The role of angiocidin in sarcomas

Abstract

BACKGROUND:

Angiocidin, first identified as a tumor-associated thrombospondin-1 (TSP-1) receptor, is a key mediator of tumor progression. TSP-1, an extracellular protein produced by stromal cells, up-regulates gelatinases and tumor cell invasion in epithelial malignancies. The authors recently developed 2 angiocidin-inhibitory peptides that block angiocidin–TSP-1 binding. They hypothesized that angiocidin mediates increased gelatinase expression and tumor cell invasion in sarcomas through its interaction with TSP-1.

METHODS:

Angiocidin, TSP-1, and gelatinase expression was evaluated in low-grade and high-grade sarcoma specimens. The authors established 3 distinct cell lines from a patient with an extraskeletal osteosarcoma: EXOS-N (normal mesenchymal), EXOS-P (primary osteosarcoma), and EXOS-M (lung metastasis). Each was evaluated for angiocidin, gelatinase, and gelatinase inhibitor (tissue inhibitors of metalloproteinase) expression and for invasive capacity. Their responsiveness to TSP-1 was determined. The role of angiocidin in up-regulating gelatinase expression and invasion was studied using the authors' angiocidin-inhibitory peptides.

RESULTS:

Expression of angiocidin, TSP-1, and gelatinases correlated with tumor grade. Angiocidin expression, gelatinase activity, and invasiveness in the EXOS cell lines correlated with phenotype; EXOS-N cells did not express angiocidin or gelatinases and were not invasive; EXOS-M cells were 5 times more invasive than EXOS-P cells and exhibited greater angiocidin and gelatinase expression. EXOS cell gelatinase activity and invasiveness increased 4- to 5-fold in response to TSP-1. Inhibition of angiocidin with the authors' inhibitory peptides blocked TSP-1–promoted increases in gelatinase activity and tumor cell invasion.

CONCLUSIONS:

Angiocidin promotes gelatinase up-regulation and tumor cell invasion in sarcomas. Angiocidin-inhibitory peptides are potent inhibitors of sarcoma cell invasion in vitro, suggesting a potential therapeutic role for these peptides in the treatment of sarcomas. Cancer 2009. © 2009 American Cancer Society.

Sarcomas are a diverse group of neoplasms that arise from mesenchymal tissues.1 They are generally regarded as aggressive malignancies with a propensity toward hematogenous spread and high recurrence rates. Surgical resection along with radiation is the mainstay of therapy. Despite considerable toxicity, adjuvant chemotherapy is largely ineffective.2-5 Newer, better-tolerated targeted molecular therapies have shown promising results in epithelial malignancies, but have not yet been successful in the treatment of sarcomas.6, 7

Angiocidin is a tumor-associated molecule that is emerging as a key mediator of angiogenesis and tumor progression.8, 9 Overexpression of angiocidin has been shown in many solid epithelial tumors, and increased levels of angiocidin have been demonstrated in the serum of patients with different epithelial malignancies.10, 11 Preliminary reports have shown low to negligible expression of angiocidin in normal adult tissues, making it an attractive molecular target for tumor-specific therapy.9, 12, 13 The protumor effects of angiocidin are the result of high-affinity binding interactions with extracellular matrix proteins including thrombospondin-1 (TSP-1), a matrix protein produced in tumors by activated stromal cells of mesenchymal origin.14, 15 These angiocidin-stromal protein binding interactions lead to an up-regulation in gelatinase expression and matrix remodeling, critical steps in the metastatic cascade.9, 16

We have recently synthesized 2 angiocidin-inhibitory peptides that block the activity of angiocidin through competitive inhibition of its stromal protein-binding interactions. Preliminary data have demonstrated that these peptides are excellent inhibitors of angiogenesis and tumor cell invasion in vitro.9, 10, 14 The first of these peptides, a 6‒amino acid residue peptide with the sequence CSVTCG, blocks the binding of angiocidin to TSP-1. The second peptide, a 25‒amino acid sequence derived from the TSP-1 binding site within the angiocidin molecule, also competitively inhibits the binding of angiocidin to TSP-1.

In the present study, we evaluate the possible role of angiocidin in sarcoma progression. We hypothesize that angiocidin mediates an up-regulation of gelatinase activity and promotes tumor cell invasion in sarcomas. We also evaluate the potential use of our angiocidin-inhibitory peptides as inhibitors of tumor cell invasion in these aggressive neoplasms.

MATERIALS AND METHODS

Reagents and Antibodies

All reagents, unless otherwise specified, were purchased from Sigma Chemical Company (St. Louis, Mo). Tissue culture supplies were purchased from Fisher Scientific Company (Pittsburgh, Pa). Polyclonal antihuman TSP-1 antibody was raised in a goat as previously described.17 Polyclonal antiangiocidin antibody was raised in a rabbit against angiocidin protein isolated from a human lung carcinoma and purified as previously described.15 Sheep antihuman matrix metalloproteinase (MMP)-9 immunoglobulin G (IgG) was purchased from The Binding Site Co. (Birmingham, UK), and goat antihuman MMP-2 IgG was purchased from R&D Systems (Minneapolis, Minn). For each antibody, the species-respective nonspecific IgG was used as a control.

TSP-1 Purification

TSP-1 was isolated from platelets activated with Ca²⁺ ionophore A23187 as previously described and was further purified to remove bound transforming growth factor β1 as previously described.18, 19 The TSP-1 preparations used for cell culture were routinely checked for endotoxin content with the E-TOXATE kit (Sigma Chemical Co.), and no detectable levels of endotoxin were found.

Synthesis of Angiocidin Inhibitory Peptides

In this study, we used 2 angiocidin inhibitory peptides, the CSVTC and 25-mer peptides that have previously been shown to inhibit angiocidin–TSP-1 binding.20, 21 The CSVTCG peptide, 25-mer peptide, K₈₁GKITFCTGIRVAHLALKHRQGKNH₁₀₅, and a 25-mer scramble peptide were synthesized by Alpha Diagnostic International, San Antonio, Texas. The peptides were >80% pure. The CSVTCG peptide is homologous to the angiocidin binding site on the TSP-1 molecule.20, 21 This recombinant CSVTCG peptide inhibits tumor cell invasion and decreases tumor metastasis in a mouse melanoma model.17 The 25-mer peptide is derived from a 20 amino acid segment in the amino terminal domain of recombinant angiocidin. This particular domain of angiocidin contains the TSP-1 binding site; an angiocidin deletion mutant missing this domain fails to bind matrix proteins and loses its protumor effects.9, 10

Cell Culture

The 3 human EXOS cell lines were developed from a patient who was treated at The M. D. Anderson Cancer Center for stage IV disease (grading determined according to the American Joint Committee on Cancer staging system). The cells of origin were from 3 tissues in the same patient: benign fibroblasts from an area of normal tissue stroma (EXOS-N), primary osteosarcoma cells from the primary tumor (EXOS-P), and metastatic osteosarcoma cells from a metastatic lesion in the lung (EXOS-M). EXOS-N cells proliferate very slowly, exhibit contact growth inhibition in vitro, and do not form tumors in nude mice. EXOS-P and EXOS-M cells proliferate faster, do not exhibit contact growth inhibition in vitro, and form tumors in nude mice with features consistent with osteosarcoma. EXOS cells were cultured in Dulbecco modified Eagle medium (DME) supplemented with 10% fetal calf serum (FCS). The cultures were maintained on plastic, incubated in 5% carbon dioxide at 37°C in a humidified incubator, and proven to be pathogen free. Tumor cells were harvested from cultures 80% to 90% confluent by a short exposure to 0.02% ethylenediaminetetraacetic acid solution and then resuspended in serum-free DME before use.

Histology

Formalin-fixed, paraffin-embedded sarcoma specimens were obtained from the tissue bank of the Department of Pathology at the Medical College of Georgia. Three high-grade and 3 low-grade tumor specimens were obtained for each sarcoma subtype analyzed: osteosarcoma, leiomyosarcoma, liposarcoma, and malignant fibrous histiosarcoma. Sections were prepared from an area of tumor most characteristic of the sarcoma subtype. Collection of human tissue specimens was performed under a protocol approved by the Medical College of Georgia Institutional Review Board, and patient identifiers were removed before release of tissue blocks from the pathology department.

Immunohistochemical Localization of Angiocidin, TSP-1, and the Gelatinases in Sarcoma Specimens

Sarcoma specimens were analyzed for expression of angiocidin, TSP-1, and the gelatinases, MMP-2 and MMP-9. Slides containing 5-μm-thick tissue sections were deparaffinized and rehydrated by sequential treatment with xylene and ethyl alcohol (70%, 80%, 95%, and 100%). Tissue sections stained for TSP-1 were treated with 0.1% trypsin in 20 mM Tris–HCl buffer (pH 7.8) for 7 minutes at room temperature. Endogenous peroxidase activity was quenched by treatment with 3% hydrogen peroxide for 10 minutes. Nonspecific binding in the tissue sections was blocked with 1% milk. Specimens were incubated in primary antibody overnight at 4°C. Primary antibodies were sheep polyclonal antihuman MMP-9 IgG diluted 1:50, polyclonal goat antihuman MMP-2 diluted 1:100, goat polyclonal antihuman TSP-1 IgG diluted 1:50, or polyclonal rabbit antihuman angiocidin IgG diluted 1:5000. Appropriate nonimmune IgG was used for negative controls. The sections were treated with secondary antibodies as described in the Vectastain Universal Quick Kit (Vector Laboratories, Burlingame, Calif). Immunohistochemical staining was performed using a biotin-streptavidin immunoperoxidase method (ABC Elite; Vector Laboratories) and 3,3′-diaminobenzidine chromogen. Counterstaining was performed with hematoxylin.

Immunocytochemical Localization of TSP-1 and Angiocidin in EXOS Cells

Endogenous peroxidase activity in EXOS cells fixed in glutaraldehyde was quenched by treatment with 3% H₂O₂ for 10 minutes. Nonspecific binding in the tissue sections was blocked with 2% milk in 0.1 M maleic acid buffer (pH 7.5) using a Vector blocking kit and following the manufacturer's recommendations. Goat polyclonal antihuman TSP-1 IgG diluted 1:50 or rabbit polyclonal antiangiocidin antibody diluted 1:5000 was used as a primary antibody. The specimens were incubated overnight at 4°C. Nonimmune goat IgG and rabbit IgG were used as negative controls. Immunohistochemical staining was performed by using a biotin-streptavidin immunoperoxidase method (ABC Elite; Vector Laboratories) and 3,3′-diaminobenzidine chromogen.

Western Blot Analysis

Protein levels of the tumor cell extracts were determined by bicinchoninic acid protein analysis (Bio-Rad Laboratories, Richmond, Calif). Sample tumor cell extracts containing equal amounts of total protein were fractionated by 8% to 25% gradient (TSP-1) or 12% (angiocidin) sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred to nitrocellulose membranes. Nonspecific binding was blocked with 5% milk in phosphate-buffered saline (PBS) containing 0.05% Tween 20 overnight. The immunoblots were then incubated with 80 μg/mL polyclonal goat antihuman TSP-1 IgG or 1:500-diluted polyclonal rabbit antihuman angiocidin serum for 3 hours at room temperature in PBS with Tween 20. After washing, the immunoblots were incubated with horseradish peroxidase-conjugated antigoat or antirabbit IgG for 45 minutes. The bound antibodies were detected with an enhanced chemiluminescence system (Amersham, Arlington Heights, Ill). Purified recombinant TSP-1 and angiocidin served as controls. Angiocidin purification has been previously described.9

Effect of TSP-1 on Sarcoma Cell Gelatinase and Tissue Inhibitors of Metalloproteinase Expression

Approximately 1 × 10⁶ cells (EXOS-N, EXOS-P, or EXOS-M) were plated in each well of a 6-well microtiter plate and allowed to attach and grow to 80% to 90% confluence in medium with 10% FCS. The cells were weaned from the FCS-containing medium to serum-free medium over 48 hours. The cells were then incubated for 48 hours at 37°C in 1 mL of either serum-free medium supplemented with 0.1% bovine serum albumin (BSA) (control) or serum-free medium supplemented with 0.1% BSA plus TSP-1 at concentrations of 10, 20, or 30 μg/mL. Cells were washed twice with PBS and then lysed with 100 μL of cold 1% Triton X-100 in Tris-buffered saline, pH 8.5, to prepare tumor cell extracts. MMP-2, MMP-9, tissue inhibitors of metalloproteinase (TIMP)-1 and TIMP-2 protein levels were measured using enzyme-linked immunosorbent assay kits purchased from Calbiochem (San Diego, Calif) following the manufacturer's recommendations.

Tumor Cell Invasion Assay

Tumor cell invasion was evaluated using a modified Boyden chamber invasion assay as previously described. This in vitro assay has proven to be a reliable surrogate marker for malignant potential in vivo.22, 23 Briefly, transwell polycarbonate filter inserts with 8-μm pores (Corning, Acton, Mass) were coated with growth factor reduced Matrigel and dried overnight at room temperature. The lower chamber was filled with 600 μL of serum-free DME (negative control) or with DME containing 10% fetal bovine serum (positive control). EXOS-N, EXOS-P, or EXOS-M cells (5 × 10⁵) were suspended in 100 μL serum-free media and placed in the upper chamber. The cells were cultured for 48 hours at 37°C. Media were aspirated, and cells on the upper surface of the filter were removed completely by wiping with a moist cotton swab. The filters were placed in 4 μg/mL calcein in Hanks balanced salt solution (HBSS) for 2 hours at 37°C, then placed in 4% paraformaldehyde in HBSS for 10 minutes. Inserts were moved to empty wells and then read with a chemiluminescence plate reader using a 480/560 filter set. To analyze the effect of TSP-1 on EXOS cell invasion, TSP-1 at concentrations of 10, 20, or 30 μg/mL were added to the upper chamber containing serum-free media. Inhibition assays were conducted similarly with the addition of 40 μg/mL of either CSVTCG or 25-mer angiocidin-inhibitory peptides to the cell suspension in the upper chamber. A 25-mer scramble peptide was used as a negative control. Each experiment was performed in triplicate.

Statistical Analysis

Immunostaining data were analyzed using contingency tables and Fisher exact test with a 2-sided P value. Correlation coefficients were calculated using Pearson correlation calculations and a 2-tailed P value. Differences between groups were analyzed using the Student t test or, for ≥3 groups, a 1-way analysis of variance. P values <.05 were considered to be significant.

RESULTS

Expression of Angiocidin, TSP-1, and the Gelatinases Correlate With Tumor Grade in Sarcoma Specimens

We evaluated 4 histological subtypes of sarcoma (leiomyosarcoma, liposarcoma, malignant fibrous histiosarcoma, and osteosarcoma) for expression of angiocidin, TSP-1, and gelatinases (MMP-9 and MMP-2). Overall, 24 low-grade and high-grade tumors were included. Ninety-two percent (11 of 12) of high-grade sarcomas stained positively for angiocidin versus only 42% (5 of 12) of low-grade tumors (odds ratio [OR], 15.4; P < .05). Even when low-grade tumors showed overexpression of angiocidin, high-grade tumors demonstrated greater staining intensity and a greater percentage of positive cells. Angiocidin expression patterns varied by tumor histology. However, in most cases angiocidin localized to the tumor cells; there was negligible angiocidin expression in the stroma (Fig. 1). High-grade sarcomas showed a tendency toward increased expression of TSP-1 over low-grade lesions, although this failed to reach statistical significance (58% or 7 of 12 vs 25% or 3 of 12; OR, 4.2 [P = .2]). Only in leiomyosarcoma specimens did low-grade lesions stain positively for TSP-1 more frequently than high-grade lesions (2 of 3 vs 1 of 3, respectively). Staining for MMP-9 and MMP-2 colocalized with angiocidin. Overexpression of MMP-9 and MMP-2 was greater in high-grade versus low-grade lesions (7 of 9 vs 0 of 9; OR, 57 [P < .01] and 7 of 9 vs 3 of 9; OR, 7 [P = .15], respectively).

Figure 1. Angiocidin expression in sarcoma specimens is shown to correlate with tumor grade. Four sarcoma subtypes were analyzed for angiocidin expression using immunohistochemistry: osteosarcoma, leiomyosarcoma, liposarcoma, and malignant fibrous histiosarcoma (MFH). Three specimens of high-grade tumor and 3 specimens of low-grade tumor were analyzed for each sarcoma subtype. The sections were prepared from an area of tumor most characteristic of the sarcoma subtype. High-grade tumors demonstrated greater angiocidin expression. Angiocidin was found to be immunolocalized to the tumor cells. When specimens stained positive for angiocidin, immunostaining was noted within the sarcoma tumor cells, with no staining noted in the associated tumor stroma.

Expression of Angiocidin in Sarcoma Cells Correlates With Their Invasive Potential

EXOS-N (normal mesenchymal), EXOS-P (primary tumor), and EXOS-M (metastatic) cells were evaluated for angiocidin expression. Angiocidin expression was evaluated by immunohistochemistry on all 3 EXOS cell lines. EXOS-N cells did not express angiocidin. EXOS-P and EXOS-M cells stained positively for angiocidin, with EXOS-M cells demonstrating greater staining intensity (Fig. 2A). Western immunoblotting confirmed these results (Fig. 2B).

Figure 2. Angiocidin expression in extraskeletal osteosarcoma (EXOS) cells was found to correlate with cell phenotype. (A) Immunocytochemical localization of angiocidin in EXOS-N (normal mesenchymal), EXOS-P (primary tumor), and EXOS-M (metastatic) cells is shown. Expression of angiocidin was negligible in EXOS-N cells. The cytoplasm and membrane of EXOS-P and EXOS-M cells stained positively for angiocidin, with EXOS-M cells demonstrating greater staining intensity. Nuclear staining was not observed. (B) Western blot analysis of angiocidin expression in EXOS cells is shown. Results from the cytostaining were confirmed by this Western blot analysis. EXOS-N cells demonstrated no angiocidin expression. EXOS-M cells demonstrated greater expression of angiocidin than EXOS-P cells. Purified, recombinant angiocidin was used as a positive control.

The invasive capacity of our cell lines was measured using a modified Boyden chamber. Invasive potential of EXOS cells correlated with their level of angiocidin expression. EXOS-N cells were not invasive; EXOS-M cells were 5 times more invasive than EXOS-P cells (0.05 vs 0.01; P = .001) (Fig. 3).

Figure 3. Extraskeletal osteosarcoma (EXOS) cell invasiveness was found to correlate with angiocidin expression. The ability of EXOS-N (normal mesenchymal), EXOS-P (primary osteosarcoma), and EXOS-M (lung metastasis) cells to invade extracellular matrix was measured using a modified Boyden chamber. EXOS-M cells, which were derived from a metastatic lung lesion and which demonstrated the highest angiocidin expression among the 3 cell lines, demonstrated the greatest invasiveness. EXOS-P cells, which came from the primary tumor and which demonstrated intermediate angiocidin expression, were 5-fold less invasive than EXOS-M cells. EXOS-N cells, derived from the patient's normal fibroblasts, did not express angiocidin and did not invade the Matrigel. Results were found to be statistically significant between all 3 groups (P = .001). OD 560 indicates optical density at 560 nanometers.

Gelatinase Activity in Osteosarcoma Cells Correlates With Their Invasive Potential

Gelatinase activity depends on the balance between gelatinase and gelatinase inhibitor (TIMPs) production. Therefore, we evaluated our EXOS cells for MMP-2, MMP-9, TIMP-1, and TIMP-2 expression. Baseline gelatinase expression by the 3 EXOS cell lines correlated with cell phenotype. EXOS-N cells showed no detectable levels of gelatinases. MMP-9 and MMP-2 levels were 3 and 7 times higher (0.33 ng/mL vs 0.133 ng/mL [P < .05] and 9.33 ng/mL vs 1.33 ng/mL [P < .001], respectively) in EXOS-M cells compared with EXOS-P cells (Fig. 4A).

Figure 4. Baseline gelatinase and gelatinase inhibitor expression by sarcoma cells indicated that gelatinase activity was correlated with cell invasive potential. (A) Gelatinase expression (matrix metalloproteinase [MMP]-2 and MMP-9) by extraskeletal osteosarcoma (EXOS) cells was measured using enzyme-linked immunoadsorbent assays (ELISA). EXOS-N (normal mesenchymal) cells did not appear to express gelatinases, whereas EXOS-M (lung metastasis) cells demonstrated 3-fold and 7-fold increases in MMP-9 and MMP-2 levels, respectively, compared with EXOS-P (primary osteosarcoma) cells. (B) Gelatinase inhibitor expression (tissue inhibitors of metalloproteinases [TIMP]-1 and TIMP-2) was measured using ELISA assays. TIMP-1 levels were found to be similar in EXOS-P and EXOS-M cells, but TIMP-2 levels were found to be 7-fold higher in EXOS-P compared with EXOS-M cells. Overall, the higher gelatinase expression and lower TIMP-2 level noted in the EXOS-M cells compared with the EXOS-P cells indicates a higher gelatinolytic activity and correlates with their more invasive phenotype.

TIMP-1 levels were similar in EXOS-P and EXOS-M cells (105 vs 116 ng/mL, respectively; P > .05), but TIMP-2 levels were 7-fold higher in EXOS-P than in EXOS-M cells (14 vs 2 ng/mL, respectively; P = .007) (Fig. 4B). Overall, the higher gelatinase expression and lower TIMP-2 level seen in the EXOS-M cells indicates higher gelatinolytic activity and correlates with their more invasive phenotype as compared with the EXOS-P cells.

EXOS Cell Gelatinase Expression and Invasiveness Increases in Response to TSP-1

EXOS-N gelatinase expression was not inducible by TSP-1. TSP-1 induced a dose-dependent increase in MMP-9 expression by EXOS-P and EXOS-M cells (r = 0.999 [P = .0006] and r = 0.9740 [P = .026], respectively). TSP-1 also induced a dose-dependent increase in MMP-2 expression by EXOS-P and EXOS-M cells (r = 0.901 [P = .09] and r = 0.927 [P = .07], respectively) (Fig. 5A and B). TIMP-1 expression by EXOS-P and EXOS-M cells increased marginally in response to TSP-1 (30% and 17%, respectively; P < .05 for both) (Fig. 6A). TIMP-2 expression by EXOS-P cells decreased significantly in response to TSP-1, whereas TIMP-2 expression by EXOS-M cells remained largely unchanged (40% [P < 0.05] and 16% [P > 0.05], respectively) (Fig. 6B). The net effect of exogenous TSP-1 on EXOS cell TIMP expression trended toward increased expression by EXOS-M cells and little to no change in expression by EXOS-P cells. Therefore, TSP-1 promoted an overall increase in gelatinolytic activity in both the EXOS-P and EXOS-M cells when its effects on both TIMP and gelatinase expression were taken together.

Figure 5. Extraskeletal osteosarcoma (EXOS) cell gelatinase expression was found to increase in response to thrombospondin-1 (TSP-1). (A) Baseline expression of matrix metalloproteinase (MMP)-9 was found to be correlated with cell phenotype and was greatest for EXOS-M (lung metastasis) cells and negligible for EXOS-N (normal mesenchymal) cells. EXOS-P (primary osteosarcoma) cells demonstrated intermediate baseline MMP-9 expression. EXOS-P and EXOS-M cells demonstrated a dose-dependent increase in MMP-9 expression in response to TSP-1. (B) Baseline expression of MMP-2 was also found to be correlated with cell phenotype and was greatest for EXOS-M cells and negligible for EXOS-N cells. EXOS-P cells demonstrated intermediate baseline MMP-2 expression. EXOS-P and EXOS-M cells demonstrated a dose-dependent increase in MMP-2 expression in response to TSP-1.

Figure 6. The effect of thrombospondin-1 (TSP-1) on tissue inhibitors of metalloproteinases (TIMP) expression is shown. EXOS-P (primary osteosarcoma) and EXOS-M (lung metastasis) extraskeletal osteosarcoma cells were cultured for 48 hours in either serum-free Dulbecco modified Eagle medium (DME) containing 0.1% bovine serum albumin (control) or serum-free DME plus 20 μg/mL of TSP-1. TIMP expression was measured from whole cell protein extracts using enzyme-linked immunoadsorbent assays. (A) TSP-1 led to marginal increases in TIMP-1 expression by EXOS-P and EXOS-M cells (30% and 17%, respectively; P < .05 for both). (B) TIMP-2 expression by EXOS-P and EXOS-M cells decreased by 40% (P < .05) and 16% (P > .05), respectively, in response to TSP-1.

Using a modified Boyden chamber invasion assay, we tested the effects of TSP-1 on EXOS cell invasiveness. At baseline, EXOS-M cells showed the greatest invasive capacity, followed by EXOS-P cells. As expected, EXOS-N cells were not invasive at baseline and showed no response to exogenous TSP-1. Both EXOS-P and EXOS-M showed an approximately 3-fold increase in Matrigel invasion in response to TSP-1 (0.01 optical density [OD] vs 0.04 OD and 0.05 OD vs 0.15 OD; P < .05 for both) (Fig. 7).

Figure 7. Extraskeletal osteosarcoma (EXOS) cell invasiveness increases in response to thrombospondin-1 (TSP-1). Tumor cell invasive capacity was measured using a Matrigel-coated modified Boyden chamber assay. The lower chamber had either Dulbecco modified Eagle medium (DME) with 10% fetal calf serum (Serum, positive control), DME alone (Serum-free, negative control), or serum-free DME containing 20 μg/mL of TSP-1. EXOS-N (normal mesenchymal) cells were not found to be invasive at baseline or in response to TSP-1. EXOS-P (primary osteosarcoma) and EXOS-M (lung metastasis) cells demonstrated 3-fold increases in Matrigel invasion in response to TSP-1. OD 560nm indicates optical density at 560 nanometers.

Angiocidin-Inhibitory Peptides Block the Effects of TSP-1 on Tumor Cell Gelatinase Expression and Invasion

To show that angiocidin mediates this TSP-1-induced increase in gelatinase expression, we inhibited the binding of TSP-1 to angiocidin by using 2 peptides previously developed in our laboratory and shown to inhibit TSP-1–angiocidin binding. The CSVTCG peptide, homologous to the CSVTCG sequence in the type I domain of TSP-1, binds with high specificity to the TSP-1–binding domain on angiocidin and inhibits A549 human lung adenocarcinoma tumor cell metastasis.15 The 25-mer peptide containing a 20 amino acid sequence from angiocidin mediates an antitumor effect by competitively binding to TSP-1.10 We measured the inhibitory effects of our peptides on TSP-1–mediated gelatinase expression and sarcoma tumor cell invasion. The CSVTCG and 25-mer angiocidin-inhibitory peptides individually down-regulated the TSP-1–induced increase in gelatinase expression by 60% to 75% (P = .0001) (Fig. 8). Our angiocidin inhibitory peptides also reversed TSP-1–induced increases in tumor cell invasion by 50% each (P = .0004) (Fig. 9).

Figure 8. Angiocidin-inhibitory peptides block thrombospondin-1 (TSP-1)-mediated increases in gelatinase expression by EXOS-P (primary osteosarcoma) cells. Matrix metalloproteinase (MMP)-9 expression was measured using an enzyme-linked immunoadsorbent assay. EXOS-P cells (1 × 10⁶) were plated in 6-well microtiter plates and grown to 80% to 90% confluency in medium with 10% fetal bovine serum (FBS). The cells were weaned from the FBS-containing medium to serum-free medium over 48 hours. The cells were then incubated for 48 hours at 37°C in 1 mL of Dulbecco modified Eagle medium (DME) containing 0.1% bovine serum albumin (BSA) and either 20 μg/mL of exogenous TSP-1, 20 μg/mL of TSP-1 plus 40 μg/mL of CSVTCG peptide, or 20 μg/mL of TSP-1 plus 40 μg/mL of 25-mer peptide (TSP + 25-mer). Cells cultured in DME containing 0.1% BSA alone (serum-free) demonstrated baseline MMP-9 expression. DME containing 10% fetal bovine serum was used as a positive control, and the scramble peptide served as a negative protein control. MMP-9 overexpression in response to TSP-1 was significantly decreased by inhibiting angiocidin with the CSVTCG or the 25-mer peptide. Nonspecific protein control demonstrated no effect on TSP-1–mediated MMP-9 expression. scr indicates scrambled peptide.

Figure 9. Angiocidin-inhibitory peptides block thrombospondin-1 (TSP-1)-mediated increases in EXOS-P (primary osteosarcoma) cell invasion. The ability of EXOS-P cells to invade the extracellular matrix was measured using a modified Boyden chamber. Cells were suspended in the upper chamber in serum-free Dulbecco modified Eagle medium (DME). The lower chamber contained serum-free DME with 20 μg/mL of TSP-1. To measure the effect of our angiocidin inhibitory peptides in TSP-1–mediated sarcoma tumor cell invasion, 40 μg/mL of either CSVTCG peptide or 25-mer peptide (TSP + 25-mer) was added to the upper chamber containing tumor cells and TSP-1. DME containing 10% fetal bovine serum (FBS) in the lower chamber was used as a positive control (10% FBS, positive control) and serum-free media alone (serum-free) served as a baseline measure of EXOS-P cell invasiveness. The scramble peptide served as a negative protein control. Increased EXOS-P tumor cell invasion in response to TSP-1 was significantly decreased by inhibiting angiocidin with CSVTCG and 25-mer peptides. Nonspecific protein control demonstrated no effect on TSP-1–mediated sarcoma tumor cell invasion. OD indicates optical density; scr, scrambled peptide.

DISCUSSION

Sarcomas are a heterogeneous family of tumors of mesenchymal origin. Despite aggressive surgical therapy, >50% of sarcoma patients will die of their disease.24 Systemic therapy with cytotoxic agents has been largely ineffective despite considerable toxicity. This reflects our extremely poor understanding of the biology of sarcomas. Thus, better understanding of sarcoma biology and investigation of new paradigms and therapeutic approaches for the systemic treatment of these tumors are an urgent priority.

TSP-1 is an extracellular matrix protein produced by activated stromal cells. A growing body of literature supports a central role for TSP-1 in tumor progression in epithelial malignancies. Our laboratory and others have shown that TSP-1 up-regulates tumor-associated proteases and promotes tumor cell invasion and metastases in epithelial malignancies.9, 15-17, 25-34 We have recently shown that TSP-1 promotes tumor cell invasion in pancreatic and gastric cancer through a mechanism that involves up-regulation of gelatinases,16, 25, 29, 35 key enzymes in the metastatic cascade.36, 37 The potential role of TSP-1 in mesenchymal tumors such as sarcomas is presently unknown.

Despite this evidence supporting a role for TSP-1 in tumor progression, some investigators suggest that TSP-1 has an inhibitory activity in cancer.38 One possible explanation for these contradictory observations could be differences in the capacity of various tumor cell lines to bind TSP-1 rather than their capacity to synthesize TSP-1. It is well known that different cell lines express different TSP-1 receptors with various affinities for TSP-1, and that the cellular functions of TSP-1 are mediated by these specific cell surface receptors and/or binding proteins.39

Angiocidin, recently cloned and characterized as a tumor cell surface binding protein for TSP-1, binds the CSVTCG sequence within the type 1 domain of TSP-1 with high affinity and specificity.9, 15 We have shown that angiocidin is overexpressed in several epithelial malignancies. Higher levels of angiocidin expression in these malignancies correlate with adverse prognostic factors and worse clinical outcomes.9, 16, 32, 40, 41 More recently, we have shown that recombinant soluble angiocidin binds TSP-1, prevents TSP-1 interaction with tumor cells, and inhibits angiogenesis and tumor cell invasion in breast cancer and melanoma models of tumor progression.9 Furthermore, reduction of angiocidin expression in human umbilical vein endothelial cells via small interfering RNA silencing inhibits angiogenesis and gelatinase expression.8 The potential role of angiocidin in mesenchymal tumors such as sarcomas is presently unknown.

In our study, angiocidin, TSP-1, and the gelatinases were expressed in sarcoma tumors. Angiocidin colocalized with the gelatinases in sarcoma tumor cells and tumor-associated microvessels. TSP-1 was present primarily in the tumor-associated stroma. We found that angiocidin was overexpressed in high-grade tumors, a major predictor of outcome and tumor behavior in sarcoma patients. We also showed that angiocidin is differentially expressed by EXOS tumor cells. Normal mesenchymal cells (EXOS-N) did not express angiocidin. Although primary sarcoma cells (EXOS-P) expressed angiocidin, it was the highly metastatic EXOS-M cells that showed the greatest amount of angiocidin expression. Furthermore, gelatinase expression by our EXOS cells mirrored their angiocidin expression, with no expression by normal EXOS-N cells and maximal gelatinase expression by the metastatic EXOS-M cells. This suggests a critical role for angiocidin in the acquisition of a more metastatic phenotype. This finding also suggests that in tumors of mesenchymal cell origin such as sarcomas, angiocidin could potentially serve as a tumor-specific target for therapy.

Gelatinase activity depends on a balance between gelatinase expression and expression of the gelatinase inhibitors, TIMP-1 and TIMP-2.42 Normal EXOS-N cells exhibited negligible gelatinase activity. Gelatinase expression was higher and overall TIMP expression was lower in our metastatic EXOS-M cells when compared with the primary EXOS-P cells. Therefore, higher overall gelatinolytic activity correlated with a more aggressive sarcoma tumor cell phenotype and greater invasive capacity in vitro. Furthermore, higher gelatinase expression levels also correlated with more aggressive high-grade sarcoma tumor specimens. Overall gelatinolytic activity in our sarcoma cells increased in a dose-dependent fashion in response to TSP-1. Moreover, TSP-1 promoted sarcoma tumor cell invasion in direct proportion to this increase in gelatinolytic activity.

We have previously developed 2 angiocidin-inhibitory peptides that inhibit TSP-1 binding to angiocidin. The first of these peptides is a CSVTCG peptide. The amino acid sequence of this peptide is homologous to the angiocidin-binding site on the TSP-1 molecule.20 The recombinant CSVTCG peptide inhibited tumor cell invasion and decreased tumor metastasis in a mouse melanoma model.17 The second angiocidin inhibitory peptide we developed is a 25-mer peptide derived from a 20 amino acid segment of recombinant angiocidin. This particular domain of angiocidin contains the TSP-1 binding site; a deletion mutant missing this domain failed to bind matrix proteins and lost its antitumor activity.9, 10 The synthetic 25-mer peptide containing this sequence mimics the antitumor and antiangiogenic activity of recombinant soluble angiocidin.

In this study, we used our angiocidin-inhibitory peptides to help elucidate the role of angiocidin on TSP-1–mediated sarcoma tumor cell gelatinase activity and tumor cell invasion. In so doing, we highlight their potential therapeutic applications in sarcoma. Inhibition of TSP-1 binding to angiocidin by our angiocidin-inhibitory peptides induced a significant down-regulation of sarcoma tumor cell gelatinase expression. Furthermore, this effect resulted on a profound inhibition on sarcoma tumor cell invasion in response to our angiocidin-inhibitory peptides. These data, together with the high levels of angiocidin expression in sarcoma tumor specimens and sarcoma tumor cells, suggest a central role for angiocidin in sarcoma tumor cell invasion. Furthermore, the higher levels of angiocidin and gelatinases observed in high-grade tumors, as well as in the metastatic EXOS-M cells in vitro, suggest a central role for angiocidin in the acquisition of a more aggressive tumor phenotype.

In summary, the results of the current study provide strong evidence that suggests a central role for angiocidin in sarcoma progression. We have shown that angiocidin exerts its proinvasive effect through interactions with the matrix protein TSP-1, leading to an up-regulation of gelatinase activity and a promotion of sarcoma tumor cell invasion. The current study data also suggest that CSVTCG and 25-mer angiocidin-inhibitory peptides may have a role as potential targeted therapies against sarcomas. Further research into angiocidin-targeted peptide therapy is warranted.

Conflict of Interest Disclosures

The authors made no disclosures.