Phenotypic Switch in Blood: Effects of Pro-Inflammatory Cytokines on Breast Cancer Cell Aggregation and Adhesion

Hematogeneous metastasis can occur via a cascade of circulating tumor cell adhesion events to the endothelial lining of the vasculature, i.e. the metastatic cascade. Interestingly, the pro-inflammatory cytokines IL-6 and TNF-α, which play an important role in potentiating the inflammatory cascade, are significantly elevated in metastatic breast cancer (BCa) patients. Despite their high metastatic potential, human breast carcinoma cells MDA-MB-231 lack interactions with E-selectin functionalized surfaces under physiological shear stresses. We hypothesized that human plasma, 3-D tumor spheroid culture, and cytokine-supplemented culture media could induce a phenotypic switch that allows BCa cells to interact with E-selectin coated surfaces under physiological flow. Flow cytometry, immunofluorescence imaging, and flow-based cell adhesion assay were utilized to investigate the phenotypic changes of MDA-MB-231 cells with various treatments. Our results indicate that plasma, IL-6, and TNF-α promote breast cancer cell growth as aggregates and induce adhesive recruitment of BCa cells on E-selectin coated surfaces under flow. 3-D tumor spheroid culture exhibits the most significant increases in the interactions between BCa and E-selectin coated surfaces by upregulating CD44V4 and sLex expression. Furthermore, we show that IL-6 and TNF-α concentrations in blood may regulate the recruitment of BCa cells to the inflamed endothelium. Finally, we propose a mechanism that could explain the invasiveness of ‘triple-negative’ breast cancer cell line MDA-MB-231 via a positive feedback loop of IL-6 secretion and maintenance. Taken together, our results suggest that therapeutic approaches targeting cytokine receptors and adhesion molecules on cancer cells may potentially reduce metastatic load and improve current cancer treatments.


Introduction
Cancer mortality is predominantly caused by the dissemination of cancer cells from the primary tumor to distant organs where secondary sites are formed via a metastatic progression. Breast cancer, one of the most diagnosed forms of cancer, still causes high mortality rates due to the emergence of invasive, therapy-resistant cancer cells [1]. There are conflicting ideas on the nature of these invasive cells, from their population size within the primary tumor [2,3] to their methods of dissemination (lymphatic [4] or hematogeous [5]) and the process of metastasis within blood vessels [6,7,8].
One mode of cancer metastasis is through the bloodstream, which involves the escape of cancer cells from the primary tumor site into the circulatory system via intravasation. Circulating tumor cells (CTCs) can then interact and adhere to the endothelial lining of the vasculature through a series of receptor-mediated events, commonly referred to as the metastatic adhesion cascade. This cascade mimics the leukocyte adhesion cascade (reviewed in [5]) where the initial contact between cancer cells and the endothelium is facilitated by a family of endothelial adhesion molecules called selectins [9,10].
Firm adhesion of CTCs on the endothelium then allow extravasation and subsequent secondary tumor site formation.
There are several factors in the blood that are secreted by lymphocytes and macrophages known to facilitate the metastatic progression and potentially interact with CTCs within the bloodstream. In particular, interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-a) have been reported to be elevated in the blood serum of patients diagnosed with advanced stage breast tumor and correlate with an increased number and size of metastatic sites [11,12]. IL-6 and TNF-a have been shown to promote the growth and invasiveness of colon and prostate cancer epithelial cells in vitro and in vivo [13,14]. Studies have also shown that cancer cell chemokine receptors also play a role in determining the destination of metastases [15].
MDA-MB-231 is a well characterized metastatic breast cancer cell line isolated from pleural effusion of a patient with adenocarcinoma suffering from widespread metastasis, and is known as a 'basal' or 'triple-negative' cell line with stem cell-like or post-Epithelial-Mesenchymal Transition (EMT) features and fails to express estrogen receptors, progesterone receptors or human epidermal growth factor receptor 2 (HER2). MDA-MB-231 cells have been previously reported to express E-selectin ligands such as CD44V4 as well as selectin-binding moieties such as sialyl lewis x (sLe x ) [16,17]. Despite their expression of functional E-selectin ligands, we have observed that MDA-MB-231 cells do not adhesively interact with E-selectin coated microtube surfaces under flow, as reported here. In the present study, we investigated the effect of plasma, cytokine treatments, and 3-D spheroid culture conditions on altering the adhesion phenotype of MDA-MB-231 cells on E-selectin functionalized microtubes. We hypothesized that human plasma, tumor spheroid culture, and exposure to proinflammatory cytokines such as IL-6 and TNF-a could induce robust interactions between metastatic breast cancer cells in circulation and the inflamed endothelium. The findings of this study suggest new therapeutic approaches targeting specific cytokine(s) and their receptor(s) to help prevent the metastatic progression of breast cancer cells in transit.

Blood plasma triggers an adhesive phenotypic switch of breast cancer (BCa) cells on E-selectin coated surfaces under flow
When grown in 2-D monolayers with culture media, highly metastatic MDA-MB-231 breast cancer cells showed no adhesive interactions with E-selectin coated surfaces under flow conditions ( Figure 1a) despite the expression of the E-selectin ligand CD44V4 and the binding moiety sLe x (Figure 1b). This is contradictory to its reported high metastatic potential from in vivo studies where MDA-MB-231 cells were found to efficiently metastasize to distant organs through the bloodstream [18], where selectin-mediated tethering and rolling events have been shown to play important roles, as reviewed in [19].
A more physiological method of culturing cancer cells is presented where human blood plasma is used in combination with culture media. The plasma treatment was found to switch the MDA-MB-231 cells from non-interacting to adhesive (Figure 1a) where conventionally cultured MDA-MB-231 cells (control) exhibited no rolling cells on an E-selectin coated surface and plasma cultured cells exhibited strong rolling behavior with rolling velocities of 5.7260.32 mm and a 10-fold increase of cell flux compared to control cells ( Figure 1c). It was found that sLe x expression was upregulated on cells cultured in plasma compared to control cells (Figure 1b), which may contribute to the resulting adhesive phenotype.

Pro-inflammatory cytokines IL-6 and TNF-a induce adhesive recruitment of BCa cells and is blocked by the anti-inflammatory drug Metformin
Pro-inflammatory cytokines have been linked to poor prognosis in metastatic breast cancer patients with elevated IL-6 and TNF-a cytokine in their blood plasma [11,12]. Similar to culture in plasma, MDA-MB-231 cells cultured in IL-6 and TNF-a supplemented media also induced the adhesive phenotypic switch resulting in cell rolling velocities of 11.9060.28 mm/s and 14.2260.85 mm/s for IL-6 and TNF-a conditions, respectively, twice that of plasma and the combination of IL-6 and TNF-a treated cell rolling velocities (Figure 2a). Treatment with IL-6 and TNF-a at physiologically relevant concentrations such as 25 pg/ mL also induced rolling on E-selectin coated microtubes at rolling velocities of 15.4361.31 mm/s and 17.0362.21 mm/s, respectively. Furthermore, whereas TNF-a upregulated both CD44V4 and sLe x expression compared to control cells, IL-6 only upregulated sLe x (Figure 2b) suggesting the importance of increased sLe x moieties. Blocking the IL-6 receptor with anti-IL-6R monoclonal antibody effectively hindered the adhesive phenotype resulting in nearly abolished cell flux on the E-selectin surfaces (Figure 2a). Treating cells with an isotype control for the monoclonal antibody did not affect the IL-6 induced MDA-MB-231 cell rolling on Eselectin coated microtubes (data not shown). In some experiments, IL-6 treated MDA-MB-231 cells were incubated with neuraminidase for 1 hr to cleave sLe x from the cell surface. Interactions between MDA-MB-231 cells and the E-selectin coated surfaces were abolished after neuraminidase treatment, indicating that the rolling behavior of MDA-MB-231 cells is E-selectin: sLe x dependent ( Figure 2a). Interestingly, the anti-inflammatory drug Metformin (1,1-dimethylbiguanide hydrochloride), a frequently used drug to treat type 2 diabetes, showed similar effects to the IL-6R antibody. Metformin has been shown to target Stat3 and induce apoptosis in triple-negative breast cancers as well as significantly decrease serum IL-6 level in breast cancer patients [21,22].
Plasma, IL-6, and TNF-a promote breast cancer cell growth as aggregates

3-D tumor spheroid culture increases the interactions between BCa cells and E-selectin coated surfaces by upregulating CD44V4 and sLe x expression
Similar to elevated cytokine concentrations, the increased occurrence of tumor spheroids and circulating tumor microemboli has been linked to poor prognosis and increased metastatic potential [23,24].

Discussion
There is compelling evidence supporting the strong interplay between various types of inflammation and cancer progression suggested by numerous clinical and epidemiological studies [25,26]. The recruitment of tumor-associated immune cells during the inflammation that accompanies tumor progression, has been shown to promote tumor growth and contribute to angiogenesis, invasion, and metastasis [27]. Infiltration of these tumor-promoting immune cells as well as various inflammatory cytokines, particularly TNF-a and IL-6, has been found in the breast tumor microenvironment [28]. A clinical study reported elevated levels of pro-inflammatory cytokines in the serum of breast cancer patients when compared to healthy individuals [29]. As one of the well characterized 'triple-negative' breast cancer cell lines, MDA-MB-231 has been reported to also be mesenchymal stem cell-like [30]. Furthermore, 'triple-negative' breast cancers have been found to correlate with inflammatory states such as obesity and diabetes where pro-inflammatory cytokines, including IL-6, are highly expressed.
The lack of interaction between culture medium grown MDA-MB-231 cells with E-selectin coated surfaces, as observed in this study, was rather unexpected considering their reported high metastatic potential from both in vitro and in vivo studies. We have shown that human plasma, IL-6, and TNF-a supplemented culture conditions induced an adhesive phenotypic switch, allowing MDA-MB-231 cells to establish stable rolling on Eselectin coated surfaces under physiological flow. Plasma factors such as fibrinogen as well as other cytokines such as IL-1b and IFN-c have also been shown to promote cell adhesion [31,32,33]. In this study, we focused on IL-6 and TNF-a due to their association with poor prognosis for breast cancer patients. The concentration of cytokines used for our in vitro experiments was based on similar studies that have investigated the effect of these cytokines on invasiveness of breast cancer cell lines [34]. CD44V4, a major E-selectin ligand for breast cancer cells, was found to be upregulated on MDA-MB-231 cells treated with TNF-a. TNF-a has been reported to increase CD44 expression on ovarian cancer cells upon the activation of the c-Jun NH 2 -terminal kinase [35] and, although not in the context of cancer, affects the glycosylation and sulfation of various glycoproteins [36]. On the other hand, IL-6 has been shown to induce a significant increase in the expression of a1, 3/4-fucosyltransferases (FUT11 (fucosyltransferase 11 gene) and FUT3) as well as the amounts of sLe x and 6-sulfo-sLe x Effect of Cytokines on Breast Cancer Adhesion PLOS ONE | www.plosone.org epitopes in human bronchial mucosa [37]. Although CD44V4 is a major E-selectin ligand for breast cancer cells, there are other ligands such as MUC1 which play an active role in rolling and firm adhesion of breast cancer cells on inflamed endothelium, as reported in our recent study [38]. The flow cytometry results of the present study comparing control and IL-6 treated MDA-MB-231 cells suggest that IL-6 also promotes the synthesis of sLe x on breast cancer cells, potentially providing more opportunity for sLe x : Eselectin mediated heterotypic interaction between cancer cells and the activated endothelium.
Selectin-based technology has been used to capture viable circulating tumor cells (CTCs) that can be further characterized by releasing the bound CTCs in the absence of calcium [39,40].
Recently, E-selectin-mediated adhesive interactions were utilized inside an E-selectin coated polydimethylsiloxane (PDMS) micro-bubble system to capture colon CTCs under flow conditions [41]. Presently, most CTC studies have focused on the survival and metastasis of individual tumor cells in the circulation. However, emerging evidence supports the importance of circulating tumor microemboli (CTM) during cancer progression [42]. CTM represent a ''collective migration'' of tumor cells and have been found to exhibit higher metastatic potential than that of individual CTCs [43]. It has been proposed that these multicellular clumps of tumor cells can metastasize by becoming mechanically trapped in the smallest capillaries due to their large size [23]. We found that tumor spheroid/aggregate grown MDA-MB-231 breast cancer cells upregulate their number of E-selectin ligands and binding moieties and demonstrated the most robust rolling behavior on Eselectin coated surfaces, compared to MDA-MB-231 cells from plasma and cytokine treatments, suggesting that CTMs may Highly metastatic tumor cells have been reported to exhibit increased homotypic aggregation, gaining resistance to apoptosis [24]. There is also evidence to suggest that breast cancer cells cultured as 3-D spheroids express a higher level of IL-6 mRNA [44]. A recent study by our group reported that homotypic and heterotypic interactions in breast cancer cells cultured as 3-D spheroids alter their adhesion phenotype. These interactions were shown to increase the interaction of breast cancer cells with human recombinant E-selectin [45]. Furthermore, when MDA-MB-231 cells were cultured as 3-D spheroids in media, high concentrations of IL-6 and TNF-a were found in the spent media and cell adhesion to E-selectin surfaces were found to be dependent on IL-6 concentration.
We propose a novel mechanism ( Figure 6) that may explain the aggressiveness of 'triple-negative' breast carcinoma. First, the proinflammatory cytokines act on single cells to induce an aggregated morphology and an increased proliferation rate. Cell aggregation then upregulates the secretion rate of IL-6 and TNF-a thereby increasing cytokine concentrations. This may cause a positive feedback mechanism where the increased cytokine concentrations cause more aggregation and proliferation that further stimulate cytokine secretion. The ultimate effect of aggregation is the increased expression of E-selectin ligands and binding moieties that facilitate stable rolling on the inflamed endothelium to enable subsequent extravasation.
In summary, our results indicate that plasma and cytokines can promote homotypic aggregation and that cell proliferation as tumor spheroids can change the adhesion phenotype of cancer cells to immobilized E-selectin under physiological flow, which may contribute to the higher metastatic potential of CTM. Taken together, these results suggest that therapeutic approaches targeting cytokine receptors and adhesion molecules on cancer cells may help reduce the likelihood of metastasis and deserve further attention.

Reagents
Recombinant E-selectin-IgG 1 chimera was purchased from R&D systems (Minneapolis, MN). Blotting grade blocker non-fat dry milk was obtained from Bio-Rad Laboratories (Hercules, CA) and Protein-G was purchased from EMD Biosciences (San Diego, CA). FITC Mouse IgG1 k isotype control, purified mouse antihuman CD15s (clone CSLEX), and APC rat anti-mouse IgM were all purchased from BD Biosciences (San Jose, CA). FITC mouse anti-human CD44V4 was obtained from AbD Serote (Germany). Ca 2+ and Mg 2+ free DPBS (Invitrogen, Camarillo, CA, USA), calcium carbonate (Sigma Chemical Co., St. Louis, MO, USA), low endotoxin (1 ng/mg), and essentially globulin-free Bovine Serum Albumin (Sigma Chemical Co., St. Louis, MO, USA) were used to prepare flow buffer for cell adhesion assays. Nuclear ID TM cell viability reagent was obtained from Enzo Life Sciences (Farmingdale, NY).

Breast Cancer Cell Culture
Breast cancer cell line MDA-MB-231 was purchased from ATCC and maintained in 10% Fetal Bovine Serum (FBS; Cellgro), 1% penicillin-streptomycin (Invitrogen), and high glucose Dulbecco's Modified Eagle Medium at 37uC with 5% CO 2 in a humidified incubator.

Plasma isolation and treatment
Whole peripheral blood was collected from informed consenting healthy female donors using sodium-heparin tubes with a venoject vacuum system. Collected whole blood was centrifuged for 25 min at 500 rpm. The plasma layer on top was carefully removed without disturbing the interface and passed through a sterile 0.2 mm filter. 50% of the isolated plasma was used for cell culture while the rest was stored in 280uC for other assays. MDA-MB-231 cells were plated at 2610 5 cells/well in 6-well plates. Isolated plasma and culture medium at a 1:1 ratio were used to treat cells for 48 h.
Tumor spheroid generation using PDMS coated plate Dow Corning's SylgardH 184 silicone elastomer kit in a 10:1 base to curing agent ratio (w/w) was used to cure PDMS [46]. The prepolymer components were manually mixed with a pipette tip in a 50 mL tube for 60 s. 300 mL of PDMS pre-polymer was pipetted

Pro-inflammatory cytokine and Metformin treatment
Human recombinant IL-6 (R&D) and TNF-a (R&D) were added to the culture media at a range of concentrations from 1-25 ng/mL. Cell rolling assay and flow cytometry were performed after 48 hr treatments (single cytokine or combined). For blocking/neutralization experiments, 0.1 mM Metformin (1,1-dimethylbiguanide hydrochloride) or 1 mg/mL of mouse anti-human IL-6R monoclonal antibody was added with the IL-6 treatment.

Neuraminidase treatment
After IL-6 treatment and prior to rolling experiments, MDA-MB-231 cells were treated with 0.1 U/ml Vibrio Cholerae neuraminidase (Roche Biochemicals, Indianapolis, IN) for 45 min at 37uC to cleave terminal sialic acid residues. After enzyme treatment cells were washed and incubated with 0.1% BSA to block nonspecific interactions.

MTT Assay
MDA cells were seeded into each well of a 24-well culture plate. After 12 hr in culture, the medium was withdrawn and replaced with 0.5 mL fresh DMEM culture medium or cytokine spiked medium. After 2 days in culture, all media was withdrawn and replaced with 0.5 mL fresh DMEM medium. Next, 50 ml MTT stock solution (5 mg MTT powder/1 ml PBS) was added into each well and incubated at 37uC for 1 hr. Media was then removed and replaced with 0.5 mL DMSO for 5 min incubation at RT. Absorbance was measured at a wavelength of 570 nm with background subtraction at 660 nm.

Cell Viability Assay
The Nuclear-ID TM Blue/Red cell viability reagent (GFP-Certified TM ), a mixture of a blue fluorescent cell-permeable nucleic acid dye and a red fluorescent cell-impermeable nucleic acid dye that is suited for staining dead nuclei was used to determine cell viability. MDA-MB-231 cells with various treatments including control, plasma, IL-6, IL-6 + Metformin, and TNF-a were grown in imaging ready glass coverslip mounted 24well plates at 5610 4 cells/well. Cells were incubated with the Nuclear-ID reagent at 1:1000 dilution for 30 min at 37uC and imaged with a Zeiss 710 confocal microscope (10X objective) with a dual filter set for DAPI (Ex/Em: 350/470 nm) and Texas Red (Ex/Em: 540/605 nm).

Flow cytometry
After 48 h in culture, untreated and treated (plasma, tumor spheroid, IL-6, and TNF-a) MDA-MB-231 cells were removed from culture plates using an enzyme-free cell dissociation buffer solution to preserve membrane proteins and prevent clumping. After washing with PBS, the cells were resuspended in 1% BSA in DPBS to a final concentration of 250,000 cells per sample. Antibodies against CD44V4 and sLe x or appropriate isotype controls were added to the cell suspensions and incubated over ice for 45 min. Following incubation, cells were washed twice with 500 mL of 1% BSA to remove any unbound antibody. Flow cytometry samples were analyzed using an Accuri C6 flow cytometer (BD Bioscience, San Jose, CA).

Cell Rolling Assay
Micro-renathane tubing with 640 mm internal diameter (Braintree Scientific) was cut to a length of 50 cm, functionalized with Protein G (10 mg/mL) and Fc chimera E-selectin (20 mg/mL, R&D), and blocked with 5% BSA or milk (Sigma). Functionalized microtubes were then secured to the stage of an Olympus IX81 motorized inverted microscope (Olympus America, Melville, NY). A CCD camera (model no: KP-M1AN, Hitachi, Tokyo, Japan) and DVD recorder (model no: DVD-1000MD, Sony Electronics) were used to record experiments for offline analysis. MDA-MB-231 cells were suspended in flow buffer at 1610 6 cells/mL and perfused through protein coated microtubes using a syringe pump (KDS 230, IITC Life Science, Woodland Hills, CA) at a wall shear stress of 1.0 dyn/cm 2 . ''Rolling'' cells were defined as those observed to translate in the direction of flow with an average velocity less than 50% of the calculated hydrodynamic free-stream velocity [47].
Quantification of cytokine concentration using enzymelinked immunosorbent assay (ELISA) Healthy donor plasma was isolated as described above. Conditioned media was harvested from both 2-D culture and 3-D tumor spheroid culture. Anti-human IL-6 and TNF-a ELISA kits from eBioscience (San Diego, CA) were used to quantify the cytokine concentration. Standard curves were performed per instruction and the concentration of cytokines in each sample was calculated accordingly.

Real-time quantitative PCR (qPCR)
10 ng of cDNA produced by the reverse transcription of total RNA was used in each quantitative PCR reaction. Also included in the 20 mL qPCR reaction system were 10 mL iQ TM SYBR Supermix (Bio-Rad), 1 mL of 2 mM forward primer and 1 mL of 2 mM reverse primer and nuclease free water.
Primer for GAPDH qPCR 59-AGAGCACAAGAGGAAGAGAGAGAC-39 (Forward) 59-AGCACAGGGTACTTTATTGATGGT-39 (Reverse). qPCR reactions were carried out in 96 well real-time PCR plates (Bio-Rad) using a Bio-Rad MyIQ Real-time PCR detection system. The qPCR reaction consisted of 5 min at 95uC to activate the polymerase and 50 PCR cycles (uncoupling step at 95uC for 20 sec followed by annealing step at 59uC for 20 sec and elongation step at 72uC for 30 sec), followed by a melting temperature analysis to test for any nonspecific amplification. The expression level of CD44 gene in MDA-MB-231 cells with different treatment was normalized to the expression level of the standard gene GAPDH.