pads of nude mice (n=4–8 for each group, 2 × 106 cells per fat pad). The tumor was monitored and evaluated every 2–3 days with calipers. Tumors were measured in 2 dimensions, and volume was calculated according to the formula: V=0.5 × (length) × (width)2. When the tumor reached a mean diameter of 4mm (day 15), tumor-bearing mice were randomized into 4 groups and intraperitoneally injected with vehicle, Paclitaxel alone (10mg/kg on days 15, 20 and 25), reparixin alone (25mg/kg on every day for 10 days starting from day 25), or paclitaxel following by reparixin for 10 days. At the end of drug treatment, mice were humanely euthanized and tumors were harvested for further analyses and secondary transplantation in a blinded manner.
Tumor tissues were dissociated mechanically and enzymatically to obtain a single-cell suspension. Tumors were minced by scalpel and incubated in Hams F12 (Invitrogen) containing collagenase/hyaluronidase (#07912, STEMCELL Technologies, Canada) at 37°C for 60min. The tissues were further dissociated by pipette trituration and then passed through a 40-μm nylon mesh to produce a single-cell suspension. Three groups of mice were implanted with tumor cells. Each athymic nude mouse was inoculated into one of the inguinal mammary fat pads with 105, 104, 103 or 102 cells from the first tumors receiving treatment of either vehicle, paclitaxel, or paclitaxel followed by reparixin. Tumor growth and size were measured every other day for 6 weeks.
Data are expressed as means±S.D. unless specified elsewhere. Statistical significance was determined using a Student's t test, ANOVA or chi-square test wherever appropriate. A Kaplan–Meier log-rank test survival plot was performed using the data available at http://www.cbioportal.org/study?id=brca_metabric#summary (Breast Cancer, METABRIC, Nature 2012 & Nature Communications 2016). The log-rank test was performed to determine whether observed differences between groups were statistically significant. Results were considered significant with a P-value<0.05.
We thank Operations Manager of Flow Cytometry Dr. Vera Tang for technical advice and support. This work is supported by operating grants from Canadian Breast Cancer Foundation-Ontario Region, the Canadian Institutes of Health Research MOP-111224, and Natural Sciences and Engineering Research Council RGPIN-2017-05020 (to LW).
Supplementary Information accompanies this paper on Cell Death and Disease website (http://www.nature.com/cddis)
Edited by R Johnstone
The authors declare no conflict of interest.