Isolation of the purified Gabra6-positive subpopulation of GCPs
RNA sequencing of cell-type marker genes of the cerebellum after the cultivation of GCPs in the presence of NAC for 7 days was performed10 (Supplementary Fig. 1A). These values (P < 0.05) were summarized relative to the total number of cells. When the expression levels of these markers in each representative cell population of Jdp2-KO GCPs were compared with those detected in WT GCPs, the expression of granule cell marker genes as the dominant population was lower than that observed in WT GCPs; in contrast, the expression of the marker genes of Purkinje cells, astrocytes, and oligodendrocytes was 1.7–2.0-fold higher in Jdp2-KO GCPs compared with WT GCPs after cultivation with NAC. These data are similar to the results obtained from the recent single-cell RNA sequencing reported by Carter et al. [13]. In addition, the gamma-aminobutyric acid A receptor (Gabra) subunit alpha 6 (Gabra6) mRNA has been reported to be expressed in the cerebellum during development [14]. Thus, we examined the expression levels of proteins such as PC-related and calcium-channel-related molecules in the presence of NAC for 7 days using Western blotting (Supplementary Fig. 1B). The expression of Cacn alpha1a (1.4-fold), calbindin (1.3-fold), Gabra1 (1.3-fold), Gabra6 (1.5-fold), Gabrb2 (1.3-fold), Grin2a (1.2-fold), Pcp4 (2.0-fold), and Vglut1 (1.8-fold) was significantly upregulated in Jdp2 KO vs. WT GCPs, indicating that Jdp2 might be the master regulator of Gaba-receptor-mediated neural differentiation into Purkinje cells [14]. Thus, we sorted the GCPs to isolate the Gabra6+ subpopulation after cultivation in the presence of NAC for 7 days using an anti-Gabra6 antibody (Fig. 1A and B); we found that more than 90% of the total cells were positive for Gabra6, whereas less than 5–10% were negative for Gabra6 (Fig. 1B). These Gabra6+ GCPs were analyzed further for stemness, pluripotency, cell cycle, differentiation, and oxidation/antioxidation.
Expression of genes related to stemness, differentiation, and the AhR–Nrf2 axis
For their characterization, the FACS-separated Gabra6-positive GCPs (Gabra6+ GCPs) were obtained after the cultivation of GCPs for 7 days in the presence of NAC and were then stained with an anti-Atoh1 antibody [15]. We found that Atoh1 expression was higher in WT compared with Jdp2 KO cells (Fig. 1C). The previous results of an RNA sequencing experiment in cerebellar cells derived from P6 mice showed that more than 95% of cells were positive for Atoh1[10]. After cultivation with NAC for 7 days, the number of Atoh1-positive cells was decreased to 70%. In fact, 70% of Gabra6+ GCPs remained positive for Atoh1 (data not shown). Furthermore, stemness markers such as Sox2 and Klf4 were upregulated in Jdp2 KO compared with WT cells. In turn, the levels of markers such as Oct4, Nanog, and c-Myc were lower, but not altered between WT and Jdp2 KO cells with NAC for 7 days (Fig. 1D). Western blotting also confirmed this finding (data not shown). Moreover, the WT and Jdp2 KO Gabra6+ GCPs were negative for alkaline phosphatase after cultivation in the presence of NAC for 7 days (data not shown). This indicates that cultivation with NAC resulted in the loss of the stemness character. The antioxidation-specific transcription factor Nrf2 and the cell-cycle regulator p21Cip1 were expressed at 1.9-fold and 1.7-fold higher levels in Jdp2 KO vs. WT cells, respectively. In contrast, the oxidative-stress-related factor AhR was expressed at 81% lower levels in Jdp2 KO compared with WT cells (Fig. 1E). The expression of Lgr5 in Jdp2 KO was higher by 1.2-fold than that detected in WT cells (Fig. 1F). Thus, the addition of NAC to Gabra6+ GCPs seems to initiate the differentiation program while maintaining the stemness character.
Cell proliferation and antioxidation
The BrdU incorporation activity of WT GCPs was higher than that of Jdp2 KO Gabra6+ GCPs (Fig. 2A). Moreover, immunocytochemistry demonstrated that the number of BrdU-positive Gabra6+ GCPs from WT mice was 1.4-fold higher than that from Jdp2 KO mice (Fig. 2B), suggesting that Jdp2 is an activator of the proliferation of Gabra6+ GCPs. In addition, a comparative cell-cycle analysis between WT and Jdp2 KO Gabra6+ GCPs showed that Jdp2 KO triggered a concomitant decrease in the number of cells in the G2/M phase (by 70%) (Fig. 2C). Furthermore, Western blotting revealed that cell-cycle-arrest–related proteins, such as p21Cip1 and p57Kip2, were expressed at 1.5-fold and 1.6-fold higher levels, respectively, in Jdp2 KO compared with WT cells. In contrast, cell-cycle-processing factors, such as E2F1, cyclin A2, Cdk4, and Cdk2, were decreased by about 60–80% in Jdp2 KO vs. WT cells (Fig. 2D). These data suggest that Jdp2 plays a critical role in the control of the cell cycle in Gabra6+ GCPs. In addition, our previous studies demonstrated that the apoptotic activity of WT GCPs was higher than that of Jdp2 KO GCPs [11].
To clarify the mechanism underlying the upregulation of GSH in Gabra6+ GCPs in the presence of NAC, we examined the effect of NAC in WT and Jdp2 KO Gabra6+ GCPs, to compare their antioxidation activities. Under the NAC condition, the levels of ARE-luciferase activity in WT and Jdp2 KO cells were increased by 1.4–3.3-fold vs. those recorded in the absence of NAC (Supplementary Fig. 2A). These results suggest that NAC plays a critical role in the antioxidation response; this effect was more sensitive in Jdp2 KO Gabra6+ GCPs compared with WT Gabra6+ GCPs.
The expression levels of Gsk3β [16] and NeuN [17] as neural markers were significantly lower in Jdp2 KO Gabra6+ GCPs in the presence of NAC. In contrast, the level of Gabra6 positivity was about 1.7-fold higher in Jdp2 KO vs. WT cells (Supplementary Fig. S2B and C). Thus, Jdp2 might exert specific effects on Gabra6+ expression, but not on Gsk3β and NeuN-specific neural differentiation. Thus, we next focused on the triggering of the early commitment toward neural differentiation activity in Jdp2-depleted Gabra6+ GCPs, which indicated that Jdp2 might be a blocker of the NAC-mediated differentiation of Gabra6+ GCPs into the neural stages.
Neuronal differentiation protocol of Gabra6+ GCPs
Here, we developed two methods to achieve the neural differentiation of Gabra6+ GCPs into PCs. One method (A) consisted in the generation of spherebodies in low-attached 96-well plates of Gabra6+ GCPs in neurobasal medium (B27 supplement, N2 supplement, mouse leukemia inhibitor factor (mLIF), and 2-mercaptoethanol) plus NAC for about 1 week, followed by commitment toward differentiation by replacing with differentiation medium (Fig. 3A) [6, 7]. After the incubation of Gabra6+-GCP-derived neurosphere bodies in the differentiation medium for 7 days, we found that the differentiation efficiency of Jdp2 KO-derived neurosphere bodies was 1.2-fold faster than that of WT-derived neurosphere bodies (left panel, Fig. 3B). After further differentiation induction up to 10‒14 days, we found that the development of neurites in Jdp2 KO Gabra6+ GCP-derived neurosphere bodies was better compared with that of neurites from WT Gabra6+ GCP-derived neurosphere bodies (right panels, Fig. 3B). The second method (B) consisted in the use of flat cultivation after digesting the sphere bodies with accutase, followed by re-cultivation for an additional 7–14 days. Specific neuronal fibers were clearly apparent after cultivation for an additional 7 days (Fig. 3C). More than 80% of the subpopulation of Gabra6+ GCPs was stained with anti-calbindin antibodies, whereas less than 8% of each population was positive for anti-Atoh1 (GC-specific), anti-GFAP (astrocyte-specific), and anti-CD45 (glia-specific) antibodies (Fig. 4A and B). This suggests that about 80% of Gabra6+ GCPs can differentiate into Purkinje cells in vitro in differentiation medium containing NAC.
To define the role of exposure to NAC in this differentiation process, we compared the expression of Purkinje progenitor cell markers, such as Neph3 [18, 19] and calbindin, between the condition of NAC exposure and that of the absence of NAC in Gabra6+ PGCs for 7 days. After NAC exposure of Gabra6+ PGCs prepared using an accutase-2D cell method, we found that PC makers, such as calbindin 3 and Neph3, were predominant (about 1.8-fold- and 1.5-fold-higher expression, respectively) in Jdp2 KO vs. WT cells after 2 weeks of the neural differentiation protocol (Fig. 4C). These results suggest that Jdp2 controls the frequency and the speed of differentiation of GCPs into PCs. Therefore, the Jdp2 protein itself can prevent the neural differentiation of Gabra6+ GCPs into Purkinje neurons.
Enhancement of Ca 2+ signals in Jdp2 KO Gabra6+ GCPs
A previous study demonstrated that the Gaba receptor is involved in the regulation of the differentiation of rodent neural progenitor cells [18]. The Gaba receptor is a G-protein-coupled receptor that is associated with inositol 1,4,5-triphosphate (IP3)-induced Ca2+ signals [20, 21, 22]. We hypothesized that the Gaba-receptor-mediated Ca2+ signals are involved in Jdp2-regulated neural differentiation. To investigate the effect of Jdp2 on Gaba-receptor-mediated Ca2+ signals, we examined the caged inositol IP3-mediated Ca2+ uptake, which leads to local calcium release, in accutase-2D-differentiated Gabra6+ GCPs (Fig. 5). In the case of calcium release in Jdp2 KO cells, the level of uncaged IP3 was 1.1-fold higher than that detected in WT cells (Fig. 5A). This uncaged-IP3-mediated Ca2+ uptake was modulated by the Gaba receptor. Thus, to confirm the Gaba-receptor-regulated IP3-mediated calcium release, the Gaba receptor agonist GABOB was added to the culture medium. The calcium uptake in Jdp2 KO accutase-2D-differentiated Gabra6+ GCPs was 1.3-fold higher than that observed in WT cells after GABOB stimulation (Fig. 5B). Taken together, these results suggest that the ability to trigger calcium signaling is evoked by a higher calcium uptake in Jdp2-depleted accutase-2D-differentiated Gabra6+ GCPs via intracellular uncaging. To examine the role of the Gaba receptor in inositol IP3-mediated Ca2+ uptake in Jdp2 KO accutase-2D-differentiated Gabra6+ GCPs, we used various Gaba receptor inhibitors, such as bicuculline, PTZ, and flumazenil (Fig. 5C). The calcium uptake in Jdp2 KO accutase-2D-differentiated Gabra6+ GCPs after GABOB stimulation was reversed to the original levels detected in control GCPs.
Expression of PC-related and Gaba-receptor-related genes
To examine the role of the Gabra6, we used various GABA receptor inhibitors [23], such as bicuculline and PTZ (specific for α/β), and flumazenil (a competitive antagonist at the benzodiazepine-binding site on α) and assessed whether they inhibited neural differentiation (Fig. 6A-D). These inhibitors blocked the expression of calbindin and βIII tubulin (Fig. 6A), Gabra6 (Fig. 6B), and the glutamate–cysteine ligase modifier subunit (Gclm, known as gamma-glutamylcysteine; glycine ligase or glutathione synthetase) [24] (Fig. 6C). These results indicate that the Gaba6 plays a critical role in NAC-mediated antioxidation, for the accumulation of GSH, by increasing the glutamine–cysteine pump including Gclm, and determines the neural network in Jdp2-depleted Gabra6+ GCPs, including PC neurons.
In an attempt to search for mutations in the GABRA6, GABRA1, GABRB2 genes, as well as the antioxidative gene NRF2, cysteine transporters (such as SLC7A11 and CD44v), and their regulators (such as JDP2 and p21Cip1 (CDKN1A) in brain tumors, the cBioPortal (http://www.cbioportal.org/faq#how-do-i-cite-the-cbioportal) data were accessed. In total, 8,139 patients and 8,597 samples from 26 studies were grouped for each item, such as signaling, apoptosis, and cell-cycle-progression pathways [25–27]. The frequency of mutation in the GABRA6, GABRA1, and GABRB2 genes was significantly higher than that detected for other cell-cycle- and cell-proliferation-related genes; furthermore, their co-occurrence in patients with cancer was examined. Thus, alterations in the GABRA gene family seem to be closely related to cancer occurrence, not only in mice, but also in human patients with brain cancer, through the regulation of the antioxidative response genes to produce GSH via SLA7A11 and CD44v, as an xCT receptor. A synthetic analogue of GABA was reported to play roles in anti-inflammation and as an antioxidant in mice [28] and pigs [29]. As shown in Figs S3A–D and S4A, B, the mutation rates of GABA receptors were significantly increased [25–27]. Accordingly, molecules that control the antioxidation response, such as GSH production-related molecules (e.g., SLA7A11, CD44v, p21Cip1 (CDKN1A), and JDP2), were also mutated significantly by gene amplification and deep deletions, similar to the gene mutation events observed in the GABA receptor family. GABRA1 might be related to NRF2 and CD44v, GABRA2 is related to CD44v, and GABRA6 is related to SLC7a11 (Fig. S3 and S4). These databases suggest that JDP2 is a possible regulator of the GABA receptor family signaling which might be concerned with ROS-antioxidation balance and neural differentiation from PGCs to PCs.