This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Erdreich-Epstein, A. Articles by Durden, D. L. Search for Related Content PubMed PubMed Citation Articles by Erdreich-Epstein, A. Articles by Durden, D. L.

Integrins _vß₃ and _vß₅ Are Expressed by Endothelium of High-Risk Neuroblastoma and Their Inhibition Is Associated with Increased Endogenous Ceramide¹

Divisions of Hematology-Oncology [A. E-E., M. L., L. S. M., R. C. S., D. L. D.] and Infectious Diseases [K. S. K., M. F. S.], Neil Bogart Memorial Laboratories, and Departments of Pediatrics and Pathology [H. S.], Children’s Hospital Los Angeles, University of Southern California School of Medicine, Los Angeles, California 90027; and Department of Preventive Medicine [S. G.], University of Southern California School of Medicine and the Children’s Cancer Group, Arcadia, California 91066

	ABSTRACT

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Inhibition of the RGD-binding integrins, _vß₃ and _vß₅, prevents endothelial cell anchorage and induces endothelial apoptosis, which results in disruption of tumor angiogenesis and inhibition of tumor growth in animal models. In this study, we demonstrate by immunohistochemical analysis that integrin _vß₃ was expressed by 61% (mean) of microvessels in high-risk neuroblastomas (stage IV and MYCN-amplified stage III; n = 28) but only by 18% (mean) of microvessels in low-risk tumors (stages I and II and non-MYCN-amplified stage III; n = 12). Integrin _vß₅ was found on 60% (mean) of microvessels in 21 Stage IV tumors. These data suggest that neuroblastomas may be targeted for antiangiogenic treatment directed against endothelial integrins _vß₃ and _vß₅. In cell culture, inhibition of integrin-dependent endothelial cell anchorage to vitronectin by RGDfV, an RGD function-blocking cyclic peptide, induced apoptosis in bovine brain endothelial cells compared with the control peptide, RADfV (37.5% versus 8.7%, respectively), as detected by chromatin condensation and nuclear fragmentation. Treatment with RGDfV but not with RADfV, which prevented attachment of endothelial cells to vitronectin or fibronectin, was associated with up to a 50% increase in endogenous ceramide, a lipid second messenger that can mediate cell death. Furthermore, exogenous C₂-ceramide was cytotoxic to bovine brain endothelial cells and induced activation of C-jun N-terminal kinase (JNK), a MAP kinase that can be activated in stress-induced apoptosis pathways. This suggests that ceramide may function in detachment-induced endothelial cell apoptosis, originating from inhibition of vitronectin binding to integrins such as _vß₃ and _vß₅. This is the first report to demonstrate expression of integrins _vß₃ and _vß₅ by microvascular endothelium of a childhood tumor and association of their expression with neuroblastoma aggressiveness. Furthermore, our data provide the first suggestion that inhibition of endothelial cell anchorage, resulting from specific blockade of RGD-binding integrins, increases endogenous ceramide, which may contribute to endothelial cell death.

	INTRODUCTION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Integrins are /ß heterodimeric cell-surface receptors that link the extracellular matrix with the cell cytoskeleton and generate intracellular signals to regulate cell survival and growth, cell differentiation, and cell motility. Vitronectin, a matrix protein comprising two hemopexin domains, a somatomedin B domain and a cell-binding region, binds integrins _vß₁, _vß₃, _vß₅, _vß₆, _vß₈, _IIbß₃, and ₈ß₁ via the RGD sequence in the cell-binding region at its NH₂-terminus (1, 2, 3, 4) . Integrin binding to vitronectin can be inhibited by specific function-blocking antibodies or by synthetic RGD peptides, which mimic the conformation of the RGD sequence (5, 6, 7, 8) . Of the synthetic RGD peptides, some of the most active in inhibiting cell attachment to vitronectin are the cyclic pentapeptides (5 , 6) . To date, the only vitronectin-binding integrins that have been described on the surface of endothelial cells are _vß₃, _vß₅ (7, 8, 9) , and to a lesser extent _vß₆ (10) .

The angiogenic integrins, _vß₃ and _vß₅, which are expressed on endothelial cells, are crucial for their survival (8 , 11, 12, 13, 14) . Survival signals transmitted by integrin _vß₃ lead to inhibition of p53 activity, decreased expression of p21WAF1/CIP1, and suppression of the bax cell death pathway in endothelial cells (15) . When plated on osteopontin, the _vß₃-dependent signals for endothelial cell survival are mediated via NFB³ (16) . These survival pathways can be blocked either by function-blocking antibodies or by peptide analogues, which block the active RGD binding site on integrins _vß₃ and _vß₅, thus inducing apoptosis of the angiogenic endothelial cells (8 , 11 , 12) . Endothelial apoptosis results in inhibition of new blood vessel formation, disruption of existing angiogenic vasculature, inhibition of tumor growth, and tumor regression (8 , 11 , 12 , 14, 15, 16) .

Apoptosis of endothelial cells can occur in response to a variety of stress stimuli such as LPS (17) , TNF (18) , ionizing radiation (19) , growth factor withdrawal (20) , and disruption of matrix binding or specific integrin ligation (11 , 21) . Ceramide, a lipid second messenger that is synthesized de novo or derived from membrane sphingomyelin, is implicated in apoptotic signaling pathways induced by stimuli such as irradiation, TNF, and LPS (22, 23, 24) and has been linked to activation of pro-apoptotic MAPKs (e.g., JNK/SAPK; Ref. 23 ). The delicate balance between cellular levels of the survival-promoting, sphingosine-1-phosphate and the apoptosis-linked ceramide, both metabolites of sphingomyelin, provides a means for sensitive regulation of pro- and antiapoptotic MAPKs (e.g., JNK/SAPK, p38, or extracellular regulated kinases), thus contributing to regulation of cell survival and death (25) . Interestingly, ceramide enhances expression of MMP-1³ in fibroblasts (26) , suggesting that it may mediate signals for matrix remodeling and possibly for angiogenesis. Although ceramide has been implicated in endothelial apoptosis induced by a variety of stimuli both in vivo (17) and in vitro (19 , 23) , it is not known whether ceramide plays a role in signaling of apoptosis resulting from inhibition of integrin-mediated endothelial cell anchorage.

Treatment of NB, the second most common solid tumor in childhood, is successful in less than half of patients with high-risk (stage IV and MYCN-amplified stage III) disease (27, 28, 29) . A high vascular index in NB correlates with poor prognosis (30) , suggesting dependence of aggressive tumor growth on active angiogenesis, the sprouting of new capillaries from existing blood vessels. The matrix-degrading enzymes MMP-2 and MMP-9 are elevated in high-risk NB compared with low-risk disease (31) and are expressed in NB cell lines, which themselves induce proliferation of endothelial cells (32) , supporting the angiogenic phenotype of this tumor. Moreover, several antiangiogenic approaches have demonstrated activity in animal models of NB (33, 34, 35) . Thus, utilization of antiangiogenic approaches, by themselves or in combination with other therapies, could potentially improve the outcome in children with NB and may warrant further study.

In view of the in vivo efficacy of antiangiogenic treatment directed against integrins _vß₃ and _vß₅ in animal models (8 , 11 , 12 , 35) and the correlation of a high vascular index with poor prognosis in NB (30) , we hypothesized that tumor endothelium in NB may present a potential target for antiangiogenic treatment directed against integrins _vß₃ and/or _vß₅. Furthermore, we hypothesized that ceramide, which can contribute to apoptotic signaling in endothelial cells, may be involved in apoptosis induced by inhibition of integrin-mediated anchorage in these cells. In this study, we report that inhibition of endothelial cell attachment and spreading and the resultant apoptosis, originating from blockade of RGD-binding integrins such as _vß₃ and _vß₅, are associated with the generation of endogenous ceramide. In addition, we show by immunohistochemical analysis that integrins _vß₃ and _vß₅ are expressed on average by 61% of microvessels in the high-risk NB (stage IV and MYCN-amplified stage III; n = 28) but _vß₃ is only expressed in 18% of low-risk tumors (stages I and II and non-MYCN-amplified stage III; n = 12). These data suggest that NB endothelium may be targeted for antiangiogenic treatment directed against integrins _vß₃ and _vß₅.

	MATERIALS AND METHODS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Cells, Reagents, and Antibodies.
All reagents were purchased from Sigma (St. Louis, MO) unless stated otherwise. BBEC transfected with large-T antigen were maintained in RPMI 1640 supplemented with glutamine, pyruvate, 10% FCS, and 10% Nu-Serum IV culture supplement (Collaborative Biomedical Products, Bedford, MA) as described (36) . These cells spontaneously expressed high levels of integrin _vß₃ on their surface (mean, 92.4%, SD, 5.8%, by fluorescence-activated cell-sorting analysis using mAb LM609 in 11 separate determinations over a period of at least 4 months in culture). Primary HUV-EC-C were purchased from the American Type Culture Collection (passage 16; Manassas, VA), maintained according to the American Type Culture Collection recommendations, and used between passages 16 and 24. Previously characterized affinity-purified mAb to integrin _vß₃ (LM609) and the _v integrin subunit (LM142) were a generous gift from D. Cheresh (7 , 12 , 14) . Monoclonal antibody clone P1F6 against integrin _vß₅ was purchased from Life Technologies (Gaithersburg, MD). Isotype-specific mouse IgG1 (control for the mAb stains) and polyclonal anti-Factor VIII were from Dako (Carpinteria, CA). Secondary antibodies for the double immunofluorescent stains were FITC-conjugated goat F(ab')₂ anti-rabbit IgG and rhodamine-conjugated goat F(ab')₂ anti-mouse IgG (BioSource International, Camarillo, CA). Human fibronectin (domains 2–11, constituting the 110-kDa cell-binding fragment containing the RGD site but not the Hep-2/CS1 region) was purchased from Upstate Biotechnology (Lake Placid, NY), and human vitronectin was purchased from Promega (Madison, WI) or prepared as described by Yatohgo et al. (37) . The cyclic pentapeptides RGDfV and RADfV were generous gifts from Merck (Darmstadt, Germany) via Alfred Jonczyk. All cell culture experiments were repeated at least three times.

Patients and Tumor Specimens.
The NB specimens included in this study were resected at institutions of the CCG (Arcadia, CA) between 1986 and 1992. Clinical staging was performed according to standard criteria used by the CCG (38) . Neuroblastoma tumor tissue was obtained surgically, snap-frozen, and shipped on dry ice to the CCG Neuroblastoma Biology Reference Laboratory. Upon receipt, a portion was removed without thawing, placed in Tissue-Tek OCT embedding compound (Miles, Elkart, IN), and maintained at -70°C. Another part of each tumor was fixed in formalin and embedded for light microscopic examination. Histological sections of tumors were evaluated for stromal cells, neuroblastic differentiation, and mitotic/karyorrhectic cell number and categorized as having favorable or unfavorable histology (39) . Forty-two tumors were analyzed. Two of the forty-two samples processed were excluded from the analysis after processing because during the pathological analysis one sample was found to be a lymph node metastasis rather than a sample from the primary tumor and one OCT block was inadequate (it contained only tumor capsule). Patient and tumor characteristics for the remaining 40 patients are summarized in Table 1 . Thirty-five of the samples were obtained at the time of diagnosis, before the patient had received any chemotherapy or radiation; five of the samples were obtained from tumors resected from patients who had already received chemotherapy (i.e., at the time of delayed resection or after relapse).

View this table: [in this window] [in a new window]   Table 1 Most microvessels in high-risk NB express integrin vß3

Immunofluorescence Analysis.
Immunofluorescent staining was performed as previously described with minor modifications (12) . Four-µm cryostat NB sections in OCT were fixed in cold acetone for 30 s, air dried, and subsequently blocked with 2.5% BSA in PBS (pH 7.4) for 1–2 h at room temperature. Sections were then washed 5–8 times in PBS and incubated with monoclonal anti-_vß₃ (LM609 1:50, 60 µg/ml) and polyclonal anti-Factor VIII (1:1000) for 1 h at room temperature. Sections were washed and then incubated for 1.5 h at room temperature with rhodamine-conjugated anti-mouse IgG (1:300) and FITC-conjugated anti-rabbit IgG (1:300). Tissue sections were then washed, mounted, and photographed using an Olympus AX70 compound microscope (Olympus America, Melville, NY) at x200.

Cell Radiolabeling and Analysis of Ceramide.
Ceramide metabolism was studied as described (41, 42, 43) . Briefly, endothelial cells were plated the day before the experiment (3 x 10⁶ cells/100-mm plate, in their usual growth medium) and metabolically labeled with 9,10 [³H]-(N)-palmitic acid (1 µCi/ml; NEN, Boston, MA) overnight at 37°C. The next day the medium was collected, and cells were briefly trypsinized, washed, and replated on non-tissue culture-treated 6-well plates that were precoated with fibronectin (10 µg/ml), vitronectin (2 µg/ml), or BSA (0.5%) and blocked with BSA (0.5%) in the medium in which they were incubated overnight (10⁶ cells/well). Where indicated, the cells were replated in the presence or absence of the cyclic RGD-blocking peptide RGDfV or control RADfV. At the end of incubation the cells were briefly trypsinized, combined with the nonadherent cells of the same well, washed (at 4°C), and lipid was extracted using equal volumes of methanol/2% acetic acid (v/v), water, and chloroform. The dried lipid was solubilized in chloroform:methanol (2:1 v/v) and analyzed using TLC plates with a solvent system consisting of chloroform:acetic acid (9:1 v/v) and using lipid standards as markers as described (41, 42, 43) . Ceramide and total cellular lipids were quantitated in a liquid scintillation counter (cpm), and ceramide was calculated and expressed as percent [ ³H]ceramide of total [³H]-labeled cellular lipids extracted.

Cell Density and Viability.
Cell viability was assessed by the uptake of MTT (Thiazolyl blue; Sigma), a process which is dependent on intact mitochondrial function, as described by Mosmann (44) . This method yields results similar to those of other methods of assessing endothelial cell viability (45) . Briefly, to assess viable cell density, MTT (250 µg/ml) was added to the cells and incubated for 2 h at 37°C. Subsequently the medium was carefully removed in such a way that any detached or loosely adherent cells were not removed, and the cells containing the trapped MTT crystals were solubilized in 200 µl DMSO at 37°C for 15 min. Absorbance was determined in a microtiter plate reader (Molecular Devices, Menlo Park, CA) at 550 nm and subtracted from the A_{650 nm}.

Cell Adhesion.
Cell adhesion was assayed by plating 100,000 trypsinized cells onto wells of a 48-well non-tissue culture-treated plate coated overnight with vitronectin (2 µg/ml) or the 110-kDa RGD-containing fragment of fibronectin (10 µg/ml), which were washed and blocked with 0.5% BSA in PBS. After 1 h, the wells were washed 3 times to remove nonadherent cells, and the remaining cells were quantitated using the MTT assay as described above.

Hoechst Stain.
BBEC were stained with Hoechst-bisbenzamide 33258 (Sigma) according to the manufacturer’s instructions to assess chromatin condensation and nuclear fragmentation. Approximately 300 cells in each sample were counted under UV filter at magnification x400.

JNK Activation.
Lysates from 5 x 10⁶ BBEC that have been exposed to C₂-ceramide were prepared in Triton X-100 extraction buffer [EB buffer: 1% Triton X-100, 10 mM Tris (pH 7.6), 50 mM NaCl, 0.1% BSA, 1 mM PMSF, 1% aprotinin, 5 mM EDTA, 50 mM NaF, 0.1% 2-ß mercaptoethanol, 5 µM phenylarsine oxide, and 100 µM vanadate], precleared (15,000 x g for 45 min at 4°C), and anti-SAPK/JNK immunoprecipitation was performed using anti-SAPK1/JNK1 rabbit polyclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) as previously described (46) . Immunoprecipitates were washed several times with buffer [20 mM PIPES (pH 7.0), 0.1 M NaCl, and 20 µg/ml aprotinin], and kinase activity of JNK was quantitated using a slight modification of an in vitro kinase assay as described (47) . JNK immunoprecipitates were incubated in a solution containing 20 mM PIPES (pH 7.0), 10 mM MnCl₂, 5 µCi [³²P]ATP (3000 Ci/mmol), 20 µg/ml aprotinin, and 1 µg of GST-c-jun fusion protein (NH₂ terminus, residues 1–154) in final volume of 20 µl. After an incubation of 30 min at 30°C, the reactions were terminated by the addition of 20 µl of 2x SDS-sample buffer and heated at 98°C for 5 min. The JNK kinase and GST-c-jun substrates were resolved by SDS-PAGE, phosphorylated GST-jun protein was visualized by autoradiography, and the amount of JNK kinase protein immunoprecipitated was verified by anti-JNK Western blotting.

Statistical Analyses.
All laboratory and histopathological analyses were performed independently and without knowledge of clinical data. Immunohistochemical slides prepared from serial sections were analyzed by two observers (A. E-E. and H. S.), who visually determined the proportion of microvessels in the whole section in which the microvascular endothelium specifically stained with the LM609 or P1F6 mAb compared with the total Factor VIII-positive vessels in the adjacent section. The observers repeated the analysis of the sections on two separate occasions and were blinded to the clinical characteristics of the patients and to their previous readings. The variation for integrin _vß₃ expression between the first and second reading in the 40 tumors examined was 15% in 2 tumors, 10% in 9 tumors, and 0–5% in the remaining 29 tumors. The variation for integrin _vß₅ between the first and second readings in the 21 tumors evaluated was 15% in 1 tumor, 10% in 6 tumors, and 0–5% in 13 tumors. The value used for statistical analysis and for plotting the figures was the average of the readings for each tumor. A simple linear regression analysis was used to evaluate the association of age (as a continuous variable and also dichotomized as 12 months versus >12 months), MKI (low + intermediate versus high; Ref. 3 ), pathology (favorable versus unfavorable), MYCN (not amplified versus amplified), and whether or not the patient had received treatment prior to tumor sampling, with expression of _Vß₃ and _Vß₅ as measured by the percent of microvessels that stained with the antibodies LM609 (for _Vß₃) and P1F6 (for _Vß₅). The P values reported are based on the F test from the regression analysis and are all two-sided (Table 1) . Means and 95% confidence intervals based on the individual SD were calculated. To summarize the association between expression of _Vß₃ and _Vß₅, scatter plots were drawn and Pearson’s correlation coefficient was calculated with its associated P value (Fig. 4) . To test whether the observed correlation could be explained by MYCN amplification, a partial correlation was calculated. To evaluate whether the association between _Vß₃ and _Vß₅ was similar or different in patients whose tumors expressed MYCN amplification or not, separate regression models were fit for the two subgroups of patients and were compared with the single model fit for all 21 patients.

View larger version (19K): [in this window] [in a new window]   Fig. 4. Expression of endothelial integrin vß3 correlates with expression of integrin vß5 in stage IV NB. Expression of integrins vß3 and vß5 on tumor microvessels was determined in frozen sections of 21 cases of stage IV NB (•, MYCN-amplified; , non-MYCN-amplified) as described in "Materials and Methods." The results are expressed as the percentage of Factor VIII-positive microvessels that stain with LM609 (vß3) plotted against the percentage of Factor VIII-positive microvessels that stain with P1F6 (vß5) in each of the tumors analyzed. The calculated Pearson’s correlation coefficient between microvascular endothelial expression of Vß3 and Vß5 in the 21 tumors analyzed was r = 0.75 (P < 0.001).

	RESULTS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Integrin _vß₃ Is Expressed on Angiogenic Endothelial Cells of High-Risk NB.

View larger version (156K): [in this window] [in a new window]   Fig. 1. Microvessels of high-risk NB express integrin vß3. Frozen sections of Stage III NBs [A and C (high-risk): MYCN-amplified, unfavorable Shimada classification, and high MKI, resected from a patient diagnosed at age 47 months; B and D (low-risk): non-MYCN-amplified, favorable histology by Shimada classification, and low MKI, resected from a patient diagnosed at age 33 months] were stained by immunoperoxidase for integrin vß3 (A and B, mAb LM609) and Factor VIII (C and D) as described in "Materials and Methods." Brown color represents positively stained microvessels. Percent positivity of microvessels for integrin vß3 and vß5 was determined as a fraction of the vessels which stained for Factor VIII in consecutive serial sections. A representative field is shown. Photographed at magnification x100.

View larger version (118K): [in this window] [in a new window]   Fig. 2. Integrin vß3 expression is localized to endothelial cells in NBs. A single cryostat section from a MYCN-amplified stage IV NB with unfavorable Shimada classification resected from a child diagnosed at age 8 months was dual-stained with both mAb LM609 directed to integrin vß3 and polyclonal rabbit anti-Factor VIII antibody. After incubation with the primary antibodies, the sample was reacted with two secondary antibodies, Rhodamine-conjugated anti-mouse IgG and FITC-conjugated anti-rabbit IgG, as described in "Materials and Methods." Immunofluorescence was detected with an Olympus AX70 compound microscope at magnification x200. Green indicates Factor VIII-expressing endothelium (A), red indicates expression of integrin vß3 (stained with mAb LM609; B), and yellow demonstrates colocalization of integrin vß3 to the Factor VIII-positive endothelial cells (achieved by using double exposure of the film, first using the green fluorescence and then the red fluorescence, resulting in a yellow fluorescence where they were superimposed; C).

View larger version (13K): [in this window] [in a new window]   Fig. 3. Integrin vß3 is expressed by microvessels of high-risk but not low-risk NBs. Serial frozen sections of 40 cases of NB of different stages and MYCN amplification, as summarized in Table 1 , were stained for integrin vß3 (mAb LM609) and Factor VIII (see Fig. 1 ), and the fraction of the vessels expressing integrin vß3 (expressed as a percentage of the vessels expressing integrin vß3) was determined as detailed in "Materials and Methods." , stage I non-MYCN-amplified tumors; , stage II non-MYCN-amplified tumors; , stage III non-MYCN-amplified tumors; , stage III MYCN-amplified tumors; , stage IV nonamplified tumors; •, stage IV MYCN-amplified tumors.

Association between Expression of Integrin _vß₃ and _vß₅ in High-Risk Neuroblastomas.

Inhibition of Integrin-mediated Endothelial Cell Anchorage Is Associated with Elevation of Endogenous Ceramide.
Integrin-mediated signals are required for anchorage-dependent survival of endothelial and epithelial cells (21 , 48 , 49) . Inhibition of integrins _vß₃ and _vß₅ on endothelial cells lining tumor neovasculature in animal models induces endothelial cell apoptosis, disruption of tumor angiogenesis, and inhibition of tumor growth (8 , 11 , 14) . Endothelial cells also require integrin-mediated attachment and spreading on specific matrix proteins to survive in culture (14 , 21 , 48) . Because we showed that endothelium in high-risk NBs expressed integrins _vß₃ and _vß₅ (Figs. 1 2 3 4) , we sought to further study the signaling pathways that may contribute to endothelial apoptosis mediated via inhibition of these integrins.

On the basis of in vitro and in vivo experiments demonstrating involvement of ceramide in UV irradiation, LPS, and TNF-induced endothelial apoptosis (17 , 19 , 23) , we hypothesized that ceramide could be involved in endothelial cell death resulting from inhibition of integrin-mediated anchorage. Because NB is a neural crest-derived tumor, in testing this hypothesis we chose to use BBEC immortalized (large-T transfected) endothelial cells derived from a neural microenvironment (the brain). Like other endothelial cells, BBEC spontaneously express high levels of integrin _vß₃ on their surface. Cellular lipids of attached, log-phase growth BBEC were labeled with [³H]palmitic acid, detached briefly by trypsin/EDTA, and then replated on BSA, fibronectin, or vitronectin for 4 h. Lipids were extracted and separated by TLC as detailed in "Materials and Methods." Prevention of attachment and spreading of endothelial cells onto extracellular matrix by plating them on a BSA-coated surface induced elevation of ceramide of up to 50% above that observed in cells plated on fibronectin or vitronectin (Fig. 5A) . The mean increase in endogenous ceramide in BBEC plated on BSA was 35% (SD, 15%; n = 5 experiments) compared with vitronectin and 29% (SD, 9%; n = 3 experiments) compared with fibronectin. These results were replicated in primary HUV-EC-C. An increase of 54% in endogenous ceramide was observed in the HUV-EC-C when they were plated on BSA compared with vitronectin (SD, 1.33 ± 0.1% [³H]ceramide on BSA compared with 0.87 ± 0.01% on vitronectin).

View larger version (32K): [in this window] [in a new window]   Fig. 5. Apoptosis induced by inhibition of integrin-mediated anchorage on vitronectin is associated with increased endogenous levels of ceramide in endothelial cells. A, BBEC (106 cells/sample) labeled with 1 µCi/ml 9,10 [3H]-(N)-palmitic acid were plated on non-tissue culture-treated 6-well plates coated with BSA (0.5%), vitronectin (2 µg/ml), or fibronectin (10 µg/ml). After 4 h, lipids were extracted and ceramide content was analyzed by TLC. TNF (500 ng/ml) was added to one set of triplicate samples plated on vitronectin 30 min before extraction of lipids (positive control). Ceramide is expressed as the percentage of total [3H]-labeled lipids extracted. Bars represent mean of the percent ceramide extracted from cells from three identical samples plated on BSA, vitronectin, or fibronectin, or treated with TNF in this experiment. Error bars represent SD. The results represent a typical experiment, repeated 10 times. B, BBEC (105 cells per well) were plated in a vitronectin-coated non-tissue culture-treated 48-well plate with concentrations of RGDfV (•) or RADfV () between 0 and 50 µg/ml and were allowed to adhere and spread for 60 min. The wells were then washed three times to remove nonadherent cells, and the remaining adherent cells were quantitated by MTT assay (see "Materials and Methods"). Data points and error bars are means and SD of six replicate samples; in data points where the error bar is smaller than the symbol, it is hidden by it. C, BBEC (105) were plated in a 48-well non-tissue culture-treated plate coated with vitronectin (2 µg/ml) in the presence of 10 µg/ml cyclic RGDfV (), RADfV (control; ), or without addition (control; ) and were incubated for 6 h. Cells were trypsinized, and a cytospin from each sample was stained with Hoechst-bisbenzamide to demonstrate nuclear fragmentation and condensation. Bars represent mean percentage of condensed nuclei per high power field of total nuclei in the field (magnification x400) counted in five different fields of each cytospin. Error bars represent the SD of the count of the five high power fields. D, The cyclic RGD-blocking peptide RGDfV [10 µg/ml () or 0.5 mg/ml ()] or control peptide RADfV [0.5 mg/ml ()] were added to 106 [ 3H]palmitic acid-labeled BBEC and then plated on vitronectin as in A. For comparison, cells were also plated on BSA [positive control ()] or vitronectin [negative control ()] without additions. After 4 h, lipids were extracted and ceramide was quantitated as above. Data are presented as mean ± SD (n = 3) of the percent ceramide extracted from three identical samples of cells treated under the same conditions; error barsrepresent the SD of the triplicate samples.

Ceramide-induced Cytotoxicity Is Associated with the Activation of JNK.
Some stress signals that lead to apoptosis in endothelial cells induce endogenous ceramide, which is associated with activation of JNK, a MAPK that functions in apoptosis (23) . We showed that inhibition of integrin-mediated endothelial cell anchorage, which results in apoptosis, induces elevation of endogenous ceramide. We asked whether ceramide itself could induce JNK activation and BBEC death. C₂-ceramide was cytotoxic to BBEC with a mean concentration which induces 50% cell death (LC₅₀) of 31.7 µM (SD, 10.5 µM) at 24 h, as determined in eight separate experiments (each done in replicates of 5–10 points for each ceramide concentration). The death response was proportional to the concentration of ceramide added (Fig. 6A) and to the length of time BBEC were exposed to the lipid (Fig. 6B) . The inactive ceramide analogue dihydroceramide elicited only a limited amount of cell death, which was constant and not dose dependent between 20 and 100 µM dihydroceramide (data not shown). Exposure of BBEC to increasing concentrations (0.1–40 µM) of exogenous C₂-ceramide for 2 h induced activation of JNK, as assayed by measurement of phosphorylation of GST-c-jun fusion protein by JNK immunoprecipitated from the cells (Fig. 7) . The increase in JNK phosphorylation of the GST-c-jun fusion protein was observed as early as 1 h after the addition of C₂-ceramide, with the largest increment in the increase in JNK activation occurring between 1 and 2 h or later (data not shown). Thus, exogenous C₂-ceramide, which induces apoptotic cell death (45) , also induces the activation of JNK in endothelial cells. From these data we conclude that ceramide is a potential participant in detachment-induced apoptotic signaling in endothelial cells and, specifically, that it may mediate some of the signals associated with integrin blockade.

View larger version (19K): [in this window] [in a new window]   Fig. 6. Exogenous C2-ceramide is cytotoxic to BBEC. A, BBEC (2 x 104 cells/well) were incubated in a 96-well plate with 0–80 µM C2-ceramide for 24 h. Cell viability was assessed by MTT assay at 24 h as detailed in "Materials and Methods." Each data point represents the mean of six replicate wells; error bars represent SD of the six replicates, and where they are smaller than the symbols, they are hidden by them. B, Details are as in A, except that BBEC were incubated with 30 µM (•) or 50 µM C2-ceramide () for different lengths of time, washed and incubated without ceramide.

View larger version (28K): [in this window] [in a new window]   Fig. 7. C2-ceramide induces activation of JNK in endothelial cells. BBEC (5 x 106 cells/sample) were exposed to C2-ceramide (0–40 µM) for 2 h, harvested, lysed, and immunoprecipitated with -JNK antibody. Kinase activity of JNK was quantitated using GST-C-jun fusion protein as substrate, which was resolved on 10% SDS-PAGE. This autoradiograph demonstrates phosphorylation of the GST-C-jun fusion protein substrate by the JNK immunoprecipitated from C2-ceramide-treated BBEC lysates. Positive control was obtained by measuring JNK activation in BBEC (5 x 106) exposed to UV radiation (25 mjoules) and then incubating for 2 h before assessment of JNK activity as described above. Equal amounts of precipitated JNK were verified by -JNK immunoblots (data not shown).

	DISCUSSION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our data showing that microvessels in MYCN-amplified NB express high levels of integrins _vß₃ and _vß₅ point to a potential role for the MYCN oncogene in angiogenesis. Because MYCN itself is a transcription factor, it could regulate angiogenic factors originating in the tumor cells, such as vascular endothelial growth factor or matrix-degrading proteins, as has been shown for mutant K-ras and polyoma middle T antigen, respectively (52 , 53) . Supporting this, NB tumors and their surrounding stromal cells express matrix metalloproteinases (31) . Moreover, MYCN is required for motility of NB cell lines and for their proteolytic activity in vitro, suggesting that MYCN may contribute to NB aggressiveness by enhancing angiogenesis (54) . Thus, it is possible that overexpression of MYCN may contribute to NB aggressiveness by promoting angiogenesis.

The immunohistochemical stains of the NBs in our series showed that integrin _vß₃ and _vß₅ expression was confined to the endothelium of tumor microvessels and did not appear on the tumor cells themselves (Figs. 1 and 2 and data not shown). Gladson et al. (55) examined paraffin sections of NBs for expression of the individual _v, ß₃, and ß₅ integrin subunits. Our immunostaining with the mAb against _vß₅ (P1F6) are in agreement with their description that the ß₅ integrin subunit is not expressed at the protein level in neuroblasts of undifferentiated tumors. However, Gladson et al. (55) observed that the _v and ß₃ subunits were associated with most of the undifferentiated neuroblasts, a finding interpreted to imply that integrin _vß₃ is expressed on those cells. It is conceivable that the binding partner(s) for the _v and ß₃ integrin chains seen in that work did not belong to integrin _vß₃ but may have originated from different integrin partners (i.e., _vß₆ or _vß₈; Refs. 56, 57, 58 ), some of which are highly expressed in the mammalian brain (58) . This could explain the specific microvascular endothelial expression of integrin _vß₃ we found using mAb LM609, which specifically recognizes the native conformation of the functional RGD-binding domain on integrin _vß₃ (7 , 59) .

We demonstrate here for the first time that ceramide is elevated in endothelial cells that are prevented from attaching and spreading on matrix, either by plating on BSA or by inhibition with the RGDfV inhibitory peptide (Fig. 5) . The increase in endogenous ceramide induced by RGDfV and plating on BSA are in a range comparable with increases observed following exposure to stressors such as TNF, sorbitol, H₂O₂, LPS, UV irradiation, or irradiation (17 , 60) . Our data suggest that ceramide may be involved in signaling pathways originating from specific inhibition of the RGD binding site on vitronectin-binding integrins such as _vß₃ and _vß₅. Although we cannot exclude possible involvement of multiple vitronectin-binding integrins under these conditions, it is likely that inhibition of adhesion and spreading on vitronectin by RGDfV was mediated mostly by inhibition of integrins _vß₃ and _vß₅, both of which are highly expressed on endothelial cells (7 , 9) . Alternatively, increased ceramide may be the result of inhibition of cell anchorage and/or of the associated cytoskeletal signals resulting from shape change per se (61) , may be unrelated to the alteration of integrin signaling, or possibly may be related to direct activation of apoptosis by an RGD-blocking peptide (62) . To implicate specific signaling intermediates that may link integrins, ceramide, and regulation of cell survival (i.e., phosphatases such as protein tyrosine phosphatase- or kinases such as AKT/PKB) and to determine the role of cytoskeletal shape change versus integrin blockade in cell survival, it will be necessary to explore these pathways in detail by using specific dominant-negative/positive intermediates or anti-integrin antibodies that block signaling without inhibiting cell attachment (63) . Although there have been no prior reports of ceramide as a second messenger in integrin signaling, recent work suggests that selectins, another family of cell adhesion molecules, may use ceramide signaling in the activation of lymphocytes (64) . Increased endogenous ceramide may contribute to endothelial apoptosis, as shown by the fact that incubation of endothelial cells with C₂-ceramide is associated with DNA laddering (data not shown and Ref. 45 ). Our results compare favorably with those described by Xu et al. (45) , as can be seen in the fact that we achieved a mean LC₅₀ of 31.7 µM ± 10.5 SD C₂-ceramide in our experiments compared with their report of 50% endothelial cell kill at 50 µM C₂-ceramide. Our data provide the first evidence that signaling pathways originating from inhibition of endothelial cell anchorage are associated with increased endogenous ceramide, which may contribute to apoptosis.

In several cell types including endothelial cells, JNK is thought to lie downstream of ceramide in apoptotic pathways induced by stress (23 , 65) . In epithelial cells, detachment from matrix induces JNK activation, which can be inhibited by the interleukin 1ß converting enzyme protease inhibitor crmA or by overexpression of Bcl-2 (66) . Our experiments, in which exogenous C₂-ceramide induced activation of JNK in endothelial cells, suggest that the increased endogenous ceramide we observed after inhibition of integrin-mediated cell anchorage (Fig. 5) may contribute to JNK activation in signaling of apoptosis (Fig. 7) . However, the ceramide-JNK-apoptosis link is not universal, and several reports describe the dissociation of ceramide production from JNK and other death signals (67, 68, 69, 70) . Recent work by Ameyar et al. (67) described lack of correlation between JNK activation and ceramide production in breast carcinoma cells, suggesting that the ceramide-JNK link may be specific to individual cell types. Adding to the complexity, Khwaja and Downward (68) show that, although JNK is activated by detachment of normal MDCK epithelial cells, it is unlikely that it plays a direct role in detachment-induced apoptosis in these cells because activated phosphatidylinositol 3'-kinase and AKT prevented apoptosis but did not prevent the associated JNK activation. Activated Raf and dominant negative SAPK/ERK kinase-1, which inhibited the detachment-induced activation of JNK, did not prevent apoptosis, suggesting induction of apoptosis by elements which are independent of JNK activation in the MDCK epithelial cells (68) . Clearly the specificity of cellular responses to ceramide and other apoptosis-mediating signals depends upon many factors including the nature of the stimulus, costimulatory signals, the cell type involved, and (most likely) the surrounding matrix proteins. Thus, although inhibition of endothelial cell anchorage induced elevation of ceramide in our studies and exogenous ceramide induced JNK activation and apoptotic cell death, a direct causal link between the three needs to be investigated further.

Cell survival is dictated by a delicate dynamic balance between pro- and anti-apoptotic signals, which in turn depend upon a complex interplay of factors such as integrins and growth factors. For example, in fibroblasts interleukin-1-mediated inflammatory responses, which induce activation of the proapoptotic JNK/SAPK, and the survival-mediating NFB are both modulated by integrin binding to RGD motifs on fibronectin (71) . Both the JNK/SAPK and NFB pathways, as well as phosphatidylinositol 3'-kinase/AKT, are tightly controlled, and several reports suggest that ceramide may help regulate them to tip the balance in favor of apoptosis (60 , 72, 73, 74, 75) . These reports support our finding that ceramide may be involved in regulation of anchorage-induced signaling in endothelial cells. Furthermore, NFB mediates integrin _vß₃-dependent endothelial cell survival (16) , whereas ceramide levels increase in endothelial cells upon loss of integrin-dependent matrix anchorage (Fig. 5) . It is possible that under conditions of inhibition of cell anchorage ceramide mediates signals to down-regulate NFB activation in the endothelial cells, thus promoting their apoptosis through activation of JNK/SAPK (Fig. 7 and Refs. 23 , 66 , and 72 ). It is also possible, that stress conditions which augment ceramide and/or JNK may synergistically increase apoptosis associated with loss of anchorage-specific survival signals, thereby increasing any antiangiogenic activity of the RGDfV peptide.

The relationship between ceramide generation and caspase activation is dependent on cell type and the apoptotic stimulus used. At present, the relationship between ceramide metabolism and regulation of caspase activation in our system has been only preliminarily defined. Both nuclear fragmentation of BBEC plated on vitronectin in the presence of RGDfV and DNA laddering in BBEC plated on BSA-coated plates are inhibited by the pan-caspase inhibitor, BOC-D-FMK (data not shown). In preliminary experiments BOC-D-FMK did not prevent the increase in endogenous ceramide in BBEC plated on vitronectin for 4 h in the presence of RGDfV, suggesting that ceramide may be upstream of the BOC-D-FMK-inhibitable caspases (data not shown). Consistent with this, initial experiments show that DNA laddering induced by exogenous C₂-ceramide can be inhibited by BOC-D-FMK (data not shown). This suggests a situation similar to that found in apoptosis induced by low-dose okadaic acid in human neuroepithelioma cells in which endogenous ceramide accumulation is insensitive to caspase inhibitors and precedes caspase activation, whereas exogenous C₆-ceramide-induced apoptosis was inhibited by the caspase inhibitors (76 , 77) . In other types of apoptotic signals, such as those mediated by TNF superfamily receptors, the initiator caspases are thought to be upstream of ceramide generation, whereas effector caspases are downstream of it (78, 79, 80, 81) . Because specific signaling cascades may differ depending on the stimulus and the cell type (78) , this needs to be examined individually for each cell type and experimental model. Ongoing investigations using BBEC as a model seek to determine the direct relationship between ceramide and endothelial anoikis.

In summary, this is the first description that suggests that integrins _vß₃ and _vß₅ are highly expressed on the microvascular endothelium in a pediatric malignant tumor, high-risk neuroblastoma. Also novel is our finding that endothelial cell apoptosis, induced by inhibition of vitronectin-binding integrins and the prevention of endothelial cell anchorage on matrix, is associated with increased endogenous ceramide. Taken together, these data suggest that RGD-dependent, vitronectin-binding integrins on endothelial cells, such as _vß₃ and _vß₅, may be potential targets for antiangiogenic treatment of NB.

	ACKNOWLEDGMENTS

 
We thank Pei-Gen Wu, Yenbou Liu, and Dennis Duncan for their excellent technical assistance. We also thank Drs. C. P. Reynolds, B. Maurer, and N. Keshelava for their expert advice and assistance with the ceramide assays. Special thanks to Dr. D. A. Cheresh for generously supplying mAb LM609 and to Dr. A. Jonczyk (Merck, Darmstadt, Germany) for the generous gift of the RGDfV and RADfV cyclic peptides. Our thanks also to the donation from members of Riviera Hall Lutheran School in Los Angeles in support of this research.

	FOOTNOTES

 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

¹ This work was supported by Grant CA75637 to D.L.D. from the National Institutes of Health and by grants to A.E-E. from the Childrens Cancer Research Fund and the Concern Foundation. This work was also supported in part by Grant UO1, CA70903-02 from NIH (to D. L. D.); Grants CA60104 and CA02649 from the National Cancer Institute, Department of Health and Human Services (to R. C. S.); Grant CA14089 (to S. G.); Grants NS26310 and HL61951 (to K. S. K.); the Neil Bogart Memorial Fund of the T.J. Martell Foundation for Leukemia, Cancer, and AIDS Research; "My Brother Joey Foundation" founded by Judi and Art Partridge, and the Michael Hoefflin Foundation for Children’s Cancer (to A. E-E.); and by the STOP Cancer Award (to D. L. D.).

² To whom requests for reprints should be addressed, at the Department of Pediatrics, Herman B. Wells Center for Pediatric Research, Cancer Research Institute, 1044 West Walnut Street, Room 468, Indiana University School of Medicine, Indianapolis, IN 46202. Phone: (317) 278-3718; Fax: (317) 274-8679; E-mail: ddurden{at}iupui.edu

³ The abbreviations used are: NFB, nuclear factor B; LPS, lipopolysaccharide; TNF, tumor necrosis factor ; MAPK, mitogen-activated protein kinase; JNK, C-jun N-terminal kinase; SAPK, stress-activated protein kinase; MMP, matrix metalloproteinase; NB, neuroblastoma; BBEC, bovine brain endothelial cells; mAb, monoclonal antibody; HUV-EC-C, human umbilical vein endothelial cells; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; GST, glutathione S-transferase; MKI, mitosis-karyorrhexis index; AKT/PKB, protein kinase B; BOC-D-FMK, BOC-aspartyl(OMe)-fluoromethylketone.

Received 4/19/99. Accepted 12/ 2/99.

	REFERENCES

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Felding-Habermann B., Cheresh D. A. Vitronectin and its receptors. Curr. Opin. Cell Biol., 5: 864-868, 1993.[Medline]
2. Huang X., Wu J., Spong S., Sheppard D. The integrin _vß₆ is critical for keratinocyte migration on both its known ligand, fibronectin, and on vitronectin. J. Cell Sci., 111: 2189-2195, 1998.[Abstract]
3. Nishimura S. L., Sheppard D., Pytela R. Integrin _vß₈: interaction with vitronectin and functional divergence of the ß₈ cytoplasmic domain. J. Biol. Chem., 269: 28708-28715, 1994.[Abstract/Free Full Text]
4. Schnapp L. M., Hatch N., Ramos D. M., Klimanskaya I. V., Sheppard D., Pytela R. The human integrin ₈ß₁ functions as a receptor for tenascin, fibronectin, and vitronectin. J. Biol. Chem., 270: 23196-23202, 1995.[Abstract/Free Full Text]
5. Pfaff M., Tangemann K., Muller B., Gurrath M., Muller G., Kessler H., Timpl R., Engel J. Selective recognition of cyclic RGD peptides of NMR defined conformation by _IIbß₃, _Vß₃, and ₅ß₁ integrins. J. Biol. Chem., 269: 20233-20238, 1994.[Abstract/Free Full Text]
6. Aumailley M., Gurrath M., Muller G., Calvete J., Timpl R., Kessler H. Arg-Gly-Asp constrained within cyclic pentapeptides: strong and selective inhibitors of cell adhesion to vitronectin and laminin fragment P1. FEBS Lett., 291: 50-54, 1991.[Medline]
7. Cheresh D. A. Human endothelial cells synthesize and express an Arg-Gly-Asp-directed adhesion receptor involved in attachment to fibrinogen and von Willebrand factor. Proc. Natl. Acad. Sci. USA, 84: 6471-6475, 1987.[Abstract/Free Full Text]
8. Friedlander M., Brooks P. C., Shaffer R. W., Kincaid C. M., Varner J. A., Cheresh D. A. Definition of two angiogenic pathways by distinct _v integrins. Science (Washington DC), 270: 1500-1502, 1995.[Abstract/Free Full Text]
9. Klein S., Giancotti F. G., Presta M., Albelda S. M., Buck C. A., Rifkin D. B. Basic fibroblast growth factor modulates integrin expression in microvascular endothelial cells. Mol. Biol. Cell, 4: 973-982, 1993.[Abstract]
10. Christofidou-Solomidou M., Bridges M., Murphy G. F., Albelda S. M., DeLisser H. M. Expression and function of endothelial cell _v integrin receptors in wound-induced human angiogenesis in human skin/SCID mice chimeras. Am. J. Pathol., 151: 975-983, 1997.[Abstract]
11. Brooks P. C., Montgomery A. M., Rosenfeld M., Reisfeld R. A., Hu T., Klier G., Cheresh D. A. Integrin _vß₃ antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels. Cell, 79: 1157-1164, 1994.[Medline]
12. Brooks P. C., Stromblad S., Klemke R., Visscher D., Sarkar F. H., Cheresh D. A. Anti-integrin _vß₃ blocks human breast cancer growth and angiogenesis in human skin. J. Clin. Invest., 96: 1815-1822, 1995.
13. Friedlander M., Theesfeld C. L., Sugita M., Fruttiger M., Thomas M. A., Chang S., Cheresh D. A. Involvement of integrins _vß₃ and _vß₅ in ocular neovascular diseases. Proc. Natl. Acad. Sci. USA, 93: 9764-9769, 1996.[Abstract/Free Full Text]
14. Brooks P. C., Clark R. A., Cheresh D. A. Requirement of vascular integrin _vß₃ for angiogenesis. Science (Washington DC), 264: 569-571, 1994.[Abstract/Free Full Text]
15. Stromblad S., Becker J. C., Yebra M., Brooks P. C., Cheresh D. A. Suppression of p53 activity and p21WAF1/CIP1 expression by vascular cell integrin