The emerging role of NG2 in pediatric diffuse intrinsic pontine glioma

Abstract

Diffuse intrinsic pontine gliomas (DIPGs) have a dismal prognosis and are poorly understood brain cancers. Receptor tyrosine kinases stabilized by neuron-glial antigen 2 (NG2) protein are known to induce gliomagenesis. Here, we investigated NG2 expression in a cohort of DIPG specimens (n= 50). We demonstrate NG2 expression in the majority of DIPG specimens tested and determine that tumors harboring histone 3.3 mutation express the highest NG2 levels. We further demonstrate that microRNA 129-2 (miR129-2) is downregulated and hypermethylated in human DIPGs, resulting in the increased expression of NG2. Treatment with 5-Azacytidine, a methyltransferase inhibitor, results in NG2 downregulation in DIPG primary tumor cells in vitro. NG2 expression is altered (symmetric segregation) in mitotic human DIPG and mouse tumor cells. These mitotic cells co-express oligodendrocyte (Olig2) and astrocyte (glial fibrillary acidic protein, GFAP) markers, indicating lack of terminal differentiation. NG2 knockdown retards cellular migration in vitro, while NG2 expressing neurospheres are highly tumorigenic in vivo, resulting in rapid growth of pontine tumors. NG2 expression is targetable in vivo using miR129-2 indicating a potential avenue for therapeutic interventions. This data implicates NG2 as a molecule of interest in DIPGs especially those with H3.3 mutation.

Keywords: DIPG; NG2; PDGF; glioma; histone 3.

Conflict of interest statement

CONFLICTS OF INTEREST

The authors declare no potential conflicts of interest.

Figures

Figure 1. Immunohistochemical and western blotting analysis validates NG2 upregulation in human DIPG

(a) In human tissue, diffuse expression of NG2 was observed in DIPG tumor localized to membrane (inset). (b) Adjacent healthy brainstem tissue. Scale bar: 100 μM. (c) Graphical representation of western blotting data of NG2 expression in 22 DIPG tumor and 16 adjacent normal specimens. GAPDH was used for protein loading normalization. Average NG2 expression was plotted after normalizing NG2 levels to GAPDH in each sample. P < 0.05 was considered significant. (d) Average of normalized NG2 expression levels in NG2 western blot samples with histone 3.1 K27M mutation or histone 3.3 K27M mutation. P < 0.05 was considered significant.

Figure 2. NG2 overexpression in PDGFB and PDGFB/H3.3 K27M/p53−/− mouse models

(a) In mouse, expression of NG2 in PDGFB mouse was limited to brainstem tumor (dotted area). (b) Sporadic expression of NG2 across brainstem and cerebellum of normal mouse. (c) Injection of NSG SCID mouse with mouse DIPG cells that are H3.3.K27 mutant and have PDGFB overexpression and p53 deletion resulted in overexpression of NG2 (dotted area). (d) In the adjacent normal brain tissue of NSG SCID mouse injected with PDGFB/H3.3.K27M/p53−/− cells, NG2 overexpression was not detected. Insets are 40 × magnifications of the corresponding panel. Scale bar: 200 μM.

Figure 3. NG2 upregulation in DIPG is partially due to hypermethylation and downregulation of its regulatory microRNA, miR129-2

(a) Luciferase vector with 3′-UTR sequence of NG2 cloned downstream of the firefly luciferase gene (Luc-3′UTR) and plasmid vector with miR129-2 sequence were used. PDGFB mouse neurospheres were transfected in triplicates with luciferase and miR129-2 vectors, as shown, and luciferase expression was determined. Co-expression of Luc-3′UTR of NG2 and miR129-2 resulted in significant downregulation of luciferase expression as compared to cells transfected with luc-3′UTR and empty vector. P < 0.05 was considered significant. (b) PDGFB mouse neurospheres were transfected with miR129-2 or NG2-shRNA plasmids. Upregulation of either miR129-2 or NG2-shRNA resulted in NG2 downregulation in vitro when compared to empty vector or control shRNA treated cells. (c) Human DIPG cells, SF8628, were transduced with GFP expressing pCDH-control or pCDH-miR129-2 lentivirus. Immunofluorescence analysis by confocal microscopy revealed upregulation of miR129-2 (indicated by GFP) resulted in NG2 (red) downregulation in vitro when compared to control transduced cells. DAPI was used to stain nuclei. Scale bar = 5 μm. (d) Average expression of miR129-2 as assessed by RT-PCR (*miR129) or illumina expression chip assay (miR129), miR129 hypermethylation (miR129 me), and NG2 mRNA levels in human DIPG compared to adjacent normal. miR129-2 is shown to be downregulated in DIPG as assessed by RT-PCR and illumina chip assays. Hypermethylation of miR129-2 at four or more CpG sites corresponding to miR129-2 promoter were detected in 57% of DIPGs. As expected, downregulation of miR129-2 due to hypermethylation resulted in upregulation of NG2 mRNA as detected by NG2 mRNA using illumina platform. (e) To reverse miR129-2 hypermethylation, PDGFB mouse neurospheres were treated with 5-Azacytidine, a DNA methyl-transferase inhibitor drug. NG2 expression decreased in mouse tumor cells when treated with 10-30 μM 5-Azacytidine as compared to untreated cells. NG2 expression in each sample was normalized to corresponding GAPDH signal and average NG2 expression from three experiments was plotted.

Figure 3. NG2 upregulation in DIPG is partially due to hypermethylation and downregulation of its regulatory microRNA, miR129-2

(a) Luciferase vector with 3′-UTR sequence of NG2 cloned downstream of the firefly luciferase gene (Luc-3′UTR) and plasmid vector with miR129-2 sequence were used. PDGFB mouse neurospheres were transfected in triplicates with luciferase and miR129-2 vectors, as shown, and luciferase expression was determined. Co-expression of Luc-3′UTR of NG2 and miR129-2 resulted in significant downregulation of luciferase expression as compared to cells transfected with luc-3′UTR and empty vector. P < 0.05 was considered significant. (b) PDGFB mouse neurospheres were transfected with miR129-2 or NG2-shRNA plasmids. Upregulation of either miR129-2 or NG2-shRNA resulted in NG2 downregulation in vitro when compared to empty vector or control shRNA treated cells. (c) Human DIPG cells, SF8628, were transduced with GFP expressing pCDH-control or pCDH-miR129-2 lentivirus. Immunofluorescence analysis by confocal microscopy revealed upregulation of miR129-2 (indicated by GFP) resulted in NG2 (red) downregulation in vitro when compared to control transduced cells. DAPI was used to stain nuclei. Scale bar = 5 μm. (d) Average expression of miR129-2 as assessed by RT-PCR (*miR129) or illumina expression chip assay (miR129), miR129 hypermethylation (miR129 me), and NG2 mRNA levels in human DIPG compared to adjacent normal. miR129-2 is shown to be downregulated in DIPG as assessed by RT-PCR and illumina chip assays. Hypermethylation of miR129-2 at four or more CpG sites corresponding to miR129-2 promoter were detected in 57% of DIPGs. As expected, downregulation of miR129-2 due to hypermethylation resulted in upregulation of NG2 mRNA as detected by NG2 mRNA using illumina platform. (e) To reverse miR129-2 hypermethylation, PDGFB mouse neurospheres were treated with 5-Azacytidine, a DNA methyl-transferase inhibitor drug. NG2 expression decreased in mouse tumor cells when treated with 10-30 μM 5-Azacytidine as compared to untreated cells. NG2 expression in each sample was normalized to corresponding GAPDH signal and average NG2 expression from three experiments was plotted.

Figure 4. Mouse and human DIPG neurospheres exhibit symmetric NG2 expression in vitro and are immature stem cells co-expressing oligodendrocyte (Olig2) and astrocyte (GFAP) markers

(a) PDGFB mouse tumor neurospheres (Mouse) and human primary DIPG cells (SF8628 and SUDIPGVI) were grown and stained using NG2 antibody (red) and DAPI (blue) nuclear staining. NG2 is equally expressed in dividing cells suggesting that its expression is defective (symmetric) in mitotic cells. Scale bar = 5 μm. (b) PDGFB mouse tumor (Mouse) and SF8628 human DIPG primary cells were expanded in culture and immunostained for DAPI (nuclear), oligodendrocyte marker (Olig2, green), NG2 (red), and astrocyte marker (GFAP, magenta). Confocal microscopy was used to assess expression of above-mentioned proteins in mitotic cells. Scale bar = 5 μM.

Figure 4. Mouse and human DIPG neurospheres exhibit symmetric NG2 expression in vitro and are immature stem cells co-expressing oligodendrocyte (Olig2) and astrocyte (GFAP) markers

(a) PDGFB mouse tumor neurospheres (Mouse) and human primary DIPG cells (SF8628 and SUDIPGVI) were grown and stained using NG2 antibody (red) and DAPI (blue) nuclear staining. NG2 is equally expressed in dividing cells suggesting that its expression is defective (symmetric) in mitotic cells. Scale bar = 5 μm. (b) PDGFB mouse tumor (Mouse) and SF8628 human DIPG primary cells were expanded in culture and immunostained for DAPI (nuclear), oligodendrocyte marker (Olig2, green), NG2 (red), and astrocyte marker (GFAP, magenta). Confocal microscopy was used to assess expression of above-mentioned proteins in mitotic cells. Scale bar = 5 μM.

Figure 5. Orthotopic injection of NG2 expressing cells causes aggressive brainstem tumors in vivo resulting in a robust murine model of DIPG

(a) NG2 expressing neurospheres were isolated from PDGFB mouse pontine tumor and expanded in culture and stained for NG2 (red) and DAPI (blue) (b and c). Neurospheres were then injected into the brainstems of two-day-old (P2) mice. Whole brain was obtained at 3 weeks post injection and subjected to H&E staining (d-g). Injected cells show an infiltrative pattern of growth (g, arrow) at the site of injection. H&E staining also showed a large infiltrating tumor within pons (region p in d). 10 × magnifications (e-f) revealed the highly cellular tumor within the pons (e) that exhibits characteristics of DIPGs including tumor vascularity (f, arrows). Injected cells were infiltrated through the pons and away from the injection site (g, arrow). This pontine tumor is positive for the astrocytic marker GFAP (h), proliferation marker Ki67 (i), PDGFRβ (j), PDGFRα (k), and NG2 (l). Scale bar = 100 μM.

Figure 6. miR129-2 targets and downregulates NG2 in vivo

Brain sections from NG2-dsRed mice were analyzed 2 weeks after the intracranial injections with control lentiviral vector on left side and lentiviral vector harboring miR129-2 on right side. Detection of the green signal (GFP) indicates in vivo transduction of either control or miR129-2. Area enclosed by dotted line in 10 × image (Scale bar: 100 μM) represents the site of injection. 40 × magnification images (Scale bar: 10 μM) of boxed area in merged images are shown to demonstrate the downregulation of NG2 (red) in miR129-2 (green) transduced cells (arrows) while control transduced cells are expressing normal levels of NG2.