Journal of Cancer Stem Cell Research
Isocitrate dehydrogenase 1 mutant (IDH1mut) gliomas have global genomic hypermethylation, are less aggressive than
IDH1 wild-type (IDH1wt) gliomas, and generally grow poorly in vitro and in vivo. Yet little data exist that connect specific
hypermethylation targets to this unique phenotype. We previously reported that the gene encoding Tissue Factor (TF), F3, is
among the most hypermethylated and downregulated genes in IDH1mut gliomas relative to IDH1wt gliomas. In addition to
promoting normal blood clotting and abnormal cancer-induced thrombosis, TF directly increases the malignancy of many
cancers through protease-activated receptor 2 (PAR2), though its activity in gliomas is not well understood. In multiple IDH1wt and
IDH1mut patient-derived glioma cell lines, F3 was hypermethylated in IDH1mut cells compared to IDH1wt cells, with consistently
reduced TF protein expression. Treatment of IDH1mut glioma cells with a demethylating agent, decitabine, increased F3
transcription 5-fold, but had no effect in IDH1wt cells. TF knockdown greatly reduced the in vitro proliferation and colony
formation of IDH1wt/EGFRvIIIamp GBM6 cells and IDH1wt/EGFRamp GBM12 cells, but not of GBM43 cells, which contain an NF1
mutation downstream of EGFR. TF knockdown specifically reduced the in vitro stemlike behavior of GBM6 and GBM12 cells, as
indicated by limiting dilution assays and expression of glioma stem cell (GSC) markers. The stemlike behavior of GBM43,
however, was unaffected by TF knockdown. In vivo, TF knockdown doubled the median survival of mice intracranially engrafted
with GBM6, and caused complete regression of GBM12 (P ¼ 0.001), whereas GBM43 xenograft growth was unimpeded.
Conversely, TF induction enhanced the proliferation and colony formation of IDH1mut GBM164 and TB09 cells, though the effect
was more pronounced in GBM164. TF induction also increased the in vivo "take rate" of intracranial GBM164 xenografts from 0%
to 100% (P ¼ 0.0001), but did not enable TB09 xenograft growth. Further investigation to explain the stronger effects of TF
manipulation in GBM6, GBM12, and GBM164 cells revealed that, in those cells, TF-PAR2 activated receptor tyrosine kinases
(RTKs), including EGFR in GBM6 and GBM12, and PDGFRb in GBM164. This occurred through a Src-dependent intracellular
pathway, even when extracellular RTK stimulation was blocked. In contrast, baseline expression of RTKs was much lower in
GBM43 and TB09 cells. RNA-Seq analysis showed that, out of the entire transcriptome, only two genes showed downregulation
after TF knockdown in GBM6 and GBM12, as well as upregulation after TF induction in GBM164: PROM1, encoding CD133, and
CTNND2, encoding d-catenin. In contrast, TF manipulation did not alter expression of either gene in GBM43 or TB09 cells.
Analysis of Cancer Genome Atlas gliomas confirmed that high F3mRNAcorrelated with enrichment of multipleGSCmarkers, and
that, among GBMs, high F3 mRNA was a significant adverse prognostic marker, independent of patient age and IDH1mut status.
Together, these data suggest that: (i) TF-PAR2 acts through RTKs to promote a GSC phenotype; (ii) CD133 and d-catenin may be
critical downstream effectors of TF-induced GSC behavior; (iii) TF suppression is one reason why IDH1mut gliomas are less
aggressive; (iv) TF-PAR2 is an attractive, novel therapeutic target in IDH1wt gliomas.
Keywords: Tissue Factor, Glioma, Stem cell, IDH1.