Matricellular proteins osteopontin and osteonectin/SPARC in pancreatic carcinoma

Pancreatic ductal adenocarcinoma (PDA) remains a devastating disease. At the time of diagnosis, a majority of tumors display local invasion or distant metastases, and patients have a 5% overall 5-year survival rate.¹ Although the molecular mechanisms contributing to the aggressive nature of PDA are not thoroughly characterized, it is well recognized that a dense desmoplastic stroma is one hallmark of this disease. Desmoplasia is a tumor-induced host response characterized by proliferation of stromal fibroblasts and production of fibrotic tissue with altered extracellular matrix properties. Induction of this desmoplastic response is one way in which tumor cells coerce normal cells into creating a microenvironment that can favor invasion, metastasis and therapeutic resistance.² The tumor microenvironment in PDA includes thick bundles of collagen fibrils, stromal fibroblasts, inflammatory cells and proliferating endothelial cells. It is possible that this dense connective tissue matrix hampers penetration of chemotherapeutic agents, contributing to treatment failure.³

There is mounting evidence that matricellular proteins play an important function in modulating cancer cell behavior and the tumor microenvironment.^4,5 Matricellular proteins are noncollagenous extracellular matrix proteins that function as modulators of cell behavior as well as regulators of matrix organization.⁶ The matricellular proteins osteopontin [bone sialoprotein I (BSP-1), early T-lymphocyte activation (ETA-1) or secreted phosphoprotein 1 (SPP1)] and osteonectin [secreted protein acidic and rich in cysteine (SPARC) or BM-40] have been reported to be upregulated in PDA lesions, and several groups are trying to understand their role in this disease.

In this issue of Cancer Biology & Therapy, Zhivkova-Galunska and coworkers examine the relative expression of osteopontin and osteonectin transcripts in a panel of 14 human pancreatic cancer cell lines, and determine whether there is any correlation between these RNA levels and characteristics associated with aggressive cancer phenotype.⁷ Briefly, the authors find a correlation between higher levels of osteopontin mRNA and the ability of the cell lines to form tumors in the liver of nude rats. In a human pancreatic adenocarcinoma cell line, knock down of osteopontin mRNA, via transient transfection of a specific antisense oligonucleotide, suppressed cell growth. Since previous studies showed that treatment of pancreatic cancer cell lines with exogenous osteopontin did not affect proliferation, these data suggest that endogenous levels of osteopontin in this cell line are necessary and sufficient for the support of tumor cell growth.^7,8 Since osteopontin can also increase the invasiveness of pancreatic carcinoma cells, these data reinforce the concept that osteopontin may be a target for therapeutic intervention.⁸ The authors also demonstrate that knock down of osteonectin mRNA, using an antisense oligonucleotide, resulted in increased cell growth, whereas growth was suppressed by exogenously added osteonectin. These results are in concert with previously published data, suggesting that osteonectin plays a tumor suppressor role in PDA cells.^9,10

To consider these findings in the framework of our current knowledge of osteonectin and osteopontin in PDA, it may be important to examine the sources of these matricellular proteins in PDA lesions. Numerous studies report the immunolocalization of osteopontin in the matrix and in association with PDA cells in human tissue samples.^8,11,12 However, in situ hybridization studies suggest that macrophages intimately associated with the tumor cells are the ones responsible for synthesizing osteopontin in this microenvironment.^13,14 It is suggested that the majority of osteopontin immunostaining of the adenoma cells may be due to association of osteopontin with the cell surface, through interaction with its receptor integrins or CD44. Although there is some controversy over whether PDA cells in situ synthesize osteopontin, expression of this matricellular protein by pancreatic adenocarcinoma cell lines in vitro is well documented and these cell lines provide a means for examining the effects of osteopontin on cell behavior, either through autocrine or paracrine mechanisms.

Understanding of the role of osteopontin in PDA is complicated by the existence of alternatively spliced variants. The variant lacking exon 4 is referred to as osteopontin-c, and the variant lacking exon 5 is osteopontin-b. In breast tissue, the osteopontin-c splice variant is expressed exclusively in cancer cells, and not normal tissue. Although full length osteopontin forms aggregates in the presence of physiological levels of calcium, osteopontin-c does not, and this osteopontin isoform induces anchorage independent growth.¹⁵ These data suggest that full length osteopontin may promote cell adhesion, whereas the smaller, soluble osteopontin-c may promote invasion. Smoking is a risk factor for the development of PDA, and it was recently demonstrated that nicotine increases the expression of osteopontin-c in PDA cells. Osteopontin-c was found in 87% of invasive PDA lesions, 73% of which came from smokers.¹² Thus, in PDA, it is likely that expression of osteopontin-c supports invasion and metastasis.

In reference to the studies performed by Zhivkova-Galunska and co-workers, it would be of interest to characterize the relative expression levels of osteopontin and osteopontin-c in the pancreatic adenocarcinoma cell lines, and determine whether there may be a correlation with their ability to grow in liver of nude rats. Although overexpression of full length osteopontin was not able to promote cell growth, it is possible that a different effect on growth and cell survival may be obtained with overexpression of osteopontin-c. Further, it is intriguing to consider the mechanisms driving the alternative splicing of osteopontin transcripts in PDA.

In regard to osteonectin and its the cellular source in PDA lesions, in situ hybridization data are not available. However, cytoplasmic staining for osteonectin is strongest in fibroblastic cells immediately adjacent to infiltrating cancer cells, and expression of osteonectin by pancreatic cancer cells themselves is less common.¹⁶ In cancer, DNA methylation is a major mechanism for silencing tumor-suppressor genes, and aberrant methylation of the osteonectin gene has been reported in a number of human pancreatic adenocarcinoma cell lines and xenograft models.⁹ It is thought that aberrant methylation of genes, including osteonectin, may be related to the progression of ductal adenocarcinoma, since the degree of aberrant methylation increases with increasing histological grade of neoplasia.¹⁷

The expression of osteonectin by fibroblastic cells in the stromal compartment in PDA is strongly associated with poor patient outcome.¹⁶ Some attribute this to the concept that osteonectin expression by stromal fibroblasts is an indicator of an activated fibroblast phenotype. However, osteonectin expressed by the stromal cells may contribute to PDA pathology. In vitro studies suggest that, although osteonectin inhibits the growth of PDA cells, it also increases invasiveness.^10,18 Indeed, osteonectin has been characterized as a “counter-adhesive” protein for some cell types, with the ability to influence integrin clustering and activation.⁵

In addition, osteonectin is important for collagen deposition and fibrilogenesis.¹⁹ Expression of osteonectin in the stroma likely facilitates the development of the dense collagenous stroma associated with PDA lesions. There is a drastic decrease in fibrous tumor capsule when murine pancreatic adenocarcinoma cells are orthotopically injected into osteonectin-null mice. With this diminished mechanical restraint, tumors in the osteonectin-null mice had faster growth, and increased permeability and perfusion.²⁰

Although expression of osteonectin in the desmoplastic stroma appears to promote a metastatic phenotype and is associated with poor prognosis, there are preliminary data suggesting that osteonectin in the tumor microenvironment may be useful for targeting chemotherapy. Recently, response of head and neck cancers to a nanoparticle albumin-bound paclitaxel (nab-paclitaxel) was shown to correlate with osteonectin expression in the tumor micoenvironment.²¹ Osteonectin is known to bind albumin, and interaction of osteonectin with nabpaclitaxel may concentrate the drug in the tumor vicinity and increase efficacy.²²

Overall, there appears to be some consensus on the effects of osteopontin and osteonectin in pancreatic adenocarcinoma (Table 1). Osteopontin and its receptors are considered targets for therapeutics aimed at limiting invasion and metastasis. In contrast, the role of osteonectin in PDA is quite context dependent, with the potential to act as a tumor suppressor, or a pro-invasive factor, or as a factor that may hamper or enhance efficacy of chemotherapy. The next step will be to understand the underlying mechanisms, which will allow for targeting components of the tumor stroma, to enhance treatment of PDA.

Table 1.

Effects of matricellular proteins osteopontin and osteonectin in pancreatic ductal adenocarcinoma (PDA)

Matricellular protein	Source	Regulation	Effect in PDA	Relevance
Osteopontin/osteopontin-c	Tumor associated macrophages	Paracrine regulator	Pro-metastatic Promotes PDA growth	Target for therapeutic intervention
	PDA cells	Autocrine regulator Induced by nicotine Alternative splicing
Osteonectin	PDA cells	Autocrine regulator Epigenetic silencing (aberrant methylation)	Inhibits PDA growth	Possible tumor suppressor
	Fibroblastic cells in tumor stroma	Paracrine regulator	Pro-metastatic	Associated with poor prognosis
		Contributes to formation of dense desmoplastic stroma	Pro-fibrotic	Dense stroma may hamper penetration of chemotherapeutic agents
			Interacts with albumin	Promote accumulation of albumin-paclitaxel nanoparticles

Commentary to: Zhivkova-Galunska M, et al. Osteopontin but not osteonectin favors the metastatic growth of pancreatic cancer cell lines. Cancer Biol Ther. 2010 This issue.