Oncogene (2014) 33, 2898–2908 & 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14 www.nature.com/onc

ORIGINAL ARTICLE CD82/KAI expression prevents IL-8-mediated endothelial gap formation in late-stage melanomas

P Khanna1, C-Y Chung2, RI Neves2,3,4,5,6, GP Robertson2,3,4,7,8,9 and C Dong1,8

Melanoma cells facilitate endothelial gap formation, the first step during tumor transendothelial migration, which is mediated by both adhesion and endogenously produced chemokines (in particular, interleukin-8 (IL-8)). Tetraspanins are localized to the cell surface in cancer and participate in various functions including invasion of tissues mediated by secretion of cytokines and matrix metalloproteinases. However, little is known about the role of CD82 tetraspanins in malignant melanomas during cancer cell invasion. In this study, we investigated the functional importance of CD82 expression in melanoma-mediated gap formation by using cDNAs to induce CD82 expression in highly invasive melanoma cell lines. Results showed that CD82 expression inhibited melanoma cell-induced gap formation, melanoma cell extravasation in vitro and subsequent lung metastasis development in vivo. Mechanistic studies showed that inducible expression of CD82 in highly metastatic melanoma cells significantly increased p21 expression upon binding of Duffy antigen receptor group (DARC), inducing tumor cell senescence and interrupting IL-8-mediated vascular endothelial (VE)-cadherin disassembly. Taken together, these studies provide a rationale for using drug therapies that restore CD82 expression and inhibit IL-8 production to inhibit late-stage melanoma cell extravasation and subsequent metastasis development.

Oncogene (2014) 33, 2898–2908; doi:10.1038/onc.2013.249; published online 22 July 2013 Keywords: melanoma metastasis; CD82; IL-8; adhesion

INTRODUCTION Previous work has shown that CD82 binds to Duffy antigen Metastasis is a complex process requiring melanoma cell receptor/chemokine group (DARC), and furthermore, that DARC detachment from its primary location and migration through expression in endothelial cells is necessary for CD82-mediated 16,17 blood or circulatory systems to secondary sites. Once cancer cells tumor suppression. Binding of CD82 on tumor cells to DARC metastasize to reach distant organs, cancers become increasingly receptors on endothelial cells prevents invasion of CD82-positive difficult to treat using current therapies warranting the need tumor cells by inducing senescence in CD82 þ cells. DARC also to elucidate molecular mechanism during this process.1–3 One binds C–X–C and C–C cytokine receptors (including interleukin 16,18 important event during cancer metastasis is melanoma cell (IL)-8) secreted by highly metastatic melanoma cells. extravasation through the endothelial vessels, which requires4,5 IL-8 has been shown to be an important factor in mediating (i) the breakdown of vascular endothelial (VE)-cadherins to form tumor cell extravasation through endothelial cells and underlying 19,20 gaps, (ii) penetration of the endothelial barrier and (iii) migration extracellular matrix. IL-8 binds with high affinity to CXCR2 and 21,22 of tumor cells through blood vessels into tissues to form DARC rather than CXCR1. Studies in fibroblasts have shown metastatic lesions. These events are mediated by adhesion that knocking down DARC diminishes cell senescence and molecules, inflammatory cytokines and proteases that degrade increases tumor cell extravasation.18,23,24 In normal physiology, the underlying extracellular matrix.5–7 DARC on endothelial cells modulates IL-8 signaling by acting as a Tetraspanins are a family of membrane proteins present in all ‘chemokine sink’ where DARC binds to chemokines, but does not tissue types that have a vital role in a wide spectrum of biological activate downstream signaling pathways.25 During tumor-induced functions.8,9 In cancers, tetraspanins (including CD82) have a vital inflammation, excess secretion of IL-8 within the tumor role in cellular invasion, particularly during cell movement.10–12 microenvironment disrupts this homeostasis. In melanoma cells, Studies have found that expression levels of CD82 tetraspanins these effects are exacerbated by synergistic secretion of IL-8 from decrease in highly metastatic cancers, increasing tumor endothelial cells bound to melanoma cells, facilitating migration cell extravasation.13,14 While the metastatic-suppressive effects of melanoma cells through the blood vessels.26–31 Upon IL-8 of CD82 have been demonstrated in cell, animal and patient binding to CXCR2 receptors, signaling pathways such as models,12,13,15 the complex molecular mechanism by which CD82 phosphatidylinositol-3 kinase and/or p38 mitogen-activated expression suppresses tumor cell extravasation remains unclear. protein kinase are activated. This signaling cascade inhibits

1Department of Bioengineering, The Pennsylvania State University, University Park, PA, USA; 2Department of Pharmacology, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 3Department of Dermatology, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 4Penn State Melanoma Therapeutic Program, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 5Cutaneous Oncology Program, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 6Department of Surgery, Division of Plastic Surgery, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 7Department of Pathology, The Pennsylvania State University College of Medicine, Hershey, PA, USA; 8Pennsylvania State Melanoma Center, The Pennsylvania State University College of Medicine, Hershey, PA, USA and 9The Foreman Foundation for Melanoma Research, The Pennsylvania State University College of Medicine, Hershey, PA, USA. Correspondence: Professor C Dong, Department of Bioengineering, The Pennsylvania State University, 205 Hallowell Building, University Park, PA 16802, USA or Professor GP Robertson, Department of Pharmacology, The Pennsylvania State University College of Medicine, Hershey, PA 17033, USA. E-mail: [email protected] or [email protected] Received 22 November 2012; revised 23 March 2013; accepted 19 April 2013; published online 22 July 2013 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2899 tumor cell senescence and trigger internalization of VE-cadherin western blot analysis (Figure 1e, right panels). As CD82 À cells junctions in the endothelial cells forming gaps on endothelial migrate under low shear stresses (0.625 dyn/cm2) rather than high monolayers.32–38 Although several studies have focused on shear stresses (2 dyn/cm2), these results show that CD82 mediates CD82 and DARC interactions, the role of CD82 in regulating IL-8 the motility and subsequent extravasation of tumor cells rather signaling remains unclear. than adhesion of tumor cells. However, the role of CD82 in tumor In melanoma cells, IL-8 binding to endothelial cells have an cell extravasation remains unknown. important role in regulating VE-cadherin disassembly via activa- CD82 expression in highly metastatic lines dramatically reduces tion of CXCR2 receptors.37 However, previous literature shows that in vitro gap formation. As VE-cadherin junctions serve as the ectopic expression of CXCR2 in tumor cells results in tumor cell primary mediators of endothelial permeability,37,41 we senescence.18 These studies show that during oncogene-induced hypothesized that CD82 regulates these endothelial junctions. senescence there is an increase in IL-8 that binds CXCR2 receptors Human umbilical vein endothelial cells (HUVECs) were grown to on tumor cells to reinforce tumor cell growth arrest 98–99% confluency on coverslips and cocultured with A375M and migration.18 In cancer cells, cell senescence can be reversed melanoma cells transfected with 3.1 vector or CD82 cDNAs. by knocking down CXCR2 expression.16,39 Although several Quantification of the total gap area shows a significant reduction studies have found that CD82 binding DARC induces tumor cell in the total gap area and gap sizes in cells transfected with CD82 senescence via a p21-dependent mechanism,40–47 the effects of (Figure 2a, left) compared with the control cases (3.1 vector), CD82 on IL-8 have not been elucidated. where total gap area and sizes of the gaps are much larger. In this work, we showed that ectopic CD82 expression These results lead to the conclusion that high levels of CD82 suppresses IL-8 secretion and that binding of CD82 tetraspanins expression prevented gap formation. Conversely, knocking down to DARC receptors interrupts IL-8 signaling in endothelial cells. The CD82 expression using small interfering RNA (siRNA) increases gap p21/Waf1 protein is a cyclin-dependent kinase inhibitor, which formation (Figure 2a, right). Gap formation is one important step binds active cyclin-dependent kinase 2 or cyclin-dependent kinase during tumor cell extravasation.37 We found that reintroducing 4 complexes.43,44 Inhibition of cyclin-dependent kinase complexes CD82 expression in highly invasive melanoma cells decreased prevents progression of cells through the G1 phase of the cell tumor cell extravasation through HUVEC monolayers after 4 h cycle, resulting in growth arrest of cells or cellular senescence. (Figure 2b, top and bottom left panels). CD82 expression in cells Consistent with previous work, melanoma cells expressing was analyzed using western blot analysis (Figure 2b, top and physiological levels of CD82 bound DARC while inducing bottom right panels). These results show that CD82 is an p21/Waf1 expression in melanoma cells. We extended previous important mediator of VE-cadherin-mediated gap formation. findings to show that CD82 expression in highly invasive melanomas binds available DARC receptors on endothelial CD82 expression in highly metastatic lines dramatically reduces cells, interrupting IL-8 binding to the endothelial cell surface gap formation in vitro and invasion of blood vessels and subsequent intracellular signaling within endothelial cells. To assess whether CD82 affected tumor cell invasion through Furthermore, these binding events induce p21-induced cell 6 48 blood vessels, 1 Â 10 GFP-A375M melanoma cells were nucleo- senescence preventing melanoma cells with mutant BRAF to fected with either empty vector (3.1 vector) or CD82 cDNAs and migrate through blood vessels, in vivo. These results show that injected into the lumen of human blood vessels, which were then CD82 expression suppresses tumor effects not only via DARC placed in Dulbecco’s modified Eagle’s medium (DMEM) with 10% binding, but via regulation of IL-8 secretion, making CD82 an fetal bovine serum (FBS) for 3 days. Harvested vessels were fixed effective therapeutic target for late-stage cancer metastasis. and stained for VE-cadherin (red regions). Figure 3a (right panels) shows the disruption of VE-cadherin and invasion of blood RESULTS vessels up to 40 mm, where GFP-labeled A375M and UACC 903M melanoma cells made contact with the vessels. Wild-type A375M Restoring CD82 expression in highly metastatic melanoma cells reduces invasion into lung tissues and UACC 903M melanoma cells attached to the lumen of blood vessels, invaded the extracellular matrix and moved to regions CD82 expression was detected in eight patient samples using near the outer walls of the blood vessel (Figures 3a and b, right western blot analysis. The majority of patients showed low levels panels). Consistent with in vitro observations, GFP-labeled A375M of CD82 comparable to invasive melanomas such as A375M and and UACC 903M melanoma cells with inducible CD82 expression UACC 903M cells (Figure 1a). To test the hypothesis that CD82 failed to invade deeply through human blood vessels (Figures 3a mediates melanoma cell invasion and formation of metastatic and b, left panels). Quantification of the number of A375M and lesions, green fluorescent protein (GFP)-tagged UACC 903M or UACC 903M melanoma cells invading the vessel lumen showed A375M cells were injected via the tail vein into nude mice. Six that GFP-A375M melanoma expressing CD82 failed to invade mice from the control or experimental group were killed 2 weeks human vessels beyond 10 mm compared with the control cells, later. Compared to vector-transfected cells (3.1 vector), in vivo where the cells invaded blood vessels up to 30 and 40 mm results showed that CD82-expressing tumor cells significantly distances from the lumen (Figures 3c and d). decreased the appearance of metastatic lesions in the lungs of nude mice (Figures 1b–d). High CD82 expression in nucleofected A375M and UACC 903M melanoma cells injected in mice was CD82 expression reduces secretion of IL-8 confirmed using western blots (Figures 1b and c, right panels). In A375M and UACC 903M melanoma cells, secretion of IL-8 in These levels were comparable to endogenous CD82 expression in 0.3 Â 106 cells was measured using enzyme-linked immunosor- primary non-invasive melanoma cell lines (WM35 and WM793 bent assay (ELISA). Control melanoma cells secreted B120 pg/ml melanoma cells) (Figures 1b and c). In blots 1a, 1b and 1c, total of IL-8 compared to melanoma cells with CD82 expression, where levels of ERK1, ERK2 or a-enolase was analyzed to ensure equal IL-8 secretion decreased by approximately 50% (Figure 4a). These loading of proteins. results were consistent in A375M and UACC 903M melanoma cells In vitro flow chamber experiments showed that inducible CD82 cocultured with HUVECs for 4 h (Figure 4b). SiRNA targeting CD82 expression in A375M and UACC 903M melanoma cells decreased and IL-8 mRNA were nucleofected into A375M melanoma cells. extravasation under low shear flow compared with the control After 24–48 h incubation, cells were lysed and proteins were cases (Figure 1e, left panels), suggesting that CD82 expression separated using sodium dodecyl sulfate–polyacrylamide gel decreased tumor cell extravasation. CD82 expression in cells used electrophoresis and cell supernatants collected for ELISA analysis for in vitro flow chamber experiments was confirmed using to assess knockout efficiency (Figures 4c and d). Compared to

& 2014 Macmillan Publishers Limited Oncogene (2014) 2898 – 2908 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2900 the buffer and scrambled cases, cells nucleofected with siRNA or IL-8 siRNA decreased melanoma cell extravasation by 60% targeting CD82 or IL-8 in CD82 þ A375M melanoma cells resulted (Figure 4e). Previous studies have shown that CD82 binds DARC in a 20% decrease in endothelial gap formation (Figure 4e). receptors on endothelial cells to induce tumor cell senescence To support our hypothesis that CD82 mediates tumor cell via p21/Waf-1 expression.16 As CD82 expression regulates IL-8 extravasation, a Boyden chamber was used to measure tumor secretion, we hypothesized that CD82 binding of DARC receptors cell extravasation when knocking down IL-8 or CD82 expression. prevented endothelial response to IL-8 and induced cell We showed that A375M melanoma cells nucleofected with CD82 senescence via an increase in p21 expression.

Figure 1. CD82 expression in highly metastatic melanoma cells dramatically reduces their invasion into lung tissues. (a, left) Western blot analysis shows that highly invasive melanoma cells, UACC 903M and A375M, have little or no CD82 expression compared with primary or less invasive WM35 melanoma cells, which have high levels of CD82 expression. Right: Samples from melanoma patients were probed for CD82 expression using western blot analysis. Seven out of eight melanoma patients show no CD82 expression. (b, left) Prominent lung lesions when A375M tumor cells transfected with empty (3.1) vectors are injected into mice compared to mice injected with CD82-expressing A375M cells (middle). Images are representative of six separate experiments ( Â 40). Right: Western blot analysis shows that nucleofected CD82 expression in A375M melanoma cells is comparable to primary melanoma cells (WM35 cells). (c, left) Prominent lung lesions when UACC 903M tumor cells with empty (3.1) vectors are injected into mice compared with CD82-expressing UACC 903M cells (middle). Images are representative of six separate experiments ( Â 40). Right: Western blot analysis shows that nucleofected CD82 expression in UACC 903M melanoma cells is comparable to primary melanoma cells (WM793 cells). (d, left and right) Lesions were quantified; A375M and UACC 903M cells with CD82 expression show a significant decrease in lung metastasis compared with the control cases. (e) Under a shear stress of 2 dyn/cm2, melanoma cell extravasation remained low when compared with the control case (top left panel). Under a shear stress of 0.625 dyn/cm2, CD82-expressing melanoma cell migration significantly decreased compared with the control case (3.1 vector) (bottom left panel). Western blot analysis shows that wild-type and siRNA-treated UACC 903M and A375M melanoma cells have no CD82 expression compared with high levels of CD82 in nucleofected UACC 903M and A375M melanoma cells. (Top and bottom right panels) Experimental and control values represent the means±s.e.m. (*Po0.05).

Oncogene (2014) 2898 – 2908 & 2014 Macmillan Publishers Limited Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2901

Figure 1. Continued.

CD82 expression inhibits IL-8 binding to HUVECs through interaction with DARC receptors Table 1. Summary of novel concepts and finding CD82 expressing A375M melanoma cells adhered to confluent Figure and Conclusions HUVECs after 1 h, the number of melanoma cells bound to results confluent HUVECs decreased when A375M melanoma cell and HUVEC cocultures were incubated with either anti-CD82 or anti- 1a–c CD82 expression disappears in most patients with DARC antibodies. To assess whether CD82 directly binds DARC late stages of metastatic melanoma receptors on endothelial cells, we measured cell adhesion to 1b and 3 Restoration of CD82 expression prevents invasion of confluent HUVEC monolayers after 15 min in the presence of CD82 melanomas into vital organs such as the lung or DARC antibodies. While CD82 expressing A375M and UACC 4a and b Restoration of CD82 expression prevents secretion of interleukin-8, a chemoattractant involved in 903M melanoma cells remained bound to HUVECs after 15 min, directing melanoma migration and movement the number of cells adhered to HUVECs decreased by nearly 50% in the presence of CD82 or DARC antibodies after 1 h (Figures 5a and b). The attachment of GFP-A375M melanoma cells transfected with empty vector (3.1 vector) showed little change in attachment found that potent inflammatory cytokines influence melanoma to HUVEC monolayers when incubated with CD82 or DARC development,49–51 the exact mechanisms by which inflammatory antibodies for either 15 min or 1 h (Figures 5a and b). These results cytokines are regulated are not well understood. Furthermore, the show that antibodies targeting DARC and CD82 interactions role CD82 and other tetraspanins have in mediating chemokines decrease melanoma cell adhesion over long time periods. To remains unknown. In particular, no studies have shown how CD82 assess whether IL-8 binds DARC receptor in the presence of CD82, regulates IL-8 expression and secretion during tumor/endothelial we immunoprecipitated DARC receptors with CD82 and/or IL-8 interactions to promote tumor migration and metastasis proteins from bound melanoma and HUVEC cells. In these development. As outlined in Table 1, this is the first study experiments, GFP-A375M or UACC 903M melanoma cells were to show that CD82 regulates IL-8 secretion and subsequent cocultured with HUVECs for 24 h (in the presence of crosslinkers, VE-cadherin disassembly to promote melanoma cell extravasation, DTSSP). These results show that in cocultures with wild-type and development of metastasis. melanoma cells and HUVECs, DARC binds IL-8 (Figure 5c). This study shows that invasive melanoma cells secrete However, in the case of CD82 expressing A375M and UACC high levels of IL-8, promoting metastasis melanomas,52 which 903M melanoma cells, the binding of DARC to CD82 activates p21/ can be modulated by CD82. Expression of CD82 in melanoma Waf 1 expression, which has been shown to induce melanoma cell cells interrupted IL-8-mediated activation of endothelial cells, senescence and prevent tumor cell extravasation through triggering tumor cell senescence. Thus, reagents promoting the endothelial monolayers (Figure 5c). These findings suggest that expression of intracellular CD82 in melanoma cells reduced the direct binding of CD82 and DARC receptors interrupt binding secretion of IL-8 and subsequent melanoma cell invasion, of IL-8 to DARC receptors on HUVECs, preventing invasion through suggesting that therapies that restore CD82 expression could endothelial cells. disrupt the metastatic process. The extravasation of melanoma cells was significantly reduced under low shear flow conditions following expression of CD82 in DISCUSSION melanoma cells. These results were confirmed using animal The progression of melanoma cells from an early benign to studies where CD82 expression diminished metastatic lesions malignant state is associated with an increase in cytokine in vivo, reducing the number of lesions by nearly 95%. Under low secretion and diminishing CD82 expression. While studies have shear stresses (0.625 dyn/cm2), late-stage melanoma cells could

& 2014 Macmillan Publishers Limited Oncogene (2014) 2898 – 2908 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2902

Figure 2. CD82 expression in highly metastatic lines dramatically reduced gap formation in vitro.(a) Percent endothelial gap significantly decrease with CD82 expression in A375M and UACC 903M melanoma cells (left panel). Conversely, in cocultures of WM35 primary melanoma cells nucleofected with CD82 siRNAs, there is a significant increase in gap formation (right panel). (*Po0.05 compared with the control cases (3.1 vector).) Values are means±s.e.m. (b) In vitro tumor cell extravasation after HUVECs were cocultured with WM35 cells for 4 h is significantly higher in WM35 cells treated with CD82 siRNA compared with the control case (top left panel). Conversely, A375M and UACC 903M cell extravasation with CD82 expression is significantly lower compared with wild-type melanoma cells cocultured with HUVECs (bottom left panel). Right: Western blot analysis shows nucleofection efficiency when WM35 cells are nucleofected with siRNAs targeting CD82 and UACC 903M/A375M cells are nucleofected with CD82 cDNAs. (*Po0.05 compared with the control cases (3.1 vector).) Values are means±s.e.m.

not extravasate through endothelial monolayers, suggesting that These studies suggest that inducible DARC expression may not CD82 tetraspanins mediate the motility of melanoma cells during be sufficient to modulate excess chemokines secretion into the tumor/endothelial interactions. Previous studies have shown tumor microenvironment. Thus, CD82 could provide a therapeutic that VE-cadherin disassembly and subsequent gap formation is target that would potentially decrease metastasis development by an important event during tumor cell extravasation.38,53 This is the reducing IL-8 secretion from melanoma cells. New technologies first study to show that p21 expression in melanoma cells with the potential to target and reduce IL-8 levels in melanoma increases after CD82-expressing melanoma cells bind to DARC cells have also been developed to carry siRNAs54 or an antibody- receptors, indicating a potential mechanism for reduced based immunotherapy specifically targeting secreted IL-8.29 extravasation under low shear stresses. This observation is The obstacles for siRNA-containing liposomes for IL-8 or CD82 supported by previous studies showing that extravasation is siRNAs would be efficiencies high enough for each cell to undergo interrupted by p21-induced cell senescence under static endocytosis and uptake the siRNA to dramatically decrease conditions.41 secreted IL-8. Similar issues arise in late-stage cancers where Consistent with previous studies, we found that knocking down targeting tumor cells and inducing DARC expression has been CD82 dramatically increased gap formation and subsequent suggested to reduce cancer metastasis.16,23 In targeting CD82 tumor extravasation. Previous studies have reported that an using siRNAs would reduce adverse side effects since CD82 is increase in tumor cell extravasation is primarily due to disruption expressed in normal cells. of binding between DARC and CD82 that prevents tumor cell In conclusion, we have shown that targeting CD82 reduces senescence.37,45 While DARC has been described as a ‘chemokine melanoma metastasis, which is mediated through a reduction of sink’ in erythrocytes, in endothelial cells DARC binds to melanoma cell extravasation through the endothelium. Mechan- chemokines and translocates these chemokines to the surface of istically, the reduction in melanoma cell extravasation following endothelial cells.25 In mice with DARC deficiency, vascular expression of CD82 is due to the binding of CD82 to DARC permeability dramatically decreased, but these effects were receptors. These events disrupt IL-8 binding to endothelial cells diminished upon the activation of CXCR2 receptors.25 and induce p21 expression in tumor cells. Therefore, our study

Oncogene (2014) 2898 – 2908 & 2014 Macmillan Publishers Limited Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2903

Figure 3. CD82 expression decreases gap formation and subsequent invasion of blood vessels in late-stage melanoma cells. (a) A375M melanoma cells with inducible CD82 expression have little invasion of blood vessels, about 10 mm from the lumen (left) compared with the control cases, where there are large gaps within blood vessel walls and invasion of blood vessels up to 20 and 30 mm (right). (b) UACC 903M melanoma cells with CD82 expression, there is little gap formation and invasion of blood vessels, about 10 mm from the lumen (left) compared with the control cases, where there are large gaps within blood vessel walls and invasion of blood vessels up to 20 and 30 mm (right). (c and d) Quantification of the number of invading tumor cells shows a dramatic decrease in A375M and UACC 903M melanoma cell invasion in cells with CD82 expression. (*Po0.05 compared with the control cases (3.1 vector).) Results are a representative of three experiments. ECM, extracellular matrix.

& 2014 Macmillan Publishers Limited Oncogene (2014) 2898 – 2908 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2904

Figure 4. CD82 expression decreased IL-8 secretion from melanoma cells. (a) IL-8 protein levels are higher in 0.3 Â 106 A375M melanoma cells, but decrease when melanoma cells are transfected with CD82 cDNAs. Levels of secreted IL-8 protein were compared with the control cases, showing decreased levels in more metastatic cells with CD82 expression (left). (*Po0.05 compared to control melanoma cells (3.1 vector).) (b) Significantly lower levels of IL-8 tend to be secreted from metastatic cells transfected with CD82 cDNAs, while in coculture with HUVECs (right). (*Po0.05 compared to control melanoma cells (3.1 vector).) (c) To assess knockout efficiencies for IL-8 and CD82 siRNAs, whole-cell lysates were assessed using western blot analysis. (d) To assess knockout efficiencies of IL-8 siRNAs, cell supernatants were collected after 24–48 h and IL-8 concentrations were measured using ELISA. (e) Decreasing IL-8 secretion or CD82 expression in late-stage melanoma cells (A375M) tend to result in a dramatic decrease in gap formation compared with the control cases (left) and subsequent tumor cell extravasation (right). (*Po0.05 compared with control melanoma cells (scramble or 3.1 vector).) Data represent mean±s.e.m.

suggests that targeting CD82 signaling may offer a potential (provided by Dr Meenhard Herlyn, Wistar Institute, Philadelphia, PA, USA) mechanism for therapeutic inhibition of IL-8 function in mela- were maintained in Roswell Memorial Park Institute (RPMI) supplemented noma cancer. with 10% FBS and 100 U/ml of penicillin–streptomycin. All cells were cultured in a humidified incubator at 37 1C and 5% CO2.

MATERIALS AND METHODS Cell lines and culture conditions Transfection of cDNAs and selection of stable clones HUVECs were purchased from the American Type Culture Collection Five micrograms of pcDNA3.1 hemagglutinin-tagged CD82 or pcDNA3.1 (Manassas, VA, USA) and were maintained in F12-K medium with 10% FBS, vector control constructs were nucleofected into A375M and UACC 903M 30 mg/ml of endothelial cell growth supplement, 50 mg/ml heparin melanoma cells. Transfection efficiencies were assessed using pMaxGFP (Mallinckrodt Baker Inc., Phillipsburg, NJ, USA) and 100 U/ml plasmid (Amaxa Biosystems, Koeln, Germany) as a GFP expression plasmid of penicillin–streptomycin (Biofluids Inc., Rockville, MD, USA). NHEM cells control. Following nucleofection, the cells were plated in 100 mm dish and (normal human epidermal melanocytes) were obtained from American grown to 60–70% confluency before selecting antibiotic-resistant clones. Type Culture Collection and maintained in melanocyte growth medium Puromycin-resistant cells overexpressing CD82 were isolated and CD82 supplemented with 100 U/ml of penicillin–streptomycin and 10% FBS. expression levels were confirmed using western blotting. Constitutively A375M and UACC 903M were kindly provided by Dr Gavin P Robertson active CD82 levels in nucleofected A375M and UACC 903M melanoma cells and cell lines were maintained in DMEM (Invitrogen, Carsbad, CA, USA) were compared with endogenous CD82 expression in primary melanoma supplemented with 10% FBS. WM35 and WM793 melanoma cells cell lines (WM35, WM793).

Oncogene (2014) 2898 – 2908 & 2014 Macmillan Publishers Limited Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2905

Figure 5. CD82 expression interrupts IL-8 binding to endothelial cells and induced melanoma cell senescence by binding DARC receptors. (a) Higher levels of CD82 expression in A375M melanoma cells results in adhesion to DARC receptors on confluent endothelial monolayers, these adhesion events decrease by nearly 50% after incubating cocultures with anti-DARC ( þ DARC monoclonal antibody (mAb)) or anti-CD82 ( þ CD82 mAb) antibodies for 1 h (left), while CD82 and DARC antibodies have no effects on melanoma cells transfected with 3.1 vector after 15 min or 1 h (right). (*Po0.05 compared experimental cases with CD82 control cases.) Data represent mean±s.e.m. (b) Higher levels of CD82 expression in UACC 903M melanoma cells result in adhesion to DARC receptors on confluent endothelial monolayers (CD82 control), these adhesion events decrease by nearly 50% after incubating cocultures with anti-DARC ( þ DARC mAb) or anti- ( þ CD82 mAb) CD82 antibodies for 60 min (left), while these antibodies (3.1 vector þ DARC mAb and 3.1 vector þ CD82 mAb) have no effects on melanoma cells transfected with 3.1 vector after 15 min or 1 h (right). (*Po0.05 compared experimental cases with CD82 control cases.) Data represent mean±s.e.m. (c) Immunoprecipitation (IP) blots show that DARC directly binds CD82 (left) to induce p21 expression (middle; whole-cell lysate). In the absence of CD82 expression, control melanoma cells cannot bind DARC receptors because there is little CD82 expression (left) and IL-8 binds these receptors to activate endothelial layers (left). siRNA targeting CD82 and IL-8 male and female patients were analyzed for CD82 expression as SiRNA (100 pmol) was introduced into 1.0  106 A375M melanoma cells via summarized in Table 2. nucleofection using an Amaxa Nucleofector using Solution R/program 52 52 K-17. Transfection efficiency was 495% with 80–90% cell viability. Animal studies. Animal experimentation was performed according to Following siRNA introduction, cells were allowed to recover for 2 days in protocols approved by the Institutional Animal Care at the Pennsylvania culture dishes before passaging cells into 96-well plates. After 5 days, cell State University as previously described elsewhere.55 Tumor formation viability was measured using an MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3- was measured in athymic female nude mice purchased from Herlan carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt) assay Sprague–Dawley (Indianapolis, IN, USA). Briefly, 1.0  106 A375M or UACC 52 (Promega, Madison, WI, USA). Duplexed Stealth siRNA (Invitrogen) were 903M melanoma cells nucleofected with empty vector (3.1 vector) or CD82 used for these studies. The following siRNA sequences were used: were collected in 0.2 ml of Hank’s balanced salt solution and then injected scrambled, 50-AAUUCUCCGAACGUGUCACGUGAGA-30; IL-8 no. 1, 50-GCAG intravenously in 4-to 6-week-old female mice. Mice were killed 17 days CUCUGUGUGAAGGUGCAGUUU-30; IL-8 no. 2, 50-CCAAGGAGUGCUAAAGAA after cell injection and the presence of fluorescent metastatic lesions was CUUAGAU-30; and CD82, 50-UCUCGAAUGAGCUCAGUCACGAUGC-30. detected using a Nikon SMZ 1500 dissecting microscope. Fifteen images of random fields were photographed at a magnification of  40 from each lung and the number of fluorescent lesions counted. Each control or Surrogate assays experimental group consisted of six mice. Patient studies. Patient samples were obtained following protocols approved by the Pennsylvania State University Institutional Review Board In vitro flow migration assay. Tumor cell migration was measured using (IRB). Patients were in the late stage of melanoma and samples from both a modified 48-well chemotactic Boyden chamber consisting of a top

& 2014 Macmillan Publishers Limited Oncogene (2014) 2898 – 2908 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2906 and bottom plate separated by a gasket.24 Before each experiment, a monolayer of HUVECs were grown on sterile polyvinylpyrrolidone-free Table 2. Site of tumor formation in patient samples polycarbonate filters (8 mm pore size; NeuroProbe, Gaithersburg, MD, USA) precoated with 2% gelatin diluted in PBS, for 2 h (Sigma-Aldrich, St Louis, Patient Gender Age Site of tumor formation MO, USA). The center 12 wells of bottom plate were filled with soluble number (years) chemoattractant type IV collagen (100 mg/ml diluted in RPMI 1640 with 1 Male Unknown Unknown 0.1% bovine serum albumin; BD Biosciences, San Jose, CA, USA) and 2 Female 50 Metastatic melanoma, lymph surrounding control wells were filled with RPMI 1640 containing 0.1% 3 Female 54 Metastatic melanoma, spleen bovine serum albumin. Previous work has shown that melanoma cells 4 Male 63 Metastatic melanoma, lymph migrate toward collagen IV acting as a chemoattractant.28 The apparatus 5 Male 54 Metastatic melanoma, lung was assembled by laying the filter on the bottom plate, followed by a 6 Male 75 Metastatic melanoma, lymph gasket and top plate. The chamber was primed with RPMI 1640 and 0.1% 7 Female 39 Metastatic melanoma, lung bovine serum albumin to eliminate bubbles from the system. For the 8 Female 76 Metastatic melanoma, lymph migration assay, 2  106 GFP-labeled melanoma cells were placed in the chamber and subjected to a shear flow of 0.625 dyn/cm2 for 4 h in a 37 1C incubator. The total number of cells that migrated after 4 h was counted by Immunoprecipitation and western blots. For co-immunoprecipitation of imaging 12 fields using an inverted NIKON microscope (NIKON CD82 and DARC, we mixed the CD82 þ tumor cells (A375M or UACC Corporation, Tokyo, Japan). Analysis was carried out using images melanoma cells) with HUVECs in the presence of the cell-impermeable captured with NIS elements software (NIKON Corporation). crosslinker DTSSP (3,30-dithiobis(sulfosuccinimidylpropionate)) for 30 min at 24 1C. Whole-cell extracts were prepared by resuspending cells in 2 ml of In vivo imaging of human vessels. All patients were given informed lysis buffer (1% NP-40, 10 mM Tris, pH 8.0, 150 mM NaCl, 3 mM MgCl2 and consent. All experiments using human material were undertaken accord- 2mM phenylmethanesulfonylfluoride). Lysates were incubated on ice for ing to protocols approved by the Institutional Review Board Committee at 30 min, followed by centrifugation at 16 000 g for 5 min at 4 1C. The pellet The Pennsylvania State University. All procedures were performed was discarded and the supernatant was immunoprecipitated with according to protocols approved by the Penn State Human Subjects antibody to DARC in the presence of protein G agarose beads. After Protection Office. Human veins from human subjects were isolated immunoprecipitation, we analyzed bound proteins by western blot using from discarded patient surgery material remaining from routine surgical CD82 (1:1000) and IL-8 (1:1000) antibodies. The lysate was mixed with procedures and divided into 10-mm sections. Approximately 2 cm of 2  sodium dodecyl sulfate running buffer (0.2% bromophenol blue, 4% human vessels were clamped at both ends before using 27-G needles, sodium dodecyl sulfate, 100 mM Tris, 200 mM dithiothreitol, 20% glycerol) in which were used to inject 1.0  106 A375M cells with stable pcDNA3.1 or a 1:1 ratio. Forty microliters was loaded onto a 10% sodium dodecyl CD82 expression. Human vessels containing GFP melanoma cells were sulfate–polyacrylamide gel electrophoresis gel and proteins were trans- maintained in DMEM (Invitrogen) supplemented with 10% FBS (Thermo ferred to 0.2-mm nitrocellulose membrane (Schleicher and Schuell, Keene, Scientific Hyclone, Logan, UT, USA) and penicillin/streptomycin (Invitrogen) NH, USA) by electroblotting. Primary antibodies included anti-CD82 for 11 days. Each section was stained using anti-human VE-cadherin (mouse monoclonal IgG; Invitrogen), anti-IL-8 (mouse monoclonal IgG; antibodies (dilution 10:1000 in PBS/calf serum (CS)/goat serum (GS) R&D Systems, Minneapolis, MN, USA), anti-b-actin IgG1 (Cell Signaling solution containing PBS, 2% GS, 5% CS and 0.3 Triton-X). Sections were Technologies, Danvers, MA, USA) and anti-p21 (human polyclonal IgG; incubated in antibody overnight at 4 1C and then washed two times with Santa Cruz Biotechnologies, Santa Cruz, CA, USA). Secondary antibodies PBS. Sections were incubated using secondary 594 Alexa Fluor goat anti- were peroxidase-conjugated goat anti-mouse IgG and goat anti-rabbit IgG. mouse antibodies (dilution 1:1000 in PBS/CS/GS) for 1 h at room Proteins were detected using the Enhanced Chemiluminescence Detection temperature before washing each section two times with PBS. Each System (Amersham Pharmacia Biotech, Arlington Heights, IL, USA). 1-mm section of the vessels was captured ranging from the glass surface of Nitrocellulose membranes were subsequently stripped and reprobed the slide towards the lumen of the vessel using an Olympus Fluoview 1000 using anti-b-actin as described above. confocal microscope (Olympus, Center Valley, PA, USA) with a PlanApo 60X/1.4 oil immersion lens and then reconstructed using 3-D Autoquant Enzyme-linked immunosorbent assay. Cell-free media from 3.1- or deconvolution software (Media Cybernetics, Rockville, MD, USA). The CD82-expressing A375M, and UACC cultures were collected after 24 h. number of melanoma cells in each section of the reconstructed vessel was The medium was spun at 430 g and supernatants were extracted from cell quantified and plotted using Sigma Plot software (Systat Software Inc, debris. Supernatants were stored at À 20 1C until use. Sandwich ELISA for San Jose, CA, USA). individual cytokines was performed at the Pennsylvania State University General Clinical Research Center using standard protocols. Briefly, each 48- Fluorescence imaging and analysis well plate was coated with the appropriate mouse anti-human capture antibody diluted in 0.1 M NaHCO3 (pH 8.2) at a final concentration of Before starting the experiments, 25-mm coverslips were washed with PBS 2 mg/ml (R&D Systems). Plates were incubated overnight at 4 1C. Next day, and then equal concentrations of HUVECs were then grown to 95–99% each plate was blocked for 2 h at room temperature using PBS with 1% confluency. HUVECs were cocultured with melanoma cells for 45 min. bovine serum albumin. Samples and standards were added at 100 ml per Cells were then fixed with 5% formaldehyde in PBS for 10 min. Cells well and incubated overnight at 4 1C (R&D Systems). Finally, wells were were permeabilized with PBS/CS/GS solution for 20 min. Coverslips were incubated for 2 h at room temperature in detection antibody (concentra- incubated for 1 h in primary VE-cadherin antibodies (10:1000 dilution in tion: 5 mg/ml) (R&D Systems). Plates were read on a Packard spectacount PBS/CS/GS solution) and washed two times with PBS. Finally, each of the plate reader at a wavelength of 405–415 nm. coverslips was incubated with Alexa Fluor 488/520 goat anti-mouse immunoglobulin G (IgG) antibody (dilution of 1:1000). The coverslips were Adhesion assay. HUVECs were seeded into 24-well plates and grown to incubated at room temperature for 1 h and rinsed with PBS three times confluency. We trypsinized A375M and UACC 903M melanoma cells before imaging using a NIKON TE-2000 microscope. For each experimental and resuspended them in DMEM with 10% FBS. Following recovery, 103 condition, one coverslip was viewed under a  40 objective and a series of melanoma cells were incubated with the confluent bottom cell layers in six images were taken of randomized fields using an EZ coolsnap camera. the presence or absence of antibody against CD82 or DARC antibody. After Each image was then analyzed using the Image J software version 1.32 15 min or 1 h, wells were washed with DMEM three times and then the (National Institutes of Health, Bethesda, MD, USA).37 cells were incubated for 12 h at 37 1C. The number of GFP-labeled melanoma cells attached to the confluent HUVEC monolayers were Analysis of gaps and disruption of VE-cadherin counted using a NIKON fluorescence microscope and the percentage of Disruption of VE-cadherin was identified by areas lacking green attached cells was calculated. For each well, we counted eight fields and fluorescence at VE-cadherin junctions between HUVECs. Gap area within averaged the number of cells in each well. disrupted VE-cadherin junctions was determined from six images (see arrows in Figure 2b in the Results section as an example). Gap area Statistical analysis. Statistical significance for multiple comparisons was quantified as the ratio of pixels within all the gaps and the total between groups was conducted using analysis of variance, followed by number of pixels in one image.37 The average percent of endothelial gaps appropriate post hoc tests. For comparisons between two samples, t-tests was calculated from six images and plotted as a function of time. were used. Results were considered significant at a P-value of o0.05.

Oncogene (2014) 2898 – 2908 & 2014 Macmillan Publishers Limited Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2907 CONFLICT OF INTEREST 22 Leong SR, Lowman HB, Liu J, Shire S, Deforge LE, Gillece-Castro BL et al. IL-8 The authors declare no conflict of interest. single-chain homodimers and heterodimers: interactions with chemokine receptors CXCR1, CXCR2, and DARC. Protein Sci 1997; 6: 609–617. 23 Zijlstra A, Quigley JP. The DARC side of metastasis: shining a light on ACKNOWLEDGEMENTS KAI1-mediated metastasis suppression in the vascular tunnel. Cancer Cell 2006; 10: 177–178. We acknowledge Chris Goodrich who kindly provided technical support. We also 24 Iiizumi M, Bandyopadhyay S, Watabe K. Interaction of Duffy antigen receptor for acknowledge Dr Meenhard Herlyn for kindly providing WM35 and WM793 cells. chemokines and KAI1: a critical step in metastasis suppression. Cancer Res 2007; We thank Dr Justin Brown for letting us use his Odyssey system to develop the 67: 1411–1414. western blots. This work was supported by NIH Grant CA-127892-01A1, American Cancer Society Grant RSG-04-053-01-GMC and The Foreman Foundation for 25 Zarbock A, Bishop J, Muller H, Schmolke M, Buschmann K, Van Aken H et al. Melanoma Research (GPR); NIH Grants CA-125707 (CD); and Johnson & Johnson Chemokine homeostasis vs. chemokine presentation during severe acute lung injury: the other side of the Duffy antigen receptor for chemokines. Am J Physiol Innovative Technology Research Seed Grant Program (CD and GPR). Lung Cell Mol Physiol 2010; 298: L462–L471. 26 Liang S, Hoskins M, Khanna P, Kunz RF, Dong C. Effects of the tumor–leukocyte microenvironment on melanoma–neutrophil adhesion to the endothelium in a REFERENCES shear flow. Cell Mol Bioeng 2008; 1: 189–200. 1 Gray-Schopfer V, Wellbrock C, Marais R. Melanoma biology and new targeted 27 Peng HH, Liang S, Henderson AJ, Dong C. Regulation of interleukin-8 expression therapy. Nature 2007; 445: 851–857. in melanoma-stimulated neutrophil inflammatory response. Exp Cell Res 2007; 2 Shevde LA, Welch DR. Metastasis suppressor pathways—an evolving paradigm. 313: 551–559. Cancer Lett 2003; 198: 1–20. 28 Dong C, Slattery MJ, Liang S, Peng HH. Melanoma cell extravasation under flow 3 Chiang AC, Massague J. Molecular basis of metastasis. N Engl J Med 2008; 359: conditions is modulated by leukocytes and endogenously produced interleukin 8. 2814–2823. Mol Cell Biomech 2005; 2: 145–159. 4 Brodland DG, Zitelli JA. Mechanisms of metastasis. J Am Acad Dermatol 1992; 29 Waugh DJ, Wilson C. The interleukin-8 pathway in cancer. Clin Cancer Res 2008; 27: 1–8. 14: 6735–6741. 5 Nicolson GL. Metastatic tumor cell attachment and invasion assay 30 Li A, Dubey S, Varney ML, Dave BJ, Singh RK. IL-8 directly enhanced endothelial utilizing vascular endothelial cell monolayers. J Histochem Cytochem 1982; 30: cell survival, proliferation, and matrix metalloproteinases production and 214–220. regulated angiogenesis. J Immunol 2003; 170: 3369–3376. 6 Germano G, Allavena P, Mantovani A. Cytokines as a key component of 31 Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation. cancer-related inflammation. Cytokine 2008; 43: 374–379. Nature 2008; 454: 436–444. 7 Nagasawa K, Chiba H, Fujita H, Kojima T, Saito T, Endo T et al. 32 Murdoch C, Monk PN, Finn A. Cxc chemokine receptor expression on human Possible involvement of gap junctions in the barrier function of tight endothelial cells. Cytokine 1999; 11: 704–712. junctions of brain and lung endothelial cells. J Cell Physiol 2006; 208: 33 Criscuoli ML, Nguyen M, Eliceiri BP. Tumor metastasis but not tumor 123–132. growth is dependent on Src-mediated vascular permeability. Blood 2005; 105: 8 Tsai YC, Weissman AM. Dissecting the diverse functions of the metastasis 1508–1514. suppressor CD82/KAI1. FEBS Lett 2011; 585: 3166–3173. 34 Dejana E, Bazzoni G, Lampugnani MG. Vascular endothelial (VE)-cadherin: only an 9 Romanska HM, Berditchevski F. Tetraspanins in human epithelial malignancies. intercellular glue? Exp Cell Res 1999; 252: 13–19. J Pathol 2011; 223: 4–14. 35 Gavard J. Breaking the VE-cadherin bonds. FEBS Lett 2009; 583: 1–6. 10 Singethan K, Schneider-Schaulies J. Tetraspanins: small transmembrane proteins 36 Peng HH, Hodgson L, Henderson AJ, Dong C. Involvement of phospholipase C with big impact on membrane microdomain structures. Commun Integr Biol 2008; signaling in melanoma cell-induced endothelial junction disassembly. Front Biosci 1: 11–13. 2005; 10: 1597–1606. 11 Xu C, Zhang YH, Thangavel M, Richardson MM, Liu L, Zhou B et al. CD82 endo- 37 Khanna P, Yunkunis T, Muddana HS, Peng HH, August A, Dong C. p38 MAP kinase cytosis and cholesterol-dependent reorganization of tetraspanin webs and lipid is necessary for melanoma-mediated regulation of VE-cadherin disassembly. rafts. FASEB J 2009; 23: 3273–3288. Am J Physiol Cell Physiol 2010; 298: C1140–C1150. 12 Zoller M. Tetraspanins: push and pull in suppressing and promoting metastasis. 38 Verin AD, Patterson CE, Day MA, Garcia JG. Regulation of endothelial cell gap Nat Rev Cancer 2009; 9: 40–55. formation and barrier function by myosin-associated phosphatase activities. 13 Liu WM, Zhang F, Moshiach S, Zhou B, Huang C, Srinivasan K et al. Am J Physiol 1995; 269(Partt 1): L99–108. Tetraspanin CD82 inhibits protrusion and retraction in cell movement by 39 Acosta JC, Gil J. A role for CXCR2 in senescence, but what about in cancer? Cancer attenuating the plasma membrane-dependent actin organization. PLoS One 2012; Res 2009; 69: 2167–2170. 7: e51797. 40 Lleonart ME, Artero-Castro A, Kondoh H. Senescence induction: a possible cancer 14 Shiwu WU, Lan Y, Wenqing S, Lei Z, Yisheng T. Expression and clinical significance therapy. Mol Cancer 2009; 8:3. of CD82/KAI1 and E-cadherin in non-small cell lung cancer. Arch Iran Med 2012; 41 Fan T, Jiang S, Chung N, Alikhan A, Ni C, Lee CC et al. EZH2-dependent sup- 15: 707–712. pression of a cellular senescence phenotype in melanoma cells by inhibition of 15 Takaoka A, Hinoda Y, Sato S, Itoh F, Adachi M, Hareyama M et al. Reduced invasive p21/CDKN1A expression. Mol Cancer Res 2011; 9: 418–429. and metastatic potentials of KAI1-transfected melanoma cells. Jpn J Cancer Res 42 Giuliano S, Ohanna M, Ballotti R, Bertolotto C. Advances in melanoma 1998; 89: 397–404. senescence and potential clinical application. Pigment Cell Melanoma Res 2011; 16 Bandyopadhyay S, Zhan R, Chaudhuri A, Watabe M, Pai SK, Hirota S et al. 24: 295–308. Interaction of KAI1 on tumor cells with DARC on vascular endothelium leads to 43 Liu S, Yamauchi H. p27-Associated G1 arrest induced by hinokitiol in human metastasis suppression. Nat Med 2006; 12: 933–938. malignant melanoma cells is mediated via down-regulation of pRb, Skp2 17 Lombardi DP, Geradts J, Foley JF, Chiao C, Lamb PW, Barrett JC. Loss of ubiquitin ligase, and impairment of Cdk2 function. Cancer Lett 2009; 286: KAI1 expression in the progression of colorectal cancer. Cancer Res 1999; 59: 240–249. 5724–5731. 44 Houben R, Ortmann S, Drasche A, Troppmair J, Herold MJ, Becker JC. Proliferation 18 Acosta JC, O’Loghlen A, Banito A, Guijarro MV, Augert A, Raguz S et al. arrest in B-Raf mutant melanoma cell lines upon MAPK pathway activation. Chemokine signaling via the CXCR2 receptor reinforces senescence. Cell 2008; J Invest Dermatol 2009; 129: 406–414. 133: 1006–1018. 45 Gartel AL, Serfas MS, Tyner AL. p21-Negative regulator of the cell cycle. Proc Soc 19 Singh S, Varney M, Singh RK. Host CXCR2-dependent regulation of melanoma Exp Biol Med 1996; 213: 138–149. growth, angiogenesis, and experimental lung metastasis. Cancer Res 2009; 69: 46 Campisi J. Cellular senescence as a tumor-suppressor mechanism. Trends Cell Biol 411–415. 2001; 11: S27–S31. 20 Gabellini C, Trisciuoglio D, Desideri M, Candiloro A, Ragazzoni Y, 47 Ishikawa F. Cellular senescence, an unpopular yet trustworthy tumor suppressor Orlandi A et al. Functional activity of CXCL8 receptors, CXCR1 and mechanism. Cancer Sci 2003; 94: 944–947. CXCR2, on human malignant melanoma progression. Eur J Cancer 2009; 45: 48 Roninson IB. Oncogenic functions of tumour suppressor p21(Waf1/Cip1/Sdi1): 2618–2627. association with cell senescence and tumour-promoting activities of stromal 21 Peiper SC, Wang ZX, Neote K, Martin AW, Showell HJ, Conklyn MJ et al. The Duffy fibroblasts. Cancer Lett 2002; 179: 1–14. antigen/receptor for chemokines (DARC) is expressed in endothelial cells of Duffy 49 Singh RK, Gutman M, Radinsky R, Bucana CD, Fidler IJ. Expression of interleukin 8 negative individuals who lack the erythrocyte receptor. J Exp Med 1995; 181: correlates with the metastatic potential of human melanoma cells in nude mice. 1311–1317. Cancer Res 1994; 54: 3242–3247.

& 2014 Macmillan Publishers Limited Oncogene (2014) 2898 – 2908 Inhibition of IL-8-mediated endothelial gap formation by CD82/KAI P Khanna et al 2908 50 Singh RK, Lokeshwar BL. Depletion of intrinsic expression of interleukin-8 in 53 Weis S, Cui J, Barnes L, Cheresh D. Endothelial barrier disruption by VEGF- prostate cancer cells causes cell cycle arrest, spontaneous apoptosis mediated Src activity potentiates tumor cell extravasation and metastasis. JCell and increases the efficacy of chemotherapeutic drugs. Mol Cancer 2009; Biol 2004; 167: 223–229. 8: 57. 54 Tran MA, Gowda R, Sharma A, Park EJ, Adair J, Kester M et al. Targeting V600EB- 51 Singh RK, Varney ML. IL-8 expression in malignant melanoma: Raf and Akt3 using nanoliposomal-small interfering RNA inhibits cutaneous implications in growth and metastasis. Histol Histopathol 2000; 15: melanocytic lesion development. Cancer Res 2008; 68: 7638–7649. 843–849. 55 Sharma A, Tran MA, Liang S, Sharma AK, Amin S, Smith CD et al. Targeting 52 Huh SJ, Liang S, Sharma A, Dong C, Robertson GP. Transiently entrapped mitogen-activated protein kinase/extracellular signal-regulated kinase kinase in circulating tumor cells interact with neutrophils to facilitate lung metastasis the mutant (V600E) B-Raf signaling cascade effectively inhibits melanoma lung development. Cancer Res 2010; 70: 6071–6082. metastases. Cancer Res 2006; 66: 8200–8209.

Oncogene (2014) 2898 – 2908 & 2014 Macmillan Publishers Limited