The Protein Arginine Methyltransferases 1 and 5 affect Myc properties in glioblastoma stem cells

Protein Arginine (R) methylation is the most common post-translational methylation in mammalian cells. Protein Arginine Methyltransferases (PRMT) 1 and 5 dimethylate their substrates on R residues, asymmetrically and symmetrically, respectively. They are ubiquitously expressed and play fundamental roles in tumour malignancies, including glioblastoma multiforme (GBM) which presents largely deregulated Myc activity. Previously, we demonstrated that PRMT5 associates with Myc in GBM cells, modulating, at least in part, its transcriptional properties. Here we show that Myc/PRMT5 protein complex includes PRMT1, in both HEK293T and glioblastoma stem cells (GSCs). We demonstrate that Myc is both asymmetrically and symmetrically dimethylated by PRMT1 and PRMT5, respectively, and that these modifications differentially regulate its stability. Moreover, we show that the ratio between symmetrically and asymmetrically dimethylated Myc changes in GSCs grown in stem versus differentiating conditions. Finally, both PRMT1 and PRMT5 activity modulate Myc binding at its specific target promoters. To our knowledge, this is the first work reporting R asymmetrical and symmetrical dimethylation as novel Myc post-translational modifications, with different functional properties. This opens a completely unexplored field of investigation in Myc biology and suggests symmetrically dimethylated Myc species as novel diagnostic and prognostic markers and druggable therapeutic targets for GBM.

Methylation of arginine (R) residues is the most common methylation event in mammalian cells, which modifies protein interacting properties, ruling a large amount of signaling pathways 13 . Three distinct types of methylated Rs occur in mammalians. Asymmetrically dimethylated arginine (ADMA) is the most frequent, followed by symmetrically dimethylated (SDMA) and monomethylated arginine (MMA). The enzymes responsible for R methylation are the Protein Arginine Methyltransferases (PRMTs). Six proteins have been reported to undoubtedly possess the R methyltransferase catalytic domain (PRMT1, 3, 4, 5, 6 and 8). PRMT2, 7 and 9 have sequence similarities with the other PRMTs and are supposed to possess an equivalent enzymatic activity 13 . Depending on the position of the methyl groups on R terminal guanidino nitrogens, PRMTs are, basically, classified into two types: type I PRMTs (PRMT1, 3, 4, 6 and 8) are responsible for MMA or ADMA whereas the type II PRMT5 catalyzes MMA and SDMA 13 . PRMT1 is the major methyltransferase in mammalian cells. It has a wide variety of substrate specificity, from histone to non-histone proteins 14 . From a transcriptional point of view, it is known as an activator, since the occurrence of asymmetrically (AS) dimethylated R3 on histone H4 residues (H4R3me2as) is associated with transcriptionally active chromatin 15 . PRMT1 is deregulated in a wide variety of cancer types, e.g. pancreatic adenocarcinoma 16 , gastric 17 and lung cancer 18 . It controls the epithelial-mesenchymal transition (EMT) in cancer cells 19 , while playing a major role in muscle 20 and B cell differentiation 21 . PRMT5 has received a greater attention and has been suggested as a putative druggable target in a wide variety of tumours, including GBM 22 . Like PRMT1, it has multiple nuclear and cytoplasmic targets and catalyzes the symmetrical (S) dimethylation of H4R3 and H3R8, usually repressing transcription 23 , and of H3R2, activating transcription 24 . PRMT5 is involved in a variety of cellular processes, such as neurogenesis 25 , myogenesis 26 , somatic cell reprogramming 27 and GSCs self-renewal 28 . Interestingly, PRMT1 and PRMT5 have been shown to interact in GBM cells, being PRMT1 the major responsible for the activation of cancer-related genes 29 . They also catalyze the S-and AS-dimethylation of NMyc 30,31 and, very recently, it has been shown that PRMT1 is required for p300 recruitment to Myc-regulated gene promoters 32 .
We previously demonstrated that PRMT5 interacts with Myc in glioblastoma cells, including GSCs, and that PRMT5 participates, at least in part, in the activation of Myc target genes 33 . Here we show that PRMT1 associates with Myc/PRMT5 in both HEK293T cells and GSCs and that Myc is both S-and AS-dimethylated by PRMT5 and PRMT1, respectively, both in vitro and in living cells. At the functional level, S-dimethylation protects Myc from degradation, while AS-dimethylation ensure Myc proper turnover. Finally, the inhibition of either PRMT1 or PRMT5 activity affects Myc recruitment at promoters and has a profound effect on GSCs biological functions, such as neurospheres formation and differentiation capacity. These findings represent the first demonstration in GSCs of the presence of differentially dimethylated Myc species, with distinct properties, opening a completely new field of investigation in Myc-dependent GBM biology. Further, they support the hypothesis that acting on S-Myc post-translational modification may represent a possible route to control its function.

Results
Myc interacts with PRMT1 and PRMT5. We have previously shown that Myc induces S-dimethylation of R3 on histone H4 (H4R3me2s, Fig. 1a, left and ref. 33 ) and associates with PRMT5 in both HEK293T and glioblastoma cells 33 . Since PRMT5 and PRMT1 were found associated in GBM cells 29 , we sought to determine whether Myc was able to promote also AS-dimethylation of R3 on histone H4 (H4R3me2as). To this aim, HEK293T cells were transfected with either a Flag-tagged Myc construct (FlagMyc/HEK293T) or an empty vector and the level of H4R3me2as was detected by western blot. Figure 1a, right, shows H4R3me2as induction in the presence of FlagMyc construct. We reasoned that these histone modifications should decrease by Myc interference. However, in both HEK293T and mesenchymal GSCs 33,34 transduced with a lentiviral, doxycycline inducible shRNA against Myc (shMyc), the level of H4R3me2s were reduced, while H4R3me2as increased (Fig. 1b), suggesting that impairing Myc-dependent PRMT5 activity is still sufficient to make H4R3 available for PRMT1 activity. Based on these data, we asked whether PRMT1, PRMT5 and Myc may interact. A series of reciprocal immunoprecipitation experiments, performed in FlagMyc/HEK293T cells, showed that FlagMyc associates with both PRMT5 and PRMT1 (Fig. 1c). No interactions were observed by transfecting the CBS-Flag vector alone, as expected (not shown). Consistently, the same result was obtained, at the endogenous level in GSCs (Fig. 1d). Overall, these data validate PRMT5/Myc interaction 33 and indicate PRMT1 as a novel partner in this protein complex. PRMT5 is required for the formation of Myc/PRMT5/PRMT1 protein complex. We next wondered which protein member was necessary for complex assembly. Therefore, PRMT5 and PRMT1 expression was blunted by specific siRNAs in HEK293T cells (Fig. 1e). In siPRMT5/HEK293T cells, PRMT5 depletion was associated with a decrease in H4R3me2s levels, as expected, and with an increase in H4R3me2as, underlying the competition between PRMT5 and PRMT1 for the same histone substrate. Intriguingly, Myc protein also decreased. No effect on PRMT1 expression was observed. In siPRMT1/HEK293T cells, PRMT1 decreased together with H4R3me2as levels, as expected, while H4R3me2s increased. Myc protein slightly increased, while no effect on PRMT5 expression was detected. Although PRMT5 depletion was incomplete, Fig. 1f shows that the anti-Flag antibody did not co-immunoprecipitate PRMT1; accordingly, the anti-PRMT1 antibody did not co-immunoprecipitate the FlagMyc protein, as outlined in the model shown in the right panel. In cells partially depleted of PRMT1 (Fig. 1g) we were still able to observe a residual amount of PRMT5/PRMT1/FlagMyc complex, when an anti-PRMT1 antibody was used for immunoprecipitation. However, in anti-Flag and anti-PRMT5 immunoprecipitates, PRMT1 was undetectable, whereas PRMT5 and FlagMyc still interacted (see the scheme in the right panel). Overall, these data suggest that Myc/PRMT1 binding is mediated by PRMT5 in FlagMyc/ HEK293T cells.
Myc is both symmetrically and asymmetrically dimethylated. We wondered whether Myc was dimethylated by both PRMT5 and PRMT1. To ask this question, in vitro methylation assays were performed e SYM SYM ASYM www.nature.com/scientificreports www.nature.com/scientificreports/ in HEK293T cells, overexpressing either myc-tagged PRMT5 or PRMT1, in the presence or absence of specific siRNAs (Fig. 2a). S-or AS-dimethylation of a recombinant human Myc protein was detected by immunoprecipitation experiments using antibodies recognizing either S-(SYM10 antibody) or AS-(ASYM24 antibody) dimethylated R residues. The recombinant human Myc was S-dimethylated in the presence of overexpressed PRMT5. S-dimethylation was restrained in the presence of a pool of PRMT5-specific siRNAs, as expected (Fig. 2a, top). Recombinant Myc AS-dimethylation was detected in the presence of a PRMT1-myc-Dkk overexpressed plasmid. No signal was observed in the presence of a specific siPRMT1 pool (Fig. 2a, bottom).
Moreover, protein extracts from HEK293T/FlagMyc cells were reciprocally immunoprecipitated by using SYM10 or ASYM24 or anti Flag antibodies. Western blots were performed with anti-Flag or SYM10 and ASYM24 antibodies. Figure 2b, middle panel, shows the presence of both S and AS-dimethylated FlagMyc in HEK293T cells. HEK293T cells transfected with pCBSFlag vector alone represented the negative control (not shown). To confirm that S-and AS-Myc dimethylation were PRMT5 and PRMT1-dependent, respectively, HEK293T/ FlagMyc cells were treated for 24 hours with either 5 μM EPZ015666 -a commercially available PRMT5 specific inhibitor -or 3 μM 1,5-bis(3,5-Br 2 -4-hydroxyphenyl)-1,4-pentadien-3-one, hereafter named inhibitor 9 (9i), which specifically inhibits PRMT1 at this concentration 35 . Thereafter, immunoprecipitation experiments were performed with SYM10 and ASYM24 antibodies. Figure 2c, left, shows that PRMT5 inhibition impairs FlagMyc S-dimethylation, as expected. Conversely, PRMT5 inhibition increases the level of AS-dimethylated FlagMyc. These results were mirrored by similar experiments performed on cells treated with 9i. Indeed, PRMT1 inhibition reduced the level of AS-FlagMyc, while enhancing S-FlagMyc (Fig. 2c, left and scheme in d). In siPRMT5 transfected HEK293T/FlagMyc cells, S-FlagMyc was found at low levels, as expected. Interestingly, also AS-FlagMyc decreased (Fig. 2e). This is not surprising, as FlagMyc/PRMT1 interaction requires the presence of PRMT5 in HEK293T cells (Fig. 1f). In siPRMT1 transfected HEK293T/FlagMyc cells (Fig. 2e) we detected higher levels of S-FlagMyc, whereas AS-FlagMyc was severely impaired, confirming the binding of PRMT5 and Myc in the absence of PRMT1, as previously shown in Fig. 1g. PRMT1 and PRMT5 modulate Myc stability. It has been shown that both PRMT1 and PRMT5 stabilize NMyc protein 30,31 . We observed that upon PRMT5 interference Myc expression decreased, while in PRMT1-depleted cells, Myc levels slightly increased (Fig. 1e). The same results were obtained inhibiting PRMT5 or PRMT1 activity, respectively (Fig. 2c, right). Therefore, two sets of experiments were performed, in order to investigate whether Myc stability was affected by either PRMT5 or PRMT1 catalytic inhibition in HEK293T/FlagMyc cells and in GSCs. To this aim, cells were treated with 5 μM EPZ015666 and 3 μM 9i for 24 hrs. Thereafter, cycloheximide (CHX, Fig. 3a,c) or the proteasome inhibitor MG132 (Fig. 3b,d) were added to the culture medium, along a time course between 0 and 100 minutes (CHX) or 0 and 360 minutes (MG132). In HEK293T/FlagMyc cells, we found that PRMT5 inhibition produced a decrease in FlagMyc protein levels upon CHX administration (Fig. 3a, middle), when compared to the control vehicle. This effect was also confirmed, and more evident, on endogenous Myc in GSCs (Fig. 3c, middle). Importantly, the enhanced degradation of Myc protein in the presence of EPZ015666 was not responsible for the decrease in Myc S-dimethylation (Fig. 2c). Indeed, immunoprecipitation experiments revealed that MG132 treatment did not recover S-Myc protein levels but only total FlagMyc ( Supplementary Fig. S6), indicating that the decrease in Myc S-dimethylation was due to PRMT5 catalytic inhibition, although a combined effect with proteasome degradation may also be hypothesized. On the contrary, PRMT1 inhibition stabilized both FlagMyc and endogenous Myc upon CHX treatment (Fig. 3a,c, right). Proteasome inhibition by MG132 produced a similar, slight, but statistically significant, increase of FlagMyc protein levels in EPZ015666 and 9i HEK293T/FlagMyc treated cells (Fig. 3b). In GSCs, control vehicle and EPZ015666 produced similar effects upon MG132 exposure (Fig. 3d, left and middle, respectively). Conversely, 9i dramatically enhanced endogenous GSCs Myc protein content upon MG132 administration (Fig. 3d, right). These experiments suggest that in both HEK293T cells and GSCs, PRMT5 have a protective effect on Myc protein, while PRMT1 may allow its proper turnover or promote its degradation. However, these effects are particularly evident in GSCs.

Myc is preferentially S-dimethylated in GSCs grown as neurospheres. To evaluate endogenous
Myc dimethylation in GSCs, neurospheres were treated for 24 hrs with either 5 μM EPZ015666 or 3 μM 9i. Cell extracts were immunoprecipitated with either SYM10 or ASYM24 antibody. An anti-Myc antibody was used to detect endogenous Myc. As shown in Fig. 4a, EPZ01566 treatment inhibited and 9i enhanced S-Myc, which represented the major Myc dimethylated species in GSCs. Consistently, 9i treatment decreased and EPZ015666 increased AS-Myc levels, which were, conversely, barely detectable in vehicle-treated GSCs. The presence of almost exclusively S-Myc species in GSCs grown as neurospheres prompted us to investigated whether Myc dimethylation pattern could change exposing GSCs to different stimuli. Indeed, we found that in GSCs grown in 2% FBS-containing medium, which promotes their astrocytic differentiation 36 , S-Myc levels were progressively reduced along a time course between 4 hours and 7 days, while AS-Myc remained quite stable (Fig. 4b, left). Consistently, almost total disappearance of PRMT5 and a scarce effect on PRMT1 expression upon FBS culture were observed, together with a reduction in Myc levels (Fig. 4b, middle). This also paralleled with immunoprecipitation experiments showing that Myc and PRMT1 could still interact in differentiating GSCs, while PRMT5 was www.nature.com/scientificreports www.nature.com/scientificreports/ only barely detected (Fig. 4b, right and model in panel c). In summary, these data suggest that also in GSCs Myc is both S-and AS-dimethylated and that PRMT5 and PRMT1 are competitors for Myc as a substrate. Furthermore, S-dimethylated Myc is the prevalent Myc species in GSCs grown as neurospheres whereas in differentiating GSCs its levels strongly decrease in a time-dependent manner. Differently, PRMT1 protein level and activity were weakly reduced during GSCs differentiation. www.nature.com/scientificreports www.nature.com/scientificreports/

PRMT1 and PRMT5 inhibition differentially modulates Myc recruitment at its target promoters in GSCs.
To evaluate the effect of PRMT1 and PRMT5 on Myc binding at its specific chromatin loci in GSCs, ChIP experiments were performed, either in stem or differentiation conditions, in the presence or absence of EPZ015666 or 9i or control vehicle (Fig. 5). Two well known Myc promoter targets -cyclin D1 (CCND1) and the Epithelial Growth Factor Receptor variant III (EGFRvIII) 37 -the stem gene promoters SRY-box 2 (SOX2), bound by Myc in GSCs 34 together with the Glial Fibrillary Astrocytic Protein (GFAP) promoter 38 , were tested. In stem conditions, either PRMT5 or PRMT1 inhibition did not affect Myc binding on CCND1 and EGFRvIII promoters. Conversely, SOX2 promoter was sensitive only to PRMT1 inhibition, whereas, either PRMT5 or PRMT1 inhibition decreased Myc binding on GFAP promoter. Interestingly, Myc detached from all these promoters in differentiating control cells. PRMT5 inhibition had no effect on Myc recruitment in these experimental conditions, whereas 9i treatment restored Myc binding above control levels only on CCND1 and EGRvIII promoters. Altogether, these data address a role to both PRMT5 and PRMT1 catalytic activity in regulating Myc loading at promoters.

PRMT5 affects GSCs cell cycle progression and neurospheres formation.
To evaluate in depth the effect of either PRMT1 or PRMT5 on GSCs biological properties, cell cycle, neurospheres formation, viability and apoptosis were evaluated. Cell cycle analysis (Fig. 6a, left) shows that a low, but statistically significant, number www.nature.com/scientificreports www.nature.com/scientificreports/ of GSCs treated for 7 days with EPZ015666 were arrested in S-phase, whereas an accumulation of cells in the subG1 phase was mostly observed upon 9i administration. Consistent with the role of PRMT5 in sustaining GSCs self-renewal 28 , neurospheres formation was extremely compromised after 10 days of EPZ015666 treatment, while 9i had no effect, when compared to control solvent (Fig. 6a, middle). This was paralleled by a decrease of SOX2 protein upon EPZ015666 exposure (Fig. 6a, right). Nevertheless, no differences in cell viability nor in apoptosis rate were observed in the presence of the two drugs, when compared to controls (not shown). PRMT1 affects GSCs differentiation. As known, mesenchymal GSCs differentiate into astrocytes when cultured in the presence of foetal bovine serum (FBS) 36 and we found that, in this condition, Myc is preferentially AS-dimethylated (Fig. 4b). Therefore we asked whether the inhibition of either PRMT1 or PRMT5 catalytic activity could modulate GSCs differentiation properties. To this aim, GSCs were pre-treated for 24 hrs with either EPZ015666 or 9i (or control vehicle); thereafter cells were shifted to FBS-containing medium and allowed to differentiate for 7 days. Drugs were refreshed every 48 hrs. Immunofluorescence analysis showed that control cells expressed GFAP but were negative for the stemness markers SOX2, Oct4, Nanog and Olig2 39 (Fig. 7a). Also TuJI, a neuronal marker, was undetectable, as expected. Comparable data were obtained with EPZ015666 (Fig. 7b,d), whereas different results were observed upon PRMT1 inhibition. Differentiating GSCs exposed to 9i still did not express Olig2, TuJI, SOX2, Oct4 and Nanog proteins, and dramatically lost GFAP (Fig. 7c,d), as also www.nature.com/scientificreports www.nature.com/scientificreports/ confirmed by a decrease in its mRNA level, as shown in panel e. mRNAs encoding for the other protein markers analyzed were not modulated, upon neither PRMT5 nor PRMT1 inhibition (not shown). Furthermore, since PRMT1 activity is associated with astrocytic differentiation 40 , we monitored H4R3me2as and H4R3me2s levels. A prevalence of H4R3me2as was observed in control and EPZ015666 treated cells. Consistently, 9i blocked H4R3 AS-dimethylation, whereas a slight increase in H4R3me2s was observed (Fig. 7d). In summary, PRMT1 appears to support GSCs astrocytic differentiation, since its inhibition led to the loss of FBS-induced GFAP-positive cells. On the contrary, PRMT5 inhibition did not have any impact on GSCs commitment into the astrocytic lineage.

Discussion
GBM still represents a great challenge for oncologists. Its high genetic, epigenetic and phenotypic heterogeneity have made impossible to identify valuable therapeutic routes, so far. PRMT1, PRMT5 and Myc have been all indicated as important players in GBM progression 5,41 . Specifically, Myc and PRMT5 have been proved as master regulators of GSC pluripotency 8,28 , while PRMT1 loss does not affect typical pluripotency markers 42 . Myc interacts with PRMT5 33 and PRMT1 is required for p300 recruitment at Myc-regulated gene promoters 32 . Further, PRMT1 and PRMT5 interact in GBM 29 . However, no direct association between Myc, PRMT1 and PRMT5 has been reported. Here we show, for the first time, that Myc, PRMT5 and PRMT1 form a ternary complex, both in recipient HEK293T cells and GSCs (Fig. 1c,d). More importantly, we show that Myc is dimethylated by both PRMT1 and PRMT5 (Figs 2 and 4), being the presence of PRMT5 required for Myc/PRMT1 binding, at least in HEK293T cells (Figs 1f and 2e). Therefore, PRMT5 seems to function not only as a catalytic enzyme but also as a fundamental platform of the complex (Fig. 1f), at least in this cell system. Indeed, PRMT5 inhibition impaired also Myc AS-dimethylation (Fig. 2e). The inhibition of the catalytic activity of either PRMT1 or PRMT5 enhances Myc S-or AS-dimethylation, respectively (Figs 2b,c and 4a), without disrupting the complex and indicating that, as occurs for H4R3, they compete for Myc as a substrate. Myc dimethylation is consistent with the S-and AS-dimethylation of other transcription factors. Indeed, PRMT5 dimethylates and activates the Nuclear Factor kB (NF-kB) p65 subunit 43 , while AS-dimethylation by PRMT1 results in an impairment of the NF-kB RelA subunit binding to DNA 44 . Similarly, E2F-1 transcriptional specificity is regulated by either PRMT5-dependent S-or PRMT1-dependent AS-dimethylation, which are specifically induced by appropriate stimuli 45 . Further, NMyc, the neural-specific member of the Myc protein family, is both S-and AS-dimethylated by PRMT5 30 and PRMT1 31 , respectively, and these modifications modulate protein stability. In this regard, our data show that, in both HEK293T cells and GSCs, PRMT5 stabilizes Myc (Fig. 3a,c, middle), while PRMT1 seems to regulate its proper turnover, as preventing PRMT1 catalytic activity maintained Myc protein levels constant upon de novo protein synthesis inhibition (Fig. 3a,c right) and induces a dramatic increase in GSCs Myc protein content, upon proteasome blockade (Fig. 3d, right). These results, together with the observation that GSCs possess an almost exclusively S-Myc when cultured in stem conditions (Fig. 4a), lead to the hypothesis that S-dimethylation may sustain Myc oncogenic function in GSCs. This is consistent with Myc and PRMT5 overexpression in GBM and with their role in supporting GSC stemness 4,28 . Inhibiting either PRMT1 or PRMT5 activity affects Myc recruitment at some of its target promoters (Fig. 5). In stem conditions, we observed that Myc recruitment was either unaffected (CCND1 and EGFRvIII promoters) or sensitive to PRMT1 inhibition (SOX2 promoter) or to both PRMT1 and PRMT5 inhibition (GFAP promoter), indicating that Myc binding activity is modulated by PRMT1 and PRMT5, either alone or in combination, in a gene-specific manner. Whether this phenomenon depends on Myc differential dimethylation, which may regulate its association to co-activators/co-repressors 32 , its promoter affinity and transcriptional specificity 45 or on chromatin accessibility, which may impact Myc promoter binding efficacy, or on additive effects, is an issue still to be determined. Interestingly, upon differentiation Myc binding at promoters was strongly reduced and PRMT5 inhibition had a little impact on its recruitment, whereas PRMT1 inhibition rescued Myc binding at CCND1 and EGFRvIII promoters above control levels. Since S-Myc strongly decreases upon 7 days of differentiation (Fig. 4b), it may be supposed that AS-Myc may have a lower affinity for chromatin, as occurs for RelA 44 , and inhibiting AS-dimethylation may partially restore Myc binding capacity, at least on a subset of promoters.
At the functional level, inhibiting PRMT5 catalytic activity led to a statistically significant S-phase cell cycle accumulation of a small GSCs population and to the failure of neurospheres generation, as also reported by others 28 (Fig. 6). On the other hand, PRMT1, which has no impact on GSCs maintenance, seems to dramatically affect GSCs differentiation (Fig. 7). Indeed, arginine methylation is critical for the onset of neural specific lineages during embryogenensis. In particular, PRMT1 activity is important for astrocytic differentiation of embryonic neural precursors 40 . Consistently, we have found that PRMT1 catalytic inhibition impairs GSCs differentiation into astrocytes (Fig. 7), while the loss of PRMT5 activity did not affect GSCs differentiation properties. Along the GSCs differentiation process, we observed a progressive decrease of S-Myc, which was paralleled by a decrease in PRMT5 and Myc protein levels, whereas AS-Myc remained stable (Fig. 4b). Therefore, upon differentiation, the S-Myc/AS-Myc ratio changes, in favour of AS-Myc. Consistently, we still detected Myc/PRMT1 binding after 7 days of differentiation (Fig. 4b, right), together with a residual amount of PRMT5 (Fig. 4b, middle). However, it has to be still determined whether Myc/PRMT1 interaction is direct or indirect and we cannot exclude the presence of other interactors to mediate Myc/PRMT1 association (see model in Fig. 4c).
We are aware that our observations lack mechanistic insights. In fact, the modifications in Myc binding ability at promoters (Fig. 5), GSCs cell cycle profile, failure of neurospheres formation (Fig. 6) and changes in the differentiation capability (Fig. 7) cannot be definitely addressed to S-or AS-Myc species. The identification of Myc dimethylated arginine(s), the production of specific, either unmethylatable or methyl-mimicking, Myc mutants, followed by depletion/reconstitution experiments to substitute the endogenous wild type Myc, is still missing and will be object of our future investigations.
In conclusion, our work reports, for the first time, that Myc associates with both PRMT1 and PRMT5 being differentially dimethylated. S-dimethylation protects Myc from degradation, while AS-dimethylation allows Myc proper turnover. We hypothesize S-Myc as typical of aggressive GSCs, as S-Myc/AS-Myc ratio decreases in differentiating, less aggressive, cells (Fig. 4b). In this view, S-Myc species may be suggested as a novel diagnostic and prognostic tool for GBM, once a specific antibody will be produced. Although thorough investigations are required to validate the hypothesis that S-or AS-Myc dimethylation may control Myc promoter specificity and/or binding (Fig. 5), this opens a completely new scenario on Myc epigenetic functions and may shed light onto some Myc properties which are not fully elucidated. Moreover, these findings suggest that acting on Myc S-dimethylation may represent a novel and unexplored therapeutic strategy for GBM, avoiding Myc complete silencing, which produces highly detrimental side effects in healthy cells.