Increased ROS production in non-polarized mammary epithelial cells induces monocyte infiltration in 3D culture

Tumors have been described as wounds that do not heal (Dvorak, 1986) because the two share some common features, such as chronic inflammation and disruption of normal tissue structure. Loss of cell polarity, an early cellular event during breast cancer development, is accompanied by substantial changes in the tumor microenvironment (Beliveau et al., 2010; Nelson and Bissell, 2006). Infiltration and differentiation of macrophages, the most abundant immune-related stromal cells in the tumor microenvironment (Allavena et al., 2008), correlate with poor prognosis in breast cancer patients, and drive cancer development and progression by inducing chronic inflammation (DeNardo et al., 2011; Medrek et al., 2012; Zhang et al., 2013). However, the connection between disruption of tissue polarity and macrophage infiltration remains to be determined.

Cellular polarity is a fundamental property of normal epithelial tissue (Gudjonsson et al., 2002; Martin-Belmonte and Perez-Moreno, 2012). It has been shown that polarized tissue structure is crucial for maintaining tissue integrity and tissue-specific function (Inman and Bissell, 2010; Liu et al., 2004; Xu et al., 2009; Zeitler et al., 2004). In addition, cell adhesion and tight-junctions in polarized cells spatially segregate ligands and their receptors in epithelial sheets so that the activation of cell surface receptors and cell proliferation can be tightly and properly controlled (Ewald et al., 2012; Xu et al., 2009, 2010). Disruption of epithelial cell polarity leads to loss of tissue organization, increased susceptibility to carcinogenesis and eventual formation of tumors (Bilder, 2004; Bissell and Bilder, 2003; Weaver et al., 2002). A number of polarity regulators, including scribble, discs-large, and lethal giant larvae (which each have a number of different isoforms in mammals) have been identified as tumor suppressors (Bissell and Bilder, 2003; St Johnston and Ahringer, 2010; Tepass, 2012), indicating that the malignant transformation is strongly linked with loss of polarized tissue architecture (Bunker et al., 2015; Lee and Vasioukhin, 2008). Using the 3D culture model with the S1 and T4-2 cells lines of a tumor progression series, we and others have shown that disruption of tissue polarity is associated with remodeling of the tumor microenvironment (Becker-Weimann et al., 2013; Xiong et al., 2012).

Reactive oxygen species (ROS) such as hydrogen peroxide, superoxide and the hydroxyl radical, are byproducts of normal metabolism through the electron transport chain (Chance et al., 1979), and phagocyte-derived oxidants serve a protective function with respect to oxidant injury under normal physiological conditions (Babior, 2000). However, ROS and associated oxidative stress are considered a driving force for cancer development and progression (Cairns et al., 2011; Kamp et al., 2011). ROS induce oxidative damages in DNA, lipids, proteins and other cellular components and the damage is sufficient to induce malignant transformation (Cooke et al., 2003; Valko et al., 2006). In addition, ROS act as secondary signaling intermediates that control cellular phenotype (Droge, 2002; Finkel, 2003; Miao and St Clair, 2009; Puri et al., 2002). For example, matrix metalloproteinase-3 (MMP-3)-induced ROS are both necessary and sufficient to induce the epithelial–mesenchymal transition phenotypic alterations in target mammary epithelial cells (Radisky et al., 2005).

Chronic inflammation is accompanied by increased production of tissue ROS, which in turn can promote the inflammatory response by inducing transcription factors such as nuclear factor κB (NF-κB) and activator protein 1  (AP-1) (Kroncke, 2003; Li and Karin, 1999; Reuter et al., 2010). Constitutive activation of NF-κB in cancer cells leads to a variety of pro-tumorigenic effects (Hoesel and Schmid, 2013; Karin, 2009). By contrast, the NF-κB pathway in immune cells is a central regulator of the immune responses and is required for tumor-immunosurveillance. Given the dual and differing roles of NF-κB in cancer development, a more-complete understanding of NF-κB pathway activation in cancer cells is crucial in order to specifically target this pathway in cancer tissue.

Our findings demonstrate that loss of polarity in mammary acini is associated with increased ROS production. Reducing ROS levels reprograms the non-polarized T4-2 cells to form polarized spheroids in 3D culture, whereas increased ROS production in polarized S1 cells is sufficient to disrupt basal polarity. Using a 3D co-culture system, we also found that elevated ROS levels in non-polarized cells induced monocyte infiltration, NF-κB activation and cytokine expression. These results established a potential link between tissue polarity, ROS generation and inflammation.

ROS promote cancer initiation and progression (Moller and Wallin, 1998), but the function of ROS in regulating tissue polarity has not been determined. We first asked whether disruption of the polarized acinar structure is associated with increased ROS production. The 3D culture system of the human breast cancer progression series HMT-3522 (the non-malignant S1 cell line and its malignant counterpart T4-2) is a physiologically relevant model to investigate epithelial cell polarity and cancer progression (Kenny et al., 2007; Lelievre et al., 1998; Petersen et al., 1992; Weaver et al., 1997). The non-malignant S1 cells, in response to cues from 3D laminin-rich basement membrane (lrBM), form phenotypically normal, polarized and growth-arrested structures similar to those found in vivo in the terminal ductal lobular units of the breast (Fig. 1A). In contrast, the tumorigenic T4-2 cells accumulate additional and numerous genomic alterations during transformation, and these cells form large non-polarized structures. Treatment with the epidermal growth factor receptor inhibitor tyrphostin AG 1478 (TYR), MMP inhibitor GM6001, or phosphoinositide 3-kinase (PI3K) inhibitor LY294002 reprograms T4-2 cells to reestablish basal polarity (T4R) (Fig. 1A). To evaluate ROS levels, non-polarized T4-2 and polarized S1 and T4R cells in 3D Matrigel were stained with CellROX Deep Red, followed by live-cell imaging and quantification. We found higher levels of ROS in the non-polarized T4-2 cells compared to polarized S1 and T4R cells (Fig. 1B,C). These results suggest that disruption of the polarized acinar structure is associated with increased ROS production.

Disruption of acinar structure during cancer development is accompanied by loss of tissue polarity and increased cell proliferation. It has been shown that Akt1, an isoform of Akt, and RAC1 are key regulators of cellular proliferation and tissue polarity in the non-polarized mammary epithelial cells, respectively (Liu et al., 2004). We confirmed that introduction of RAC1 L61 (which has a Q61L mutation and is constitutively active) into T4-2 cells inhibited the restoration of polarity in response to TYR (Fig. 2A–C), but had little effect on cell proliferation (Fig. 2D,E). Importantly, we found that expression of RAC1 L61 significantly elevated ROS levels in TYR-treated T4-2 cells (Fig. 2F,G). Similar results were obtained in S1 cells (Fig. 2H,I; Fig. S1) and MCF-10A cells (Fig. 2J,K; Fig. S2). However, expression of the dominant-active Akt1 (Myr-Akt) only slightly increased ROS levels in TYR-treated T4-2 cells, although cell proliferation was significantly increased (Fig. 2D–G). These results suggest that increased ROS levels are associated with disruption of cell polarity and that RAC1 is a major regulator of ROS production in non-polarized mammary epithelial cells.

To investigate whether increased ROS levels contribute to the disruption of polarized acinar structures in 3D culture, we lowered ROS in T4-2 cells using antioxidant N-acetyl-L-cysteine (NAC) or the reduced form of L-glutathione (GSH). NAC and GSH neutralize intracellular and extracellular ROS, providing a means to investigate the role of ROS in biological processes. T4-2 cells were cultured in Matrigel with NAC (4 mM) or GSH (4 mM) for 4 days to allow spheroid formation. The treatment significantly reduced intracellular levels of ROS (Fig. 3A,B). Immunofluorescence and confocal analysis showed that T4-2 cells formed polarized spheroids in the presence of NAC or GSH (Fig. 3C,D). The size of the T4-2 colony was significantly reduced in the NAC- or GSH-treated cells. Ki-67 staining showed that scavenging ROS attenuated T4-2 cell proliferation (Fig. 3E,F).

Glucose oxidase is an oxi-reductase, catalyzing the conversion of glucose into glucolactone, which is further converted into glucuronic acid and H₂O₂. It has been shown that treatment with glucose oxidase increases ROS levels in high-glucose medium (Kumar and Sitasawad, 2009; Wang et al., 2012). To determine whether increased ROS levels are sufficient to disrupt tissue polarity in 3D culture, we treated S1 cells with glucose oxidase. Treatment increased ROS levels (Fig. 4A,B), disrupted basal polarity (Fig. 4C,D), and induced cell proliferation in S1 cells (Fig. 4E,F). The same treatment also disrupted basal polarity in MCF10A cells (Fig. 4G–J) and TYR-treated T4-2 cells (Fig. S3). Importantly, reducing ROS levels with NAC (Fig. 4A,B) rescued cell polarization and inhibited cell proliferation in the glucose-oxidase-treated S1 cells (Fig. 4C–F). Collectively, these results indicate that increased ROS production is sufficient to impair polarized acinar morphogenesis.

Monocytes in peripheral circulation are recruited to tumor sites by chemotactic cytokines from tumor tissue, and are then differentiated to tumor-associated macrophages (Murdoch et al., 2004). Macrophage infiltration and differentiation has also been detected in ductal carcinoma in situ (Sharma et al., 2010; Wang et al., 1989). Macrophages also accumulate around the terminal end buds of mammary glands rather than near polarized ductal epithelial cells (Gouon-Evans et al., 2000; Ingman et al., 2006). One common feature shared by the terminal end buds and mammary tumor tissue is the presence of multilayered non-polarized epithelial cells. Therefore, we asked whether non-polarized mammary epithelial cells induce recruitment of monocytes. To recapitulate the interaction of mammary epithelial cells with monocytes during breast tumorigenesis, we developed a 3D co-culture assay (Fig. 5A). S1, T4-2, and T4R cells were cultured in 3D Matrigel for 72 h, followed by addition of THP-1 cells into the culture medium and incubation for an additional 24–48 h. A substantial number of THP-1 (GFP-expressing) cells invaded into the Matrigel and adhered to the T4-2 colonies, but very few THP-1 cells adhered to S1 or T4R cells (Fig. 5B,C). THP-1 cells are derived from a patient with acute monocytic leukemia, and this cell line has been used to study the function and differentiation of monocytes (Auwerx, 1991; Tsuchiya et al., 1982, 1980). To confirm our finding using primary human monocytes, CD14-positive monocytes (Lonza) were labeled with Cell Tracker Green (CMFDA) and co-cultured with S1, T4-2 and T4R cells as outlined above. As with the monocyte cell line, we found that these primary monocytes were recruited to the T4-2 colonies after 12 h of co-culture, whereas few monocytes migrated to S1 or T4R cells (Fig. 5D,E).

To decouple the effect of cell polarity and cell proliferation on monocyte infiltration, we performed co-culture experiments using RAC1 L61- and Myr-Akt-expressing T4-2 cells. We found that RAC1 L61 but not Myr-Akt rescued monocyte infiltration in TYR-treated T4-2 cells (Fig. 5F,G). These results suggest that monocyte infiltration is associated with disruption of polarity of mammary epithelial cells.

Next, we asked whether reducing ROS levels in T4-2 cells suppressed monocyte recruitment in the 3D co-culture assay. We found that treatment with NAC or GSH in T4-2 cells blocked THP-1 (Fig. 6A,B) and primary monocyte recruitment (Fig. 6C,D). We showed that treatment with glucose oxidase elevated ROS levels in S1 cells. To determine whether increased ROS generation is sufficient to induce THP-1 infiltration, we performed 3D co-culture with control and glucose-oxidase-treated S1 cells. We found that increased ROS generation, induced by glucose oxidase, significantly enhanced THP-1 recruitment (Fig. 6E,F). More interestingly, NAC prevented THP-1 infiltration induced by glucose oxidase (Fig. 6E,F). These results suggest that increased ROS production in non-polarized mammary epithelial cells induces monocyte infiltration during cancer progression.

Monocyte infiltration and differentiation is largely regulated by cytokines (Coussens and Werb, 2002; Nathan, 2002). To determine whether loss of tissue polarity enhances cytokine expression and subsequently promotes monocyte infiltration, we compared expression of cytokine genes in non-polarized T4-2 cells with expression in polarized S1 and T4R cells. Quantitative real-time (q)RT-PCR demonstrated that the cytokines interleukin (IL)-24, CXCL1, CXCL3 and CXCL8 were significantly upregulated in non-polarized T4-2 cells compared to polarized S1 and T4R cells (Fig. 7A). Although expression of IL-6 and IL-32 was upregulated in T4-2 cells compared to S1 cells, mRNA levels of IL-6 and IL-32 were not significantly altered in T4R cells compared to T4-2 cells (Fig. 7A). We also found that disruption of tissue polarity with constitutively activated RAC1 in TYR-treated T4-2 cells significantly enhanced the cytokine mRNA expression, whereas dominant-activated Akt only slightly increased the expression (Fig. 7B). To examine whether the secretion of cytokines is increased in non-polarized cells, we evaluated protein levels of IL-6 and CXCL-1 by examining conditioned medium collected from S1 and T4-2 cells with enzyme-linked immunosorbent assay (ELISA). We found that secretion of IL-6 and CXCL-1 was significantly increased in the T4-2 cells compared to the S1 cells (Fig. 7C).

We have recently shown that NF-κB is activated in the non-polarized T4-2 cells (Becker-Weimann et al., 2013). We measured luciferase activity in T4-2 cells stably transfected with the 3X-κB-L reporter plasmid, and results showed that NF-κB activity was significantly reduced in T4R cells (Fig. 7D). Nuclear levels of the NF-κB p65 subunit (also known as RELA) also decreased in T4R cells compared to control T4-2 cells (Fig. 7E). NF-κB is a redox-sensitive transcription factor regulating cytokine expression and inflammation (Baldwin, 1996; Morgan and Liu, 2011). We found that treatment with the NF-κB inhibitor Wedelolactone in T4-2 cells significantly reduced mRNA levels of cytokines (Fig. 7F). We also showed that RAC1 L61 but not Myr-Akt induced the NF-κB-driven luciferase activity in TYR-treated T4-2 cells (Fig. 7G). These results suggest that activation of the NF-κB cascade in non-polarized T4-2 cells at least partially mediates induction of inflammation-related cytokine expression.

To determine whether induction of cytokine expression is regulated by ROS in non-polarized mammary epithelial cells, we assessed the cytokine expression in control and NAC-treated T4-2 cells. We found that NAC treatment significantly inhibited mRNA expression of IL-6, IL-24, CXCL1, CXCL2, CXCL3 and CXCL8 in T4-2 cells (Fig. 8A). In addition, secretion of IL-6 and CXCL1 into the culture medium was significantly reduced in NAC-treated T4-2 cells (Fig. 8B). To determine whether ROS are sufficient to induce cytokine expression in mammary epithelial cells, ROS production was induced in S1 cells through glucose oxidase treatment. Increased ROS production significantly enhanced expression of cytokine genes (Fig. 8C), and Wedelolactone treatment repressed the cytokine expression induced by glucose oxidase (Fig. 8C). Next, we determined whether NAC or GSH treatment inhibits NF-κB activity in T4-2 cells. Luciferase reporter analysis data showed that the transactivation activities of NF-κB were significantly inhibited in T4-2 cells when ROS production was inhibited by NAC or GSH (Fig. 8D). Immunoblot analysis revealed that nuclear p65 protein levels were decreased in T4-2 cells when ROS were reduced by NAC (Fig. 8E). These results suggest that scavenging of ROS inhibits cytokine expression through downregulation of NF-κB activities.

Disruption of the polarized epithelial tissue structure occurs at the early stage of breast cancer development (Bissell et al., 2002; Boghaert et al., 2014; Debnath and Brugge, 2005; Muthuswamy and Xue, 2012) and is associated with substantial remodeling of the tissue microenvironment. Breast cancer is characterized by tumor-associated macrophage infiltration (DeNardo and Coussens, 2007; Leek et al., 1996; Mahmoud et al., 2012). Monocytes in the peripheral circulation are recruited to the tumor site by chemotactic cytokines from tumor tissue, and are then differentiated to tumor-associated macrophages (Murdoch et al., 2004). Macrophage accumulation has been detected in ductal carcinoma in situ (Sharma et al., 2010; Wang et al., 1989). Interestingly, macrophages also accumulate around the terminal end buds of mammary glands rather than near the polarized ductal epithelial cells (Gouon-Evans et al., 2000; Ingman et al., 2006). Mammary epithelial cells in the terminal end bud are multilayer and non-polarized (Ewald et al., 2012; Rejon and McCaffrey, 2015). These results suggest that macrophage infiltration is associated with loss of tissue polarity. Using the 3D culture model of mammary tumor progression, we demonstrate that loss of tissue polarity is associated with ROS generation, increased cytokine expression and monocyte infiltration (Fig. 5C), providing new insight into the regulation of cancer-associated inflammation.

Deregulated ROS generation contributes to development of cancer and inflammation (Colotta et al., 2009; Hanahan and Weinberg, 2011), but its function and regulation in the disruption of tissue polarity has not been determined. Here, we demonstrate that ROS levels are significantly upregulated in non-polarized breast cancer cells compared to in polarized non-malignant cells (Fig. 1). Activation of RAC1 leads to the disruption of cell polarity in mammary epithelial cells. We show that introduction of a constitutively activated RAC1 is sufficient to induce ROS generation in mammary epithelial cells. It has been shown that activated RAC1 binds to and forms a complex with NOX1. NOX1 is a homolog of the phagocyte NADPH-oxidase component gp91^phox (also known as CYBB). NOX1 has the capacity to transport electrons across the plasma membrane and to generate superoxide and other downstream ROS. The binding of RAC1 induces activation of NOX1 and triggers the NOX1-dependent ROS generation (Cheng et al., 2006). These results indicate that RAC1 is a major regulator that integrates non-polarized tissue formation and ROS production.

ROS play crucial roles in cancer development and progression (Karin, 2006; Waris and Ahsan, 2006; Wiseman and Halliwell, 1996). We identified a new function of ROS in regulating the basal polarity of mammary epithelial cells. We show that increased ROS production is necessary and sufficient to disrupt the polarized acinar structure. ROS act as secondary signaling intermediates modulating multiple pathways (Droge, 2002; Finkel, 2011), and many of these pathways are involved in cell polarization. For instance, increased ROS production enhances the activity of FAK (also known as PTK2), an important regulator of cell polarity (Chiarugi et al., 2003; Vepa et al., 1999). In addition, H₂O₂ induces activation of both myosin light chain kinase (MLCK, also known as MYLK) and protein kinase C (PKC) in endothelial cells (Zhao and Davis, 1998). However, the downstream target of ROS that induces the disruption of polarized acinar structure in mammary epithelial cells is presently unknown.

Our findings demonstrate that ROS are important mediators of the cancer-cell–monocyte interaction in 3D co-culture. We show that ROS induce expression of multiple cytokine genes in non-polarized malignant cells. These cytokines might promote recruitment and infiltration of monocytes in 3D culture. We showed previously that disruption of polarized acinar structure induces the NF-κB pathway (Becker-Weimann et al., 2013). The constitutive activation of NF-κB impairs polarized acinar morphogenesis (Becker-Weimann et al., 2013). NF-κB is a redox-regulated sensor for oxidative stress and is activated by hydrogen peroxide (Bubici et al., 2006; Kim et al., 2014; Morgan and Liu, 2011). We found that reduction of ROS levels in T4-2 cells significantly inhibits NF-κB activity, indicating that the induction of NF-κB pathway in non-polarized mammary epithelial cells is ROS dependent. The primary regulation of the NF-κB pathway is through the association of NF-κB complexes with their inhibitor, IκB proteins; it binds to the p65 or p50 (encoded by NFKB1) subunit and inhibits nuclear translocation. ROS induce NF-κB activation through a modulation of DNA-binding activity or upstream NF-κB signaling pathways (Gloire et al., 2006; Morgan and Liu, 2011). It has been shown that ROS induce Src activation, and subsequently enhances PKD and IKKβ activation (Gloire et al., 2006; Morgan and Liu, 2011). The NF-κB pathway is a crucial regulator of cytokine expression and inflammation (Karin, 2009; Schreck et al., 1991). We also found that inhibition of the NF-κB pathway represses ROS-induced cytokine expression. These results suggest that ROS enhance cytokine expression and monocyte infiltration by inducing the NF-κB pathway in non-polarized breast cancer cells.

In summary, the finding that ROS contribute to the disruption of tissue polarity and monocyte infiltration significantly advances our understanding of cancer-associated inflammation. However, additional in vivo studies are needed to corroborate the importance of ROS in the regulation of tissue polarity as well as the central role of polarity in monocyte infiltration and inflammation.

Dulbecco's modified Eagle's medium (DMEM) with F12 and RPMI 1640 media were obtained from Sigma-Aldrich (St Louis, MO). Culture additives fetal bovine serum (FBS), transferrin, β-estradiol, and prolactin were obtained from Sigma; insulin was from Boehringer Mannheim (Indianapolis, IN), and sodium selenite, hydrocortisone and epidermal growth factor (EGF) were from BD Biosciences (Bedford, MA). N-acetyl-L-cysteine (NAC; A7250), reduced L-glutathione (GSH; G6013), Wedelolactone (Wede; W4016), G418 (A1720), D-(+)-glucose (G7021), glucose oxidase (G7141) were obtained from Sigma-Aldrich. DAPI was from Molecular Probes (D1306; Eugene, OR). Matrigel was obtained from BD Biosciences (354230; San Diego, CA).

The antibodies used were obtained from the following companies: monoclonal rat anti-integrin α6 (555734, 1:500 dilution) from BD Biosciences; monoclonal rabbit anti-Ki-67 from Spring Bioscience (M3060; 1:250 dilution; Pleasanton, CA); Alexa-Fluor-488-conjugated goat anti-rat-IgG and Alexa-Fluor-488-conjugated goat anti-rabbit-IgG from Invitrogen (A11006, A11034; 1:500 dilution; Eugene, OR); polyclonal anti-p65 from Santa Cruz Biotechnology (sc-109; 1:1000 dilution; Santa Cruz, CA); monoclonal mouse anti-FLAG (F3165, 1:1000 dilution) from Sigma-Aldrich; polyclonal anti-lamin A/C (LAC; 1:1000 dilution; sc-20681) from Santa Cruz Biotechnology; monoclonal mouse anti-tubulin from Abcam (ab7291; 1:1000 dilution; Cambridge, MA); and Dylight-680-conjugated goat anti-rabbit-IgG from Thermo Scientific (35569; 1:1000 dilution; Rockford, IL).

The HMT-3522 mammary epithelial cells (a kind gift from Mina Bissell, Lawrence Berkeley National Laboratory, Berkeley, CA) were cultured in T4 medium consisting of DMEM/F12 with 250 ng/ml insulin, 10 μg/ml transferrin, 2.6 ng/ml sodium selenite, 10⁻¹⁰ M β-estradiol, 1.4 μM hydrocortisone and 5 μg/ml prolactin. The non-neoplastic S1 cells were cultured in T4 medium with 10 ng/ml EGF. The malignant T4-2 cells were cultured on collagen type I pre-coated dishes in the absence of EGF. T4 medium was routinely changed every 2–3 days. Cells were grown in a humidified incubator at 37°C with 5% CO₂.

For the 3D laminin-rich extracellular matrix (3D lrECM) ‘On-Top’ assay, 250 μl of cell suspension in T4 medium was seeded on top of a thin gel of Engelbreth–Holm–Swarm (EHS) tumor extract (Matrigel; 125 μl) in a 24-well plate. S1 cells were seeded at a density of 5.0×10⁴ cells/cm² and T4-2 cell lines were seeded at 4.0×10⁴ cells/cm². This cell suspension, on top of Matrigel, was incubated at 37°C for 60 min. To reverse the malignant phenotypes of T4-2 (T4R) cells, 250 μl T4 medium containing 10% Matrigel with different inhibitors were added to the culture with a final concentration of 100 nM (Tyrphostin AG1478, Sigma), 10 μM (LY 294002, Sigma) or 50 μM (GM6001, Calbiochem). Media with the inhibitors were changed every 2 days in the next 4 days.

THP-1 cells (a kind gift from Philip A. Kern, University of Kentucky, Lexington, KY) were cultured in RPMI 1640 medium with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Sigma), supplemented with 2-mercaptoethanol (0.05 mM, MP Biomedicals). The PCDH–GFP construct (System Biosciences) was transfected into THP-1 cells, and then selected over 2 weeks with puromycin until more than 90% of the cells were GFP positive.

MCF-10A cells (a kind gift from Michael W. Kilgore, University of Kentucky, Lexington, KY) were cultured in 2D or 3D as previously described (Debnath et al., 2003).

The cells were cultured in 3D Matrigel for 3 days using the ‘On Top’ protocol in a 24-well plate as described above. To remove TYR, NAC, GSH or glucose oxidase in 3D culture, the medium was replaced with T4 medium (chemical free) and washed twice before seeding THP-1 (GFP expressing) cells (40,000 cells/well). Bright-field and fluorescence images were taken 24–48 h later with an Eclipse Ti workstation (Nikon). The pictures were taken under fixed exposure conditions.

Human peripheral blood CD14⁺ monocyte cells (Lonza, 2W-400C) were labeled with Cell Tracker Green CMFDA (Molecular Probes, C2925) according to the manufacturer's instructions. Briefly, CD14⁺ monocyte cells were harvested by centrifugation and resuspended in a 15-ml tube with 3 ml pre-warmed serum-free RPMI 1640 medium containing 10 μM CMFDA. The labeled CD14⁺ monocyte cells were placed on top of mammary epithelial cells in 3D culture (24-well plate, 40,000 cells/well). Bright-field and fluorescence images were taken 12 h later with an Eclipse Ti workstation (Nikon).

FLAG-tagged RAC1 L61 and FLAG-tagged Myr-Akt Δ4-129 were amplified from constructs pcDNA3-EGFP-RAC1-Q61L (Addgene plasmid # 12981) and myrAkt delta4-129 (Addgene plasmid # 10841) respectively, and cloned into the pCDH-CMV-MCS-EF1-Puro (PCDH) expression vectors (System Biosciences). 293 FT cells were transfected with PCDH, PCDH-RAC1 L61 or PCDH-Myr-Akt, plus packaging lentivector using the FuGENE HD Transfection Reagent to produce lentivirus stocks. HMT-3522 mammary epithelial cells or MCF-10A cells were infected with lentivirus and selected by puromycin after 48 h of infection.

ROS measurement was conducted using a live-cell imaging method with CellROX Deep Red reagent (Grinberg et al., 2013, 2012). Cells cultured in lrECM gel were smeared on slides. CellROX reagent at a final concentration of 5 μM in T4 medium or MCF-10A medium was added and the slides were incubated for 30 min at 37°C. Fluorescence images were acquired using an Eclipse 80i microscope (Nikon) with 532–587-nm excitation and 608–683-nm emission filters. The pictures were taken under fixed exposure conditions. Fluorescence intensity was quantitated using the NIS-Elements AR software (Nikon).

Cells cultured in lrECM gel were smeared on slides, dried briefly, fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. After blocking with 10% goat serum at room temperature for 60 min, specimens were incubated with primary antibody (anti-Integrin α6 antibody, 1:500 dilution; anti-Ki-67 antibody, 1:250 dilution) overnight at 4°C. Specimens were washed three times with immunofluorescence buffer (130 mM NaCl, 3.5 mM NaH₂PO₄, 7 mM Na₂HPO₄, pH 7.2, 7.7 mM NaN₃, 0.1% bovine serum albumin, 0.2% Triton X-100, 0.05% Tween-20) and incubated in fluorochrome-conjugated secondary antibody (1:500 dilution) for 1 h at room temperature in the dark. Stained samples were imaged with an Eclipse 80i microscope (Nikon) or an Olympus FV1000 confocal microscope.

Cell proliferation was assessed by quantification of proportion of cells with Ki-67-positive staining. The number of nuclei positive for Ki-67 antigen was counted and divided by the total number of nuclei, which were stained with DAPI.

Total RNA was extracted from cells using Trizol reagent (Invitrogen). cDNA was synthesized by using a SuperScript First Strand Synthesis kit (Invitrogen) from 1.0 μg RNA samples. qRT-PCR reactions were carried out using SYBR Green PCR master mix reagents (Thermo Scientific) on an ABI StepOnePlus Real-Time PCR System (Applied Biosystems). Thermal cycling was conducted at 95°C for 30 s, followed by 40 cycles of amplification at 95°C for 5 s, 55°C for 30 s and 72°C for 15 s. The relative quantification of gene expression for each sample was analyzed by the ΔCt method. The following primers were used to amplify: IL-6, 5′-ACTCACCTCTTCAGAACGAATTG-3′ and 5′-CCATCTTTGGAAGGTTCAGGTTG-3′; IL-8, 5′-TTTTGCCAAGGAGTGCTAAAGA-3′ and 5′-AACCCTCTGCACCCAGTTTTC-3′; IL-24, 5′-TGTGAAAGACACTATGCAAGCTC-3′ and 5′-GTGACACGATGAGAACAAAGTTG-3′; IL-32, 5′-GCTTCCCGAAGGTCCTCTCT-3′ and 5′-ATAAGCCGCCACTGTCTCCA-3′; CXCL1, 5′-GCTCTTCCGCTCCTCTCACA-3′ and 5′-GGGGACTTCACGTTCACACTT-3′; CXCL2, 5′-TCGCACAGCCGCTCGAA-3′ and 5′-GGGGACTTCACCTTCACACTTTG-3′; CXCL3, 5′-TGAATGTAAGGTCCCCCGGA-3′ and 5′-CACCCTGCAGGAAGTGTCAA-3′; CXCL6, 5′-GACAGAGCTGCGTTGCACTT-3′ and 5′-CGGGTCCAGACAAACTTGCTT-3′; 18S rRNA, 5′-ACCTGGTTGATCCTGCCAGT-3′ and 5′-CTGACCGGGTTGGTTTTGAT-3′.

After HMT-3522 mammary epithelial cells were cultured in 3D lrECM gel for 3 days, culture medium was replaced with fresh T4 medium and incubated for an additional 48 h. Conditioned medium was collected, centrifuged at 4058 g for 5 min to pellet floating cells and debris. The supernatant was analyzed for secreted IL-6 (Biolegend, 430504) and CXCL1 (Abcam, ab100530) by ELISA assay according to the manufacturer's instructions.

T4-2 cells were co-transfected with NF-κB-luciferase (Addgene plasmid # 26699) and pNeo plasmid (1:10) according to previously described methods (Becker-Weimann et al., 2013; Mitchell and Sugden, 1995). Stably transfected clones were isolated by G418 selection and selected by luciferase reporter assay (50 μg/ml, 2 weeks), and then one positive clone was stably transfected with virus containing Rac1 L61 or Myr-Akt expression constructs. Following treatment, luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) and a Sirius luminometer (Berthold Detection Systems, Pforzheim, Germany).

Western blotting was performed as previously described (Xiong et al., 2012). Protein expression was visualized using a LI-COR Odyssey infra-red imaging system at a 700-nm wavelength, and quantitative densitometry was performed using Image Studio Lite software.

Inferential statistics were used to compare data sets from different experimental groups and reported data are the mean±s.e.m. Student's t-tests (two groups) and one-way ANOVA (three or more groups) were used to determine the significant differences between means, and were performed with SigmaPlot 12.3 (Systat Software Inc., San Jose, CA). The minimum statistical significance was set at P<0.05.

The authors are grateful to Mina J. Bissell for her support in this study and providing HMT-3522 cell lines. The authors thank the Research Communications Office at the Markey Cancer Center for assistance with manuscript preparation.