Differential expression of VEGF-A and angiopoietins in cartilage tumors and regulation by interleukin-1β

Authors


Abstract

BACKGROUND

Vascular endothelial growth factor (VEGF)-A and angiopoietin (Ang)-1 and Ang-2 are key factors in angiogenic signaling. In this study the expression of these factors was identified in cartilage tumors. As interleukin (IL)-1β has been found to be an indispensable factor in angiogenic signaling, we further analyzed the effect of IL-1β on the expression of VEGF-A, Ang-1, and Ang-2 using a previously established cell culture model.

METHODS

Surgical specimens of enchondromas, conventional chondrosarcomas, and dedifferentiated chondrosarcomas were obtained from 72 patients. VEGF-A, Ang-1, and Ang-2 mRNA expression was detected by conventional and quantitative reverse transcription polymerase chain reaction (PCR). VEGF-A expression was also detected by immunohistochemistry or Western blot.

RESULTS

Differential expression of VEGF-A, Ang-1, and Ang-2 was clearly demonstrated in cartilage tumors. VEGF-A expression was positively correlated with the tumor type. Higher VEGF-A expression levels were detected in conventional chondrosarcomas Grades II and III (using a 3-tier grading system) than in dedifferentiated chondrosarcomas (P < .05). A typical pattern of VEGF-A isoforms was identified, including VEGF121, VEGF145, VEGF165, and VEGF189. Ang-1 presented as a low-level transcript with slightly elevated levels in chondrosarcomas (P < .05). Highly variable Ang-2 expression levels were detected in solitary cases of conventional chondrosarcomas. IL-1β regulated VEGF-A and Ang-1 expressions in a dose-dependent manner. Whereas low IL-1β concentrations increased VEGF-A and Ang-1 transcription, high IL-1β concentrations had the opposite effect. IL-1β did not activate Ang-2 expression.

CONCLUSIONS

Angiogenic signaling in cartilage tumors is variable and at least partly regulable by IL-1β. The findings are of therapeutic relevance, either as a desired effect or a side effect in medical treatment. Cancer 2006. © 2006 American Cancer Society.

Proliferating cells depend on an adequate supply of oxygen and nutrients. Due to diffusion limits in three-dimensional cell formations, vascularization is a prerequisite for the growth of neoplastic tissue.1 The activation of angiogenesis has been attributed to an imbalance between positive and negative regulatory molecules, such as the vascular endothelial growth factors.2 Subsequent events, including local matrix degradation, endothelial cell migration, proliferation, capillary formation, and maturation critically depend on a coordinated action of further molecules.3 Angiopoietins have been thought to contribute essentially to endothelial cell functions during these processes.4

Vascular endothelial growth factor A (VEGF-A) is the most important mediator of angiogenesis and is overexpressed by a multitude of solid human tumors.5 VEGF-A increases microvascular permeability and promotes survival, migration, and proliferation of endothelial cells through interaction with transmembrane tyrosine kinase receptors VEGFR-1 and VEGFR-2.2 Due to alternative mRNA splicing, 6 isoforms of VEGF-A (VEGF121-206) have been identified, consisting of 121, 145, 165, 183, 189, or 206 amino acids.6 The VEGF-A isoforms have different abilities to bind to cell surface and extracellular heparin-sulfate proteoglycans.5 Whereas VEGF121 is weakly acidic and freely diffusible, VEGF189 and VEGF206 are basic and completely bound to the cell surface and the extracellular matrix.7 VEGF145 and VEGF165 have intermediate properties.8 Although VEGF-A isoforms share similar biological activities, their availability is therefore variable.

Angiopoietins act through the endothelial cell-specific Tie2 receptor tyrosine kinase.9 Angiopoietin (Ang)-1 is an agonist of the Tie2 receptor and promotes angiogenic remodeling as well as vessel maturation and stabilization.4 Ang-2 is an antagonist of Ang-1 and Tie-2 and may result in vessel regression or continued remodeling, depending on the presence of VEGF-A.4 The expression of Ang-1 and Ang-2 has been identified in several human tumors, with Ang-2 being more frequently up-regulated than Ang-1.9 The mechanisms by which Ang-1 and Ang-2 are regulated have not been fully elucidated yet.10

Recently, interleukin-1β (IL-1β) has been identified as an indispensable factor in angiogenesis of different tumor cells.11 Endogenous IL-1β is abundantly secreted by activated blood monocytes and macrophages and exhibits a broad range of functions.12 The proangiogenic effects of IL-1β have been attributed to a regulation of VEGF-A expression, but other angiogenic mediator molecules may be regulated by IL-1β.11 The effects of IL-1β on angiogenic signaling should be investigated further.

Cartilage tumors comprise a heterogeneous group of lesions with distinct morphological and biological properties. Unlike adult joint cartilage tissue, which is avascular, cartilage tumors possess blood vessels. In a previous study, a positive correlation was found between the grade of malignancy and the microvessel density in cartilage tumors.13 However, the role of angiogenic molecules in cartilage tumors has not been extensively studied yet. In the present study, we intended to fill these gaps with expression analyses of VEGF-A isoforms, Ang-1, and Ang-2 in cartilage tumors, including previously unchecked dedifferentiated chondrosarcomas. Furthermore, we used a cell culture model to study the effect of IL-1β stimulation on the regulation of these factors in cartilage tumor cells. The refinement of expression analyses on angiogenic factors in cartilage tumors, as well as functional investigations on the regulation of these factors, may contribute to the understanding of angiogenic processes in these tumors. As the inhibition of angiogenesis is characteristic of normal cartilage tissue, we believe that angiogenesis is a significant factor supporting the growth and progression of these tumors. The regulation of angiogenic processes may open up further therapeutic opportunities.

MATERIALS AND METHODS

Patient Samples

Surgical specimens of cartilage tumors were obtained from 72 patients corresponding to enchondromas, conventional chondrosarcomas (using a 3-tier grading system), and dedifferentiated chondrosarcomas (Table 1). Formalin-fixed and paraffin-embedded material was available from 69 cases. In 18 cases, tumor specimens were snap-frozen immediately after resection using liquid nitrogen and stored at −80°C. Additionally, frozen specimens of healthy adult joint cartilage were available from 3 other patients. Informed consent to the investigations was obtained from the patients.

Table 1. Patient Characteristics
 Cases (n = 72)Mean AgeMaleFemaleLocalization*Mean Tumor Size, cm3
Long BonesSmall BonesOthers
  • n.d.: not detected.

  • *

    Long bones, including samples from humerus, femur, tibia, and fibula; small bones, including samples from hands and feet; others include samples from skull base, vertebra, scapula, ribs, and pelvis.

  • [number of cryopreserved samples]

  • Approximated [number of informative cases]

Enchondroma17 [1]31.710761016.4 [10]
Chondrosarcoma        
 Grade I18 [5]51.3996012284 [7]
 Grade II23 [8]58.812111319563 [10]
 Grade III2 [0]57.502101n.d.
Dedifferentiated chondrosarcoma12 [4]66.5481002346 [2]
 Malignant fibrous histiocytoma8 [2]6744602422 [1]
 Osteosarcoma4 [2]65.504400270 [1]

Cell Culture

In all, 2.5 × 105 cells of the human chondrosarcoma cell line C384214 were seeded in 25-cm2 culture flasks and stimulated with recombinant human IL-1β (Tebu-bio, Offenbach, Germany) for 72 hours using different concentrations (0.1, 1, 10 ng/ml). A negative control without IL-1β stimulation was included.

RNA Isolation and Purification

Total RNA was extracted from 200-300 mg of fresh-frozen tumor tissue and adult joint cartilage and cultured cells using guanidinium thiocyanate-acid phenol-chloroform according to the protocol of Chomczynski and Sacchi.15 Before RNA extraction, cartilage and tumor tissues were pulverized using a mortar and a pestle exposed to liquid nitrogen. After extraction, RNA probes were purified using a NucleoSpin RNA II kit (Macherey & Nagel, Duren, Germany), including a DNase I digestion.

cDNA Synthesis

Reverse transcription was performed in five separate 20-μl reactions for each sample using BioScript Moloney Murine Leukemia Virus Reverse Transcriptase RNase H Minus (Bioline, Luckenwalde, Germany), Oligo(dT)15 Primer, and Random Primer (Promega, Mannheim, Germany) according to the manufacturer's recommendations. After cDNA synthesis, the reactions were mixed together.

Conventional Polymerase Chain Reaction (PCR)

Amplification was done in 12.5-μl reactions using Biotaq DNA Polymerase (Bioline) according to the manufacturer's recommendations. The following primers were used for the detection of GAPDH: 5′-ATCATCCCTGCCTCTACTGG-3′ and 5′-CCCTCCGACGCCTGCTTCAC-3′ (product size: 187 basepairs [bp]); VEGF-A: 5′-TGCCTTGCTGCTCTACCTCC-3′ and 5′-TCACCGCCTCGGCTTGTCAC-3′ (410 bp for VEGF121, 481 bp for VEGF145, 542 bp for VEGF165, 595 bp for VEGF183, 613 bp for VEGF189, and 664 bp for VEGF206); Ang-1: 5′-CCACAACCTTGTCAATCTTT-3′ and 5′-GTAAGATCAGGCTGCTCTGT-3′ (488 bp); Ang-2: 5′-GGAACACTCCCTCTCGACAA-3′ and 5′-CCTCACGTCGCTGAATAATT-3′ (506 bp). The PCR conditions consisted of 35 cycles at 95°C for 30 seconds, 52°C (Ang-2), 56°C (GAPDH), or 58°C (VEGF-A, Ang-1) for 40 seconds, and 72°C for 1 minute. PCR products cloned into pCRII (Invitrogen, Karlsruhe, Germany) served as a positive control. PCR products were observed by agarose gel electrophoresis. Cloned VEGF-A PCR products were directly sequenced using an ABI PRISM 310 sequencer (Perkin-Elmer, Weiterstadt, Germany).

Quantitative PCR

Real-time quantitative PCR was performed using a LightCycler and a LightCycler FastStart DNA MasterPLUS SYBR Green I kit (Roche, Mannheim, Germany). The primers used for quantitative PCR of GAPDH, Ang-1, and Ang-2 were the same as for conventional PCR. Primers used for quantitative PCR of VEGF-A were: 5′-GGGCAGAATCATCACGAAGT-3′ and 5′-TGGTGATGTTGGACTCCTCA-3′ (product size: 211 bp). The PCR conditions consisted of 45 cycles at 95°C for 10 seconds, 52°C (Ang-2), 56°C (GAPDH), 58°C (Ang-1), or 60°C (VEGF-A) for 20 seconds, and 72°C for 8 seconds (GAPDH), 9 seconds (VEGF-A), or 20 seconds (Ang-1, Ang-2). Calculated amounts of GAPDH-, VEGF-A-, Ang-1-, and Ang-2-PCR products cloned into pCRII (Invitrogen) were used for quantification. Negative controls without templates were included. Ratios of VEGF-A/GAPDH, Ang-1/GAPDH, and Ang-2/GAPDH transcript amounts were calculated from mean values of repeated reactions.

Western Blot

Cell culture supernatants were concentrated 10-fold using Vivaspin 500 columns (molecular weight cutoff: 30 kDa; Vivascience, Hannover, Germany), reduced by adding 5× Laemmli buffer containing 20% DTT and boiled. Then 25 μL of the samples were separated by SDS-PAGE and blotted onto nitrocellulose membranes. The membranes were blocked with 3% dry milk in TBS/Tween and incubated for 2 hours at room temperature (RT) with polyclonal rabbit antibodies against VEGF-A (Santa Cruz, Heidelberg, Germany), diluted 1:200. A peroxidase-conjugated antirabbit antibody, diluted 1:25,000, was used for a second incubation (30 minutes). West Pico Supersignal substrate (Perbio, Bonn, Germany) was left on the membrane until distinct bands had developed. A MagicMark standard (Invitrogen) was used to identify the molecular weights. The enhanced choriluminescence membrane images were quantified using the GeneGnome and GeneTools image scanning and analysis package (Syngene, Cambridge, UK).

Immunohistochemistry

Sections of formalin-fixed and paraffin-embedded tumor tissue (4.0 μm thick) were mounted on SuperFrost plus glass slides (Menzel, Braunschweig, Germany), deparaffinized, and rehydrated. The slides were exposed to rabbit antibodies against VEGF-A (Santa Cruz) diluted 1:1000, von Willebrand factor (Dakocytomation, Hamburg, Germany) diluted 1:400, and arbitrarily CD68 (Dakocytomation) diluted 1:400, in a buffer containing 10% fetal bovine serum. Before incubation for 60 minutes at 37°C, slides for anti-VEGF-A reactions were exposed to 10 mM citrate buffer (pH 6.0) in a microwave oven. Biotinylated antirabbit immunoglobulins (Dakocytomation) were used for a second incubation (30 minutes at RT). Anti-VEGF-A reactions were detected by a Vectastain ABC alkaline phosphatase kit (Axxora, Grunberg, Germany) and Fast Red substrate (Dakocytomation). A NexES system (Ventana, Illkirch, France) and an iVIEW DAB detection kit (Ventana) were used for observation of anti-von Willebrand factor and anti-CD68 reactions. The slides were counterstained with hematoxylin. Sections of human placenta were used as positive controls in each reaction series.

Evaluation of Staining Results

VEGF-A expression was classified according to the following grading system: staining extensity was categorized as 0 (no positive cells), 1 (≤ 25% positive cells), 2 (> 25% and ≤ 50% positive cells) or 3 (> 50% positive cells), and staining intensity was categorized as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). The individual categories were added to give a total immunoreactive score (IRS).

Microvessel density (MVD) was assessed using von Willebrand factor-stained slides. The number of highlighted vessels was counted in whole areas of 0.25 cm2 using a microscopic grid. Tumor margins and surrounding tissue were excluded from evaluation. Specimens containing few or fragmented tumor tissue with an area of less than 1 cm2 were not evaluated. Mean values from repeated counts were recorded for each case.

Statistical Analysis

Data from quantitative PCR and immunohistochemistry were analyzed by Student t-test, Kruskal-Wallis test, Mann-Whitney U-test, and χ2 test. A P-value of <.05 was considered statistically significant. All analyses were performed using SPSS 10.0 software (SPSS, Chicago, IL).

RESULTS

Expression of VEGF-A, Ang-1, and Ang-2 mRNA in Cartilage Tumors

At least small amounts of VEGF-A transcripts were detected in 17 of 21 samples obtained from adult joint cartilage and cartilage tumors (Figs. 1, 2A). Using conventional PCR, 4 isoforms of VEGF-A were identified, including VEGF121, VEGF145, VEGF165, and VEGF189, as approved by direct sequencing of cloned PCR-products. VEGF183 and VEGF206 were detectable neither in adult joint cartilage nor in cartilage tumors. VEGF121 and VEGF165 presented the most abundant PCR products. Quantitative PCR revealed the highest levels of total VEGF-A transcript amounts in conventional chondrosarcomas Grade II with a three times higher mean value than in chondrosarcomas Grade I (.332 vs. .106). The lowest VEGF-A transcript amounts were found in dedifferentiated chondrosarcomas (.028). Using Student t-test, VEGF-A expression levels in chondrosarcomas Grade II higher than in other cases proved to be statistically significant (P = .006).

Figure 1.

Gel electrophoresis of GAPDH, VEGF-A, Ang-1, and Ang-2 RT-PCRs. Four isoforms of VEGF-A were detected, including VEGF121, VEGF145, VEGF165, and VEGF189. VEGF121 and VEGF165 were the most abundant PCR products. VEGF183 and VEGF206 were not identified. Ang-1 PCR was mainly positive in conventional and dedifferentiated chondrosarcomas. Ang-2 PCR was positive in 4 cases of chondrosarcomas Grades I and II, which were simultaneously positive for VEGF-A and Ang-1.

Figure 2.

Results of quantitative RT-PCRs for (A) VEGF-A, (B) Ang-1, and (C) Ang-2 in cartilage tumors. Open circles indicate ratios of VEGF-A, Ang-1, or Ang-2/GAPDH transcript amounts of individual samples. Bars indicate mean values for each tumor type.

Ang-1 transcripts were detected in 17 of 21 samples, including adult joint cartilage, conventional, and dedifferentiated chondrosarcomas (Figs. 1, 2B). Quantitative PCR revealed no difference in mean Ang-1 transcript amounts in chondrosarcoma subtypes (.0025 each), whereas mean Ang-1 quantity was about 4 times lower in adult joint cartilage (.0006). No Ang-1 expression was detectable in enchondroma. VEGF-A expression in conventional and dedifferentiated chondrosarcomas higher than in adult joint cartilage and enchondroma turned out to be statistically significant (P = .006).

Ang-2 PCR was positive in 4 of 21 samples, including 1 chondrosarcoma Grade I and 3 chondrosarcomas Grade II (Figs. 1, 2C). Ang-2 quantity was highly variable in the 4 samples. All Ang-2-positive samples were simultaneously positive for Ang-1 and VEGF-A.

VEGF-A Immunohistochemistry and MVD in Cartilage Tumors

Positive immunostaining for VEGF-A was observed mainly in the cytoplasm of cartilage tumor cells (Fig. 3). The staining pattern varied in the different tumor types. In enchondromas and chondrosarcomas Grade I, staining was absent or limited to small foci (Fig. 3A,C), whereas it was more extensive and strong in chondrosarcomas Grades II and III (Fig. 3D,H). In dedifferentiated chondrosarcomas, staining was diffusely weak or virtually absent in malignant fibrous histiocytoma-like areas (Fig. 3I) or was slightly positive in and around cartilaginous areas (Fig. 3K). Focal positivity was found in osteosarcomatous areas around osteoid formations (Fig. 3L). Occasionally, VEGF-A immunostaining was also observed in fibrous septa between the tumor nodules of enchondromas and conventional chondrosarcomas (Fig. 3E). The positive cells were identified as pericytes of blood vessels and macrophages, as confirmed by anti-CD68 immunohistochemistry (Fig. 3F). Blood vessels were mainly localized in fibrous septa of enchondromas and conventional chondrosarcomas (Fig. 3B,G). In single cases, intracartilage tumor vessels were also identified (Fig. 3G). The vessel arrangement in enchondromas and conventional chondrosarcomas proved to be highly variable. In dedifferentiated chondrosarcomas, vessels were almost evenly distributed in the sarcomatous area (Fig. 3J), whereas cartilaginous areas appeared to be avascular. The MVD was more than 5 times higher in high-grade malignant areas of dedifferentiated chondrosarcomas than in enchondromas and conventional chondrosarcomas. The results of VEGF-A immunohistochemistry and MVD evaluation are summarized in Table 2. Using Kruskal-Wallis test for the analysis of VEGF-A immunoreactivity scores in the different tumor types, the results proved to be statistically significant (P < .0001). Using Mann-Whitney U-test, microvessel densities in dedifferentiated chondrosarcomas higher than in enchondromas and conventional chondrosarcomas also proved to be statistically significant (P < .0001). We could not find any correlation between VEGF-A-IRS or MVD and clinicopathologic data, except for VEGF-IRS and age (Table 3).

Figure 3.

Immunohistochemical results. (A) Focal positivity in tumor cells of an enchondroma, anti-VEGF-A. (B) Typical distribution of microvessels in the fibrous septa of an enchondroma, including calcifications, anti-von Willebrand factor. (C) Focal weak positivity of tumor cells in a chondrosarcoma Grade I, anti-VEGF-A. (D) Marked positivity of tumor cells in a chondrosarcoma Grade II, anti-VEGF-A. (E) Positivity of tumor cells, pericytes of blood vessels and infiltrating cells in the fibrous septa of another chondrosarcoma Grade II, anti-VEGF-A. (F) Identification of infiltrating cells as macrophages in the same case, anti-CD68. (G) Detection of intracartilage vessels in the same case, anti-von Willebrand factor. (H) Strong immunostaining of tumor cells in a chondrosarcoma Grade III, anti-VEGF-A. (I) Negligible staining in a malignant fibrous histiocytoma-like area of a dedifferentiated chondrosarcoma, anti-VEGF-A. (J) High MVD with almost evenly distributed vessels in the same case, anti-von Willebrand factor. (K) Focal weak immunostaining in the cartilaginous area and the surrounding sarcomatous area of the same case, anti-VEGF-A. (L) Focal positivity around osteoid formations in osteosarcomatous area of another dedifferentiated chondrosarcoma, anti-VEGF-A. Original magnification ×100 (B–D, G, I–K); ×200 (A, E, F, H, L).

Table 2. VEGF-A Immunoreactivity and Microvessel Density
 VEGF-A ImmunoreactivityMicrovessel Density*
Cases n = 69Positive Cases (%)Mean IRSP<.0001Cases n = 28Mean Vessels/mm2
  • n.d.: not detected.

  • *

    P<.0001 for MVD (microvessel density) of dedifferentiated chondrosarcomas versus MVD of other groups.

  • P<.0001

  • Mean vessel/mm2 ± standard deviation.

  • §

    Mean IRS (immunoreactive score) of highly malignant dedifferentiated component (mean IRS of low malignant cartilaginous component).

Enchondroma177 (41.2)1.1851.99 ± 1.21
Chondrosarcoma     
 Grade I1714 (82.4)2.41101.56 ± 0.79
 Grade II2322 (95.7)4.0472.09 ± 0.92
 Grade III22 (100)60n.d.
Dedifferentiated chondrosarcoma     
 Malignant fibrous histiocytoma76 (85.7)2.43 (1.57)§510.7 ± 2.06
 Osteosarcoma33 (100)3.33 (2.33)§110.1
Table 3. χ2 Analysis of the Relation between VEGF-A Immunoreactivity, Microvessel Density, and Clinicopathologic Parameters
 VEGF-A ImmunoreactivityMicrovessel Density
Low IRS (0–3)High IRS (4–6)PLow MVD (≤ 2.00 vs/mm2)High MVD (>2.00 vs/mm2)P
  1. vs: vessels; MVD: microvessel density; IRS: immunoreactive score.

Age      
 Lower age, ≤ 50 y265.0294.151
 Higher age, > 50 y1820 69 
Sex      
 Male269.8287.978
 Female1816 76 
Localization      
 Long bones1917.0868.401
 Small bones91 22 
 Others167 73 
Tumor size      
 ≤ 250 cm3146.471.119
 > 250 cm344 34 

Expression of VEGF-A, Ang-1, and Ang-2 in Chondrosarcoma Cells and Regulation by IL-1β

VEGF-A transcripts were also detected in cultured chondrosarcoma cells. Using conventional PCR, 4 VEGF-A isoforms were identified, including VEGF121, VEGF145, VEGF165, and VEGF189 (Fig. 4A). IL-1β stimulation did not alter the VEGF-A isoform composition. Stimulation with low IL-1β concentrations (0.1 and 1 ng/ml) increased VEGF-A mRNA expression, whereas high concentrations (10 ng/ml) had an opposite effect (Fig. 4B). Using Western blot analysis, 4 VEGF-A species with molecular weights of ∼11, 14, 19, and 22 kDa were detected in the culture medium (Fig. 4C). IL-1β stimulation increased total VEGF-A amounts up to 3.4-fold. High IL-1β concentrations changed the relative amounts of VEGF-A species. A shift to the 11 kDa band was observed.

Figure 4.

(A) Gel electrophoresis of GAPDH, VEGF-A, Ang-1, and Ang-2 RT-PCRs of chondrosarcoma cell cultures treated with IL-1β (0, 0.1, 1, and 10 ng/ml). (B) Results of quantitative RT-PCRs for VEGF-A. (C) Western blot analyses of VEGF-A expression in chondrosarcoma cell cultures treated with IL-1β. Four different VEGF-A species were detected, corresponding to ∼11, 14, 19, and 22 kDa molecular weight. Total amounts of VEGF-A protein (all bands) analyzed by GeneTools Software. (D) Results of quantitative RT-PCRs for Ang-1 and Ang-2. Open circles and squares with error bars indicate ratios of VEGF-A, Ang-1, or Ang-2/GAPDH transcript amounts (arbitrary units).

Low IL-1β concentrations (0.1 and 1 ng/ml) also increased Ang-1 mRNA expression, whereas high concentrations (10 ng/ml) down-regulated Ang-1 (Fig. 4D). In contrast to this, Ang-2 mRNA was detectable neither in stimulated nor in unstimulated chondrosarcoma cells (Fig. 4D).

DISCUSSION

In our present study, we clearly demonstrate differential expression of VEGF-A and angiopoietins in cartilage tumors. In addition to previously reported studies using immunohistochemistry,13, 16 we identified distinct VEGF-A isoforms and determined VEGF-A transcript amounts with conventional and quantitative RT-PCR. We detected VEGF-A expression not only in malignant cartilage tumors, but also in enchondromas and in healthy adult joint cartilage. The latter has also been reported by other investigators.17 Remarkably, high VEGF-A expression has been described in hypertrophic chondrocytes of the epiphyseal growth plate,18 suggesting a regulation of VEGF-A under certain microenvironmental conditions. Several cytokines and growth factors have been found to influence VEGF-A expression, among them IL-1β.7 A source of IL-1β can be macrophages, which we found scattered across the fibrous septa and the surrounding tissue of enchondromas and chondrosarcomas. Using a previously established cell culture model,14 we demonstrated the regulation of VEGF-A mRNA and protein by IL-1β in human chondrosarcoma cells. These results indicate that macrophages may not only support tumor angiogenesis by VEGF-A expression themselves,19 but also by stimulating VEGF-A expression in cartilage tumor cells through IL-1β secretion. Using high IL-1β concentrations, we observed a decline in VEGF-A transcription, suggesting a negative feedback mechanism.

We found a typical pattern of VEGF-A isoforms in adult joint cartilage and in cartilage tumors composed of VEGF121, VEGF145, VEGF165, and VEGF189. Although the simultaneous production of different VEGF-A isoforms has been reported in several cell types, variable isoform compositions have been found, predominantly consisting of VEGF121, VEGF165, and VEGF189, e.g., in nonsmall cell lung carcinomas or thyroid neoplasms.20, 21 The occurrence of VEGF145 has been identified in cultured cells originating from the female reproductive system.8 Recently, a rare VEGF145 expression was reported in a subset of breast and ovarian tumors.22 In contrast to this, we identified VEGF145 as a typical isoform in adult joint cartilage and cartilage tumors of both sexes as well as in cultured chondrosarcoma cells. The latter has also been reported in the chondrosarcoma cell line OUMS-27.23 However, other investigators failed to detect VEGF145 in normal and osteoarthritic cartilage17 or cartilage tumors.24 In agreement with other studies dealing with this topic, we did not detect VEGF183 and VEGF206 isoforms either in adult joint cartilage or in cartilage tumors. These results suggest a special fine-tuning of VEGF-A isoforms in cartilage tumor cells in vivo and in vitro. However, little is known about the need for alternative splicing of VEGF-A transcripts. As the translated VEGF-A isoforms have different biological properties regarding their binding abilities to cell surface and extracellular matrix proteins,5 their simultaneous production may ensure a differential release of bound VEGF-A species under certain conditions such as extracellular matrix remodeling during angiogenesis. Furthermore, different binding affinities of VEGF-A isoforms to their receptors may modulate VEGF-A signaling. Previously, a lower binding affinity to VEGFR-1 was described for VEGF121 than for VEGF165.25 Although VEGFR-2 has been identified as the major mediator of endothelial cell proliferation, migration, and survival, the biological functions of VEGFR-1 are still unclear. As VEGFR-1 was supposed to be a negative regulator of VEGFR-2 function,6 the production of VEGF-A isoforms with low affinity to VEGFR-1 may enforce VEGFR-2 effects. Recently, progesterone was identified to modify VEGF-A isoform composition by selectively increasing VEGF189 in human uterus.26 Other factors that influence alternative splicing of VEGF-A transcripts need to be investigated. As IL-1β has been identified as an indispensable factor in angiogenesis,11 we presumed modulatory effects on alternative VEGF-A splicing. However, we did not detect any change in the VEGF-A isoform composition in chondrosarcoma cells. Using Western blot analysis, we detected soluble VEGF-A species with a molecular weight of ∼11-22 kDa in the culture medium, at least partially indicating a proteolytic cleavage of native VEGF-A proteins. The latter may be attributed to the expression of several proteases such as urokinase plasminogen activator (uPA) in the chondrosarcoma cell line C3842 used in this study.14 It still needs to be clarified whether the cleaved VEGF-A protein species have particular biological functions in cartilage tumor cells. In previous reports, the proteolytic processing of VEGF189 by uPA was found to be a prerequisite for its mitogenic effect by enabling binding to VEGFR-2,27 whereas the cleavage of VEGF189 and VEGF165 by plasmin decreased the mitogenic activity and the VEGFR-2 binding ability of these isoforms.26, 27 As the overexpression of uPA has been significantly correlated with higher local recurrence rates and distant metastases in conventional chondrosarcomas,28 our findings suggest that the maturation of VEGF189 by uPA cleavage, which enables angiogenic signaling through VEGFR-2, may contribute to the progression of these tumors by supporting vascularization and blood supply.

Using quantitative RT-PCR, we detected the highest VEGF-A transcript amounts in conventional chondrosarcomas Grade II. The results of VEGF-A immunohistochemistry were equivalent, suggesting VEGF-A expression increasing with the grade of malignancy. Similar results have been reported for other cancers such as ovarian carcinomas or renal cell carcinomas,29, 30 but also for chondrosarcomas.13, 16 However, we did not detect significant differences in MVD of enchondromas and conventional chondrosarcomas. Moreover, we found a significantly lower VEGF-A expression and a significantly higher MVD in high-grade malignant areas of dedifferentiated chondrosarcomas. Due to the lack of cartilage production in the high-grade malignant component of dedifferentiated chondrosarcomas, the results corresponded to previous reports on inhibitory effects of cartilage matrix constituents on angiogenesis,31 the more so as the low-grade malignant cartilaginous areas proved to be avascular. These results suggest that in cartilage-free tumors, comparatively lower VEGF-A levels are required to cause an imbalance of pro- and antiangiogenic factors, which is essential for angiogenesis.1 Such imbalances may also be the cause of the occurrence of intracartilage vessels in solitary cases of cartilage tumors, which have also been reported by other investigators.13 However, intracartilage vessels proved to be a focal phenomenon, susceptible to sampling errors, and without clear-cut evidence of a correlation to the VEGF-A immunoreactivity in the adjacent tissue. These results indicate heterogeneity in angiogenic signaling and capillary formation in cartilage tumors. As vascularization and blood supply are prerequisites for tumor growth,1 foci of intracartilage vessel formation may be a starting point of progressive disease in conventional chondrosarcomas.

Besides VEGF-A, angiopoietins have been found to be critical mediators of angiogenesis.10 Using quantitative RT-PCR, we identified Ang-1 expression at comparatively low levels in conventional and dedifferentiated chondrosarcomas, independent of the grade of malignancy. However, Ang-1 levels were significantly higher in chondrosarcomas than in adult joint cartilage and enchondroma. Stronger expression of Ang-1 has also been described in other cancers, such as gliomas, nonsmall cell lung carcinomas, or hepatocellular carcinomas, whereas unchanged or not significantly increased Ang-1 expression levels have been reported in renal cell carcinomas, gastric carcinomas, Kaposi sarcomas, or angiosarcomas.9 The effect of Ang-1 has been attributed to maturation and stabilization of vessels, but unrealized wider roles for Ang-1 have been suggested.10 This notion can be supported by our findings that Ang-1 is also expressed in avascular healthy adult joint cartilage.

In contrast to Ang-1, Ang-2 expression has been found to be more frequently up-regulated in several tumors.9 However, we detected Ang-2 expression only in four samples of conventional chondrosarcomas, which were simultaneously positive for Ang-1 and VEGF-A. The amount of Ang-2 mRNA in these cases was highly variable, suggesting an irregular spatial distribution of Ang-2 expression, as described for breast cancers or hepatocellular carcinomas.9 Our results are in line with the known biological functions of Ang-2, which have been attributed to local destabilization of vessels at sites of regression and sprouting, antagonizing the stabilizing effect of Ang-1.10 Lack of Ang-2 expression in avascular adult joint cartilage also supports this notion. Whether Ang-2 protein expression may be associated with the occurrence of intracartilage vessels in cartilage tumors needs to be elucidated. However, due to the lack of suitable antisera for immunohistochemistry, analyses of the spatial distribution of angiopoietins are generally rare.9

The mechanisms by which Ang-1 and Ang-2 expressions are regulated are not clear. Using quantitative RT-PCR, we demonstrated a dose-dependent regulation of Ang-1 transcription by IL-1β in chondrosarcoma cells similar to VEGF-A regulation. Whereas lower IL-1β concentrations increased Ang-1 expression, high concentrations led to a decrease in Ang-1 expression. The latter has also been described in human endothelial cells.32 In contrast to Ang-1, we did not detect Ang-2 expression in C3842 chondrosarcoma cells, either stimulated by IL-1β or not. However, these results fit with the absence of Ang-2 expression in the majority of cartilage tumors. All the results suggest that Ang-2 expression in chondrosarcoma cells is independent of VEGF-A expression. In contrast to this, an up-regulation of Ang-2 by VEGF-A has been described in endothelial cells.33 The effect of other factors that may potentially influence Ang-2 expression in cartilage tumors such as the hypoxia inducible factor-1α, the von Hippel-Lindau gene, or the tumor necrosis factor-α10 needs to be investigated.

The effects of IL-1β on the regulation of VEGF-A and Ang-1 in cartilage tumor cells may be of therapeutic relevance. The administration of recombinant IL-1β may support angiogenesis in cartilage tumors. This may enhance the effect of cytostatic drugs or other active agents that are applied intravenously. However, the induction of endogenous IL-1β may occur as a side effect in medical treatment. To prevent angiogenic signaling, the administration of an IL1 receptor antagonist may be considered. Further investigations are needed to analyze these effects in vivo. Future studies should also determine the role of macrophages in cartilage tumors, which may be involved in angiogenic signaling, either by producing angiogenic factors such as VEGF-A themselves or by regulating the production of angiogenic factors in tumor cells through cytokines such as IL-1β.

Ancillary