Glioblastoma is a deadly cancer that is largely unresponsive to treatment, due in part to high inter- and intratumor heterogeneity. A variety of immune cells, including microglia and peripheral macrophages/monocytes, infiltrate the tumor and the tumor microenvironment, and have been shown in some cases to promote tumor growth. These cells are a potential target for therapy, but the interactions between these immune cells and the tumor cells are poorly understood. One molecule thought to play a role in these interactions is osteopontin (OPN), which is encoded by the gene SPP1. Osteopontin is an extracellular matrix protein that has been implicated in the pathogenesis of glioblastoma as well as other cancers. In addition to being present in the microenvironment of the tumor, some glioblastoma cells produce OPN, and these cells have more stem cell-like characteristics, more invasive tendencies, and are often radiation resistant. In a recent paper published in Neuro Oncology from Eric Holland’s lab in Human Biology, in collaboration with colleagues from the University of Washington, the Max Delbrueck Center for Molecular Medicine in Berlin, and Emory University, the authors show that glioblastoma-associated microglia/monocytes (GAMs) are the main source of OPN in glioblastoma and examine the role of microenvironment-derived OPN on glioblastoma progression.
Searching The Cancer Genome Atlas (TCGA) database for markers co-expressed with OPN, the authors found that several markers for microglia and macrophages/monocytes were positively correlated with OPN levels. By contrast, OPN expression was not correlated with tumor cell markers. When the authors labeled OPN in 9 human glioblastoma tissue sections, OPN was observed in all of them. OPN mainly co-localized with Iba1, a marker of microglia cells. 62% of OPN was localized to microglia cells. However, there was also OPN staining in Iba1-negative cells in some tumors. In tissue samples from normal human brain samples, OPN staining was much lower. The authors also characterized OPN abundance in normal mouse brains and in a mouse glioblastoma model from implanted GL261 cells. OPN abundance was highest in the tumor core, but the tumor border regions also had detectable OPN staining.
After characterizing intra-tumor OPN expression, the authors went on to evaluate the role of microenvironment-derived OPN on tumor growth. Mice lacking OPN and wild-type mice were intracranially injected with glioblastoma-causing GL261 cells, which express low levels of OPN. Mice lacking the OPN gene had faster disease progression and tumor growth compared with wild type mice. While the number of proliferating tumor cells measured by Ki67 was similar between wild type and OPN-negative mice, there were a much higher percentage of apoptotic cells in the tumors of wild type animals, as measured by TUNEL. The authors also noticed a larger number of visible hemorrhages in tumors in OPN-negative mice compared to wild type mice. When they measured the diameter of tumor blood vessels, they found that OPN-negative mice had lower numbers of blood vessels per square millimeter, reduced numbers of small vessels, and higher numbers of large vessels. When characterizing OPN abundance in glioblastoma tissue, the authors noticed that some pericytes, contractile cells lining blood vessels, expressed OPN. Ninety percent of tumor vessels in tumors from wild type mice were lined by pericytes, while only 50% of tumor vessels in tumors from OPN-negative mice were lined with pericytes. The lower abundance of pericytes in the vessels could be linked to reduced small vessels and increased larger vessels in OPN-negative mice.
The brains of OPN-negative and wild type mice were examined for immune cell infiltration. In the brains before tumor cells were injected, there was no difference in brain-resident microglia between mouse genotypes. After tumors began to form, however, there was a higher abundance of brain-resident microglia in OPN-negative mouse brains than in wild type. The amount of periphery-derived macrophages/monocytes remained the same between genotypes, nor was there a difference in the number of T cells between genotypes. When the authors looked at cytokine levels to determine activation status of the microglia in these tumors, they found no apparent differences in activation between OPN-negative and wild type mice.
The authors conclude that in normal brain tissue, microglia express OPN only at very low levels, and expression is up-regulated once glioblastoma is established. Lack of OPN in the host microenvironment, however, leads to faster disease progression in mice. The previously-reported link between OPN expression in glioblastoma tumor cells and increased tumor progression is puzzling when opposed with the reported finding here that low microenvironment OPN levels also speed tumor progression. One explanation the authors propose is that different cell lines may be to blame. The GL261 cells do not express OPN or require it for growth. Therefore, when implanted in a mouse model lacking OPN, they are still able to grow rapidly. By contrast, in a model where glioblastoma cells are dependent on OPN signaling, the results may be different. Several studies in other cancer types have also reported faster disease progression in OPN-negative mouse models, including studies in hepatocellular carcinoma and squamous cell carcinoma. However, another study in hepatocellular carcinoma and one in melanoma found a decrease in disease burden in OPN-negative mice. Therefore, the effect of OPN on tumor progression appears to differ in tumors and tissue types.
The increased levels of microglia in the OPN-negative mice with glioblastoma tumors might in part explain the more rapid disease progression. Although the authors found no significant differences in cytokine levels between OPN-negative and wild type mice, further in-depth analysis will be necessary to fully characterize the activation status of these immune cells. Furthermore, the difference in tumor vasculature observed between tumors in OPN-negative mice and wild-type mice requires further exploration. Pericytes lining the blood vessels may be affected by OPN produced in the GAMs, leading to fewer micro vessels and more large vessels.
Frank Szulzewsky, Nina Schwendinger, Dilansu Güneykaya, Patrick J Cimino, Dolores Hambardzumyan, Michael Synowitz, Eric C Holland, Helmut Kettenmann. Loss of host-derived osteopontin creates a glioblastoma-promoting microenvironment, Neuro-Oncology, November 2017, 19:6(252).
Funding for this work was provided by the Deutsche Forschungsge meinschaft, NeuroCure, the National Institutes of Health, and the Berlin Institute of Health.