Aberrant heparan sulfate proteoglycan localization, despite normal exostosin, in central chondrosarcoma

Abstract

The tumor suppressor genes EXT1 and EXT2 are involved in the formation of multiple osteochondromas, which can progress to become secondary peripheral chondrosarcomas. The most common chondrosarcoma subtype is primary central chondrosarcoma, which occurs in the medullar cavity of bone. The EXT1/EXT2 protein complex is involved in heparan sulfate proteoglycan (HSPG) biosynthesis, which is important for signal transduction of Indian hedgehog (IHH), WNT, and transforming growth factor (TGF)-beta. The role of EXT and its downstream targets in central chondrosarcomas is currently unknown. EXT1 and EXT2 were therefore evaluated in central chondrosarcomas at both the DNA and mRNA levels. Immunohistochemistry was used to assess HSPG (CD44v3 and SDC2), WNT (beta-catenin), and TGF-beta (PAI-1 and phosphorylated Smad2) signaling, whereas IHH signaling was studied both by quantitative polymerase chain reaction and in vitro. mRNA levels of both EXT1 and EXT2 were normal in central chondrosarcomas; genomic alterations were absent in these regions and in 30 other HSPG-related genes. Although HSPGs were aberrantly located (CD44v3 in the Golgi and SDC2 in cytoplasm and nucleus), this was not caused by mutation. WNT signaling negatively correlated with increasing histological grade, whereas TGF-beta positively correlated with increasing histological grade. IHH signaling was active, and inhibition decreased cell viability in one of six cell lines. Our data suggest that, despite normal EXT in central chondrosarcomas, HSPGs and HSPG-dependent signaling are affected in both central and peripheral chondrosarcomas.

Figure 1

A: EXT1 mRNA expression levels of central chondrosarcoma are comparable with growth plate levels whereas peripheral tumors (published previously16) show a threefold decrease. B: EXT2 mRNA levels are slightly higher in central and slightly lower in peripheral chondrosarcoma, compared with growth plate. Expression levels of the tumors are log2 transformed and expressed as relative to the growth plate levels. C: 8q tiling array comparative genomic hybridization pattern of central chondrosarcoma L803 is shown. Genomic alterations are absent, despite low EXT1 expression by qPCR. L1689 showed similar results as L803 (data not shown). For comparison, an osteochondroma with homozygous loss of multiple clones covering the EXT1 gene and hemizygous loss of a larger part of 8q is shown in D (figure composed of data published previously16).

Figure 2

A: Central chondrosarcoma showing cytoplasmic dot-like accumulation of CD44v3 at immunohistochemistry, suggestive for Golgi retention. B: This was confirmed by IF using a Golgi-specific marker, 58K protein, in green, and CD44v3 in red, resulting in a yellow color when co-localization occurs. The white line indicates the position where the staining profile (inset) has been taken. C: Central chondrosarcoma showing nuclear staining of SDC2, which was found in 50% of chondrosarcomas in addition to cytoplasmic staining. D: Nuclear localization of SDC2 was verified by immunoblotting in two chondrosarcomas. In the isolated nuclear fraction heterochromatin protein 1γ was present, verifying the procedure for separating the nuclear from the cytoplasmic staining. Original magnifications, ×40.

Figure 3

A and B: The activity of canonical WNT signaling was increased in grade I chondrosarcomas compared with enchondromas. However on further increase in histological grade, the activity decreased again (Pearson χ², P = 0.038). C and D: PAI-1 expression, indicating active TGF-β signaling, was found in the majority of central chondrosarcomas. PAI-1 expression was correlated to histological grade (χ² test, P = 0.002). Original magnifications, ×40.

Figure 4

mRNA levels of GLI1 (A), GLI2 (B), and PTCH (C) were measured as a readout of IHH signaling. Levels of all three genes in chondrosarcomas were in the same range of growth plate samples and large variations within the groups were observed. No relation to histological grade was found. Expression levels are log10 transformed.

Figure 5

A: In vitro the GLI1 mRNA expression decreases to almost half of the level of the dimethyl sulfoxide (DMSO) control in PANC1 and in CH2879 on 24-hour treatment with 10 μmol/L of cyclopamine (CP), although this was not found in the other five chondrosarcoma cell cultures. B: None of the chondrosarcoma cell cultures showed a negative effect on PTCH mRNA expression on 24-hour 10 μmol/L CP treatment. Both experiments were performed in duplicate. C: Both PANC1 and CH2879 demonstrated a decrease in cell viability measured by WST-1 assay, on 72 hours of 5 μmol/L and 10 μmol/L CP treatment. An increase in viability was observed in L784, whereas the other four chondrosarcoma cell cultures did not show an effect of cyclopamine treatment. The black bars represent the inactive compound tomatidine (TM), which demonstrate the selective effect of CP. Data represent three individual experiments performed in quadruplicate. Student’s t-test for paired data, *P < 0.05.

Figure 6

Overview of the EXT/HSPG pathway involved in both peripheral and central chondrosarcomas. Despite the absence of EXT alterations in central chondrosarcomas there is cytoplasmic retention of HSPGs, which is similar to peripheral chondrosarcomas and nuclear SDC2 localization that was specific for central chondrosarcomas. In addition, whereas IHH signaling decreased with grade in peripheral chondrosarcomas, it is still active in high-grade central chondrosarcomas. Active canonical WNT signaling might be important for the transition from benign to malignant central cartilaginous tumor type, however it is not crucial for tumor progression, whereas TGF-β is. Our results therefore suggest that disturbed HSPG function is involved in the histogenesis of both central and peripheral chondrosarcoma, although different steps in HSPG biosynthesis seem affected. (Figure based on Hameetman et al.22).