hGFAP-mediated GLI2 overexpression leads to early death and severe cerebellar malformations with rare tumor formation

Abstract

The zinc-finger transcription factor GLI2 is frequently amplified in childhood medulloblastoma of the Sonic-hedgehog type (SHH-MB), with or without amplification of NMYC or deletion of TP53. Despite the aggressive tumor behavior, tumorigenesis is not well understood, and adequate mouse models are lacking. Therefore, we generated mice with a GLI2 overexpression under control of the hGFAP-promoter. These mice died within 150 days. The majority only survived until postnatal day 40. They displayed severe cerebellar hypoplasia, cortical malformations, but no brain tumors, except for one out of 23 animals with an undifferentiated hindbrain lesion. Additional loss of p53 did not result in cerebellar tumors, but partially rescued the cerebellar phenotype induced by GLI2 overexpression. Similarly, the combination of GLI2 and NMYC was neither sufficient for the development of SHH-MB. We therefore assume that the development of childhood SHH-MB in mice is either occurring in cellular origins outside the hGFAP-positive lineage or needs additional genetic drivers.

Keywords: Biological sciences; Cancer; Molecular biology.

Graphical abstract

Figure 1

Vector plasmid construct, breeding schemes for Gli2 mice, and survival analysis (A) Illustrated is the vector plasmid construct, which was applied for pronuclear injection. Containing a lac-promoter part, the strong CAG-promoter (Cytomegalovirus early enhancer, the first intron of chicken beta-actin gene, the splice acceptor of the rabbit beta-globulin gene), T7 RNA-Polymerase promoter for transcription, the STOP cassette comprising three SV40-derived poly(A) signal repeats flanked by two loxP sites, the truncated GLI2N insert, a ZsGreen sequence (Zoanthus green fluorescent protein), a mammalian terminator bGH (bovine growth hormone), a lacZ gene coding for beta-galactosidase, f1_ori (DNA replication of bacteriophage f1 functions as a signal for initiation of viral strand synthesis and for its termination), and the Amp-promoter for ampicillin resistance. (B) The breeding schemes for the Gli2 mouse models are demonstrated. The transgenic line expressing Cre recombinase under the control of the human glial fibrillary acidic protein (hGFAP) promoter was crossed with CAG-lsl-Gli2N mice. The CAG promoter was applied to drive expression of a construct composed of floxed sites followed by an active form of GLI2 lacking the N-terminal repressor domain (GLI2N), resulting in hGFAP-cre::CAG-lsl-Gli2N transgenic mice. To obtain double-mutant mouse models, hGFAP-cre::p53 mice were intercrossed with CAG-lsl-Gli2N⁺;p53 mice to achieve hGFAP-cre::p53-CAG-lsl-Gli2N⁺ (heterozygote or homozygote). hGFAP-cre::lsl-NMYC mice were crossed with CAG-lsl-Gli2N;⁺lsl-NMYC mice, resulting in hGFAP-cre::lsl-NMYC-CAG-lsl-Gli2N⁺mice, in heterozygous or homozygous results. More simplistic names were given to the mouse models as demonstrated in brackets. (C) Survival analysis of hG::Gli2N⁺ compared to hG::p53^Fl/+;Gli2N⁺ (∗0.012 Log rank (Mantel-Cox) test); 95% CI of ratio 0.1967 to 1.056)) and hG::p53^Fl/Fl;Gli2N⁺ (∗0.014 Log rank (Mantel-Cox) test); 95% CI of ratio 0.1774 to 1.126)). (D) In a second graph, the comparison of the survival of hG::Gli2N⁺ to hG::NMYC^Fl/+;Gli2N⁺ (ns); 95% CI of ratio 0.2422 to 2.482 and hG::NMYC^Fl/Fl;Gli2N⁺(ns); 95% CI of ratio 0.7242 to 2.662 is displayed.

Figure 2

Brain sections after H&E, NeuN, and Ki67 staining Sagittal brain sections of paraffin-embedded tissues after H&E, Ki67, and NeuN staining. First columns display whole cerebellum (scale bar: 200 μm), second columns higher magnifications of the cerebellum (scale bar: 50 μm), and last columns the Ki67 staining (scale bar: 50 μm) and NeuN staining (scale bar: 50 μm). (A) hGFAP-cre control animal at postnatal day 10. (B) A representative hG::Gli2N⁺ mouse at postnatal day 17. Severe cerebellar disorder is demonstrated and proliferation with Ki67-positive staining as well as positive NeuN staining. (C) Heterozygous hG::p53^Fl/+;Gli2N⁺ animal at postnatal day 17 with a similar disorder and proliferating cells within the cerebellar structure. (D) Homozygous hG::p53^Fl/Fl;Gli2N⁺ animal at postnatal day 17 shows comparable cerebellum than the hGFAP-cre control mouse, but so-called giant cell structures both in the cerebellum and other parts of the brain (arrows). These giant cell structures are also proliferating (Ki67 staining) but did not show NeuN expression (arrows). (E) In 4 out of 13 hG::NMYC^Fl/+;Gli2N⁺ mice, a cerebellar disorder was observed as exemplarily demonstrated here, but also tumor-like structures or no abnormalities in the whole brain at all. (F) Homozygous hG::NMYC^Fl/Fl;Gli2N⁺ mice did not show any abnormalities in the brain structure at all.

Figure 3

Brain sections of CAG::Gli2N⁺, hG::Gli2N⁺, hG::p53^Fl/+;Gli2N⁺, hG::p53^Fl/Fl;Gli2N⁺, and hG::NMYC^Fl/+;Gli2N⁺mice at p7 after H&E, Ki67, NeuN, and Calbindin staining The mouse lines CAG::Gli2N⁺, hG::Gli2N⁺, hG::p53^Fl/+;Gli2N⁺, hG::p53^FlFl+;Gli2N⁺, and hG::NMYC^Fl/+;Gli2N⁺ at p7 are demonstrated after H&E, Ki67, NeuN, and Calbindin staining. (A) CAG::Gli2N⁺ control mouse at p7 is demonstrated. The cerebellum and adjacent brain regions are labeled with their subunits in the first row (scale bar: 200 μm). In the H&E staining, the granule layer (GL) and Purkinje cell layer (PCL) are shown (scale bar: 50 μm) and in a higher magnification (scale bar: 20 μm) in the second and third row. Ki67, NeuN, and Calbindin staining is presented as specific staining (scale bar: 20 μm). (B) H&E, Ki67, NeuN, and Calbindin staining of hG::Gli2N⁺ mice. The cerebellum is unstructured and PCL and GL cannot be delineated as precisely as in the control. (C) Also in the demonstrated hG::p53^Fl/+;Gli2N⁺ mouse, the cerebellum is unstructured, but shows Ki67, NeuN, and Calbindin staining of the PC as the hG::Gli2N⁺ mouse. (D) hG::p53^Fl/Fl;Gli2N⁺ mice show giant cells in the H&E and Ki67 staining (white arrows) and the PCL and GL looks comparable to the one in the control (A). (E) H&E (scale bars: 200, 50, and 20 μM), Ki67 (20μM), and NeuN as well as Calbindin (20 μM) staining are demonstrated and are comparable to the control.

Figure 4

Representative images and staining of giant cells, tumors and tumor-like structures (A–H) Representative H&E and Ki67 stains of a hG::p53^Fl/Fl;Gli2N⁺animal (Figures 4A–4D). Giant cells were found near the piriform cortex (Figure 4B), near the hippocampus area and caudate putamen (Figure 4C), and the cerebellum (Figure 4D). Scale bar is 1 mm for images of the whole brain demonstrated on the left, 20 μm for H&E in B, C, and D and 50 μm for Ki67 staining. E: hG::p53^Fl/Fl;Gli2N⁺ mouse showed so-called giant cells, stained with Olig2 (Figure 4E, scale bar: 200μm) and Cyclin-D1 (Figure 4F, scale bar: 200 μm), both show expression. High-power images of Olig-2 expression giant cells (Figure 4G, scale bar: 20 and 50 μm) and Cyclin-D1 expressing giant cell (Figure 4H, scale bar: 20 and 50 μm) are demonstrated. (I–L) In one out of 23 hG::Gli2N⁺ mice, a tumor in the cerebellum (scale bar: 1 mm) was observed and is illustrated here. Besides H&E staining (Figure 3J, scale bar: 20 and 50 μm) and Ki67 staining (Fig, 3K, scale bar: 20 and 50 μm), also Nmyc staining was applied (Figure 3L, scale bar: 20 and 50 μm). (M–O) The single of the 12 transgenic hG::NMYC^Fl/+;Gli2N⁺ mice with tumor-like structures is displayed (scale bar: 1 mm). Tumor-like structures demonstrated in the brain stem (Figure 3N) and in the subventricular zone (SVZ) (Figure 3O) are shown after H&E, Ki67, and Nmyc staining (scale bar: 20 μm H&E and 50 μm Ki67 and Nmyc). (P) Mouse body weights of hG::Gli2N⁺, hG::NMYC^Fl/+;Gli2N⁺, hG::p53^Fl/Fl;Gli2N⁺, and control mice (p53^Fl/Fl;Gli2N⁺, NMYC^Fl/+;Gli2N⁺, and Gli2N⁺) are displayed. For statistical analysis, a grouped analysis with multiple t test was applied compared with the sum of the control mice (p ∗0.03 for of hG::NMYC^Fl/+;Gli2N⁺ mice). Data are shown as mean ± standard error of mean (SEM), and p values less than 0.05 were considered significant.

Figure 5

Representative H&E stains of the spinal cord Representative H&E stains of the spinal cord of all five mouse models demonstrate normal appearance and no metastasis or tumors-like structures. Spinal cords were cut longitudinal or transversal (scale bar: 200 μm).

Figure 6

Western blot analyses (A) The lysates of fresh frozen whole cerebellum tissue were fractionated by SDS-PAGE and processed to a western blot membrane. The membrane was incubated with the GLI2-antibody PA1-28838 and the secondary antibody anti-rabbit HRP-conjugated P044801-2. α-Tubulin was used as housekeeping control. Expected band at ∼190 kDa was detected. (B) Western blot analysis using a Gli2 antibody which detects the middle region of human Gli2 and can so detect Gli2N and human full-length Gli2. The truncated Gli2N was detected in all mouse models, the human full-length Gli2 in the control (human breast cancer cell line MDA-MB-231). β-Actin was demonstrated as housekeeping gene for the analysis. (C) In this western blot analysis, Gli1 was detected in all mouse models, even though hG::p53^Fl/+;Gli2N⁺ and hG::NMYC^Fl/+;Gli2N⁺ show a distinct weaker band than the homozygous mouse model. (D) NMYC expression was demonstrated in all mouse models, GFAP was used as housekeeping control.