Mutation in EHMT1 causes reduced spheroid formation and growth arrest. To generate the 3D neurospheroid culture for the KS cell line we used our well-established protocol for toxicology assays in non-diseased, control cell lines 16. Although the spheroid formation has occurred in both cell lines, while the control spheroids manifested a progressive growth during the culture, the growth of the KS spheroids was overtly arrested and their size at the termination of the 3D culture (at day 57th ) was equal to the starting size (Fig. 1a-b). To measure EHMT1 activity in TD57 spheroids we detected the expression of its product H3K9me2 using ICC (Fig. 1c-d). The lower H3K9me2 immunoreactivity detected in KS denotes EHMT1 deficiency and thus downregulated activity. To address the question if the smaller spheroid size of the KS line is the result of a lower proliferation rate and cell cycle arrest, we measured the gene expression of KI67, marker gene of proliferation, and CDKN1A, indicative of cell cycle arrest using qRT-PCR on TD57 day spheroids. We found that the expression of KI67 was significantly lower in KS than in the control (p = 0.004), while CDKN1A showed slightly higher but statistically non-significant expression in the KS (Fig. 1e). To identify whether the lower proliferation rate is evoked by the 3D culture system or is associated with the impaired EHMT1 expression, we quantified the expression of both genes in NPC stage as well. Figure 1f shows that although the trend is similar to the expression levels seen in 3D (lower KI67 expression and higher CDKN1A expression in KS) the difference is not significant. Similar changes were reported by Lee et al 20 in lung cancer cell lines where they used siRNA to silence the EHMT1. The authors found that EHMT1 knockdown reduced spheroid formation in 3D culture and caused cell-cycle arrest by increasing CDKN1A expression. Another aspect of the spheroid formation the authors investigated was the expression of the CDH1 gene encoding E-cadherin responsible for cell adhesion and they have found that its expression decreases when EHMT1 is silenced accounting for the structurally loose spheroid. To verify if cadherin expression is affected in the KS neurospheres we measured both CDH1 (E-cadherin) and CDH2 (N-cadherin) in TD57 spheroids. Conversely to the findings of Lee et al 20 in KS spheroids, there was a massive, more than one-hundred-fold increase in CDH1 expression (p = 0.02) while CDH2 was downregulated (p = 0.003) (Fig. 1e). At the NPC stage, E-cadherin was not detectable, and N-cadherin showed a similar trend to the 3D culture, being strongly downregulated in KS cells (p < 0.0001) (Fig. 1f).
Next, we characterized the cellular composition of the spheroids using qRT-PCR and ICC on terminally differentiated spheroids on day 57. Concerning the distribution of the three main cell types there was no significant difference in the two cell lines neither at transcript nor at protein level. The number of MAP2-positive neurons, astrocytes (identified by AQP4 in qRT-PCR or GFAP in ICC) and oligodendrocytes (identified by CALU11 in qRT-PCR or MBP in ICC) was similar (Fig. 2a-c). Regarding the neuronal subtypes, we have found that NMDA receptor (NMDAR) subunit 1 encoded by GRIN1 is significantly overexpressed both on transcript (p = 0.002) and protein level (p = 0.02) (Fig. 2d-f). This result is consistent with the findings of Frega et al. who quantified the NMDAR1 expression in two patients with different EHMT1 loss-of-function mutations and in neurons with CRISPR-engineered disruption of EHMT1 15. Furthermore, glutamatergic VGLUT1/2 (vesicular glutamate transporters 1 and 2), cholinergic ChAT (choline o-acetyltransferase) and dopaminergic TH (tyrosine hydroxylase) neurons were detected using ICC and qRT-PCR. VGLUT1/2 protein expression was significantly higher in KS spheroids (p = 0.02). ChAT, although higher in KS did not reach a significant difference on protein level (p = 0.05) but it was significantly higher on RNA level (p = 0.006) (Fig. 2f). TH showed significant downregulation in KS both on protein (p = 0.02) and transcript level (p = 0.0002) (Fig. 2d-f). The expression of two additional markers was detected using qRT-PCR. Although both GAD1 (glutamate decarboxylase 1) and SLC6A4 (solute carrier family 6 member 4) showed higher expression in KS spheroids the difference was not significant (Fig. 2f). To detect and quantify the presence of synapsis we used immunostaining against synaptophysin (SYNP) and vesicle-associated membrane protein 2 (VAMP2) (Fig. 2d). The level of synaptophysin was slightly elevated, while VAMP2 was significantly overrepresented in KS spheroids (p = 0.02) (Fig. 2e).
Neurite outgrowth assay shows growth advantage in KS neurospheres. To investigate the impact of EHMT1 deficiency in KS on neurite growth we performed a neurite outgrowth assay using TD57 spheroids by plating them on laminin-coated tissue culture plates and culture for three days and neurites were visualized by immunostaining of tubulin beta 3 (TUBB3). We were intrigued by the results as opposite to the growth behavior of the KS spheroids, the neurites of the KS were significantly longer (p = 0.002) and presented a more abundant branching than the control (Fig. 3A-C). To verify if the same difference was true for the soma, we measured TUBB3 expression on cryosections of TD57 spheroids. Similar to the results obtained with the neurite outgrowth assay, KS spheroids also contained significantly higher TUBB3 expression (p = 0.02) indicating longer neurites (Fig. 3d-e).
The growth and branching of neurites, among others, is determined by the neurotrophin brain-derived neurotrophic factor (BDNF) 21. To investigate if BDNF is responsible for the longer neurites in KS we used immunostaining to detect BDNF expression in the neurites of TD57 spheroids. We have found that indeed, BDNF expression is significantly higher in KS than in the control (p = 0.008, Fig. 3f-g) accounting for differences in length and branching of neurites.
To quantify the synapsis formation between neurites, we performed immunostaining against postsynaptic density protein (PSD95) and synaptophysin (SYNP). PSD95, as an essential scaffolding protein during synaptogenesis and neurodevelopment involved in the stabilization of the NMDA receptors by direct biding and anchoring 22. Synaptophysin is the most abundant synaptic vesicle membrane protein and facilitates the endocytosis of synaptic vesicles in central neurons 23. The expression of both proteins is lower in KS, with only PSD95 reaching significance (p = 0.01) (Fig. 3h-j).
Compound- and cell type-specific toxicity effects. To assess the sensitivity of 3D neutrospheres and neuronal progenitors we exposed them to various doses of four compounds with different mechanisms of toxicity for 72 hours. The herbicide paraquat (PQ2+,N, N′-dimethyl-4–4′-bipyridinium) is one of the environmental toxicants responsible for Parkinson's disease (PD) characterized primarily by the loss of dopaminergic neurons 24. PQ undergoes redox cycling with cellular diaphorases such as nicotinamide adenine dinucleotide phosphate (NADPH) oxidase and nitric oxide synthase (NOS) to yield the monovalent cation PQ+ 25. From this redox cycle, superoxide is generated, leading to oxidative stress-related cytotoxicity. In our test system, when NPCs were treated with PQ, although the KS had an EC50 value of 66.26 µM, and the control was EC50 = 88.71 µM the difference did not reach statistical significance, and the dose response curves were almost overlapping (Fig. 4a, Table 1). Conversely, the mature spheroids showed elevated sensitivity in both cell lines as compared to NPCs (EC50 = 23.7 µM for the control and EC50 = 4.8 µM for KS), with significantly lower cell viability of the KS spheroids at 30 and 10 µM (p < 0.0001) (Fig. 4b). The second compound tested was rotenone, a commonly used insecticide in agriculture, known to cause mitochondrial dysfunction and ROS by inhibiting the mitochondrial complex I which, due to dopaminergic-neuron selective toxicity, contributes not only to Parkinson’s disease pathophysiology but it was also shown to exert developmental neurotoxicity in a human brain spheroid model already at non-cytotoxic concentrations 24,26. The NPCs of the KS and control cell lines treated with rotenone showed overlapping dose curves, with no significant difference at any concentration tested (Fig. 4a). The EC50 values for rotenone in the case of NPCs was 0.108 µM for the control and 0.095 µM for the KS. When spheroids were exposed to the toxicant significantly higher cell death was noted at 0.1 µM and 0.01 µM rotenone treatment in the KS versus the control (p = 0.0007 and p = 0.01 respectively) (Fig. 4b). The EC50 values obtained for rotenone in 3D cultures were 0.239 µM for the control and 0.111 µM for the KS. When comparing EC50 values for NPC vs 3D, the trend is opposite to what we measured using paraquat, as both cell lines are more sensitive at NPC stage than the mature spheroids, although the difference was below statistical significance (Table 1).
Table 1
EC50 values calculated from the dose response analysis performed for paraquat (PQ), rotenone (ROT), bardoxolone (CDDO-Me) and doxorubicine (DOX).
Toxicant | CONT | KS |
| NPC | 3D | NPC | 3D |
PQ | 92.7 (± 4.2) | 23.7 (± 5.3) | 61.14 (± 8.5) | 4.8 (± 1.6) |
ROT | 0.1 (± 0.01) | 0.23 (± 0.08) | 0.07 (± 0.01) | 0.11 (± 0.02) |
CDDO-Me | 0.39 (± 0.04) | 1.76 (± 0.15) | 0.31 (± 0.04) | 1.6 (± 0.3) |
DOX | 0.01 (± 0.006) | 17.2 (± 5.7) | 0.01 (± 0.005) | 10.3 (± 2.5) |
The third compound tested was bardoxolone methyl (CDDO-Me), a synthetic triterpenoid that is a robust inducer of the Nrf2 pathway, which by inhibiting NF-κB, leads to antioxidant and anti-inflammatory effects 27. At the NPC stage the cell lines are equally sensitive to the tested bardoxolone doses (EC50 = 0.381 µM for the control and EC50 = 0.331 µM for the KS) (Fig. 4a, Table 1), while in the 3D stage, the dose-response curve of the KS is slightly lower than the control’s, although statistical significance could be detected only at 1 µM (p = 0.04) (Fig. 4b). The EC50 values for both cell lines in spheroid stage was 1.5 µM. Overall, the NPCs of both control and KS were significantly more sensitive than the 3D stage (p = 0.0004 for CONT NPC vs CONT 3D and p = 0.0002 for KS NPC vs KS 3D) (Table 1).
Finally, we used doxorubicin (DOX) as it is one of the most effective and widely used chemotherapeutic anti-cancer drugs, crucial to the treatment of a range of neoplasms, while causing severe neurotoxicity 28. In activaley proliferating cells DOX inhibits topoisomerase II, leading to the formation of double-stranded breaks of DNA or direct intercalation with DNA, which in turn inhibits DNA duplication and transcription to mRNA 29.In differentiated cells its cytotoxic effect is conferred mainly by mitotoxicity, the formation of free radicals, and lipid oxidation 28. At the NPC stage, KS shows slightly higher tolerance at some of the concentrations tested although the difference is not significant and the EC50 value is 0.01 µM for both lines (Fig. 4a, Table 1). At 3D TD57 the two cell lines have similar dose-response curves, and overall, both cell lines show significantly higher tolerance (p = 0.001 for CONT NPC vs CONT 3D; p = 0.03 for KS NPC vs KS 3D) than in the NPC stage (Table 1), which again is in line with the different proliferation rates of NPC vs TD57 with NPCs proliferating at a much higher pace (Fig. 4b).
Regarding the effect of PQ Naspolini et al 30 showed that PQ exposure and the oxidative stress response resulted in decreased glucose transport into neural cells associated with a compensatory increase in lactate dehydrogenase (LDH) activity suggesting the importance of lactate as energy fuel. To address the question if EHMT1 haploinsufficiency impacts LDH expression we measured the transcript levels of the two subunits of LDH, LDHA which has higher affinity for pyruvate and, LDHB with higher affinity for lactate. Figure 5 shows that the expression of LDHA is significantly downregulated in the KS line in both maturation stages tested (p = 0.03 for the NPC, and p = 0.002 for the 3D spheroid), while LDHB is not affected by the mutation.