Cysteine-linked dimerization of BST-2 confers anoikis resistance to breast cancer cells by negating proapoptotic activities to promote tumor cell survival and growth

Almost all breast tumors express the antiviral protein BST-2 with 67%, 25% and 8.2% containing high, medium or low levels of BST-2, respectively. Breast tumor cells and tissues that contain elevated levels of BST-2 are highly aggressive. Suppression of BST-2 expression reprograms tumorigenic properties of cancer cells and diminishes cancer cell aggressiveness. Using structure/function studies, we report that dimerization of BST-2 through cysteine residues located in the BST-2 extracellular domain (ECD), leads to anoikis resistance and cell survival through proteasome-mediated degradation of BIM—a key proapoptotic factor. Importantly, BST-2 dimerization promotes tumor growth in preclinical breast cancer models in vitro and in vivo. Furthermore, we demonstrate that restoration of the ECD cysteine residues is sufficient to rescue cell survival and tumor growth via a previously unreported pathway—BST-2/GRB2/ERK/BIM/Cas3. These findings suggest that disruption of BST-2 dimerization offers a potential therapeutic approach for breast cancer.

Adhesive interaction between fibroblasts and breast cancer cells upregulate BST-2 expression. To further evaluate the significance of BST-2-mediated adhesive interactions, we show that adhesion between monolayers of Cal51 ( Figure 1f 38 (j) Representative images and colony size of crystal violet-stained MCF-7 cell growth in soft agar assay showing anchorage-independent growth of Vector-or OE-BST-2-expressing MCF-7 cells. Colony diameters from five different fields of six different wells were measured following a 30-day transformation assay. The colony sizes were averaged and a percent calculated relative to MCF-7 Vector-expressing cells, which was set up to 100%. All experiments were repeated at least three times and similar results were observed. Error bars correspond to S.E.M. Significance was taken at *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001 Cal51 compared with MDA-MB-231. In addition, fibroblasts contain varying levels of BST-2 ( Figure 1i), but their BST-2inducing capability is comparable (Figures 1f and g). These data indicate that interactions between cancer cells and fibroblasts may regulate cancer cell BST-2 and may promote cellular reprograming.
BST-2 expression is required for efficient growth of breast cancer cells in suspension. To assess the biological relevance of BST-2-mediated adhesive interactions, we first examined the effects of BST-2 on anchorage independency and spheroid formation. MCF-7-OE-BST-2 cells produced larger spheroids compared with MCF-7-vector cells ( Figure 1j). These data indicate that BST-2 promotes survival and growth of cancer cells in suspension; suggesting that breast tumor cells that are anchorage independent due to high levels of BST-2 may undergo anoikis in circulation in the absence of BST-2.
BST-2 dimers are present in breast cancer cells and dimerization is regulated by ECD cysteine residues. Owing to the effective role of BST-2 in cell adhesion, we evaluated the structural property of BST-2 that has a role in cell adhesion. We engineered MCF-7 cells (Figure 2a) stably expressing WT BST-2 that is predominantly dimer (OE-BST-2D) and dimerization-deficient BST-2 that is expressed as monomers (OE-BST-2M). PCR analysis shows efficient expression of BST-2D and BST-2M mRNA (Figure 2b), whereas western blots confirm presence of BST-2D or BST-2M in these cells (Figure 2b). Functionally, BST-2D and BST-2M increase viability, proliferation and invasion of MCF-7, albeit with subtle differences (Figure 2c).
BST-2 dimers mediate adhesion of breast cancer cells to components of the tumor microenvironment. Here, we assessed whether the variant of BST-2 in cancer cells is critical for BST-2:BST-2 interactions that mediate cell to cell and/or cell to matrix adhesion. 3 (Figure 2j). Induction of BST-2 in macrophages with the BST-2 agonist, IFNα, 12 results in increased adherence of cancer cells expressing the different variants of BST-2 to IFNα+ macrophages compared with IFNα-macrophages (Figure 2i, white and gray backgrounds). The increased adherence of OE-BST-2M cells to IFNα+ macrophages could be attributed to enhancement of endogenous BST-2D. Furthermore, monocytes, irrespective of their level of BST-2, adhere efficiently to monolayers of shCTL cells compared with shBST-2 cells (Figure 2j, pink and blue bars). The effect of BST-2 is dependent on the variant of BST-2, as monocytes expressing shCTL and shBST-2 adhere more efficiently to BST-2D-expressing monolayers (Figure 2j). These data indicate that the level and variant of BST-2 in cancer cells may determine the rate of immune cell adherence.
Next, we performed adhesion in the presence and absence of recombinant BST-2 (rBST-2). Results show that rBST-2 efficiently blocks adhesion of BST-2-expressing cancer cells but has no effect on adhesion of BST-2-suppressed cells (Figure 2k), indicating that BST-2 is responsible for the observed adhesion. Furthermore, rBST-2 specifically blocks adhesion of OE-BST-2D cells but has no effect on OE-BST-2M cells (Figure 2l), confirming that the variant of BST-2 (D or M) is crucial in cancer cell adhesion.
We confirmed the role of BST-2 dimerization on adhesion by seeding equivalent numbers of cells on rBST-2 pre-coated plates. Compared with shCTL cells, shBST-2 cells were unable to adhere efficiently to rBST-2-coated plates ( Figure 2m). Importantly, OE-BST-2D increases cell adherence, whereas OE-BST-2M did not ( Figure 2n). These data suggest that recombinant human BST-2 binds to both murine and human BST-2 in cancer cells and blocks cancer cell to cancer cell adhesion.
BST-2 dimerization regulates anchorage independency. As BST-2 dimerization is critical for cellular and matrix interactions, we showed that BST-2 dimerization is crucial for colony formation and anchorage-independent growth. As expected, 4T1 shCTL cells form significantly larger colonies compared with 4T1 shBST-2 cells (Figure 3a). OE-BST-2D but not OE-BST-2M efficiently rescues colony formation in shBST-2 cells (Figure 3a), indicating that BST-2 dimerization is required for growth of cancer cells independent of anchor. These findings were confirmed with MCF-7 cells (Figure 3b). The difference in the ability of OE-BST-2D and OE-BST-2M cells to grow in suspension is not because of the level of BST-2 (Figures 3c and d) but can be attributed to the variant of BST-2 ( Figure 3d, red brackets-BST-2 shifts in nonreducing gels). These data indicate that BST-2 expressed as dimers may endow cancer cells the ability to cluster, survive and grow in suspension-a characteristic of aggressive epithelial-derived tumor cells.
BST-2 dimerization promotes adherent-independent survival of cancer cells by inhibiting anoikis. If BST-2 dimerization is involved in protection of cancer cells from anoikis, cells expressing BST-2D will survive under anoikis conditions. Indeed, following poly-HEMA-mediated induction of anoikis, shBST-2 cells have significant reduction in viability compared with shCTL cells (Figures 4a and b). OE-BST-2D but not OE-BST-2M rescues viability of shBST-2 cells (Figures 4a and b). The inability of BST-2M cells to survive under anoikis conditions is due to the variant of BST-2 because BST-2 mRNA is higher in BST-2M cells compared with shBST-2 cells (Figure 4c).  BST-2 dimerization results in phosphorylation of BST-2 in cancer cells. To explore the mechanism by which BST-2 promotes cellular interactions, we showed that dimerization of BST-2 molecules activates BST-2. Western blot analysis of input protein following exposure of cells to vehicle or rBST-2 shows that the levels of GAPDH, phosphor-tyrosine (p-Tyr), and growth factor receptor-bound protein 2 (GRB2) are similar (Figure 5a). In contrast, immunoprecipitation with anti-BST-2 antibody reveals that BST-2 is tyrosine phosphorylated in cancer cells in a manner that is dependent on BST-2 dimerization (Figure 5b). These data suggest that BST-2Dexpressing cells contain activated and more functionrelevant BST-2.
To determine the role of activated BST-2 in cancer cells, we started by investigating the level of GRB2a docking protein that binds to phospho-tyrosine residues of activated receptors and recruits ERK1/2 to the signaling complex. The level of GRB2 remained the same in quiescent cells (Figure 5b). But upon BST-2 activation, higher GRB2 and ERK1/2 were bound to phospho-BST-2 in shCTL and OE-BST-2D cells (Figure 5b). Remarkably, GRB2 and ERK1/2 bound to BST-2 in OE-BST-2M cells did not increase upon BST-2 activation (Figure 5b).
We confirmed that BST-2 is tyrosine phosphorylated using a mutant form of BST-2 that is able to form dimers but the cytoplasmic tail tyrosine residues at positions 6 and 8 had been substituted with alanine residues (OE-BST-2DΔTy). The level of anti-BST-2-precipitated p-Tyr, GRB2 and ERK1/2 did not change between vehicle and rBST-2-treated OE-BST--2DΔTy cells (Figure 5b), suggesting that phospho-Y6/Y8 recruits GRB2.
BST-2 dimerization induces proteasomal degradation of BIM. As BST-2D downregulates BIM, we examined whether this downregulation occurs via proteasomal degradation. Although TPA activates/phosphorylates ERK1/2 and decreases BIM levels in OE-BST-2D cells, MG132 treatment results in accumulation of BIM in TPA-treated cells (Figures 5e and f). Importantly, TPA, MG132 or TPA/MG132 has no effect on the levels of pERK1/2 and BIM in OE-BST-2M cells (Figures 5e and f). The concentration of inhibitors used were non-toxic (Figure 5g) and not responsible for the observed BST-2-independent reduction of pERK1/2 in MG132-treated cells (Figures 5e and f). Together, results in Figure 5 suggest that BST-2 dimerization promotes ERK1/2 activation, BIM phosphorylation/degradation and inhibition of Cas3 activation that culminate in enhanced anoikis resistancea phenotype required by cancer cells to survive in circulation.
CTC clusters of metastatic breast cancer patients are enriched in BST-2. The clinical significance of BST-2mediated cell clustering and survival was evaluated using data from a publicly available dataset 9 to compare the levels of BST-2 in circulating tumor cells (CTCs). Intrapatient comparison of BST-2 in CTC singlets versus CTC clusters shows that 8 of 10 patients have CTC clusters that express higher BST-2 than their respective CTC singlets (Figure 6a). On the average, CTC clusters express higher (~2 fold) BST-2 compared to CTC singlets (Figure 6b). Further analysis shows that BIM RNA inversely correlates with BST-2 RNA in CTCs (Figure 6c), supporting the findings in Figures 4d and e and further suggest that BST-2 may facilitate cancer cell clustering, thus protecting cancer cells from hemodynamic shear stress in circulation.
BST-2 dimerization regulates the growth of triplenegative breast cancer cells in mice. Compared with shCTL, shBST-2 cells have a significant decrease in primary tumor growth (Figures 7a-c). Analysis of tumor volume (TV)  (Figures 7a-c). OE-BST-2D and shCTL but not shBST-2 and OE-BST-2M cells efficiently metastasize as evidenced by increased luciferase expression over time (Figures 7a and d).
In addition, we observed increase spontaneous pulmonary metastases of shCTL and OE-BST-2D but not shBST-2 and OE-BST-2M tumors (Figures 7e and f). The decrease in lung metastasis in shBST-2 and OE-BST-2M tumor-bearing mice could be attributed to smaller primary tumors, although the effect of BST-2 on primary tumor is distinct from its effect on lung metastasis. 3 Alternatively, dimerization-competent OE-BST-2D cancer cells could associate with BST-2expressing lung-associated cells, and such association may protect cancer cells from apoptosis. Indeed, BIM levels in the lungs of shCTL and OE-BST-2D tumor-bearing mice were  (Figure 7g). These data indicate that BST-2 dimerization is required for tumor growth and that disruption of BST-2 dimerization may render metastatic cells susceptible to apoptotic insult in the lungs.
According to Kaplan-Meier's survival analysis, growth of shCTL and OE-BST-2D tumors culminates in death with mean overall survival of 37.5 and 41.0 days for OE-BST-2D and shCTL tumor-bearing mice, respectively (Figure 7h). At variance, the mean overall survival for shBST-2 and OE-BST-2M tumor-bearing mice was undefined (Figure 7h). Together, these data suggest that disruption of BST-2 dimerization may serve to prevent tumor growth and metastatic colonization of the lungs, thus increasing overall survival of tumor-bearing hosts.
On the basis of these observations, we propose a new model for cancer cell survival and growth in which dimeric BST-2 orchestrates pro-adhesive and anti-anoikis stimuli ( Figure 8). The principle of this new model is that dimeric BST-2 allows interaction between cancer cells and the tumor microenvironment that promotes the survival, growth and metastasis of tumor cells.

Discussion
Here we provide evidence for structural and molecular link between BST-2 and breast cancer by highlighting the following: First, the BST-2 ECD cysteine residues mediate formation of BST-2 dimers in cancer cells. Previous studies demonstrated the effect of BST-2 dimers in protection against viral infection. 18 However, these studies did not evaluate the involvement of BST-2 dimerization in altering cancer cell behavior. Our study identifies BST-2 dimerization as critical in the promotion of cancer cell adhesion. We found that cancer cells expressing dimeric BST-2 efficiently adhere to other The r 2 value depict strong inverse correlation between BST-2 and BIM in CTCs. Lack of significance is attributed to lack of enough data points. Samples with RPKM values of zero were excluded from the study. Data were from GEO dataset GSE51827. 22 Error bars correspond to S.E.M. Significance was taken at **Po0.01 BST-2 dimers promote anoikis resistance WD Mahauad-Fernandez and CM Okeoma cancer cells, potential stromal cells, and ECM proteins. Thus, cancer cells expressing BST-2 dimers may serve as a target or docking sites for other cells and ECM proteins to bind to tumors. The interaction between BST-2-expressing primary tumor cells and other resident stromal cells may regulate the expression of other factors in secondary organs, thus conditioning metastatic sites for subsequent arrival of tumor cells, especially tumor cells that express dimeric BST-2. It remains to be determined how BST-2 mediates the adhesion of cancer cells to components of the ECM. Perhaps, the cysteine residues involved in BST-2 dimerization may associate with cysteine residues found on fibronectin type II domain. 19 Second, similar to the role of BST-2 in adhesion, the ability of BST-2 to promote survival and growth of cancer cells in suspension is controlled by BST-2 dimerization. Cancer cell adhesion is intricately related to the ability of such cells to survive in suspension. We report that cells expressing BST-2 dimers and not monomers activate intracellular signaling that result in the degradation of BIM and blockade of Cas3 activation, culminating in anoikis resistance and cell survival. Therefore, BST-2:BST-2 dimerization may transmit survival signals or suppress proapoptotic factors in breast cancer cells, creating a microenvironment that allows cells to grow independent of anchor. In our studies, we identified BST-2/ GRB2/ERK/BIM/Cas3 as an important pathway in anoikis evasion by breast cancer cells. This BST-2-directed cell reprograming allows cancer cells to survive in circulation. Evidently, BST-2 is present in circulating breast cancer cells, and levels are elevated in cells that circulate as clusters. Whether BST-2 is directly linked to cancer cell clustering and survival in human blood is yet to be determined. Meta-analysis of data from CTCs shows an inverse correlative association between BST-2 and BIM. This concept was experimentally validated with breast cancer cells in vitro and with lung tissues from tumor-bearing mice. Of note, the tumorigenic activity of BST-2 dimerization is operative across species (mouse and human) and is independent of the aggressive nature of the cells.
Third, our study extends our knowledge of the molecular mechanism of anoikis evasion and the positive impact of the proteasome on tumor growth. The loss of cell viability and growth arrest observed in vitro and in vivo following expression of monomeric BST-2 is dependent on blockade of GRB2 recruitment and ERK1/2 activation, proteasomal degradation of BIM and activation of Cas3. Whether or not other anti-or pro-apoptotic factors are involved is yet to be determined. Also unknown are the kinases that catalyze and co-ordinate this complex BST-2/GRB2/ERK/BIM/Cas3 pathway. Activation of ERK1/2 in cells expressing BST-2 dimers may be orchestrated by serine/threonine kinases, such as Src or Ras known to phosphorylate ERK. 20,21 In addition to accumulation of BIM protein, BIM mRNA was upregulated in cells expressing reduced levels of BST-2 or monomers of BST-2. It is unclear how BST-2 dimerization can lead to reduced BIM at the RNA level. In immune cells, BST-2 is negatively regulated by MYD88/PI3K. 22 Possibly in cancer cells, BST-2 may activate PI3K to phosphorylate FOXO3A-a transcription factor that induces BIM expression 23,24 upon its dephosphorylation and nuclear translocation. 25 Although the identity of the BST-2 tyrosine residues that are phosphorylated is yet to be revealed, it is known that the cytoplasmic tail of BST-2 contains two tyrosine residues at positions 6 and 8 that become phosphorylated upon virus-mediated BST-2 activation. 26 Possibly, these or other tyrosines present in the different domains of BST-2 are phosphorylated. Indeed, in silico analysis using the PPSP software (http://ppsp.biocuckoo.org/) 27 revealed that BST-2 contains four phosphorylatable tyrosines at positions 6, 8, 153 and 154. 28 Aside from tyrosines, the cytoplasmic tail of BST-2 contains phosphorylatable serines and a threonine.
Fourth, we provide evidence that cells expressing monomeric BST-2 are unable to grow in the mammary gland. As monomeric BST-2 is deficient in adhesion and anchorage independency, it is possible that these cells were unable to make contact with each other or with mammary gland resident cells. The lack of increased metastatic tumor growth in the lungs persuades us to speculate that tumor cells expressing monomeric BST-2 may alter the tumor environment landscape by changing the type of immune cells that are recruited to the tumor microenvironment because of changes in the expression of signaling cytokines and chemokines. [29][30][31] It is also possible that cells expressing monomeric BST-2 may not survive in circulation thus limiting the number of cancer cells that may reach metastatic sites. Noteworthy, although expression of BST-2M in cells almost completely repressed Figure 8 Hypothetical model of BST-2-mediated anoikis resistance. Wild-type dimer-forming BST-2 (BST-2D) in cancer cells is activated upon cell to cell or cell to ECM interaction. BST-2 activation results in phosphorylation of the cytoplasmic tail (CT), presumably at the tyrosines residues located at positions 6 and 8. Other phosphorylation events independent of these tyrosines are possible. Phosphorylated BST-2 recruits GRB2 (an adaptor protein that recognize p-Tyr), facilitating activation of yet to be identified kinase(s), such as Src or Ras, which in turn phosphorylates ERK. Phospho-ERK then phosphorylates BIM resulting in subsequent proteasomal degradation and removal of BIM. In the absence of BIM, mitochondrial membranes remain intact and pro-Cas3 is not cleaved and activated (cCaspase-3). The end result is that cancer cells overcome anoikis, survive under harsh conditions and grow/metastasize tumor formation, shBST-2 cells (containing low levels of BST-2D) were able to form primary tumors, albeit small. Although the reason for BST-2M-mediated repression of tumor growth is yet to be determined, it is possible that functionally distinct signals may be elicited by BST-2M and shBST-2 cells and that BST-2M signals may have a negative growth effect on tumor cells (autocrine). It is also plausible that BST-2M signals may be transmitted to other distal cells to inhibit cell growth (paracrine).
In summary, we have demonstrated how BST-2 activity shapes the function of breast cancer cells. We identify BST-2/ GRB2/ERK/BIM/Cas3 as the pathway regulating BST-2mediated cancer cell adhesion, anoikis resistance, anchorage-independency, cell survival and growth. Our findings may motivate development of new targeted treatments based on disruption of BST-2 homodimerization in tumors.

Materials and Methods
Cell lines. The murine triple-negative breast cancer cell line-4T1 and the luminal A breast cancer cell line-MCF-7, respectively, are kind gifts from Drs. Lyse Norian and Weizhou Zhang of the University of Iowa, Iowa City, IA, USA. All cells were maintained according to ATCC guidelines (Manassas, VA, USA).
Animals. Five-week-old female BALB/cAnNCr mice purchased from Harlan (Indianapolis, IN, USA) were used. Tumor-bearing mice were killed when they became moribund. TV was calculated as: TV = 0.5(length × width 2 ). 32 Experiments involving mice were approved by the University of Iowa Animal Care and Use Committee (IACUC).
Mice injections and live animal imaging. Orthotopic mammary tumors were generated by implanting 300 000 cancer cells into the 10th mammary fat pad of 5-week-old female mice. Before imaging, mice were anesthetized, weighed and injected intraperitoneally with D-luciferin (Sigma-Aldrich, St. Louis, MO, USA). Mice were imaged using the Xenogen IVIS three-dimensional optical imaging system (Caliper Life Sciences, Hopkinton, MA, USA). Luciferase expression was quantified with Living Image Software (Caliper Life Sciences). Primary tumors were weighted and photographed post-mortem. Pulmonary nodules were quantified by manual counting.
Generation of BST-2-overexpressing cancer cells. MCF-7 cells, which contain low levels of endogenous BST-2 3 or 4T1 shBST-2 cells in which endogenous mouse BST-2 was downregulated, 3 were stably transfected with either empty pcDNA3.1 (Vector for MCF-7 cells or shBST-2 for 4T1 cells), pcDNA3.1 containing dimerization-competent wild-type human BST-2 (BST-2D) or pcDNA3.1 containing dimerization mutant BST-2 in which cysteine residues at positions 53, 63 and 91 were replaced with alanine residues (BST-2M). These BST-2 constructs are a kind gift from Dr. John Guatelli of UCSD (La Jolla, Ca, USA) and Dr. Klaus Strebel of NIH (Bethesda, MD, USA). 18,33 Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA) was used for the transfections and the amounts used were adjusted according to the manufacturers' instructions. Transfected cells were selected with G418 at 500 μg/ml and stable cells were used in all experiments. Note that shBST-2 is a shRNA specific for mouse BST-2 and does not affect the expression of human BST-2.