Ptch1^+/− mice were obtained in CD1 background through ablation of exons 6 and 7 (45). The Btg1 knockout mice were generated in the C57BL/6 strain by replacing the exon I of the gene with the neomycin resistance cassette, as previously described (11).

The crossing of Ptch1^+/− with Btg1^−/− mice generated Ptch1/Btg1 double-mutant mice, which were interbred for at least six generations to obtain an isogenic progeny of the different genotypes under study. To identify the preneoplastic lesions, Ptch1/Btg1 double-mutant mice were crossed with Math1-GFP transgenic mice, which express the green fluorescent protein (GFP) driven by the Math1 enhancer in proliferating GCPs (46, 47), kindly provided by Jane Johnson (UT Southwestern Medical School, Dallas, TX, USA); the Ptch1/Btg1/Math1-GFP mice were then interbred four or more times before starting any analysis.

The Ptch1/Btg1 pups were routinely genotyped by PCR analysis using genomic DNA extracted from tail tips, as described (12, 40, 42, 45); the mice which carry the transgene Math1-GFP have been previously identified using a “GFP flashlight” (Nightsea) that made the GFP⁺ pups glow.

All animal experiments were performed with mice of both sexes, in accordance with the current European Ethical Committee guidelines (directive 2010/63/EU) and with the protocol approved by the Italian Ministry of Health (authorization 193/2015-PR).

Mice were observed daily for MB formation for 12 months after birth. On the appearance of tumor symptoms, such as head doming, ruffling of fur, hunched posture, impaired balance, severe weight loss, posterior paralysis, or lethargy, they were euthanized under anesthesia and autopsied. MBs were either snap frozen in liquid nitrogen for analyses of mRNAs and proteins or fixed in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) by overnight immersion.

For quantification of the lesions, asymptomatic Ptch1^+/−/Btg1^+/+/Math1-GFP⁺ and Ptch1^+/−/Btg1^−/−/Math1-GFP⁺ mice were killed at different times (2 or 6 weeks of age), and their cerebella were fixed by immersion overnight in 4% PFA in PBS. For the EGL studies, cerebella of postnatal day 7 (P7) mice were dissected out and kept overnight in 4% PFA in PBS.

The mice at P7, 2 and 6 weeks of age were intraperitoneally injected with 5-bromo-2′-deoxyuridine (BrdU) (95 mg/kg; Sigma Aldrich, St. Louis, MO, USA) 1 h before sacrifice to visualize the GCPs entering in S-phase, according to existing protocols (48, 49).

All the samples (with the exception of MBs employed for analyses of mRNAs and proteins) were cryoprotected in 30% sucrose in PBS and frozen at −80°C until use.

Samples were embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA, USA) and cut in free-floating 40-μm-thick sagittal sections at −25°C on a rotary cryostat.

To detect the BrdU incorporated into cellular DNA during cell proliferation, the sections were treated with 2N HCl for 45 min at 37°C to denature the nucleic acid, then with 0.1 M sodium borate buffer, pH 8.5, for 10 min. All the slices were permeabilized with 0.3% Triton X-100 in PBS and incubated overnight at 4°C with primary antibodies diluted in 3% normal donkey serum in PBS. The following primary antibodies were used: rat monoclonal antibody against BrdU (AbD Serotech, Raleigh, NC, USA; MCA2060; 1:400); rabbit monoclonal antibody against Ki67 (LabVision Corporation, Fremont, CA; clone SP6; 1:150); rabbit polyclonal antibody against cleaved (activated) caspase-3 (Cell Signaling Technology, Danvers, MA, USA; 9661; 1:100); mouse monoclonal antibodies raised against NeuN (Millipore, Temecula, CA, USA; MAB377; 1:300) or CD15 (Santa Cruz Biotechnology, Santa Cruz, CA, USA; sc-19648; 1:100); goat polyclonal antibody against NeuroD1 (R&D Systems, Minneapolis, MN, USA; AF2746; 1:200). To visualize the antigens, the samples were reacted with the appropriate secondary antibodies, all from Jackson ImmunoResearch (West Grove, PA, USA; 1:200): a donkey anti-rat TRITC-conjugated (BrdU), a donkey anti-rabbit Cy3-conjugated (Ki67, caspase-3), a donkey anti-mouse TRITC-conjugated (CD15) or conjugated to Alexa-488 (NeuN), a donkey anti-goat Cy3-conjugated (NeuroD1).

The samples were counterstained by Hoechst 33258 (Sigma-Aldrich; 1 mg/ml in PBS) to visualize the nuclei.

Digital images of the immunostained sections were collected with a TCS SP5 confocal laser scanning microscope (Leica Microsystems, Wetzlar, Germany). The CD15 immunolabeled MB slices were examined with an Olympus FV1200 laser scanning confocal microscope using a PlanApoN 60× Oil SC (NA = 1.40) Olympus objective. All the digital pictures were analyzed by the IAS software (Delta Sistemi, Rome, Italy).

To evaluate the presence of preneoplastic lesions in mice of 2 or 6 weeks of age, we have examined for each cerebellum 12 sections, at 400 μm of distance, in order to allow a representative sampling. The lesions were identified by visualizing the GFP-positive GCPs.

The images of the lesions in cerebellar sagittal sections (Figures 2A,B) were obtained by an Olympus Optical (Tokyo, Japan) BX53 fluorescence microscope connected to a Spot RT3 camera (Diagnostic Instruments Inc., Sterling Heights, MI, USA).

BrdU⁺ or cleaved caspase-3⁺ cells in lesions were counted in the whole lesion area on at least five non-adjacent midsagittal sections per lesion. Planimetric measurements of lesion area were performed for the total extension of the lesion in each photomicrograph field by tracing the outline of the whole lesion on the digital picture and were measured with the IAS software. We analyzed the lesions present in all lesioned mice whose number for each genotype and time point is indicated in Figure 2C.

The proliferating (BrdU⁺), differentiating (NeuroD1⁺ or NeuN⁺), and apoptotic (cleaved caspase-3⁺) cells in the EGL of mice at P7 were measured on at least five non-adjacent midsagittal sections per mouse at the midpoint of the fifth, seventh, and ninth folia of each section, according to published protocols (12, 42); three mice for each genotype were analyzed. The EGL cell numbers were expressed as percentage ratio of proliferating, differentiating, or apoptotic cells to the total number of cells (labeled by Hoechst 33258), counted for the entire length of the EGL in each photomicrograph field.

Also in MBs, the proliferating (Ki67⁺), apoptotic (cleaved caspase-3⁺), or CD15-positive cell numbers were expressed as a percentage ratio to the total number of cells, labeled by Hoechst 33258. The cell counts were performed on about 40 randomly representative images, collected under the same parameters from five different tumors for each genotype.

Total cellular RNA was extracted from Ptch1^+/−/Btg1^+/+ (n = 4) and Ptch1^+/−/Btg1^−/− (n = 4) tumors with Trizol Reagent (Invitrogen, San Diego, CA, USA) and reverse-transcribed, as previously described (42). Real-time qPCR was carried out with a 7900HT System (Applied Biosystems, Foster City, CA, USA) using SYBR Green I dye chemistry in duplicate samples. The mRNA relative expression values, obtained by the comparative cycle threshold method (50), were normalized to expression levels of the TATA-binding protein gene, used as endogenous control. Specific qRT-PCR primers were designed by the Beacon Designer 7.1 software (Premier Biosoft International, Palo Alto, CA, USA) or by the BLAST software (Primer Blast NCBI: http://www.ncbi.nlm.nih.gov/tools/primer-blast) (51); their sequence is available on request.

The Ptch1^+/−/Btg1^+/+ (n = 3) and Ptch1^+/−/Btg1^−/− (n = 3) tumors were lysed into lysis buffer [50 mM Tris–HCl (pH 7.5), sodium deoxycholate 0.1%, 150 mM NaCl, 5 mM EDTA (pH 8.0), 10% glycerol, 1% NP40], containing 1 mM phenylmethylsulfonyl fluoride (PMSF), 10 μg/ml of leupeptin and aprotinin, 10 mM β-glycerophosphate, 4 mM NaF, 0.1 mM Na₃VO₄, 5 mM Na(C₃H₇COO), and sonicated. Proteins (80 μg per MB) were electrophoretically analyzed by sodium dodecyl sulfate (SDS)−10% polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes; these were incubated for 2 h in blocking buffer {TBS [10 mM Tris–HCl (pH 8), 150 mM NaCl]−0.05% Tween−5% powdered milk} and then incubated overnight in the same buffer with the indicated primary antibodies.

The antibodies used for immunoblots were as follows: the rabbit polyclonal antibody anti-Prmt1 (07–404; Millipore; 1:500); the mouse monoclonal antibody anti-cyclin D1 (sc-20044; Santa Cruz Biotechnology; 1:200); the mouse monoclonal antibody anti-phospho-BAD (Ser99, sc-271963; Santa Cruz Biotechnology; 1:50); and the mouse monoclonal antibody DM1A anti-αTubulin (T6199; Sigma Aldrich; 1:10,000).

The densitometric analysis was performed using the Scan Analysis software (Biosoft, Cambridge, UK). Mean ± SEM densitometric values were obtained by averaging the percentage change of each MB relative to the control sample (one Ptch1^+/−/Btg1^+/+ tumor) and normalizing each sample value to the corresponding α-tubulin densitometric value, used as an endogenous control.

All statistical analyses were performed using StatView 5.0 software (SAS Institute, Cary, NC, USA).

The χ²-test was used to compare the MB incidence (Table 1) and the percentage of positivity for preneoplastic lesions (Figure 2) in mice of the different genotypes studied.

Data of MB time latency (Table 1) and number of lesions per cerebellum (Figure 2) were analyzed as average ± SEM and analyzed by the Student's t-test. Also the statistical analyses of mRNA and protein expression values were performed by Student's t-test.

The immunohistochemistry data of cell proliferation, differentiation, and apoptosis, as well as of CD15-positive cells (shown in Figures 1, 4–7, respectively) were presented as mean% of total cells ± SEM. The experiments where the data were calculated as percentage of total cells were tested for the equality of variance between groups by the Levene's test; if equality was not satisfied, the data were analyzed with non-parametric tests, namely, Kruskal–Wallis test to analyze the main effects and Mann–Whitney U-test to analyze the simple effects (as for measurements of MB proliferation and apoptosis in Figure 1, of NeuN-positive cells in Figure 5, caspase-3-positive cells in Figure 6, and of CD15-positive cells in Figure 7). Data with equality of variance (i.e., proliferating cells in Figure 4 and NeuroD1-positive cells in Figure 5) were analyzed using two-way ANOVA to test the main effects of the two genotypes (Btg1 and Ptch1) on cell numbers. Individual between-group comparisons to test simple effects were carried out by Fisher's protected least significant difference (PLSD) ANOVA post-hoc test.

Moreover, the quantification by immunohistochemical analysis of apoptotic cells in preneoplastic lesions was obtained as mean ± SEM/mm² of lesion and was statistically analyzed by Student's t-test (Figure 3).

All statistical tests were considered significant with p < 0.05.

It is known that the Ptch1^+/− mice develop spontaneous MBs arising from GCPs within 10 months of birth, with an incidence that is background-dependent (52). Moreover, we have observed that the ablation of Btg1 increases the proliferation of GCPs, with consequent formation of a hyperplastic EGL (12).

Thus, we tested whether the Btg1 deletion could affect the tumorigenic potential of GCPs. For this, we crossed the Ptch1^+/− mice with the Btg1 knockout mice and analyzed the frequency of MB development in double-mutant mice during 1 year of life. Unexpectedly, we observed that in Ptch1^+/− background, the ablation of one or both alleles of Btg1 did not significantly favor tumorigenesis, although in the double-mutant mice there was a trend to increase the MB incidence relative to the Ptch1^+/−/Btg1^+/+ mice (referred throughout this report to as Ptch1^+/−/Btg1^WT). In fact, the Ptch1^+/−/Btg1^+/− and Ptch1^+/−/Btg1^−/− (the latter hereafter referred to as Ptch1^+/−/Btg1^KO) mice developed MB with a frequency of 38.89 and 42.42%, respectively, relative to the MB incidence of Ptch1^+/−/Btg1^WT mice (the latter incidence was 34.62%; Ptch1^+/−/Btg1^+/− vs. Ptch1^+/−/Btg1^WT p = 0.648; Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT p = 0.469; χ²-test; Table 1). Also the tumor latency in double-mutant mice, either Ptch1^+/−/Btg1^+/− or Ptch1^+/−/Btg1^KO, did not significantly differ from that of control Ptch1^+/−/Btg1^WT mice (Ptch1^+/−/Btg1^+/− vs. Ptch1^+/−/Btg1^WT p = 0.852; Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT p = 0.386; Student's t-test; Table 1).

To assess the effect of Btg1 ablation on proliferation and apoptosis of neoplastic GCPs within the tumors, we carried out an immunohistochemical analysis of MB sections with the antibodies for Ki67, a protein expressed by cycling cells (53) or for the active proteolytic fragment of caspase-3, indicator of the apoptotic process (Figures 1A,C). Surprisingly, we found that the percentage of mitotic cells to the total number of cells (detected by Hoechst 33258; proliferation index) in MBs of Ptch1^+/−/Btg1^KO mice was equivalent to that in tumors of Ptch1^+/−/Btg1^WT mice (p = 0.8003; Mann–Whitney U-test; Figures 1A,B). Conversely, we observed in the Ptch1^+/−/Btg1^KO MBs a highly significant increase in the percentage of caspase-3-positive cells to the total number of cells (apoptotic index), with respect to the Ptch1^+/−/Btg1^WT tumors (p < 0.0001 and 42% increase; Mann–Whitney U-test; Figures 1C,D).

Next, we examined the effect of Btg1 absence on the earliest stages of tumorigenesis by analyzing the premalignant lesions in the cerebella of Ptch1^+/−/Btg1^+/+/Math1-GFP⁺ and Ptch1^+/−/Btg1^−/−/Math1-GFP⁺ mice (referred throughout this report to as Ptch1^+/−/Btg1^WT/Math1-GFP⁺ and Ptch1^+/−/Btg1^KO/Math1-GFP⁺, respectively) sacrificed at different times, i.e., at 2 or 6 weeks. Two weeks of age is a stage corresponding to the end of proliferative development of GCPs within the EGL, when the lesions become morphologically detectable and are quite frequent (about 75%) (54); at 6 weeks of age, the EGL has disappeared but the lesions remain frequent, although only a limited number will develop into MB (47). The mouse model Ptch1^+/− is a powerful tool to identify the early stages in the MB development because within 2–6 months of birth, more than 50% of mice present at the surface of the cerebellum regions of ectopic cells, which represent a preneoplastic stage of tumorigenesis. These regions are usually nodular formations, extended into the fissures between the cerebellar lobules, and contain highly proliferating preneoplastic GCPs (pGCPs) (47, 55, 56). The pGCPs inside the lesions express the transcription factor Math-1, and then they can be identified in Math1-GFP⁺ mice by GFP expression (47, 56).

The screening of the cerebella of the asymptomatic Ptch1^+/−/Btg1^WT/Math1-GFP⁺ and Ptch1^+/−/Btg1^KO/Math1-GFP⁺ mice did not reveal differences between the two genotypes under study in the incidence of MB lesions (i.e., number of mice with lesions) or in the number of lesions per cerebellum neither at 2 weeks nor at 6 weeks of age (Figure 2). In particular, the ablation of the Btg1 gene did not influence the expected frequency of GFP⁺ lesions in Ptch1^+/− mice at 6 weeks (Ptch1^+/−/Btg1^KO/Math1-GFP⁺ 70% frequency, and Ptch1^+/−/Btg1^WT/Math1-GFP⁺ 80% frequency, p = 0.6056, χ²-test; Figures 2B,C), which resulted in agreement with previous data (56).

Then, we analyzed by immunohistochemistry the proliferation (after a 1-h pulse of BrdU) and the survival of pGCPs within the lesions. Similar to the MBs, no differences were detected between Ptch1^+/−/Btg1^WT/Math1-GFP⁺ and Ptch1^+/−/Btg1^KO/Math1-GFP⁺ mice at 2 and 6 weeks of age in the proliferative rate of pGCPs, expressed as the number of BrdU⁺ cells present in the GFP⁺ lesion area (data not shown). On the other hand, we observed a significant increase of apoptosis, measured as cleaved caspase-3⁺ cell number per lesion area, in Ptch1^+/−/Btg1^KO/Math1-GFP⁺ mice, relative to Ptch1^+/−/Btg1^WT/Math1-GFP⁺ mice, either at 2 weeks (p = 0.008 and 38% increase; Student's t-test; Figures 3A,C) or at 6 weeks of age (p = 0.0003 and 49% increase; Student's t-test; Figures 3B,C).

Together, these results indicate that the Btg1 deletion in Ptch1^+/− mice did not enhance the neoplastic transformation of pGCPs and, then, the MB incidence, and suggest that this might be a consequence of a significant increase of apoptosis.

Subsequently, we sought to assess the effects of the genetic ablation of Btg1 in background Ptch1^+/− during the cerebellar development. To this aim, we measured the proliferation, differentiation, and survival of the GCPs within the EGL at P7, a stage corresponding to the age of highest expansion of GCPs.

At P7, the percentage of proliferating GCPs to the total number of cells, identified by incorporation of BrdU after 1-h pulse, was significantly higher in Ptch1^+/+/Btg1^KO mice than in Ptch1^+/+/Btg1^WT mice (hereafter referred to as Btg1^KO and Btg1^WT, respectively; 27.2% increase, p < 0.0001, Fisher's PLSD ANOVA post-hoc test; Figures 4A,B), as expected (12). Also in Ptch1^+/−/Btg1^WT mice, the GCP proliferation was increased by constitutive activation of the Shh pathway (Ptch1^+/−/Btg1^WT vs. Btg1^WT p = 0.0012, 11.1% increase, Fisher's PLSD ANOVA post-hoc test; Figures 4A,B). The latter increase of proliferation agrees with previous observations (42) and is significantly lower than that induced by Btg1 deletion alone (Ptch1^+/−/Btg1^WT vs. Btg1^KO p < 0. 0001, Fisher's PLSD ANOVA post-hoc test; Figures 4A,B). Notably, no positive synergy was observed between the Btg1 and Shh pathways since the ablation of Btg1 did not increase the percentage of proliferating GCPs above the level induced by the constitutively activated Shh signaling pathway (Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT p = 0.5737, Fisher's PLSD ANOVA post-hoc test; Figures 4A,B).

Next, we measured the differentiation of GCPs by labeling these cells with the neural markers NeuroD1 and NeuN (Figure 5). NeuroD1 is highly expressed in early post-mitotic GCPs and is implicated in their differentiation (57), while NeuN labels fully differentiated granule neurons (58). At P7, we observed a significant decrease in the percentage of NeuroD1⁺ GCPs in all conditions of augmented proliferation (Btg1^WT vs. Btg1^KO p = 0.0001; Btg1^WT vs. Ptch1^+/−/Btg1^WT p = 0.0063; Btg1^WT vs. Ptch1^+/−/Btg1^KO p = 0.0002; Fisher's PLSD ANOVA post-hoc test; Figures 5A,B). Furthermore, no difference was observed in NeuroD1 expression between mice with the genetic ablation of Btg1 alone or combined with the deletion of the Ptch1 allele, indicating that no evident synergy occurred between the Btg1-null and the Ptch1^+/− background (Figures 5A,B). Similarly, the percentage of NeuN-positive GCPs was significantly lower in all groups relative to the Btg1^WT mice, with a slightly more pronounced decrease in the Btg1^KO mice (Btg1^WT vs. Btg1^KO p < 0.0001; Btg1^WT vs. Ptch1^+/−/Btg1^WT p = 0.0200; Btg1^WT vs. Ptch1^+/−/Btg1^KO p = 0.0117; Mann–Whitney U-test; Figures 5A,C). Ptch1^+/−/Btg1^WT and Ptch1^+/−/Btg1^KO mice presented equivalent percentages of NeuN⁺ GCPs, significantly higher than that in Btg1^KO mice (Ptch1^+/−/Btg1^WT vs. Btg1^KO p = 0.0048; Ptch1^+/−/Btg1^KO vs. Btg1^KO p = 0.0154; Mann–Whitney U-test; Figures 5A,C). All this indicates that the ablation of Btg1, although associated to a decrease of differentiation, did not interfere with the inhibition of differentiation exerted by the Ptch1^+/− genotype.

Then, we analyzed the survival of GCPs by detection of cleaved caspase-3⁺ cells in the EGL. At P7, Btg1 knockout significantly increased the percentage of GCPs undergoing apoptosis either in Ptch1 wild-type mice (Btg1^KO vs. Btg1^WT p = 0.0011 and 55% increase; Mann–Whitney U-test; Figures 6A,B), as expected (12), or in Ptch1^+/− mice (Ptch1^+/−/Btg1^KO vs. Btg1^WT p < 0.0001 and 122% increase; Mann–Whitney U-test; Figures 6A,B). Interestingly, although the ablation of one Ptch1 allele alone did not affect the apoptosis of the GCPs (Ptch1^+/−/Btg1^WT vs. Btg1^WT p = 0.3723; Mann–Whitney U-test; Figures 6A,B), the combined genetic deletion of Btg1 gene significantly augmented the percentage of caspase-3⁺ GCPs (Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT p < 0.0001; Mann–Whitney U-test; Figures 6A,B), far above the level induced by Btg1 deletion alone (Ptch1^+/−/Btg1^KO vs. Btg1^KO p = 0.0002; Mann–Whitney U-test; Figures 6A,B).

Overall, this indicates that during cerebellar development, the ablation of Btg1 in Ptch1 heterozygous mice does not influence the proliferation and differentiation processes, which are only dependent on the constitutive activation of the Shh pathway, but significantly increases the percentage of caspase-3-positive GCPs, suggesting the existence of a synergy between the Btg1 and Shh pathways in the apoptotic process.

To investigate the molecular mechanisms of increased apoptosis following the genetic ablation of Btg1, we first analyzed by qRT-PCR the expression levels of three components of the apoptotic pathway, namely, B-cell lymphoma 2 (Bcl-2), which is an antiapoptotic protein, p53 upregulated modulator of apoptosis (Puma), and Bcl-2-associated X protein (Bax), identified as apoptotic inducers. No significant changes of these mRNAs were detected in Ptch1^+/−/Btg1^KO tumors with respect to Ptch1^+/−/Btg1^WT tumors (data not shown). Very recently, the involvement of Prmt1 in the apoptosis of tumor cells has been demonstrated, although with opposite roles in different types of cancer (59, 60). Since Prmt1 is a known molecular partner of Btg1, we investigated the Prmt1 level in the MBs and we observed a significant increase of its expression in Ptch1^+/−/Btg1^KO tumors relative to Ptch1^+/−/Btg1^WT tumors (54.4% increase, p = 0.0021; Student's t-test; Figure 7A). Moreover, we analyzed the mRNA level of cyclin D1, target gene of Btg1, and we found that in Ptch1^+/−/Btg1^KO MBs the cyclin D1 expression was significantly higher than in Ptch1^+/−/Btg1^WT MBs (p = 0.0117; Student's t-test; Figure 7A). In our experience, this increase should be associated to great changes in cellular proliferation (12); however, Ptch1^+/−/Btg1^KO MBs show a proliferation index equivalent to that of Ptch1^+/−/Btg1^WT MBs (see Figures 1A,B). Yet, we also observed an increased p53 expression within the tumors lacking the Btg1 gene (Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT p = 0.0070; Student's t-test; Figure 7A), which, given the proapoptotic action of p53, may explain the lack of increase of proliferation in MBs.

We have previously demonstrated that in the neurogenic niches, Btg1 controls the quiescence and self-renewal of the stem cells (11, 29), then by qRT-PCR, we also evaluated in the MBs the mRNA levels of some markers of brain tumor stem cells, such as nestin, glial fibrillary acidic protein (GFAP), CD133, and CD15 (38, 61–68). In Ptch1^+/−/Btg1^KO MBs, we observed an increase of the levels of nestin and GFAP mRNAs relative to the Ptch1^+/−/Btg1^WT MBs, with the GFAP expression attaining statistical significance (p = 0.0467; Student's t-test; Figure 7A). Furthermore, we recorded a statistically significant increase of CD15 expression in Ptch1^+/−/Btg1^KO tumors with respect to Ptch1^+/−/Btg1^WT tumors (p = 0.0057; Student's t-test; Figure 7A), while no changes in mRNA level of CD133 were detected between the tumors of the two different genotypes under study (p = 0.2455; Student's t-test; Figure 7A).

Finally, no significant differences were observed for the main downstream Shh-responsive genes, namely, Gli1, Gli2, Ptch1, and N-Myc, in Ptch1^+/−/Btg1^KO MBs relative to Ptch1^+/−/Btg1^WT MBs (data not shown), indicating that the deletion of Btg1 does not interfere with Shh signaling in cerebellar tumors.

Also, at the protein level, we found a significant increase of Prmt1 in Ptch1^+/−/Btg1^KO MBs with respect to Ptch1^+/−/Btg1^WT MBs (p = 0.028; Student's t-test; Figure 7B). Moreover, we observed an increase of the level of cyclin D1 protein in Ptch1^+/−/Btg1^KO tumors relative to the Ptch1^+/−/Btg1^WT tumors, without attaining statistical significance (p = 0.089; Student's t-test; Figure 7B).

It is known that the methylation of the proapoptotic BAD protein at Arg-94 and Arg-96 catalyzed by Prmt1 inhibits Akt-mediated BAD phosphorylation at Ser-99, thus counteracting the Akt-dependent survival signaling and inducing the apoptotic process with concomitant activation of caspase-3 (44). Then, we analyzed the protein level of BAD phosphorylated at Ser-99 (pBAD) in MBs of Ptch1^+/−/Btg1^KO and Ptch1^+/−/Btg1^WT mice and we observed a decrease, although not significant, in BAD phosphorylation in tumors lacking the Btg1 gene (p = 0.397; Student's t-test; Figure 7B). This suggests that the Btg1 ablation could increase the apoptotic process by inducing the Prmt1-dependent methylation of the BAD protein.

We also analyzed the expression of GFAP protein in the MBs of both genotypes by measuring the area of GFAP staining detected by immunohistochemistry (see Supplementary Material); consistently with the mRNA expression, we observed a significant increase of GFAP protein in Ptch1^+/−/Btg1^KO vs. Ptch1^+/−/Btg1^WT MBs (p = 0.0074; Mann–Whitney U-test; Figure S1).

The increase of expression of CD15, a marker of brain tumor stem cells, detected by qRT-PCR in the Ptch1^+/−/Btg1^KO MBs compared with Ptch1^+/−/Btg1^WT MBs (Figure 7A), was confirmed at protein level by immunohistochemical analysis of tumor sections from MBs of the two different genotypes under study (Figures 7C,D). In MBs lacking the Btg1 gene, we observed a two-fold increase of the percentage of CD15-positive cells with respect to Ptch1^+/−/Btg1^WT tumors (p < 0.0001; Mann–Whitney U-test; Figures 7C,D), with a greater tendency of CD15⁺ cells to aggregate into clusters.

Noteworthy, by immunohistochemical analysis, we also evaluated in tumors and preneoplastic lesions of Ptch1^+/−/Btg1^WT and Ptch1^+/−/Btg1^KO mice the expression of Sox2, a marker for tumor stem cells in Shh-induced MB (69), without finding any difference (data not shown).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.