Transforming growth factor-β pathway activity in glioblastoma

Abstract

Transforming growth factor (TGF)-β is a central molecule maintaining the malignant phenotype of glioblastoma. Anti-TGF-β strategies are currently being explored in early clinical trials. Yet, there is little contemporary data on the differential expression of TGF-β isoforms at the mRNA and protein level or TGF-β/Smad pathway activity in glioblastomas in vivo.Here we studied 64 newly diagnosed and 16 recurrent glioblastomas for the expression of TGF-β1-3, platelet-derived growth factor (PDGF)-B, and plasminogen activator inhibitor (PAI)-1 mRNA by RT-PCR and for the levels of TGF-β1-3 protein, phosphorylated Smad2 (pSmad2), pSmad1/5/8 and PAI-1 by immunohistochemistry.Among the TGF-β isoforms, TGF-β1 mRNA was the most, whereas TGF-β3 mRNA was the least abundant. TGF-β1-3 mRNA expression was strongly correlated, as was the expression of TGF-β1-3 mRNA, and of the TGF-β1-3 target genes, PDGF-B and PAI-1. TGF-β2 and TGF-β3 protein levels correlated well, whereas the comparison of the other TGF-βisoforms did not. Positive correlation was also observed between TGF-β1 and pSmad1/5/8 and between pSmad2 and pSmad1/5/8. Survival analyses indicated that a group of patients with high expression levels of TGF-β2 mRNA or pSmad1/5/8 protein have inferior outcome.We thus provide potential biomarkers for patient stratification in clinical trials of anti-TGF-β therapies in glioblastoma.

Keywords: PAI-1; PDGF-B; TGF-β; biomarker; glioblastoma.

Figure 1. Expression of TGF-β isoforms in glioblastoma in vivo

A-C, Relative mRNA expression levels for TGF-β₁, TGF-β₂ and TGF-β₃ were assessed in all glioblastoma tissues (pooled, newly diagnosed or recurrent). The black bar marks the mean in each group. Values are represented on a logarithmic scale. Statistical significances of p < 0.05 (*) and p < 0.001 (***) were determined using the Mann-Whitney test. D-F, Correlation of TGF-β isoform mRNA expression among all samples pooled. Values are represented on a logarithmic scale. Two-tailed Spearman test coefficients (r) and significances are indicated (open circles, newly diagnosed; closed circles, recurrent). TGF-β₁, TGF-β₂ or TGF-β₃ protein levels were assessed by immunohistochemistry and median H scores determined and presented for all patients pooled (G), newly diagnosed tumor tissues (H) and recurrent tumor tissues (I) separately. The black bar marks the mean in each group. Statistical significances of p < 0.01 (**) and p < 0.001 (***) were determined using the Mann-Whitney test. (J-L) Correlation of TGF-β protein levels among all samples pooled (open circles, newly diagnosed; closed circles, recurrent). (M) Correlation analyses of the three TGF-β protein levels with mRNA expression of the respective TGF-β isoform are shown. Two-tailed Spearman test coefficients (r) and significances are indicated.

Figure 2. Immunohistochemical studies of the TGF-β pathway in glioblastoma

Representative stainings for TGF-β₁, TGF-β₂, TGF-β₃, pSmad2, pSmad1/5/8, and PAI-1 (score < 50 left, score 50-150 middle, score > 150 right). Size bars correspond to 100 μm.

Figure 3. Assessment of TGF-β pathway activation in glioblastoma: Smad phosphorylation

A, pSmad2 or pSmad1/5/8 protein levels were assessed by immunohistochemistry and median H Scores are shown for all patients pooled, newly diagnosed tumor tissues or recurrent tumor tissues separately. The black bar marks the mean in each group. B, Correlation is shown for the H scores of pSmad2 and pSmad1/5/8 for all samples pooled. Paired correlation analyses of pSmad2 (C,E) or pSmad1/5/8 (D,F) protein levels and TGF-β isoform mRNA (C,D) or protein (E,F) levels are shown for all samples pooled. Two-tailed Spearman test coefficients (r) and significances (p) are indicated (open circles, newly diagnosed; closed circles, recurrent).

Figure 4. Assessment of TGF-β pathway activation in glioblastoma: expression of TGF-β response genes

A, PDGF-B or PAI-1 mRNA expression data were assessed by RT-PCR for all patients pooled, newly diagnosed tumor tissues or recurrent tumor tissues separately. The black bar marks the mean in each group. Values are represented on a logarithmic scale. B, Correlation is shown for the mRNA data of PDGF-B and PAI-1 for all samples pooled. Correlation was assessed of mRNA data of TGF-β₁ or TGF-β₂ or TGF-β₃ and mRNA data of PDGF-B (C) or PAI-1 (D) for all patients pooled. Correlation was assessed of protein data of TGF-β₁ or TGF-β₂ or TGF-β₃ and mRNA data of PDGF-B (E) or PAI-1 (F) for all patients pooled. G, Correlation is shown for pSmad2 or pSmad1/5/8 mRNA data and PDGF-B or PAI-1 mRNA data. Two-tailed Spearman test coefficients (r) and significances (p) are indicated (open circles, newly diagnosed; closed circles, recurrent tumor specimens).

Figure 5. TGF-β pathway activity: an analysis of the Cancer Genome Atlas (TCGA) network

A, Correlation is shown among the TGF-β isoforms. Correlation was assessed using mRNA data of TGF-β₁ or TGF-β₂ or TGF-β₃ and mRNA data of PDGF-B (B) or PAI-1 (C). Two-tailed Spearman test coefficients (r) and significances (p) are indicated. Data are obtained from the TCGA network.