Background
The majority of the cancer stroma compartment is comprised of cancer-associate fibroblasts (CAFs) which are a prognostic factor in invasive breast cancers [
1‐
3]. These fibroblasts differ from the normal resident fibroblasts in that they show a persistent activated phenotype that leads to secretion of a variety of pro-tumorigenic cytokines, growth factors and extracellular matrix (ECM) molecules. These stroma cells show great potential as drug targets in breast cancer, given their genomic stability and abundance. Increasing evidence also demonstrates that targeting CAFs could be a promising clinical approach. However, targeting these stromal cells remains a challenge due to their heterogeneity. Therefore, a better characterization of the underlying mechanisms that promote the activated phenotype of CAFs is necessary for targeting these cells with therapy. One such mechanism that we have previously shown to influence the activated fibroblast phenotype is an upregulation of a pro-oxidative enzyme, NADPH oxidase 4 (Nox4) [
4].
While mitochondria are considered the major source of cellular ROS generation, significant levels of ROS can be contributed through the activation of Nox enzymes. These membrane-associated enzymes function in generating superoxide (O
2•−) or its derivative, hydrogen peroxide (H
2O
2), when triggered by specific stimuli, which include a number of growth factors and cytokines [
5]. Depending on their subcellular distribution, the Nox family of proteins execute a wide range of ROS-mediated biological functions. By activating/inactivating oxidative-sensitive kinases or protein tyrosine phosphatases, Nox-derived ROS are now recognized to be important mediators that provide “fine-tuning” of a variety of signaling cascades. Recently, research suggested that Nox activation is linked to the etiology of cancer [
6]. Among the seven Nox family members, Nox4 is the only one that is constitutively active in generating ROS (mainly H
2O
2 and a small amount of O
2−*) without the need of additional accessory proteins, as are required by the other Nox enzymes [
7‐
9].
To date, not much is known about the role of Nox4 in breast cancer despite a few studies showing that overexpression of this enzyme in cancer cells can promote an aggressive phenotype [
10]. In fibroblasts, Nox4 has been shown to be one of the downstream effectors of TGFβ in mediating fibroblast activation during fibrosis in cardiac and pulmonary fibroblasts [
11,
12] but the role of Nox4 in CAFs is not fully understood. We have previously shown that Nox4 is essential in mediating the myofibroblast phenotype in RMF-HGF [
4]. RMF-HGF is an experimentally-generated mammary CAF model isolated from reduction mammoplasty and transduced with hepatocyte growth factor (HGF) [
13]. These pro-tumorigenic fibroblasts showed a significant upregulation of Nox4 as compared to its parental non-malignant fibroblasts (RMF: isolated from reduction mammoplasty). Moreover, inhibition of Nox4 attenuated the collagen contraction ability, myofibroblast marker expression, and pro-invasive properties of RMF-HGF [
4], suggesting a potentially important role of Nox4-generated ROS in CAFs.
The nuclear factor erythroid 2-related factor 2 (Nrf2) is a transcription factor that is activated in response to oxidative stress and electrophilic stress. It trans-activates antioxidant genes and detoxification genes to restore cellular redox balance and provides phase II detoxification response in eukaryotes [
14]. Under basal conditions, Nrf2 is kept at very low levels where Keap1 functions as an adapter for the E3 ubiquitin ligase Cullin-3 (CUL3) constitutively targets Nrf2 for ubiquitination and degradation. Under oxidative stress, Keap1 is inactivated, leading to a translocation of Nrf2 to the nucleus, where it activates genes that contain an antioxidant response element (ARE) [
15]. Nrf2 has dual functions in cancer, where its activation prior to tumor initiation or progression is preventive, but activation of Nrf2 in an established tumor enables increased proliferation and aggressiveness, as well as resistance to therapies [
16,
17], making Nrf2 a promising anti-cancer target.
In this study, we have further investigated the influence of Nox4 on fibroblast activation and their tumor-promoting function. We found that stroma deletion of Nox4 and administration of a pharmaceutical Nox4 inhibitor (GKT137831) resulted in suppression of orthotopic mammary tumor growth and metastasis in two syngeneic models, suggesting a prominent role of stroma Nox4 in oncogenesis. To show that this pro-oxidant is clinically relevant, we obtained a panel of patient-derived breast CAFs and found a significant upregulation of Nox4 expression levels in all CAFs compared to the normal mammary fibroblasts (also referred to as RMFs). Furthermore, Oncomine analysis showed that Nox4 is ranked in the top 1–3% of most significantly upregulated genes in breast carcinoma stroma versus normal stroma, whereas only a slight upregulation is seen in breast cancer epithelial cells. In addition, we showed that Nox4-generated ROS provides a pro-survival advantage in CAFs via promoting an autophagic phenotype. We later found that activation of the Nrf2 antioxidant pathway contributes to this pro-survival feature in CAFs. Taken together, our study strongly supports a role of Nox4 in breast CAFs and that this pro-oxidant could be a promising stroma target to interfere with the tumor supporting network in breast cancer.
Methods
Cell lines, breast CAFs, growth conditions, and reagents
MDA-MB231 (ATCC, HTB-26, RRID:CVCL_0062) and 4T1 (ATCC, CRL-2539, RRID:CVCL_0125) were cultured as previously described [
18]. Authentication was verified by short tandem repeat DNA profiling. Breast CAFs were obtained from Asterand Biosciences (now BioreclamationIVT) and were certified and accrued with strict consensual control, quality assurance and accurate clinical data. All pathology diagnosis were confirmed by board certified pathologists and each specimen is scored for condition, cancer percentage, necrosis and other physical factors. The E0771 mouse breast cell line was obtained from CH3 Biosystems and were cultured as recommended by the supplier. Reduction mammoplasty fibroblasts expressing human HGF (RMF-HGF) and their parental normal fibroblasts (RMF) were generated and gifted by Dr. Charlotte Kuperwasser (Tufts University, Boston, MA, USA) [
13]. All cultures were regularly tested for Mycoplasma.
Antibodies and reagents
Antibodies used were: anti-Nox4 (Abcam Cat# ab133303, RRID:AB_11155321), anti-Nrf2 (Abcam Cat# ab62352, RRID:AB_944418), anti-Keap1 (Proteintech Cat# 10,503–2-AP, RRID:AB_2132625), anti-beta-actin (Cell Signaling Technology Cat# 3700, RRID:AB_2242334), anti-p62 (Cell Signaling Technology Cat# 5114, RRID:AB_10624872), and Phospho-p62 (Ser349) (Cell Signaling Technology Cat# 16,177, RRID:AB_2798758). Secondary horseradish peroxidase conjugated antibody used were anti-rabbit (1:5000, Thermo Fisher Scientific, Cat# G-21234, RRID AB_2536530) and anti-mouse (1:6000, Thermo Fisher Scientific Cat# A24524, RRID:AB_2535993).
Brusatol was purchased from Sigma Aldrich (SML1868) and was prepared in DMSO (5 mg/ml). DMF was purchased from Sigma-Aldrich (242,926) and was dissolved in DMSO (5 mg/ml). The selective Nox4 inhibitor GKT137831 was kindly supplied by Genkyotex (S.A., Geneva, Switzerland).
Real time PCR primers
SDF1: Forward—5′ GAC CCA ACG TCA AGC ATC TC 3′.
Reverse—5′ CGG GTC AAT GCA CAC TTG TC 3′.
αSMA: Forward—5′ GCG TGG CTA TTC CTT CGT TA 3′.
Reverse 5′ TCA GGC AAC TCG TAA CTC TTC TC 3′.
PDFGRα: Forward—5′ TGC CTG ACA TTG ACC CTG T 3′.
Reverse—5′ CCG TCT CAA TGG CAC TCT CT 3′.
Birc5: Forward—5′ ACCACTTCCAGGGTTTATTCC 3′.
Reverse—5′ CAGGCAGAAGCACCTCTG 3′.
FAP: Forward—5′ TCCAGAATGTTTCGGTCCTG 3′.
Reverse—5′ CTATATGCTCCTGGGTCTTTGG 3′.
IL10: Forward—5′ AGG CTG AGG CTA CGG CGC 3′.
Reverse—5′T TAG ATG CCT TTC TCT TGG AG 3′.
Macrophage differentiation and polarization
Peripheral blood mononuclear cells (PBMCs) were isolated from human donor whole blood. Primary human monocytes and peripheral blood leukocytes were separated via elutriation by the UNMC Elutriation core. Monocytes, and PBLs were used immediately after separation or were cryopreserved in liquid nitrogen before use. PBMCs and monocytes were maintained at 37˚C in 5% CO2 in RPMI media with glutamine, 10% fetal bovine serum, penicillin and streptomycin added. Polarization of macrophages was induced as we previously described (). Specifically, monocytes were differentiated and polarized to M2 macrophages with M-CSF (100 ng/mL, BioLegend #574,806) for 7 days to promote monocyte differentiation and growth. Then, the M-CSF stimulated macrophages were polarized to M2 by addition of IL-4 (20 ng/mL, BioLegend #574,002) for 24 h. 20uM of GKT137831 was added to the monocytes during IL-4 treatment.
Collagen contraction assay
Activity of CAFs was determined and analyzed by collagen contraction assay as previously described [
4].
Invasion and migration assay
Invasiveness of MDA-MB231 was performed as described [
4]. Culture-inserts (Ibidi) were used to measure cell migration. Cell suspension at density of 1 × 10
6/ml breast cancer cells and 1.6 × 10
5/ml of fibroblast (70 μL volume) was applied to each chamber of the cell culture insert. The cell culture insert was removed after 16 h leaving a defined cell-free gap of 500 μm. The cells were then treated with brusatol (40 nM) for 24 or 48 h. The cells were fixed and stained with 2% crystal violet. Images were captured every 6 h using an inverted phase contrast microscope. Images were taken at 10X magnification, and the cell-free space was analyzed by using the ImageJ software (RRID:SCR_003070). The percent of wound closure in five randomly chosen fields was calculated.
Western blot analysis
Cell lysates preparation and Western blot analysis was performed as previously described [
19].
RT–PCR
Total RNA isolation, and reverse transcription, and RT-PCR analysis were performed as previously described [
4]. Primer sequences are listed in supplemental Materials and Methods, except for Nox4, which have been previously described [
4].
ROS Detection and glutathione assay
Extracellular H
2O
2 levels were measured using the Amplex Red assay as previously described [
20]. Glutathione levels were measured by the GSH/GSSG-Glo assay (Promega #V6611) as previously described [
20].
In Situ RNA hybridization for tissue labeling
In situ detection of Nox4 mRNA transcription was performed on the breast cancer tissue microarray (TMA-BR8013) using a RNAscope kit (Advanced Cell Diagnostics, Hayward, CA, USA) with verified probes (Probe-Hs-NOX4) and RNAscope FFPE Reagent Kit, 2.0 HD-Brown Detection Kit, according to the manufacturer's protocol. Specificity of the probe was verified using a positive and a negative control, as provided by the supplier. Housekeeping gene Peptidylpropyl isomerase B (PPIB) was used as an internal-control. Tissues were blindly scored by a board-certified pathologist. Positive staining was determined by brown punctate dots in the nucleus and/or cytoplasm, as recommended by the supplier.
Immunofluorescent autophagic flux assay
Fibroblasts (normal or cancer associated) were plated on coverslips and transfected with LC3 tandem expressing GFP-RFP using Fugene 6 according to the manufacturer’s protocol. After 2 days, steady-state cells were fixed with 4% paraformaldehyde and Vectashield containing DAPI. Respective filters were used to image red and green LC-3II puncta. All confocal images were captured using Zeiss 710 Confocal Laser Scanning Microscope and analyzed using Zeiss Zen software.
siRNA transfection
All siRNA transfections were performed as previously described [
4]. The siCon (#4,390,843) and the siNox4 (ID#s224159) were purchased from Thermo Fisher Scientific, Waltham, MA, USA. A positive control and a negative control were used to verify the specificity of the siRNAs, by real time-PCR and western blot analysis.
Viability assay
10,000 cells/well were seeded in triplicate in 96-well plates (Corning, USA) and allowed to attach for 18 h prior to drug treatments. On the day of assay, 10 μl of the Presto Blue ™ Cell Viability Reagent was added to each well and incubated for 30 min at 37 °C. The signal of each well was detected at 540 nm using Tecan-M200 plate reader.
Generation of Nox4 inducible fibroblasts
RMF was constructed to overexpress Nox4 under doxycycline induction via the Lentiviral Tet-On 3G inducible expression systems (Takara Bio USA, Inc.).
In vivo tumor study
Balb/c mice and C57BL/6 J mice (wild type and Nox4
−/−) at 8–10 weeks old were purchased from The Jackson Laboratory (Bar Harbor, ME). Only female mice were used. The animal protocol was reviewed and approved by the Institutional Animal Care and Use Committee of the University of Nebraska Medical Center. The number of animals used per group was determined in consultation with a biostatistician, based on Power calculation to ensure that the desired 80% power is achievable. Orthotopic tumor implantation was performed as previously described [
18]. After tumors were palpable, mice were randomly grouped prior to drug administration. In the treatment group, mice were given a daily dose of GKT137831 at 60 mg/kg in 200 uL of vehicle via oral gavage starting 1 week post-cell injection. Control group received the same daily volume of vehicle. Tumor volume was measured with a caliper within a week after the treatment started.
Immunofluorescent analysis
Fixed tumor tissue that also contained stroma of the mammary fat pad and adjacent skin were paraffin-embedded and sectioned by the Tissue Science Facility at the University of Nebraska Medical Center. Tissues were de-paraffinized in xylenes and rehydrated through graded alcohols. For antigen retrieval, slides were boiled for 20 min in 10 mM EDTA (pH 9.0). Slides were then allowed to cool for 10 min. Tissues were blocked in 10% horse serum in PBS for 3 h. Following blocking, tissue sections were incubated with a primary antibody for alpha smooth muscle actin (Novus Biological, NB300-978, 1:200) overnight at 4 °C in a humidified chamber. The following day, slides were washed in PBS and stained with a secondary antibody conjugated to AlexaFluor555 (1:250, donkey anti-goat, Invitrogen, cat. A21432). Slides were mounted under coverslips with ProLong™ Gold Antifade with DAPI (Invitrogen, cat. P36931). Slides were imaged using a Leica DM 4000B LED fluorescent microscope, followed by analysis with ImageJ. For the primary tumor region, areas of surrounding stroma were imaged, capturing between 6–10 fields of view. The stromal regions were manually traced to eliminate any tumor tissue or dead space and the number of positive alpha smooth muscle actin cells were normalized to the area analyzed. For lung sections, the sites of tumor metastases were specifically imaged, 7 fields of view were analyzed per lung. Dead space and blood vessels were removed from the analysis. Mean fluorescent intensity per unit area was determined for each field of view.
Statistical analysis
Statistical analyses were performed using GraphPad Prism 7 software version 7.03 (RRID:SCR_002798). For some experiments, a single-factor ANOVA followed by post hoc Tukey test was used to determine statistical differences between means. Statistical analyses were assessed using a two-tailed Student’s t test. Results shown are representative of at least three separate experiments each performed in triplicate.
Discussion
It is well recognized that an oxidative tumor microenvironment (TME) contributes to tumor progression, metastasis, recurrence, and therapeutic resistance [
31]. While ROS or oxidative stress can originate from various sources in neoplastic epithelial cells, including from altered metabolism, oncogene expression, and heightened proliferative signals, there is a gap in our understanding of the role played by the stromal components in these phenomena. Our study here points to CAFs as one of the major contributors to the oxidative TME, via upregulation of Nox4. Hanley et al. [
32] have suggested Nox4 as a potentially promising CAF target based on Nox4 and αSMA IHC staining in stromal regions of human samples (head and neck squamous cell carcinoma, esophageal adenocarcinoma, and colon adenocarcinoma), as well as in vivo findings using TGFβ-activated skin fibroblasts. Increased Nox4 gene expression has also been reported in prostate cancer-associated stroma [
33]. While these small number of studies support a tumor-promoting role of Nox4 derived from activated fibroblasts, direct evidence linking this pro-oxidant to the tumor-supporting CAF phenotype and the mechanisms involved is lacking, particularly in breast cancer. This report, by utilizing a panel of patient-derived breast CAFs, in addition to an experimental-CAF model [
4] (RMF-HGF) support a role of Nox4 in inducing an autophagic CAF phenotype and a pro-survival Nrf2-Birc5 stress response, thereby promoting mammary tumorigenesis and metastasis.
High Nox4 activation can be detrimental to cells during severe stress. An adaptive antioxidant mechanism in counteracting this chronically high Nox4 context is likely involved to maintain an optimum range of ROS for CAF survival. Indeed, we observed the levels of the master stress regulator, Nrf2 are upregulated in all CAFs (Fig.
6). Nrf2 is a transcription factor that is activated in response to oxidative stress and electrophilic stress. Hyperactivation of Nrf2 in tumors creates an environment that favor the survival of cancer cells by protecting them from excessive oxidative stress, chemotherapeutic agents, or radiotherapy. Upregulation of Nrf2 is therefore, strongly associated with poor patient prognosis and therapeutic resistance [
17]. While much is known about the role of Nrf2 in cancer cells, very little is known regarding the implication of this adaptive response pathway in CAFs. Although a proteomic analysis showed upregulation of some Nrf2-targeted proteins in chemo-induced activated fibroblasts [
34] and CXCL14-induced activated fibroblasts [
35], demonstrating that hyperactivation of Nrf2 occurs in breast CAFs has not been reported.
Cancer cells rely on autophagic flux to promote growth and survival [
36]. This has led to numerous clinical trials focusing on inhibiting autophagy in cancer. Chloroquine and hydrochloroquine (CQ and HCQ) are currently the only clinically available drugs that target autophagy [
37]. Despite promising outcomes, further investigation is needed to elucidate the molecular mechanisms of these compounds not just in cancer cells but in other non-cancerous cell types in the TME. The dependency of CAFs on autophagy is not well established. Heightened basal autophagy has been reported in stroma fibroblasts of head and neck cancers [
38] and prostate cancer [
39]. However, the molecular mechanism underlying these phenomena is not well defined. In another study, normal fibroblasts that were co-cultured with breast cancer cells showed an autophagic phenotype [40), but these studies did not investigate the significance of autophagy in patient-derived CAFs. It is not surprising that CAFs rely on autophagy-mediated pro-survival mechanisms as these fibroblasts are under tremendous metabolic demands and the addition of autophagy allows for a tolerance to higher energy stress while increasing bioavailability of biosynthesis materials. A few reports have shown an increase in autophagy in fibroblasts (mouse MEF or skin fibroblasts) when exposed to cancer cells [
41,
42], suggesting that this process is not only exploited by cancer cells but also frequently adopted by stroma fibroblasts.
Sequestosome 1 (A.K.A. p62) has been shown to be another factor that can activate Nrf2 by directly binding to Keap1 and targeting Keap1 for selective autophagic degradation [
24,
25]. It is overexpressed in many types of human cancers including breast cancer [
43,
44] and its expression correlates with poor prognosis in patients with triple-negative breast cancer [
45]. Further supporting a role of p62 in oncogenesis and as a potential CAF target, a p62-encoding DNA vaccine exhibited strong antitumor and anti-metastasis activity in four mouse tumor models, including mammary carcinoma [
46]. This observation is in contrast to some studies reporting a down regulation of p62 expression in prostate tumor-associated stroma [
47], experimentally activated hepatic stellate cells [
48] and skin CAFs [
49]. In this study, we clearly observed an increase in the levels of p62 protein expression (both total and S351-phosphorylated p62) in patient-derived CAFs. This S351-phosphorylation site was shown to be critical in regulating selective autophagy and Nrf2 activation [
50]. Furthermore, Oncomine analysis also showed a significant upregulation of p62 gene expression in breast tumor stroma vs. normal stroma as shown in Fig.
8G [
29] as well as in another breast tumor stroma microarray analysis [
51]. In addition, Kang et al.[
52] have recently showed that the p62-Nrf2 axis contributes to fibroblast activation and tumor progression of lung cancer. Together, these contradictory observations likely reflect the CAF heterogeneity and imply a context-dependent function of p62 in CAFs.
Although Nrf2 is well known to be a promising anti-cancer target, no FDA-approved drugs targeting Nrf2 activity in cancer have been realized to date [
17]. The plethora of downstream targets (more than 100 identified so far) of this master regulator of cellular stress response also complicates the specific use of Nrf2-targeting approaches for cancers. Our data presented here suggest that Birc5 is a key downstream modulator of Nrf2 that promotes CAF phenotype and their survival. This is not surprising, as Birc5 has recently been shown to prevent excessive autophagy as a pro-survival mechanism in breast cancer [
53,
54]. Moreover, Birc5 is overexpressed in aggressive cancers [
55] where its presence correlates with increased resistance to chemotherapy [
56] and irradiation [
57]. We suspect that Nrf2 directly participates in transactivating gene expression of Birc5 in CAFs. This is based on an analysis of transcription binding motifs using the Eukaryote Promoter Database (JASPAR Core matrix profile MA0150.1). There are 2 predicted Nrf2 binding sites in the − 2 K upstream of TSS. Importantly, these sites also overlap with the MafG binding motifs (− 1227 and -414). Nrf2 does not act alone but can form heterodimers with MafG to co-occupy functional Nrf2 binding sites to participate in the transcriptional activation [
58]. Further supporting this, in a lung tissue transcriptome analysis,
Birc5 was found to be one of the genes downregulated in Nrf2-/- mice when exposed to cigarette smoke [
59]. The direct involvement of Nrf2 on Birc5 transcription was, however, not confirmed in this study. Additional studies will be needed in establishing the Nox4-Nrf2-Birc5 axis as a mediator of CAF survival to provide the rationale to exploit this redox vulnerability of CAFs.
There is a recent report showing that tissue-specific deletion of Nox4 gene promoted tumor formation in carcinogen-induced colorectal cancers and fibrosarcomas (61). The authors demonstrated that Nox4-generated ROS are critical for inducing DNA damage response upon exposure to carcinogens and that depleting Nox4 promoted genetic instability and tumor initiation. This study further highlights the biphasic effects of ROS and their differential roles during various stages of tumor initiation, malignant transformation, and progression. Based on the fact that multiple studies have shown the anti-tumor effect of Nox4 targeting approaches in established tumors (62,63), including breast cancers (64,65), in addition to the evidence presented in the present study, we strongly believe that Nox4 remains a promising CAF target for solid tumors and its role in breast cancer warrants more studies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.