ERK Phosphorylation Regulates the Aml1/Runx1 Splice Variants and the TRP Channels Expression during the Differentiation of Glioma Stem Cell Lines

The identification of cancer stem cells in brain tumors paved the way for new therapeutic approaches. Recently, a role for the transcriptional factor Runx1/Aml1 and the downstream ion channel genes in brain cancer development and progression has been suggested. This study aimed to explore the expression and the role of Runx1/Aml1, its Aml1b and Aml1c splice variants and the downstream TRPA1 and TRPV1 ion channels in undifferentiated and day-14 differentiated neural stem cells (NSCs and D-NSCs) and glioblastoma stem cells (GSCs and D-GSCs) lines with different proneural (PN) or mesenchymal (MES) phenotype. Gene and protein expression were evaluated by qRT-PCR, cytofluorimetric, western blot and confocal microscopy analyses. Moreover, by western blot, we observed that ERK phosphorylation enhances the Aml1b and Aml1c protein expression during glioma differentiation. Furthermore, the agonists of TRPA1 and TRPV1 channels stimulated apoptosis/necrosis in GSCs and D-GSCs as evaluated by Annexin V and PI staining and cytofluorimetric analysis. Finally, by qRT-PCR, the modulation of Wnt/β catenin, FGF, and TGFβ/SMAD signaling pathways in PN- and MES-GSCs was reported. Overall, our results provide new evidence regarding Runx1/Aml1 isoform overexpression and modulation in TRP channel expression during gliomagenesis, thus offering new directions for glioblastoma therapy.


Introduction
Malignant gliomas remain one of the more deadly human brain tumors with a poor prognosis despite years of research in anti-tumor therapeutic strategies. A hallmark characteristic of gliomas is their molecular and cellular heterogeneity, which is considered the reason for their high malignancy, recurrence and resistance to therapy [1]. Neoplastic transformation of differentiated glial cells was the most accepted hypothesis to explain the origin of gliomas for many years [1,2]. However, recent reports support the existence of a stem cell-derived origin for different types of cancers, including brain tumors [3].
Neural stem cells (NSCs) and glial progenitor cells are present in multiple regions of the adult brain (e.g., the dentate gyrus, within the hippocampus and the subcortical white NSCs (NSC#1, NSC#2), proneural-like (PN-GSC#1, #23C, #28, #68 and #70) and mesenchymal-like (MES-GSC#30 and #83) GSC lines, previously characterized in [14], were isolated from surgical samples of seven adult patients with primitive brain tumor undergoing complete or partial surgical resection at the Institute of Neurosurgery, Catholic University School of Medicine, in Rome, Italy. Patients were eligible for the study if a diagnosis of glioblastoma multiforme was established histologically according to the WHO classification [29]. Informed consent was obtained before surgery according to the Ethical Committee of Catholic University School of Medicine. GSC cultures were established from tumor specimens through mechanical dissociation and culturing in DMEM/F12 serumfree medium containing 2 mM glutamine, 0.6% glucose, 9.6 g/mL putrescine, 6.3 ng/mL progesterone, 5.2 ng/mL sodium selenite, 0.025 mg/ml insulin, and 0.1 mg/mL transferrin sodium salt (Sigma-Aldrich, St. Louis, MO, USA), supplemented with EGF and bFGF. GSC lines grown as floating spheres in serum-free medium supplemented with mitogens showed an undifferentiated state, as indicated by their rounded morphology, high nuclear/cytoplasm ratio. For differentiation, GSC lines were grown in medium supplemented with 5% fetal bovine serum (FBS, EuroClone, Milan, Italy). Human GSC lines were authenticated by short tandem repeat (STR) profiling according to the American National Standards Institute/American Type Culture Collection Standard ASN-0002-2011.12 using the Cell line Integrated Molecular Authentification database (CLIMA), 13 and Cellosaurus STR database (CLASTR) of the Cellosaurus database (ExPASy) at the IRCC Ospedale Policlinico San Martino, Interlab Cell Line Collection (ICLC), Biological Resource Center (CRB-HSM), Genova, Italy [30]. All cell lines were isocitrate dehydrogenase (IDH) 1/2 wild-type.

Confocal Laser Scanning Microscopy Analysis
For immunophenotyping, GSC and D-GSC lines were treated and fixed as previously published [31]. Then, cells were incubated with 5% of bovine serum albumin (BSA, Sigma-Aldrich) and 0.1% of Tween-20 in PBS for 1 h at room temperature and then stained with anti-human TRPV1 Ab (1:1000) overnight at 4 • C. Then, samples were washed with 0.3% of Triton X-100 in PBS, incubated with Alexa Fluor 594-conjugated secondary Ab for 1 h at 37 • C. In co-labeling experiments, cells were also stained with anti-human TRPA1 Ab (1:50) overnight at 4 • C. Finally, samples were washed with 0.3% of Triton X-100 in PBS, incubated with Alexa Fluor 488-conjugated secondary Ab for 1 h at 37 • C. Cells were also co-labeled with anti-β III -tubulin and GFAP Abs.
As described above, cells were incubated with 5% of BSA and 0.1% of Tween-20 in PBS for 1 h at room temperature and then stained with anti-β III -tubulin Ab (1:50) overnight at 4 • C. Then, samples were washed with 0.3% of Triton X-100 in PBS, incubated with Alexa Fluor 488-conjugated secondary Ab for 1 h at 37 • C. Cells were also stained with anti-GFAP Ab (1:500) overnight at 4 • C. Finally, samples were washed with 0.3% of Triton X-100 in PBS, incubated with Alexa Fluor 594-conjugated secondary Ab for 1 h at 37 • C. Then, nuclei were stained with 4 ,6-diamidino-2-phenylindole (DAPI, BioRad). Slides were then analyzed with C2 Plus confocal laser scanning microscope (Nikon Instruments, Firenze, Italy). Optimized emission detection bandwidths were configured by Zeiss Zen control software.

Gene Expression Analysis
Total RNA was extracted with the RNeasy Mini Kit (Qiagen, Milan, Italy), and cDNA was synthesized using the High-Capacity cDNA Archive Kit (Applied Biosystems, Foster City, PA, USA) according to the manufacturer's instructions. Quantitative RT-PCR (qRT-PCR) was performed by using the IQ5 Multicolor real-time PCR detection system (BioRad). The reaction mixture contained the Advanced Universal SYBRGreen Supermix (BioRad). Human TRPV1, TRPA1, Aml1b, Aml1c and GAPDH RT 2 qPCR Primer assay (Qiagen) were used. The PCR parameters were 10 min at 95 • C followed by 40 cycles at 95 • C for 15 s and 60 • C for 40 s. All samples were assayed in triplicate in the same plate. The relative amount of target mRNA was calculated by the 2 −∆∆Ct method, using GAPDH as a housekeeping gene.

qRT-PCR and PCR Array
The Human Stem Cell Signaling Pathway Finder PCR Arrays with related reagents were purchased from SABiosciences (Frederick, MD, USA). Total RNA was extracted from GSC and D-GSC lines. Two µg of total RNA from each sample were subjected to reverse transcription in a total volume of 20 µL using the Reaction Ready First Strand cDNA (SABiosciences). cDNAs were analyzed by qRT-PCR performed using a IQ5 Multicolor Real-Time PCR Detection system, the Super Array's RT 2 real-time SYBR Green PCR Master Mix and the relative array, according to manufacturer's instructions. Measurement of two housekeeping genes (GAPDH; Ribosomal protein, large, P0; RPLP0) on the samples was used to normalize mRNA content. Data acquisition was performed using the web-based integrated PCR Array Data Analysis Tem-235 plate provided by SABiosciences.

Statistical Analysis
The data presented represent the mean and standard deviation (SD) of at least three independent experiments. The statistical significance was determined by analysis of variance (ANOVA) or Student's t-test; * p < 0.01.

Expression of the Aml1 Spliced Variants and TRP Channels in NSC and D-NSC
The expression of Aml1a, Aml1b and Aml1c isoforms of Aml1/Runx1 were investigated in the NSC#1 and NSC#2 by qRT-PCR. As shown in Figure 1A, mRNA for Aml1a, Aml1b and Aml1c was evidenced in the NSCs; the differentiation of NSCs for 14-days (D-NSCs) strongly increased the expression of all Aml1 variants, compared to NSCs, with Aml1b showing the higher fold increase (62-fold).
The transcriptional factor Runx1/Aml1 regulates the expression of many ion channels and receptors, including some members of the TRP thermal and chemical receptors [21]. Thus, the expression of the TRPA1 and TRPV1 channels, belonging to the anchyrin and vanilloid TRP subfamilies, respectively, was evaluated in NSC#1 and NSC#2 and D-NSC#1 and D-NSC#2. Transcripts for TRPV1 and TRPA1 were found in both NSC#1 and NSC#2 and their expression markedly increased in D-NSC#1 and D-NSC#2, compared to NSC#1 and NSC#2, respectively, with TRPA1 showing the higher increase (164-fold) ( Figure 1B).
Multicolor Real-Time PCR Detection system, the Super Array's RT 2 real-time SYBR Green PCR Master Mix and the relative array, according to manufacturer's instructions. Measurement of two housekeeping genes (GAPDH; Ribosomal protein, large, P0; RPLP0) on the samples was used to normalize mRNA content. Data acquisition was performed using the web-based integrated PCR Array Data Analysis Tem-235 plate provided by SABiosciences.

Statistical Analysis
The data presented represent the mean and standard deviation (SD) of at least three independent experiments. The statistical significance was determined by analysis of variance (ANOVA) or Student's t-test; * p < 0.01.

Expression of the Aml1 Spliced Variants and TRP Channels in NSC and D-NSC
The expression of Aml1a, Aml1b and Aml1c isoforms of Aml1/Runx1 were investigated in the NSC#1 and NSC#2 by qRT-PCR. As shown in Figure 1A, mRNA for Aml1a, Aml1b and Aml1c was evidenced in the NSCs; the differentiation of NSCs for 14-days (D-NSCs) strongly increased the expression of all Aml1 variants, compared to NSCs, with Aml1b showing the higher fold increase (62-fold).
The transcriptional factor Runx1/Aml1 regulates the expression of many ion channels and receptors, including some members of the TRP thermal and chemical receptors [21]. Thus, the expression of the TRPA1 and TRPV1 channels, belonging to the anchyrin and vanilloid TRP subfamilies, respectively, was evaluated in NSC#1 and NSC#2 and D-NSC#1 and D-NSC#2. Transcripts for TRPV1 and TRPA1 were found in both NSC#1 and NSC#2 and their expression markedly increased in D-NSC#1 and D-NSC#2, compared to NSC#1 and NSC#2, respectively, with TRPA1 showing the higher increase (164-fold) ( Figure 1B).

Overexpression of Aml1b and Aml1c Variants and TRP Channels in Distinct PN-and MES-like GSCs Compared to NSCs
We previously reported the expression of Aml1a in GSCs and the role of the TRPV2 agonist, CBD, in mediating its expression and functions [28].

Overexpression of Aml1b and Aml1c Variants and TRP Channels in Distinct PN-and MES-like GSCs Compared to NSCs
We previously reported the expression of Aml1a in GSCs and the role of the TRPV2 agonist, CBD, in mediating its expression and functions [28].
Herein, the expression of Aml1b and Aml1c splice variants was analyzed at mRNA level by qRT-PCR in seven different GSC lines, with PN-(GSC#1, #23C, #28, #68 and #70), and MES-like phenotype (GSC#30 and GSC#83) [14,15,32]. An Aml1b and Aml1c mRNA upregulation, although at different levels, was observed in all the GSCs analyzed, compared with NSCs. The Aml1b, with respect to the Aml1c isoform, showed the higher mRNA increase in MES-like GSC#83 and GSC#30 lines, respectively, compared with NSCs ( Table 1). The higher Aml1c mRNA expression was evidenced in MES-like GSC#83, whereas lower mRNA expression was observed in the PN-like GSC#68 line, compared with NSCs (Table 1). In regard to TRP channels, the TRPA1 mRNA expression was increased in all the GSC lines, independently of the phenotype, with the exception of the GSC#23C cells expressing lower levels ( Table 2). Higher TRPA1 values, compared to NSCs, were found in MES-like GSC#83; the TRPV1 expression was mainly increased in GSC#83 and strongly downregulated in PN-GSC lines, except in GSC#70 cell line, compared to NSCs (Table 2). Overall, our results showed a marked overexpression of Aml1b and Aml1c variants, compared with NSCs in all the PN-and MES-GSC lines; moreover, a marked increase of TRPA1 and TRPV1 mRNA was demonstrated in MES-GSCs as well as a marked reduction of TRPV1 expression in PN-GSCs with respect to NSCs.

Differentiation of GSC Lines Increases the Aml1b/Aml1c and TRP Ion Channel mRNA Levels
The expression of Aml1b and Aml1c, as well as TRPA1 and TRPV1, was investigated during GSCs differentiation. The differentiation of PN-and MES-GSCs for 14 days (D-GSCs) resulted in the cells showing glial morphology ( Figure S1 in Supplementary Materials) [28,31], and, as shown in Table 3, the Aml1b and Aml1c mRNA levels were upregulated in all D-GSCs compared with the respective GSCs with the main increase in MES-like D-GSC#30 line (Table 3). This enhancement was even more evident when comparing D-GSCs with D-NSCs (Table S1). In addition, TRPA1 mRNA expression was mainly increased in the MES-like D-GSC#83 line and was reduced in D-GSC#28, #68 and #70 PN-GSC lines, compared to GSCs. Conversely, TRPV1 mRNA expression decreased in both PN-and MES-like D-GSCs (Table 4). This enhancement was even more evident when comparing D-GSCs with D-NSCs (Table S2).

TRPA1 and TRPV1 Channel Expression in PN-and MES-like GSCs and D-GSC Lines
Among the seven different GSC lines, we studied the GSC#1 and GSC#83 lines, as representative cell lines previously well characterized for the expression of a proneural and mesenchymal phenotype [14,32] (Figure S1). In these cell lines, the expression of TRPA1 and TRPV1 proteins was evaluated by confocal microscopy and western blot analyses. Data showed that TRP channels are expressed in both cell lines and, partially colocalized in the plasma membrane in GSCs and during the differentiation (Figure 2).

ERK Phosphorylation Increases the Aml1b and Aml1c Protein Expression and Induces Changes in TRP Channel Expression during GSC Differentiation
The ERK pathway has been found to regulate proliferation and differentiation in GBM [33,34] and GSCs [35]. Thus, the level of pERK and total ERK was evaluated in GSC#1/D-GSC#1 and GSC#83/D-GSC#83 lines at different times of differentiation, by western blot, using an anti-pERK and anti-ERK mAbs. ERK was basally phosphorylated in GSC#1 and its phosphorylation level increases at seven days of differentiation, then declines at later time-point (14 days) ( Figure 3); no ERK phosphorylation was evident at basal level in GSC#83; ERK phosphorylation was induced at day 7 and progressively increased until day 14 of differentiation ( Figure 3). To determine ERK involvement, the GSC#1 and GSC#83 were treated for 24 h at different times of differentiation with the MEK1 inhibitor, PD98059 (50 µM), able to completely inhibit ERK phosphorylation in both GSC lines at any time tested (Figure 3). analyses. Data showed that TRP channels are expressed in both cell lines and, partially colocalized in the plasma membrane in GSCs and during the differentiation (Figure 2). were fixed, permeabilized, and stained with anti-human TRPA1 and TRPV1 Abs followed by Alexa Fluor-488andAlexa Fluor-594 secondary Abs, respectively. 40,6-diamidino-2-phenylindole (DAPI) was used to counterstain nuclei. Magnification 60×.

ERK Phosphorylation Increases the Aml1b and Aml1c Protein Expression and Induces Changes in TRP Channel Expression during GSC Differentiation
The ERK pathway has been found to regulate proliferation and differentiation in GBM [33,34] and GSCs [35]. Thus, the level of pERK and total ERK was evaluated in GSC#1/D-GSC#1 and GSC#83/D-GSC#83 lines at different times of differentiation, by western blot, using an anti-pERK and anti-ERK mAbs. ERK was basally phosphorylated in GSC#1 and its phosphorylation level increases at seven days of differentiation, then declines at later time-point (14 days) ( Figure 3); no ERK phosphorylation was evident at basal level in GSC#83; ERK phosphorylation was induced at day 7 and progressively increased until day 14 of differentiation ( Figure 3). To determine ERK involvement, the GSC#1 and GSC#83 were treated for 24 h at different times of differentiation with the MEK1 inhibitor, PD98059 (50 μM), able to completely inhibit ERK phosphorylation in both GSC lines at any time tested (Figure 3). were fixed, permeabilized, and stained with anti-human TRPA1 and TRPV1 Abs followed by Alexa Fluor-488 and Alexa Fluor-594 secondary Abs, respectively. 40,6-diamidino-2-phenylindole (DAPI) was used to counterstain nuclei. Magnification 60×.
Thus, we evaluated if ERK activation was capable of affecting the Aml1b and Aml1c protein expression during GSC#1 and GSC#83 differentiation by using an anti-Aml1 mAb, which recognizes all the Aml1 splice variants. Parallelly to ERK phosphorylation kinetic, the Aml1b and Aml1c protein levels increase at day 7 of differentiation (mainly in GSC#83) and thereafter declines at day 14 ( Figure 4). Finally, the enhancement of Aml1b and Aml1c protein levels at seven days of differentiation was reverted by the treatment with the PD98059 inhibitor in both GSC#1 and GSC#83 lines ( Figure 4) In addition, in the same manner of ERK phosphorylation, TRPV1 mRNA levels increased at 7 days in both GSCs during differentiation, and this upregulation was reduced, mainly in GSC#83, by the PD98059 inhibitor. On the contrary, PD98059 reduces the TRPA1 upregulation at 14 days of differentiation in PN-like D-GSC#1, whereas increased TRPA1 levels were found in MES-like GSC#83 ( Figure 5), suggesting that in this GSC, the enhanced TRPA1 protein expression is ERK-independent.  Thus, we evaluated if ERK activation was capable of affecting the Aml1b and Aml1c protein expression during GSC#1 and GSC#83 differentiation by using an anti-Aml1 mAb, which recognizes all the Aml1 splice variants. Parallelly to ERK phosphorylation kinetic, the Aml1b and Aml1c protein levels increase at day 7 of differentiation (mainly in GSC#83) and thereafter declines at day 14 ( Figure 4). Finally, the enhancement of Aml1b and Aml1c protein levels at seven days of differentiation was reverted by the treatment with the PD98059 inhibitor in both GSC#1 and GSC#83 lines (Figure 4)   Thus, we evaluated if ERK activation was capable of affecting the Aml1b and Aml1c protein expression during GSC#1 and GSC#83 differentiation by using an anti-Aml1 mAb, which recognizes all the Aml1 splice variants. Parallelly to ERK phosphorylation kinetic, the Aml1b and Aml1c protein levels increase at day 7 of differentiation (mainly in GSC#83) and thereafter declines at day 14 ( Figure 4). Finally, the enhancement of Aml1b and Aml1c protein levels at seven days of differentiation was reverted by the treatment with the PD98059 inhibitor in both GSC#1 and GSC#83 lines (Figure 4)  In addition, in the same manner of ERK phosphorylation, TRPV1 mRNA levels increased at 7 days in both GSCs during differentiation, and this upregulation was reduced, mainly in GSC#83, by the PD98059 inhibitor. On the contrary, PD98059 reduces the TRPA1 upregulation at 14 days of differentiation in PN-like D-GSC#1, whereas increased TRPA1 levels were found in MES-like GSC#83 (Figure 5), suggesting that in this GSC, the enhanced TRPA1 protein expression is ERK-independent.

.Specific TRPA1 and TRPV1 Agonists Stimulate Apoptotic/Necrotic Cell Death in PN-and MES-GSCs and D-GSCs
We then analyzed whether the increased TRPA1 and TRPV1 expression at day 7 of differentiation corresponded to an increased functionality of these receptors. Thus, we determined the sensitivity of GSCs to the TRP agonists-induced cell death. GSC#1 and GSC#83 cells were treated with different doses of the TRPA1 agonist, trans-Cinnamaldehyde (CINN) [36] and TRPV1 agonist, capsaicin (CPS) [37] for 24 h. The percentage of dead cells was evaluated by staining the GSCs with PI or FITC-conjugated Annexin V and cytofluorimetric analysis. We found that PN-like GSC#1 line was sensitive to high CINN and CPS doses, with 21% and 12% of apoptotic Annexin V-positive cells, respectively. Seven days differentiation increases the GSC#1 sensitivity, as shown by the enhancement in Annexin V-positive cells both at low and high CINN and CPS doses ( Figure 6A). No PI-positive cells were evidenced in CINN/CPS-treated GSC#1 and D-GSC#1 (data not shown). Regarding the GSC#83 line, these cells were completely resistant to the CPS cytotoxic effects and only 10% of 100 μM CINN-treated cells were PI-positive necrotic cells. Differentiation increases the sensitivity to CPS treatment (20% PI-positive cells at 200 μM of CPS) and for both doses of CINN (8% and 33% of PI-positive necrotic cells upon 50 and 100 μM treatment, respectively) ( Figure 6B). No Annexin V-positive cells were evidenced in CPS or CINN-treated GSC#83 and D-GSC#83 (data not shown).
Overall, GSC#1 are more sensitive compared to GSC#83 to CINN and CPS-induced cytotoxic effects. GSC differentiation increases the sensitivity to CINN or CPS treatment mainly in PN-like GSC line.

Specific TRPA1 and TRPV1 Agonists Stimulate Apoptotic/Necrotic Cell Death in PN-and MES-GSCs and D-GSCs
We then analyzed whether the increased TRPA1 and TRPV1 expression at day 7 of differentiation corresponded to an increased functionality of these receptors. Thus, we determined the sensitivity of GSCs to the TRP agonists-induced cell death. GSC#1 and GSC#83 cells were treated with different doses of the TRPA1 agonist, trans-Cinnamaldehyde (CINN) [36] and TRPV1 agonist, capsaicin (CPS) [37] for 24 h. The percentage of dead cells was evaluated by staining the GSCs with PI or FITC-conjugated Annexin V and cytofluorimetric analysis. We found that PN-like GSC#1 line was sensitive to high CINN and CPS doses, with 21% and 12% of apoptotic Annexin V-positive cells, respectively. Seven days differentiation increases the GSC#1 sensitivity, as shown by the enhancement in Annexin V-positive cells both at low and high CINN and CPS doses ( Figure 6A). No PI-positive cells were evidenced in CINN/CPS-treated GSC#1 and D-GSC#1 (data not shown). Regarding the GSC#83 line, these cells were completely resistant to the CPS cytotoxic effects and only 10% of 100 µM CINN-treated cells were PI-positive necrotic cells. Differentiation increases the sensitivity to CPS treatment (20% PI-positive cells at 200 µM of CPS) and for both doses of CINN (8% and 33% of PI-positive necrotic cells upon 50 and 100 µM treatment, respectively) ( Figure 6B). No Annexin V-positive cells were evidenced in CPS or CINN-treated GSC#83 and D-GSC#83 (data not shown).
Overall, GSC#1 are more sensitive compared to GSC#83 to CINN and CPS-induced cytotoxic effects. GSC differentiation increases the sensitivity to CINN or CPS treatment mainly in PN-like GSC line.

NGF Levels Increase during GSC Differentiation
It was demonstrated that the effect of ERK phosphorylation depends on activation of the NGF pathway [38,39]. Thus, NGF was evaluated at mRNA levels in GSCs and D-GSCs of both PN and MES phenotypes at different times during differentiation (3, 5, 7, 11 and 14 days) by qRT-PCR. We found that the mRNA levels of NGF progressively increase until day 7 and decline thereafter, although with different kinetic in both GSC#1 and GSC#83 (Figure 7).

NGF Levels Increase during GSC Differentiation
It was demonstrated that the effect of ERK phosphorylation depends on activation of the NGF pathway [38,39]. Thus, NGF was evaluated at mRNA levels in GSCs and D-GSCs of both PN and MES phenotypes at different times during differentiation (3, 5, 7, 11 and 14 days) by qRT-PCR. We found that the mRNA levels of NGF progressively increase until day 7 and decline thereafter, although with different kinetic in both GSC#1 and GSC#83 (Figure 7).

NGF Levels Increase during GSC Differentiation
It was demonstrated that the effect of ERK phosphorylation depends on activation of the NGF pathway [38,39]. Thus, NGF was evaluated at mRNA levels in GSCs and D-GSCs of both PN and MES phenotypes at different times during differentiation (3, 5, 7, 11 and 14 days) by qRT-PCR. We found that the mRNA levels of NGF progressively increase until day 7 and decline thereafter, although with different kinetic in both GSC#1 and GSC#83 (Figure 7).  Figure 7. NGF modulation during GSC differentiation. Relative NGF mRNA expression during GSCs differentiation was evaluated by qRT-PCR. NGF mRNA levels were normalized for GAPDH expression and were expressed as relative fold with respect to the corresponding GSC lines used as calibrator. * p < 0.001 vs. undifferentiated NSCs; # p < 0.01 vs. 3, 5, 11, 14 days of differentiation.

GSC Differentiation Is Accompanied by Changes of Gene Pathways Involved in GSC Differentiation
The modulation of Aml1b/c mRNA and protein expression during the GSC differentiation is associated with changes of different genes for intracellular signals. Thus, the expression of 84 genes involved in signal transduction pathways important for stem cell maintenance, proliferation and/or differentiation, in PN-like GSC#1and MES-like GSC#83lines in proliferative condition and at seven days of differentiation, was evaluated by a Stem Cell Signaling Pathway Finder PCR array.
The results showed that 19 (10 upregulated and nine downregulated) and 14 (seven upregulated and seven downregulated) genes belonging to the TGFβ/Smad (BMP1RB), CDX2, SMAD2, SMAD3, SMAD9, TGFBR2, FGF (FGFR2), NFAT (NFAT5), Wnt/β-catenin (FZD1, FDZ3 and FDZ9) and NOTCH (NOTCH-3 and -4) were modulated in GSC#1 and GSC#83 at day 7 of differentiation, respectively (Tables 5 and 6). These findings suggest that the modulation of Aml1b/Aml1c in GSCs induced by ERK activation is a part of a differentiative network that controls the genetic transcriptional program of GSC differentiation. 2.67 ± 0.10 STAT3 −5.06 ± 0.50 Stem Cell Signaling Pathway Finder PCR array in mRNA samples extracted from GSC#1 and D-GSC#1 after 7 days of culture. Values represent fold differences of individual gene expression in D-GSC#1 compared to GSC#1. The expression levels were normalized to the average Ct value of two housekeeping genes (GAPDH and RPLP0) and calculated by the ∆∆Ct method. Genes, whose expression is, at least, 2.00-fold up/downregulated, are shown. 4.70 ± 0.10 Stem Cell Signaling Pathway Finder PCR array in mRNA samples extracted from GSC#83 and D-GSC#83 after 7 days of culture. Values represent fold differences of individual gene expression in D-GSC#83 compared to GSC#83. The expression levels were normalized to the average Ct value of two housekeeping genes (GAPDH and RPLP0) and calculated by the 2 −∆∆Ct method. Genes, whose expression was, at least, 2.00-fold up/downregulated, are shown.

Discussion
Malignant brain tumors, including GBM, are known for their high degree of cellular heterogeneity, invasiveness, aggressiveness and lethality. GSCs are crucial for the malignancy of GBM and likely represent the consequence of transformation of the normal NSC compartment [40]. The GSCs respond to differentiation signals both at transcriptional and post-transcriptional levels. Among the transcriptional factors, Aml1/Runx1 is thought to be involved in the control of the balance between cell proliferation and differentiation [41], and the cellular decision to continue proliferating or to differentiate depends on the relative expression levels of the Aml1 variant forms [42]. We previously reported that Aml1a isoform is involved in GSCs proliferation, stemness and pluripotency, and differentiation [28].
Herein, by qRT-PCR analyses, we found that NSCs express the Aml1 splice variants of Aml1/Runx1 and the downstream TRP ion channel genes, TRPA1 and TRPV1, with Aml1c and TRPV1 showing the higher expression. Serum-induced differentiation enhances the Aml1a, Aml1b and Aml1c and TRPV1 mRNA expression, whereas TRPA1 was reduced in D-NSCs. A significant amount of data on Aml1/Runx1 mutations and cancers and the relationship between Aml1/Runx1 overexpression and poor prognosis have been reported [43,44]. An Aml1/Runx1 protein accumulation, correlating to tumor aggressiveness has been reported in astrocytomas [45]. In addition, a marked accumulation of wild-type Aml1b variant, and of a novel Aml1b isoform lacking the exon 6 in ovarian cancer patients has been reported, which correlates with poor patient outcomes [46].
In this paper we analyzed seven different GSC lines having a PN-or MES-like phenotype. We found a strong overexpression of both Aml1b and Aml1c variants in all the PN-and MES-GSC lines compared to NSC. The increase of TRPA1 and TRPV1 was more evident in MES-GSCs. Instead, the reduction of TRPV1 mRNA expression was evidenced mainly in PN-GSCs, compared to NSCs. Serum-induced GSC differentiation enhanced the Aml1b and Aml1c levels in D-GSC lines, compared to GSCs. Comparative analysis of Aml1b and Aml1c mRNA expression in D-GSCs vs. GSCs or D-NSC confirms the substantial increase of Aml1b/Aml1c levels, more evident in the GSC#30 and GSC#83 MES lines, and a reduction of Aml1c in D-GSC#68 and #70 PN-GSC lines. The analysis of the expression of Aml1b and Aml1c variants at protein levels in GSCs and D-GSCs evaluated by western blot analyses evidenced that both Aml1b/c proteins accumulate at day 7 and are sustained upon 14 days of differentiation.
The differentiation program of GSCs is accompanied by the change of TRPA1 and TRPV1 expression compared to D-NSC. High levels of TRPA1 were found in NSCs and their expression decreased in D-NSCs; on the contrary, increased TRPA1 mRNA expression in MES-D-GSCs, whereas similarly to NSCs a reduction of TRPA1 and TRPV1 mRNA expression was reported in the majority of the D-GSCs.
TRPA1 is mainly expressed in primary sensory neurons, but it has been also described in non-excitable cells, including those derived from neural crest stem cells, such as glial cells and oligodendrocytes and astrocytes. In pain-sensitive nerve endings, TRPA1 senses the oxidative stress and transduces it into neural signals [47]; similarly, in GBM, oxidative stress activates TRPA1 channels [48]. Chemical stimulation of TRPA1 seems to activate stem cells [49] and at least in skeletal muscle, TRPA1 agonists stimulate migration and early stem cell differentiation [50]. Moreover, recently, a crosstalk between Runx1 and TGFβ signaling pathway regulating the TRPA1 and TRPV1 expression and an increased Ca 2+ activity was evidenced during megakaryocyte differentiation and silencing of Aml1/Runx1 markedly reduced the TRPA1 and TRPV1 mRNA expression [51]. In addition, in lung adenocarcinoma, direct binding of TRPA1 to FGFR2 resulted in a constitutive receptor activation, thereby promoting tumor progression [52].
Regarding TRPV1, it is expressed in NSCs and its level increases in D-NSCs. The MES-like GSCs express high TRPV1 mRNA levels, whereas a reduced TRPV1 mRNA levels was evidenced in PN-like GSC and D-GSC lines. It was already demonstrated that TRPV1 is expressed in neural precursor cells (NPCs) [53] and it plays a major role in NSC survival [54] and in retinoid acid-induced neuron differentiation [55]. TRPV1 loss promotes proliferation of NPCs in vivo and in vitro. Knockout of TRPV1 increases post-natally proliferating cells in stem cell niches (e.g., the dentate gyrus and SVZ). Under mild differentiating conditions in vitro, NPCs from TRPV1 knockout mice express stem cell genes and at less extend GFAP and β III -tubulin differentiation markers. However, no stemness reduction was evidenced in NPCs by TRPV1 loss. More importantly, NPCs migrate into high-grade astrocytomas to reduce glioma expansion and prolong survival by release of fatty acid ethanolamides by triggering a TRPV1-dependent astrocytoma death, via the activation of ATF3 of the ER stress pathway [56]. On this basis, we can hypothesize that TRPV1 loss in differentiated GSCs, which is different from high TRPV1 expression in D-NSC, represents an escape mechanism of tumor cells from death mediated by TRPV1 agonists. Moreover, the correlation between the expression of a most stable TRPV1 variant-3 (TRPV1v3) in GSCs and survival suggests an important role played by TRPV1 in glioma [57]. Similarly, we also found that the higher expression of TRPV1 and TRPA1 in D-GSCs is associated to CPS-and CINN-induced cell death, mainly in PN-like D-GSC line.
The transcriptional activity of Aml1/Runx1 is regulated by controlling the MAPKmediated phosphorylation of Aml1/Runx1. Activated ERK enters the nucleus and promotes the clustering of transcription factors, including Aml1/Runx1, to function. ERKdependent phosphorylation of Ser-249 and -266 disrupts the interaction of Runx1 with the mSin3A transcriptional corepressor and potentiates the transactivation and the transforming abilities of Aml1/Runx1 [58]. ERK signaling controls the neural genetic programs in cortical development and gliomagenesis [33,34,59]. ERK signaling does not promote neoplastic proliferation alone, but influences neural cell fate selection. On the other hand, ERK is commonly activated in glioma and glioneuronal tumor, raising the question of how this pathway differentially influences oncogenic cell fates. In NSCs, ERK phosphorylation inhibited differentiation [59]. Herein, we found that treatment of D-GSC lines with the MEK1 inhibitor PD98059 reverted the accumulation of Aml1b and Aml1c proteins at seven days of differentiation.
Similarly, TRPV1 protein levels increased at seven days in both the PN-and MESlike GSCs, in an ERK-dependent manner. In accordance, CPS-induced ERK activation is reversed by the TRPV1 antagonist, capsazepin or MEK inhibitor [60][61][62], and CPS triggered ERK phosphorylation in sensory neurons [63]. The increase of TRPA1 protein expression observed during PN-like GSC#1 was inhibited by PD98059 at 14 days of differentiation. Instead, in MES-like GSC#83, PD98059 did not revert the TRPA1 increase, suggesting that this event is ERK-independent. In this regard, a role for p38MAPK in TRPA1 mRNA increase in trigeminal ganglia and sensory neurons has been reported [64]; moreover, intrathecal administration of the p38MAPK inhibitor, SB203580 reduced the TRPA1 mRNA levels in DRGs [65].
NGF signaling is essential for the sustained expression of Runx1. Runx1 induction in olfactory neuroblastoma cell line JFEN is associated with TrkA expression, and this correlates with specific in vivo expression of Runx1 and TrkA in DRG nociceptive neurons [66]. A Runx1/Aml1 mRNA accumulation found in astrocytomas, correlates with tumor aggressiveness [45]. In developing neurons, binding of NGF to its receptor TrkA triggers receptor dimerization, autophosphorylation, and MAPK signaling pathway activation [67]. NGF has been reported to increase GBM cell growth via TrkA receptor phosphorylation [68] and ERK promotes the proliferation and inhibits neuronal differentiation of NSC [59]. Finally, NGF has been found to increase the TRPA1 mRNA expression in a p38MAPKdependent manner [64]. Overall, our data are in agreement with previously reported data. We found in both GSC#1 and GSPC#83 lines that NGF mRNA levels progressively increase until day 7, and decline thereafter. Thus, it seems possible that the binding of NGF to TrkA receptors triggers ERK phosphorylation in GSCs that stimulates the increase of Aml1b/Aml1c proteins.
Runx1 contributes to the acquisition of aggressive and invasive phenotype in MESand PN-like GSC phenotypes in a TGFβ-dependent manner [69,70]. In this regard, both PNand MES-like GSC differentiation modulates the TGFβ/Smad and Wnt/βFGF signaling pathways. Thus, increased anti-Mullerian hormone (AMHR2) and Bone morphogenetic protein (BMPR2 and BMPR1B) mRNA expression was evidenced in both D-GSCs at 7 days of differentiation compared to GSCs. AMHR2 is a potent regulator of TGFβ/BMP signaling; AMHR2 hetero-dimerizes with ACVR1, regulates the BMPR2-and BMPR1B-mediated differentiation signals and SMAD survival signaling [71]. In this regard, reduced SMAD3 and increased SMAD9 and SMAD2 mRNA expression were evidenced in GSC#83 and GSC#1, respectively. The caudal-type homebox 2 (CDX2) gene was increased in both GSCs. CDX2 promotes the malignant behavior of glioma cells; CDX2 knockdown suppresses glioma cell proliferation, migration and invasion and facilitates the apoptosis of glioma cells [72].
The notch pathway seems to be downregulated during GSCs differentiation, with GSC#83 and GSC#1 cells downregulating the Notch4 and Notch3, and Notch4 expression, respectively [73]. The Wnt signaling pathway controls neurogenesis from NSCs by regulating maintenance, proliferation and differentiation of progenitor cells into newborn neurons. Here, we found that FZD1, FZD3 and FDZ9 were increased in D-GSC#1 and D-GSC#83, respectively, compared to GSCs. FZD1 is highly expressed in NSCs [74] and its knockdown decreases neuron differentiation and increases astrocyte generation; FZD3 is associated with non-canonical β-catenin-independent signaling, and stimulates the neural crest formation, suggesting a role in cell fate commitment [75].
IL-6 expression increases with glioma malignancy grade and is associated with shorter survival of GBM patients. IL-6 and LIF are crucial mediators of self-renewal capacity of GSC [76]. Herein, we found that in GSC#83, which increases the expression of the CD133 stem cell marker in D-GSC#83 [14], thus the IL-6/LIFR levels were increased, whereas the reduction of LIFR levels was observed in GSC#1, showing CD133 reduction in D-GSCs, in comparison to GSCs.
Moreover, increased levels of immunoglobulin kappa J region (RBPJ) mRNA, which contributes to GBM malignancy and promotes proneural-mesenchymal transition via the IL-6/STA3 pathway [77], were evidenced in MES-like GSC#83 cells, suggesting that mesenchymal GSC#83 may be derived from a GSC with a proneural phenotype [15]. In the GBM stem cell compartment, the proneural-mesenchymal transition, the equivalent of the epithelial-mesenchymal transition associated with aggressive cancers, under the selective pressure of treatment agents is responsible for the acquisition of multitherapy resistance phenotype [78]. Finally, due to their role in promoting the cell cycle of gliomas, NFAT5 and E2F5 TFs are downregulated in seven days for the D-GSC#83 and D-GSC#1 lines, respectively [79].

Conclusions
The overexpression of Aml1b and Aml1c splice variants of the Runx1/Aml1 transcriptional factor was detected in GSC lines, compared to NSCs. GSC heterogeneity was also evidenced, with GSCs mainly expressing a proneural or a mesenchymal-like phenotype. Even if we analyzed only one GSC cell line for each phenotype, our data suggest that GSC differentiation accompanied by ERK phosphorylation is able to induce: (a) accumulation of Aml1b and Aml1c proteins, (b) changes in the downstream TRPA1 and TRPV1 ion channels expression and consequently in the TRP agonists-mediated cytotoxic functions, (c) increase in NGF mRNA levels, (d) activation of distinct TGFβ/Smad and Wnt/β-catenin and bFGF signaling pathways in proneural-and mesenchymal-like GSCs. Thus, the need to increase the knowledge on the intercellular signals controlling GSC differentiation may shed light on GSC clonal heterogeneity and improve the development of more specific target therapy in GBM patients.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/cells10082052/s1, Figure S1: Phenotype characterization of GSCs, Table S1: Aml1b and Aml1c splice variants mRNA fold expression in D-GSCs expressed as relative fold with respect to D-NSC, Table S2: TRPA1 and TRPV1channels mRNA fold expression in D-GSCs expressed as relative fold with respect to D-NSC.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.