The effect of soluble E-selectin on tumor progression and metastasis

Background Distant metastasis resulting from vascular dissemination of cancer cells is the primary cause of mortality from breast cancer. We have previously reported that E-selectin expression on the endothelial cell surface mediates shear-resistant adhesion and migration of circulating cancer cells via interaction with CD44. As a result of shedding, soluble E-selectin (sE-selectin) from the activated endothelium is present in the serum. In this study, we aimed to understand the role of sE-selectin in tumor progression and metastasis. Methods We investigated the effect of sE-selectin on shear-resistant adhesion and migration of metastatic breast cancer cells and leukocytes in vitro and in vivo. Results We found that sE-selectin promoted migration and shear-resistant adhesion of CD44+/high breast cancer cell lines (MDA-MB-231 and MDA-MB-468) to non-activated human microvessel endothelial cells (ES-HMVECs), but not of CD44-/low breast cancer cell lines (MCF-7 and T-47D). This endothelial E-selectin independent, sE-selectin-mediated shear-resistant adhesion was also observed in a leukocyte cell line (HL-60) as well as human peripheral blood mononuclear cells (PBMCs). Additionally, the incubation of MDA-MB-231 cells with sE-selectin triggered FAK phosphorylation and shear-resistant adhesion of sE-selectin-treated cells resulted in increased endothelial permeabilization. However, CD44 knockdown in MDA-MB-231 and HL-60 cells resulted in a significant reduction of sE-selectin-mediated shear-resistant adhesion to non-activated HMVECs, suggesting the involvement of CD44/FAK. Moreover, functional blockade of ICAM-1 in non-activated HMVECs resulted in a marked reduction of sE-selectin-mediated shear-resistant adhesion. Finally, the pre-incubation of CD44+ 4 T1 murine breast cancer cells with sE-selectin augmented infiltration into the lung in E-selectin K/O mice and infusion of human PBMCs pre-incubated with sE-selectin stimulated MDA-MB-231 xenografted breast tumor growth in NSG mice. Conclusions Our data suggest that circulating sE-selectin stimulates a broad range of circulating cells via CD44 and mediates pleiotropic effects that promote migration and shear-resistant adhesion in an endothelial E-selectin independent fashion, in turn accelerating tissue infiltration of leukocytes and cancer cells. Electronic supplementary material The online version of this article (doi:10.1186/s12885-016-2366-2) contains supplementary material, which is available to authorized users.


Background
Tissue infiltration of circulating cells is critical for multiple aspects of tumor progression and metastasis. First, extravasation of circulating leukocytes into the primary tumor aids in the development of a tumor stroma that provides a fertile environment for cancer cells to survive and grow. Second, extravasation of circulating tumor cells into the tissue parenchyma of distant organs gives rise to metastasis. Despite the recognized importance of tissue infiltration in tumor progression and metastasis as well as pathologic alterations in the global serum protein profiles of cancer patients, it is uncertain whether circulating factors in the vascular space extrinsically stimulate circulating cells to infiltrate tissues. During tissue infiltration, inflamed vascular endothelium, characterized by expression of E-selectin and CAMs, functions as a gateway to tissue parenchyma. Tissue infiltration by circulating cells is governed by a multi-step adhesion cascade, mediated by sequential interactions and crosstalk between adhesion molecules expressed on the endothelial cell surface and their counter-receptor ligands expressed on circulating cells [1][2][3][4]. A robust interaction is required for this attachment to resist swift blood flow. Weak interaction prevents effective attachment of the circulating cells to the endothelial surface. The initial adhesion of circulating cells subsequently leads to firm attachment that opens the endothelial junction enabling transendothelial migration into tissue parenchyma toward cytokine or chemokine stimuli [5,6].
Adhesion molecules, such as selectins, ICAM, and VCAM, expressed on the endothelial surface in response to inflammatory stimuli are the primary mediators of tissue infiltration [7,8]. E-selectin (CD62E, ELAM-1, or LECAM-2) is expressed exclusively on the luminal surfaces of inflamed vasculature [9] and mediates rolling and vascular adhesion through interaction with carbohydrate ligands [10,11]. A variety of ligands for E-selectin have been identified in different types of cells; these include CD44 (HCELL) [12][13][14], Mac-2 binding protein [15], Muc-1 [16], death receptor-3 [17], and ESL-1 [18]. The expression of E-selectin is temporally and spatially limited. E-selectin is shed into circulation (soluble Eselectin: sE-selectin) or internalized rapidly after activation. An abundance of sE-selectin in the serum is indicative of inflammation [19][20][21][22][23], and circulating sE-selectin levels are consistently high among individuals with chronic inflammation compared to their healthy counterparts [24]. Similarly, elevated serum levels of sE-selectin were reported and positively associated with tumor grade, tumor stage, and metastases in breast cancer [3,25]. Despite the positive associations between sE-selectin and breast cancer progression and metastasis, it is unclear whether sE-selectin plays any functional role in tumor progression. In the present study, we demonstrate that sE-selectin functions as a signaling molecule in circulation with the ability to 1) enhance shear-resistant adhesion and migration of CD44 +/high breast cancer cells (BCs) and leukocytes in an endothelial E-selectin independent fashion, 2) activate focal adhesion kinase (FAK), 3) promote homing of CD44 +/high BCs to the lung, and 4) stimulate tumor growth and intratumoral infiltration of PBMCs. These findings identify a possible extrinsic mechanism by which an inflammation-associated factor (i.e., sE-selectin) present in circulation pleiotropically stimulates circulating tumor cells toward tumor progression and metastasis.

Reagents and antibodies
Recombinant human sE-selectin was purchased from R&D Systems. Monoclonal antibodies for pFAK and total FAK were purchased from Cell Signaling. Blocking antibodies for ICAM and VCAM were purchased from Santa Cruz and Biolegend, respectively. FAK inhibitor II was purchased from Calbiochem.

In vitro cell adhesion assay
HMVECs were grown to confluence in the flow chamber (μ-Slide I 0.4 Luer for cancer cells and 0.2 for leukocytes; Ibidi, Madison, WI) coated with collagen I and fibronectin. The cells (10 5 cells/ml) were pre-incubated with 100 nM sE-selectin for 10 min and perfused over a 5 min period into the flow chamber in 1 % FBS containing DMEM medium at a shear stress rate of 1 dyn/cm 2 . Unbound cells were washed off with DMEM containing 1 % FBS. Using a light microscope at a final magnification of 100×, the number of cells that adhered to endothelial cells was determined by counting cells in at least five random fields. The data are expressed as the mean of triplicate experiments.

In vitro permeability assay
HMVECs were grown to confluence on the top-facing side of a collagen-coated 0.4-μm-pore transwell chamber (Corning). MDA-MB-231 cells were pre-incubated with 100 nM sE-selectin for 10 min at 37°C. MDA-MB-231 cells (2 × 10 3 cells) suspended in EBM-2 media and 2000-kDa FITC-dextran (final concentration of 1 mg/ml, Sigma-Aldrich) were seeded into the upper chamber, and 600 μl EBM-2 media was added to lower chamber. Then, at the indicated time points, 50-μl aliquots were removed from the lower chamber and the fluorescence was measured at 485/530 nm with a CLARIOstar fluorimeter (BMG Labtech).

Cell migration assay
Transwell migration assays were performed using a 12.0-μm or 5.0-μm pore transwell chamber (Becton Dickinson). Cancer cells (2.5 × 10 4 cells) or leukocytes (1.0 × 10 5 cells) suspended in serum-free media were applied to the upper chamber and 100 nM sE-selectin was added to the lower chamber. After overnight incubation at 37°C, migrated cells were stained using a HEMA 3 staining kit (Pierce) and counted under a light microscope.

Anoikis assay
Anoikis resistance was measured using ultra-low attachment plates (Corning). Cancer cells (10 4 cells) were suspended in 1 % FBS containing medium with or without sE-selectin, seeded into 96-well ultra-low attachment plates, and incubated for 72 h. The number of viable and non-viable cells was determined by trypan blue exclusion using a hemocytometer.

Animal models
All animal housing and handling procedures were in accordance with institutional guidelines at the University of Oklahoma and Thomas Jefferson University. Sixweek-old female Balb/C mice (Jackson Laboratory) or Eselectin knockout mice (C.129S4-Sele tm1Dmil /J; Jackson Laboratory) were used for the in vivo cell adhesion assay. Female Balb/C mice or E-selectin knockout mice (n = 3) were injected intravenously with 4 T1-luc murine breast cancer cells (3 × 10 4 in 100 μl saline) via tail vein. Cancer cells were pre-incubated with 100 nM sEselectin for 10 min at 37°C and washed with PBS to remove sE-selectin carry over before injection. Seven days later, whole lungs were harvested and total RNA was isolated using TRIzol (Life Technologies). Total RNA was reverse transcribed using random hexamer primers and SuperScript® III First-Strand Synthesis Supermix (Life Technologies). qPCR reactions (80 cycle) were performed using TaqMan gene expression master mix (Life Technologies) and the Biorad CFX96™ Touch Real-Time PCR Detection System (Biorad). Luciferase and mouse GAPDH primers were used (Life Technologies). Data are presented as delta-Ct relative to an endogenous control, GAPDH. For leukocyte infiltration, MDA-MB-231 breast cancer cells were injected into the mammary fat pads of 7-week-old female NOD SCID IL2 receptor gamma chain knockout mice (NSG mice: NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ; Jackson Laboratory, Bar Harbor, ME). Freshly isolated PBMCs from healthy donors were incubated with 10 μM Calcein AM (BD bioscience, San Jose, CA) for 30 min. After brief a wash, the Calcein positive PBMCs (2 × 10 7 ) were incubated with sE-selectin or saline and slowly infused over 1 min via tail vein into NSG mice bearing tumors of approximately 300 mm 3 (LxWxH). Tumors were harvested to analyze the tissue migration of Calcein-labeled PBMCs. For tumor growth, MDA-MB-231-TGL cell line (a kind gift from Dr. Massague, Memorial Sloan-Kettering Cancer Center) was used. Human PBMCs (1 × 10 7 cells) that were treated with or without sE-selectin were infused via tail vein once a week. Tumor growth was monitored by measuring bioluminescence using IVIS Lumina XR (Caliper, MA) once a week by intraorbital injection of 100 μl luciferase (15 μg/ml).

Statistical analysis
All experiments were carried out in triplicate and repeated at least twice independently. The results are presented as Mean ± SD. Comparisons between groups were analyzed using Student's t-test. Statistical significance is expressed as *p < 0.05; **p < 0.01; ***p < 0.001.

Results
Soluble E-selectin enhances the shear-resistant adhesion, migration, and anoikis-resistant survival of CD44 +/high breast cancer cells To investigate the effect of sE-selectin on factors pertinent to hematogenous metastasis, we evaluated shear-resistant adhesion, migration, and anoikis-resistant survival using two estrogen receptor (ER)-negative/ CD44 +/high BCs (MDA-MB-231 and MDA-MB-468), two ER-positive/CD44 -/low BCs (T-47D and MCF-7), and a line of Tet-on inducible, E-selectin expressing, human microvessel endothelial cells (ES-HMVECs) [26]. A confluent monolayer of ES-HMVECs was grown on a collagen/fibronectin-coated parallel flow chamber and was incubated with or without doxycycline for 4 h to induce the expression of E-selectin. A single cell suspension of BCs was infused into a chamber to recapitulate the shear-resistant adhesion of disseminated/circulating tumor cells to vascular endothelial cells. The adherent BCs were counted. Incubation of MDA-MB-231 and MDA-MB-468 BCs with sE-selectin increased the shear-resistant adhesion to doxycycline-untreated ES-HMVECs (i.e., absence of E-selectin expression on ES-HMVECs) 1.9 fold (p < 0.05) and 2.0 fold (p < 0.001), respectively, compared with adhesion of saline-treated BCs (Fig. 1a). To further examine the role of sE-selectin on migration, we performed a Boyden chamber migration assay. BCs were added to the upper chamber, while the lower chamber contained culture media, with or without sEselectin. The presence of sE-selectin enhanced migration of CD44 +/high MDA-MB-231 BCs (1.8 fold, p = 0.006) and MDA-MB-468 BCs (1.7 fold, p = 005) (Fig. 1c). However, sE-selectin did not affect the migration of CD44 -/low MCF-7 BCs or T-47D BCs (Fig. 1c). Soluble Eselectin-induced migration was also significantly reduced in CD44 knockdown clones of MDA-MB-231 BCs (Fig. 1d). Moreover, sE-selectin increased anoikis resistance in MDA-MB-231 BCs (1.36 fold, p < 0.001) and MDA-MB-468 BCs (1.16 fold, p = 0.02), but this effect was not observed in MCF-7 and T-47D BCs (Additional file 1: Figure S1). Therefore, these data suggest that sEselectin specifically enhances anoikis resistant survival, shear resistant adhesion, and migration of CD44 +/high BCs.

Soluble E-selectin enhances shear-resistant adhesion and migration in leukocytes
To further assess the effect of sE-selectin on circulating cells, leukocyte cell line HL-60 and human peripheral blood mononuclear cells (PBMCs) were used in adhesion and migration assays. Leukocytes (10 5 cells/ml) were infused into a confluent monolayer of parental HMVECs (p-HMVECs) grown on a collagen/fibronectin-coated parallel flow chamber at 1 dyn/cm 2 for 5 min and the adherent cells were counted. Consistent with the data from analysis of BCs, sE-selectin enhanced the shear-resistant adhesion of HL-60 (1.6 fold, p < 0.001) and PBMCs (2.1 fold, p < 0.001) to non-activated pHMVEC (Fig. 2a). Additionally, sE-selectin enhanced the migration of HL-60 (2.4 fold, p < 0.001), and PBMCs (3.6 fold, p < 0.001) (Fig. 2b). Functional blockade of CD44 in HL-60 using CD44 blocking antibody or a knockdown resulted in reductions of sE-selectin-mediated adhesion to non-activated p-HMVECs (p < 0.001). Taken together, these results indicate that sE-selectin may have biological function via CD44 in a broad range of circulating cells including leukocytes and cancer cells. Adhesion of cancer cells to the activated endothelium is reported to induce endothelial permeability for subsequent transendothelial migration [27,28]. We next evaluated whether the shear-resistant adhesion of sE-selectin-treated MDA-MB-231 BCs enhanced the permeability of nonactivated p-HMVECs. A single cell suspension of MDA- Adhesion was expressed as a percent relative to adhesion of salinetreated control cells. b Migration assay was performed using Boyden chamber. HL-60 cells (5 × 10 5 cells) were seeded in upper chambers (5-μm pores) and 100 nM sE-selectin was supplemented to the lower chambers. The number of migrated cells was counted after 16 h. Migration is expressed as a percent relative to saline-treated control.
c Effect of CD44 knockdown of HL-60 on sE-selectin-induced adhesion. The data represent Mean ± S.D. *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test. Data were summarized as % of respective control from an experiment conducted in triplicate and repeated twice MB-231 BCs was pre-incubated with sE-selectin or saline. After a brief spin followed by washing with PBS, the cells were layered onto a confluent monolayer of non-activated p-HMVECs with 2000 kDa FITC-dextran as a fluorescent tracer. Following the addition of sE-selectin-treated MDA-MB-231 cells, a steep increase in fluorescent leakage through the HMVEC monolayer was observed (Fig. 3a). In contrast, the adhesion of saline-treated MDA-MB-231 BCs to non-activated p-HMVECs did not have a significant effect on endothelial permeabilization over the course of the 60 min assay (Fig. 3a). Incubation of p-HMVECs with sE-selectin alone had no effect on endothelial permeability (data not shown). To confirm increased endothelial permeability, the endothelial junction was immunostained after the adhesion of BCs to non-activated p-HMVECs. A single cell suspension of MDA-MB-231 BCs was treated with sEselectin and then washed with PBS to remove sE-selectin. The cells were infused into a flow chamber at a rate of 1 dyn/cm 2 for 5 min. After a brief wash, the cells were fixed for immunostaining with VE-cadherin. The shearresistant adhesion of sE-selectin-treated MDA-MB-231 BCs to non-activated p-HMVECs resulted in the appearance of a visible gap and disappearance of VE-cadherin surface expression (Fig. 3b). These data suggest that the adhesion of sE-selectin treated CD44 +/high BCs permeabilizes non-activated endothelium.

Soluble E-selectin activates focal adhesion kinase
To determine the underlying mechanism of sE-selectinmediated adhesion and migration, a single cell suspension of MDA-MB-231 BCs was incubated with sE-selectin for 5 to 60 min. Soluble E-selectin triggered the phosphorylation of focal adhesion kinase (FAK) at Tyr-397 after 5 min of incubation; whereas there were no changes in total FAK expression (Fig. 4a). However, sE-selectin did not affect FAK phosphorylation in the CD44sh 28 clone (Fig. 4a), suggesting the involvement of CD44 in sE-selectin-mediated FAK phosphorylation. To confirm the involvement of FAK signaling in sE-selectin-mediated effects, adhesion assays were performed using pathway-specific inhibitors at a concentration that did not affect cell viability (Additional file 1: Figure S2). Suspensions of MDA-MB-231 or HL-60 cells were co-incubated with sEselectin and FAK inhibitor (FAK inhibitor II, 5 μM) for 30 min. After a brief spin followed by washing with PBS, the cells were then infused into parallel flow chambers. FAK inhibitor II abolished sE-selectinmediated shear-resistant adhesion to non-activated HMVECs (Fig. 4b).

Pre-activation of 4 T1 BCs with sE-selectin results in enhanced homing in vivo
To validate the effect of sE-selectin on hematogenous metastasis, we used a mouse model of forced lung metastasis. Murine 4 T1 BCs expressing luciferase (4T1-Luc) were pre-incubated with mouse sE-selectin and injected into wild-type Balb/C or E-selectin knockout mice via tail vein. Seven days after the injection, the mice were perfused with warm saline to remove residual ** a b  (Fig. 5a). The increased extent of lung homing in wild-type mice may be a consequence of endogenous E-selectin expression on the vessel or sE-selectin shed after activation in wild-type mice. To identify cellular adhesion molecules (CAMs) that support endothelial E-selectin-independent/sEselectin-mediated shear-resistant adhesion, functional blocking antibodies against ICAM or VCAM were used to pretreat non-activated p-HMVECs for 1 h prior to the flow adhesion assay. The pre-incubation of non-activated p-HMVECs with ICAM-1 blocking antibody resulted in a significant reduction of sEselectin-mediated shear-resistant adhesion; whereas, VCAM-1 blockade did not affect sE-selectin-mediated shear-resistant adhesion (Fig. 5b).

Pre-activation of human PBMCs with sE-selectin stimulates tumor growth
To confirm pleiotropic effects of sE-selectin, we next explored the biological consequences of leukocyte activation by sE-selectin on tumor growth in vivo. Human PBMCs were freshly isolated from healthy donors and labeled with Calcein AM, then pre-treated with sE-selectin for 30 min at 37°C. The PBMCs were infused via tail vein into NSG mice bearing human breast tumors derived from MDA-MB-231. Infiltration of Calcein positive PBMCs was measured by fluorescence-activated cell sorting (FACS) and visualized by fluorescent microscopy. Following a single intravenous infusion of sE-selectin-treated PBMCs, 1.45 ± 0.08 % of tumor cells were Calceinpositive, whereas, only 0.14 ± 0.1 % of tumor cells were Calcein-positive in mice receiving saline-treated PBMCs. This amounts to a ten-fold increase in PBMC infiltration when PBMCs are exposed to sE-selectin prior to infusion (p < 0.001) (Fig. 6a). Fluorescence microscopic analysis further confirmed that pre-incubation of PBMCs with sEselectin resulted in an increased infiltration of Calceinpositive PBMCs into tumors, compared with tumor from mice infused with saline-treated PBMCs (Fig. 6a). Next, we evaluated whether the infiltration of sE-selectintreated PBMCs into the tumor affected tumor growth. PBMCs treated with or without sE-selectin were infused at day 18 and 25 into NSG mice bearing MDA-MB-231 breast tumors carrying the luciferase gene (n = 5). Tumor growth was measured weekly at day 10, 16, 22 and 29 by bioluminescence. PBMCs were infused, and on day 22 (4 days post-PBMC infusion), we found that infusion of sE-selectin-treated PBMCs had increased tumor growth by 3.6 fold when compared to day 16 (prior to infusion). In contrast, in the mice receiving saline-treated PBMCs, only 1.4 fold tumor growth was noted. The differences in tumor growth rates between these groups were significant at four days after infusion (2.6 fold, p < 0.05). This trend remained the same at 29 days (Fig. 6b). FACS analysis of tumors further confirmed the presence of human CD45+ cells after two consecutive infusions of human PBMCs (Fig. 6c). Infusion of sE-selectin treated of PBMCs increased intratumoral infiltration by human CD4+ Tlymphocytes (20.2 %) significantly more than infusion of saline treated control PBMCs (11.5 %), but did not affect infiltration of CD8+ T-lymphocytes or CD68+ cells. Of note, the CD4/CD8 ratios of PBMCs used in this study were almost the same (CD4: 39.1, CD8: 37.2; Additional file 1: Figure S3). These data suggest that sE-selectin is a biologically functional signaling molecule in circulation after shedding and enhances the shear-resistant adhesion and migration of a broad range of CD44+ circulating cells (metastatic cancer cells or leukocytes), which additively promote disease progression via stromal development and homing of metastatic BCs to the lung.

Discussion
Circulating E-selectin, sE-selectin, is a byproduct of endothelial E-selectin that originates from the external domain of E-selectin expressed on activated endothelium as a result of enzymatic cleavage or shedding from damaged or activated endothelial cells [29]. Several studies have reported chemotactic effects of sE-selectin in different cell types, including endothelial cells and leukocytes [30][31][32]. However, the effects of sE-selectin in tumor progression and metastasis is largely unknown. We and others have previously reported that the engagement of CD44 with endothelial E-selectin mediates shear-resistant adhesion of BCs and leukocytes to endothelial cells [13,14,33]. In addition to describing the roles of endothelial E-selectin, we found that sEselectin functions as a circulating signaling molecule that potentiates shear-resistant adhesion and migration of CD44 +/high BCs and leukocytes in an endothelial E-selectin independent fashion and facilitates tumor progression and metastasis. The adhesion cascade is governed by cell-cell interactions between circulating cells and endothelial cells via a sequential affinity interaction between adhesion molecules and the counter-receptor ligand. Thus, the ablation of adhesion molecules, such as selectins, ICAM-1, or VCAM-1, significantly impairs leukocyte infiltration. Given the E-selectin's integral role in tissue infiltration as the first step mediator of adhesion, endothelial Eselectin-independent homing and shear-resistant adhesion of BCs and leukocytes were unexpected. Our data demonstrate that sE-selectin-mediated activation of the CD44/FAK axis of BCs enables ICAM-1-dependent adhesion at low shear stress, bypassing the requirement for engagement with endothelial E-selectin. A bypass of initial adhesion via endothelial E-selectin may occur when abundant CD44 + cells and circulating sE-selectin encountered endothelium expressing ICAM-1 at low shear stress. Similar bypass mechanism was reported by Thankamony and Sackstein as a novel paradigm of cell migration termed the "step 2-bypass pathway'". The step 2-bypass pathway permits transendothelial migration of mesenchymal stem cells through interaction of endothelial VCAM-1 with activated VLA-4 without engagement of the chemokine receptors (CXCR4 and SDF-1) [34]. These data suggest the presence of alternative adhesion mechanisms that are supported by CAMs. In fact, the expression of CAMs is primarily induced by inflammatory cytokines; however, low levels of constitutive expression occurs in some organs and endothelial cell lines [35][36][37]. Additionally, double knockout of ICAM/ VCAM is lethal [38,39], while triple knockout of selectins (E-, P-, and L-selectin) does not completely ablate tissue infiltration of leukocytes to wounded areas [40]. These collectively indicate the presence of a selectin-independent alternative adhesion, which may be compensated by constitutive CAMs expression. This poses the question of whether individuals with breast cancer and synchronous chronic inflammation who have elevated sEselectin and CD44+ circulating tumor cells (CTCs) are predisposed to increased risk of organ metastasis. Our data suggested that serum sE-selectin levels were associated with tumor size (Additional file 1: Figure S4). Additionally, several studies demonstrated positive correlations between high sE-selectin levels and worse clinical outcome [25,[41][42][43]. Sheen-Chen et al., reported that an elevated level of sE-selectin is significantly associated with advanced tumor size, lymph node involvement, metastasis, high histology grade, and reduced overall survival in breast cancer. The study further indicated that women with ER-negative breast tumors had a significantly higher level of sE-selectin than those with ER-positive breast tumors [41]. Similarly to cancer, elevated sE-selectin was also noted in individuals with chronic inflammation disorders including arthritis [44], diabetes [45], atherosclerosis [46], and alcoholism [47]. Our data demonstrated cellular responses to sE-selectin were concentration dependent and reached a plateau at 100 nM (Additional file 1: Figure S5); therefore, we used 100 nM sE-selectin in this study (equal to approx. 5.8 μg/ml when calculated as MW: 60 kDa). While the association of tissue inflammation and cancer progression has been reported [48,49], future study should address a possible link regarding how chronic inflammation affects cancer cell behavior in vascular space.
The effect of sE-selectin was highly specific to CD44 +/high BCs; its effect was minor or undetectable in CD44 -/low BCs and CD44 K/D MDA-MB-231 BCs. CD44 (HCELL) [34], belongs to a family of cell-surface glycoproteins, and differential forms of CD44 (e.g., CD44s and CD44v) are expressed in a variety of human leukocytes and solid tumors. Despite its lack of intrinsic catalytic activity, CD44 plays a role in the activation of signaling pathways through association with intracellular and extracellular proteins. CD44 interacts with various soluble extracellular components or ECM. Ligand binding to CD44 has been associated with numerous cellular processes such as migration, adhesion, signaling, and proliferation [50]. For example, the interaction of CD44 and its ligand, hyaluronic acid (HA), results in proliferation of eosinophils [51], B-cells [52], T-cells [53], keratinocytes [54], and human melanoma cells [55]. In addition to HA, L-selectin and E-selectin are reported to be ligands for distinct glycoforms of CD44 or CD44v, respectively [56]. The interaction of CD44 with Src family protein kinase, Rho-kinase, Tiam 1, and Vav2 triggers intracellular activation of PI3K, Ras, and Rac1, which promote cell migration, proliferation, and survival [57][58][59][60][61]. Although mounting evidence suggests the involvement CD44 in tumor progression, the roles of CD44 as well as its utility as a biomarker for primary tumors remains controversial due to its dual functions as a tumor promoter or suppressor and altered expression in CD44v [37,38]. However, recent studies highlighted the positive association of CD44 + CTCs with hematogenous metastasis. The presence of CD44 + CTCs is linked to bone metastasis [62] and correlates with poor prognosis in breast cancer [63,64]. While the CD44 has been used as a stem cell marker, our data suggest that CD44 may have functional role in hematogenous metastasis in the circulation 1) as a ligand for endothelial E-selectin at the pre-metastatic niche [13] and 2) as a signaling molecule that mediates sE-selectininduced extrinsic stimuli in circulating cells.
Abundant leukocyte infiltration into the tumor is associated with poor prognosis in many types of cancer including breast, prostate, lung cancer, and melanoma [48,65,66]. Our results demonstrate that sE-selectin enhancement of leukocyte migration is partially mediated by CD44. We used humanized NOD SCID IL2R-γ chain knockout mice, which are deficient in T-lymphocytes, B-lymphocytes, and NK cells, to recapitulate the effect of crosstalk between the human immune system and human cancer cells. Mice receiving human PBMCs treated with human sEselectin showed rapid tumor growth, which coincided with increased intratumoral infiltration of PBMCs, in particular, CD4 + T-lymphocytes (Fig. 6c). Although we expected the CD4+/CD8+ balance shift to trigger proinflammatory signaling that would lead to an increase in the human CD68+ macrophage population, no change was noted in CD68+ cells or CD8+ T-lymphocytes. A separate study showed that sE-selectin-mediated tumor growth was not as significant when mouse splenocytes were infused into athymic nu/nu mice bearing xenograft tumors derived from MDA-MB-231 cells (data not shown), although infiltration of mouse PBMCs into the tumor was more extensive than that of human PBMCs (Additional file 1: Figure S6). Possibly, there is a slight discordance between adaptive immune cells from human and innate immune cells and vessels of mice.
In conclusion, we proposed possible simultaneous and multifaceted effects of E-selectin in tumor growth and metastasis in Fig. 7: endothelial E-selectin mediates the shear-resistant adhesion of circulating tumor cells and immune cells to inflamed vessels (a), but not to noninflamed vessels (b); sE-selectin shed into circulation accelerates homing of CD44 + BCs into tissue via noninflamed vessels, endothelial E-selectin independent/ ICAM dependent manner (c); sE-selectin accelerates the infiltration of CD4+ lymphocytes into the tumor to contribute to the development of inflammatory stromal (d).

Conclusion
Soluble E-selectin (sE-selectin) is shed into the circulation from the activated endothelium. We investigated the effect of sE-selectin on shear-resistant adhesion and migration of metastatic breast cancer cells and leukocytes. Our data suggest that sE-selectin promotes migration and shear- Fig. 7 Scheme of sE-selectin-mediated shear-resistant adhesion and migration: a The expression of adhesion molecules such as E-selectin, ICAM, and VCAM is induced in the luminal surface of inflamed vessel. Margination of leukocytes and CD44 +/high BCs toward the vessel surface is followed by tethering, rolling, and adherence to inflamed vessels. The adhesion of the cells increases the permeability of endothelial junctions facilitating tissue migration of circulating cells. b By contrast, in a non-inflamed vessel, circulating cells remain in circulation and no random adhesion occurs. c Soluble E-selectin is shed from activated endothelium of inflamed tissues and activates both leukocytes and CD44 +/high BCs, which enhances their shear resistant adhesion to non-inflamed vessels and migration in the absence of E-selectin but via interaction with ICAM at low shear stress. Excessive infiltration of CD4 + T-lymphocytes into the tumor leads to development of pro-tumorigenic stroma. Infiltration of CD44 +/high BCs into distant organ increases the risk of colonization by metastatic cells resistant adhesion of circulating cells to non-activated HMVECs via CD44/FAK. This sE-selectin-mediated pleiotropic effects result in enhancement of homing of CD44 +/high BCs and immune cells to the lung and tumor, respectively. These findings identify a possible extrinsic mechanism by which serum factor stimulates circulating tumor cells toward tumor progression and metastasis.

Additional files
Additional file 1: Supplementary Figure 1. Soluble E-selectin enhances the anoikis resistance of ER-negative BCs: BCs were suspended (104 cells) in 1% FBS media with or without sE-selectin, and seeded in ultra-low attachment plates. The data represent the increased percentage of surviving cells with sE-selectin. The data shown are Mean ± SD.