Morphologic characterization of osteosarcoma growth on the chick chorioallantoic membrane

Osteosarcoma is the most common primary malignant bone tumour that usually affects children and adolescents. After the introduction of cytotoxic polychemotherapy to the treatment of osteosarcoma tremendous advances were achieved, increasing the survival rate from under 20% to over 60%[1]. But over the past 20 years no substantial improvements were made. There is still a high failure rate especially when metastases are present[2]. It is therefore of critical importance to identify new targets and new molecules with efficient antitumour activity. New targets might be identified by large-scale microarray analysis or array-based genomic hybridization[3] to identify different gene expressions in patient samples or suitable in vivo animal models. It is expected that anti-angiogenic agents will have clinical benefits for patients with osteosarcoma, probably by reducing metastasis[4]. Thus especially models that allow precise monitoring of tumour angiogenesis might be promising in finding new targets to treat osteosarcoma[5].

An in vivo assay to serve these purposes is the chick chorio-allantoic membrane (CAM) assay. The CAM is formed on embryonic day 4 - 5 by fusion of the chorion with the allantoic vesicle[6]. The resulting membrane consists of two mesodermal layers: the somatic mesoderm of the chorion and the splanchnic mesoderm of the allantois. In this structure, an extremely dense vascular network develops with bigger vessels located within the mesoderm and the capillaries situated directly under or within a very thin ectodermal layer. The physiological role of the CAM is to serve as a respiratory organ until hatching, to store waste products and to absorb calcium from the shell for bone development[7]. On the molecular level critical genes which participate in the vascularisation of complex human tissues such as placenta and the lung are significantly regulated during the CAM vascular development[8]. The CAM assay is characterized by several major advantages such as the ease of access, the extensive vascularisation and the relatively simple experimental approach. This opens up the possibility to screen many samples in an inexpensive way[9].

Since its use to demonstrate normal embryonic blood vessel development more than 100 years ago, the CAM served as a host for transplantations of embryonic tissue as well as for bacteria and viruses. During the last 20 years the CAM is extensively used in angiogenesis research but a reliable bone tumour model has not yet been reported.

Here we present the results of the screening of the human osteosarcoma cell lines HOS, MG63, MNNG-HOS, OST, SAOS, SJSA1, U2OS and ZK58 for their use in the CAM assay and provide evidence that the MNNG-HOS cell line reproducibly simulates key features of human osteosarcoma growth.

Experimental in vivo tumour models are essential for comprehending the dynamic process of human cancer progression, identifying therapeutic targets, and evaluating antitumour drugs. Besides the promising efforts made in the research on osteosarcoma using in vitro assays[11] there is a lack of potent and reproducible in vivo models to evaluate new therapies in a pre-clinical setting.

The chick chorioallantoic membrane assay is a commonly used method for studying angiogenic or anti-angiogenic activities in vivo[12]. Although first tumour transplantations to the CAM were described more than 100 years ago[13] CAM tumour models are still rarely developed and characterized, compared to murine models[14]. Taking into account the easy access, relatively simple experimental approach and the natural immonodeficient environment of the developing embryo (for review see[15]) its rare use for tumour grafting is surprising. Recently, a reliable experimental glioblastoma CAM model using the human U87 cell line has been established[10]. The experimental glioma model simulates key features of human glioma growth in a few days. Treatment with the tyrosine kinase inhibitors Gleevec (imatinib) or PTK787/ZK 222584, siRNA-mediated knock down of Vascular Endothelial Growth Factor (VEGF) and/or IL6[16, 17] potently inhibited tumour angiogenesis and growth in this model.

Another group inoculated single cell suspensions of LNCaP, PC-3 and Tsu-Pr1 human prostatic cancer cell lines on the chorio-allantoic membrane. They demonstrated a reliable tumour growth allowing the evaluation of proliferation and apoptosis induction after intravascular or topic application of anticancer drugs[18].

There are very few publications about the use of the CAM assay in human bone or soft tissue sarcoma[19, 20]. The literature does not provide any reports that specifically characterise the tumour forming potential of different human osteosarcoma cell lines. Therefore, we established a protocol allowing monitoring of the growth of several osteosarcoma lines on the CAM. The MNNG-HOS, U2OS and SAOS cell lines consistently developed vascularized tumours after 4 days, which then progressively consolidated throughout the growth process. All tumours were derived from established human osteosarcoma cell lines [21–24] and were more or less able to form a solid tumour, but only the MNNG-HOS, SAOS and U2OS reached the consistency as well as average tumour size (Figure 1G) to allow further investigations. The MNNG-HOS cells clearly provided the most reliable and consistent results. The reasons for these differences remain speculative. The difference of up to two million cells between CAM's from different experiments might partly explain the size variances, but it is rather unlikely as this would affect all cell lines. It might be due to the fact that each osteosarcoma cell line produces a differently composed extracellular matrix[21] and thus might have a different potential to degrade the CAM tissue. Possibly only the more aggressive cell lines are able to invade the CAM and induce angiogenesis, which is essential for tumour growth. The MNNG-HOS cell line is derived from HOS cells by treatment with the carcinogenic nitrosamine MNNG[25], and was shown to have a high degree of similarity but also some characteristic differences from the maternal cell line[22] that did not grow well in our model. The fact that a significantly higher mortality rate was seen after grafting of SAOS and U2OS cells (both reliably formed tumours) may be due to tumour cell dissemination in the embryo or secretion of factors which cause blood coagulation. Secretion of such molecules by cancer cells is favoured under metabolic stress such as hypoxia[26], a condition which might well occur during tumour cell implantation on the CAM. Tsuchiya et al.[20] examined 24 human tumour cell lines for their potential to form liver metastases after injecting them into a larger CAM vessel. They have shown that, among others, the MNNG-HOS cells proliferated in the liver. The metastatic potential of osteosarcoma cell lines (143B cells and to a lesser extent MNNG-HOS cells) has also been shown in an orthotopic mouse model[27] as well as in chick embryos after intravascular injection in CAM veins[28]. Thus, this model might also serve to study the development of distant metastases. Pulmonary metastases are very common in osteosarcoma and are the typical cause of death in human[1]. Also speculative at this point, they might also be responsible for the significantly higher death rate in the U2OS and SAOS treated embryos in our study.

The MNNG-HOS cells might be most suitable for further inhibition studies in our model, since this line reliably simulated key features of human osteosarcoma growth such as angiogenesis, necrosis and haemorrhage. Osteoid formation however, did not occur in this model. This might be due to the lack of mature bone in the developing chick embryo or due to the short time span of only 7 days of tumour growth. The experimental osteosarcoma described here are hybrid tumours: they develop out of human cells and secondarily get invaded by newly formed chicken capillaries and stromal cells, therefore allowing further studies aiming at the molecular characterization of osteosarcoma angiogenesis. It might for example be possible to separately study gene expression in the newly formed vessels using Affymetrix chicken GeneChips, in parallel with human GeneChips, which will measure gene expression in the tumour cells. This could lead to the development of new antitumour drugs. The CAM model also is an attractive tool to follow the fate and visualize microscopically the behaviour of the grafted cells[14]. In vivo microscopy is possible at every time point during the growth process, thus allowing direct observations of tumour development and its reactions to intravascular or topical application of anticancer drugs. Additionally, intravital videomicroscopy of the chorioallantoic microcirculation enables studies focussing on metastasis formation. MacDonald et al. used epifluorescence to identify labelled B16F1 melanoma cells, and studied successive stages of metastasis formation in the CAM in vivo[29]. This might facilitate the screening of new therapeutic approaches[30] such as molecules targeting the RANK/RANKL/OPG axis[31].

A limit of this model is the short time span of only 7 days for tumour growth which might not have been long enough for the other cell lines to grow to bigger tumours. However, especially this time frame might be of importance for the development of metastases or tumour recurrence. These first steps are heavily dependent on invasion and angiogenesis, which is a promising target for new drugs against solid tumours, including osteosarcoma[32, 33].

Our model could serve as a preclinical screening assay for anticancer molecules against osteosarcoma and therefore increase the speed and efficacy of the development of new drugs. More, the model is an excellent alternative to rodent models and fits perfectly into current recommendations of an ethically appreciable use of live animals in cancer research[34].

Last, the experimental tumour model on the CAM might be adapted for other malignant bone and soft-tissue tumours such as chondrosarcoma, Ewing's-sarcoma, malignant fibrous histiocytosis, and benign but locally aggressive lesions such as giant cell tumour of bone.