Periostin as a potential mediator of invasion and metastasis in Human Neuroblastoma

Antonio A. Giaimo and Dr. Robert Ross

Introduction

Human neuroblastoma is a cancer which arises from neuroepithelial cells of the embryonic neural crest. It is the most common extracranial solid tumor in children, often appearing in the adrenal, neck, chest, or spinal cord. Stage I or II tumors are generally localized, benign and well-differentiated and can be successfully treated by surgical resection alone. Stage III and Stage IV patient tumors are locally invasive and metastatic, respectively, with poor clinical prognoses. Metastases normally occur in lymph nodes, liver, bone marrow and bones. Little is known about the process of this metastasis. Favorable features include stromal rich tumors that are evidence of intermediate cell differentiation (Sugiura et al., 1998). Cellular heterogeneity is a prominent feature of this cancer. Three cell types have been identified in culture. The most tumorigenic is the intermediate stem cell-like (I-type) phenotype, which can be induced in vitro to differentiate into neuroblastic cells (N-type) and stromal-like cells (S-type). The N-type cells retain malignant potential and are poorly substrate adherent. S-type cells are not malignant and adhere well to substrate (Ross et al., 2004).

The secreted protein periostin (PostN) has been shown to be a mediator in cell adhesion and migration in normal cell processes but is implicated in several cancers as well (Tai et al., 2005). PostN levels are significantly increased in pancreatic cancer patients compared to control. The protein promotes tumor cell invasiveness as well as increased cell resiliency to hypoxic conditions (Baril et al., 2007). The protein is also overexpressed in 80% of human colon cancers and enhances metastatic growth (Bao., et al 2004).

We first identified PostN as possibly significant to neuroblastoma in our evaluation of histone modifications in I-type cells. A promoter tiling array using I-type cell DNA isolated by chromatin immunoprecipitation (ChIP) revealed equal amounts of histone trimethylation on lysine 27 (K27) and lysine 4 (K4) in the promoter region of PostN. High levels of K27 trimethylation are correlated with heterochromatin or non-transcribed genes while high levels of K4 trimethylation indicate an actively transcribed gene. Of significance here, embryonic stem cells contain bivalent chromatin domains with both histone marks that are primed for expression upon cell differentiation (Bernstein, 2006).

RT-PCR studies showed higher expression of PostN in S-cells and mostly absent in N and I cells. PostN mRNA levels are higher in tumor samples from stage IV patients as compared with stage I so expression levels are correlated with tumor progression and therefore prognosis (Sasaki et al., 2002). We hypothesize that PostN plays a role in the complex mechanisms leading to tumor metastasis and invasion. To this end, we have induced PostN expression in I-type lines with both BUdR and TGFβ-1 treatment. Migration and invasion assays have demonstrated that cells expressing PostN can more effectively breakdown and migrate through an extracellular matrix coated membrane. This effect may be mediated by matrix metalloproteinase 9 (MMP-9), a periostin-inducible protein that can digest the extracellular matrix enabling an otherwise stationary cell to migrate (Suguira et al., 1998). This paracrine signaling may induce MMP-9 in I and N-type cells as well as increase their ability to migrate and form metastatic colonies elsewhere. Our data also indicate a positive relationship between PostN expression and oxidative stress response. The protein may also increase resistance to hypoxia-induced cell death, a necessary adaptation for cells attempting to metastasize. We propose a role for the stromal cells in the tumor microenvironment as mediators of malignant cell invasion by secretion of periostin.

Materials and Methods

Tissue Culture, cell lines and treatment. The neuroblastoma cell lines used in this study include: (I-type lines) CB-JMN, SK-N-HM, SK-N-JD, BE(2)-C, SK-N-ER, NBL-S, SK-N-LP; (N-type lines) SH-SY5Y, BE(2)-M17V, LAI-55N, IMR32, SMS-LHN, SK-N-CH, KCN-83N, SMS-SAN; (S-type lines) SH-EP1, LAI-5s, SMS-KCNs. Cells were maintained according to previously described methods (Lazarova et al., 1998). Culture media contained a 1:1 mixture of Eagle’s Minimum Medium with non-essential amino acids and Ham’s Nutrient Mixture F12 (Life Technologies, Gaithers-burg MD) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) and no antibiotics (Thomas, 2003). Differentiation was induced by two weeks exposure to BUdR (10-5 M) or all-trans retinoic acid (10-5 M). The TGFβ-1 was reconstituted in sterile 4mM HCl containing 1 mg/ml BSA at a final concentration of 1 ug/ml and used to treat cells for 10 days (5 ng/ml).

Isolation of mRNA. Cells were collected in logarithmic phase, washed with PBS, spun down and frozen in liquid nitrogen. All mRNAs were isolated with mirVana miRNA Isolation Kit (Ambion, Inc) according to the manufacturer’s recommended protocol.

RT-PCR. Reverse transcription-polymerase chain reaction was completed with 0.5-1.0ul of cDNA in a 25ul reaction. cDNA’s were synthesized in a two step process with Superscript reverse transcriptase III according to the manufacturer’s protocol (Invitrogen). GoTaq Green Master Mix (Promega Corp) was used to amplify the desired product according to the manufacturer’s protocol. Cycling conditions were as follows: 94ْ C for 5 minutes, and 25-40 cycles at 95ْ C for 30 seconds, 57۫ C for 30 seconds, and 72۫ C for 1 minute. The final extension period was 7 minutes at 72۫ C before the products were run on a 1% agarose gel and visualized under Ultraviolet light. The primers were used are as follows:

Periostinforward 5’- CACCACAACGCAGCGCTATT-3’

reverse 5’- TGGAAGTTTCTCAAAAGCCT- 3’

MMP-2forward 5’- GGCTGGTCAGTGGCTTGGGGTA- 3’

reverse 5’-AGATCTTCTTCTTCAAGGACCGGTT- 3’

MMP-9forward 5’- TGTACCGCTATGGTTACAC- 3’

reverse 5’- CCGCGACACCAAACTGGA- 3’

GAPDHforward 5’-GTGAAGGTCGGAGTCAACGGATTTGG-3’

reverse 5’- ATGCCAGTGAGCTTCCCGTTCAGCT- 3’

Migration and Invasion Assays. In vitro cell migration and invasivity was measured using BioCoat Matrigel Invasion Chambers (Becton Dickinson). Nine cell lines were evaluated for their ability to move through a control uncoated filter membrane to determine their migration ability. Additionally, they were evaluated on their ability to move through a Matrigel Matrix coating similar to extracellular matrix to determine their invasion ability. The assays were performed as previously described (Boukerche et al., 2000). Briefly, 24-well invasion chambers were used to study the correlation of periostin expression on invasion. 0.5ml of 1% FBS media with 5x104 cells/ml was used in each assay and added to the upper face of the filter. 0.75ml of 10% FBS media was used in the well as a chemoattractant. After 22 hours the non-invading cells were removed from the upper portion of the membrane by scrubbing and the cells on the lower portion fixed for 5 minutes in 100% methanol. The membrane was then stained with 1% Toluidine blue in 1% Borax for 2 minutes, rinsed in distilled water and allowed to dry. Migrating cells were counted by light microscopy and counts represent the average of three representative fields done in duplicate. The percentage invasion was calculated by the number of cells invading through the Matrigel membrane divided by the number of cells migrating through the control membrane for each cell line. The invasion index is a measure of the percentage invasion of a treated test cell as compared to an untreated control cell.

CONCLUSIONS AND SIGNIFICANCE

Identifying the complex mechanisms of tumor invasion and metastasis are important because metastases, rather than primary tumors, are responsible for most cancer deaths. It is critical to identify proteins important in this process.

Expression of the secreted protein periostin:

  • Is present in S-type stromal cells and correlates with their ability to adhere well to substrate.
  • Can be induced by treatment with BUdR (S-like phenotypic change) and TGFβ-1 (no phenotypic change).
  • Correlates with a high invasion efficiency as compared to periostin null lines.
  • By inducement, in I-type cells, with BudR and TGFβ-1, increases the cell’s ability to digest and invade a Matrigel coated membrane.
  • Is linked to a cell’s ability to resist hypoxia-induced cell death.

These observations suggest that stromal cells may mediate malignant cell invasion in the neuroblastoma tumor microenvironment by their secretion of periostin. Further studies are warranted both into the mechanism by which this effect occurs and the prospect of periostin as a therapeutic target.

References

Baril, P., Gangeswaran, R., Mahon, P., Caulee, K., Kocher, H., Harada, T., Zhu, M., Kalthoff, H., Crnogorac, T., Lemoine, N. (2007) Oncogene. 26, 2083-2094

Bao, S., Gaoliang, O. (2004) Cancer Cell. Vol. 5, 329-341

Bernstein, B., Mikkelson, T. (2006) Cell. 125, 315-326

Ross, R., Walton, J., Kattan, D., Thomas, S., Spengler, B., Guo, H., Biedler, J., Cheung, N. (2004) Neoplasia. Vol. 6, 838-845

Sasaki, H., Sato, Y., Yamakawa, Y. (2002) Journal of Pediatric Surgery. Vol. 37, 1293-1297

Sugiura, Y., Shimada, H., Seeger, R., Laug, W., DeClerck, Y. (1998) Cancer Research. 58, 2209-2216

Tai, I., Meiru, D., Chen, L. (1995) Carcinogenesis. Vol. 26. 908-915

Acknowledgements

Great thanks to Dr. Ross and Barbara Spengler for both the chance to work with them and their guidance. This literally would not have been possible without assistance from Dan Han, Zi Yan, and Leleesha Samaraweera. I would also like to thank Dr. Melissa Henriksen, with whom I began my research, for her guidance and the opportunity to work with her.