Type I PRMT Inhibition Protects Against C9ORF72 Arginine-Rich Dipeptide Repeat Toxicity

Repeat expansion mutations in the C9ORF72 gene are the most common genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Repeat-associated non-AUG translation of this expansion produces dipeptide repeat proteins (DRPs). The arginine containing DRPs, polyGR and polyPR, are consistently reported to be the most toxic. Here we demonstrated that small molecule inhibition of type I protein arginine methyltransferases (PRMT) protects against polyGR and polyPR toxicity. Furthermore, our findings suggest that asymmetric dimethylation of polyGR and polyPR by Type I PRMTs plays important roles in their cytotoxicity.


INTRODUCTION
A mutation in the C9orf72 gene is the most common known cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) (DeJesus-Hernandez et al., 2011;Renton et al., 2011). The mutation consists of an abnormal expansion of a repeated hexanucleotide sequence (GGGGCC) in the first intron of the C9orf72 gene (DeJesus-Hernandez et al., 2011;Renton et al., 2011). In ALS and FTD, the expanded nucleotide tract is translated through an unconventional mechanism known as repeat-associated non-AUG (RAN) translation (Ash et al., 2013;Mori et al., 2013). Depending on what reading frame RAN translation takes place in, along either the sense or antisense RNA strand, it leads to the generation of five different dipeptide repeat proteins (DRPs) of variable lengths: poly-Glycine-Arginine (polyGR), poly-Proline-Arginine (polyPR), poly-Proline-Alanine (polyPA), poly-Glycine-Alanine (polyGA), and poly-Glycine-Proline (polyGP) (Ash et al., 2013;Mori et al., 2013).
The arginine-containing DRPs in particular have been demonstrated to have detrimental effects in several model systems and to interact with several different pathways (Kwon et al., 2014;Wen et al., 2014; Abbreviations: ALS, amyotrophic lateral sclerosis; FTD, frontotemporal degeneration; DRP, dipeptide repeat protein; RAN, repeatassociated non-AUG; GAR, glycine-and arginine-rich; MMe, monomethylation; ADMe, asymmetric dimethylation; SDMe, symmetric dimethylation; PRMT, protein arginine methyltransferase; SAM, S-adenosyl methionine , ICW, In-Cell Western; IVM, In Vitro Methylation. Kramer et al., 2018). For example, when administered exogenously to U2OS cells, synthetic GR 20 and PR 20 are shown to bind to nucleoli, disrupt RNA splicing and processing, and decrease cell viability (Mori et al., 2013). Our lab has previously demonstrated that exogenous application of synthetic GR 15 and PR 15 to mouse spinal cord neuroblastoma hybrid cells  induces cellular toxicity, as measured by various cell health and function assays and that this toxic effect becomes more severe as the cells are further differentiated toward neurons, with primary neurons exhibiting the greatest toxicity (Gill et al., 2019). In addition, a series of studies involving the expression of the repeat expansion in Drosophila have demonstrated polyGR and polyPR related toxicity (Mizielinska et al., 2014;Freibaum et al., 2015;Lee et al., 2016), with one study revealing the disruption of stress granule assembly due to the presence of polyGR and polyPR (Lee et al., 2016). Other pathways that have been implicated in arginine-containing DRP toxicity include those involved in nucleocytoplasmic transport (Freibaum et al., 2015) and RNA-binding (Lee et al., 2016), though the complete nature of the pathogenesis of polyGR and polyPR remains unclear. Of particular interest, recent studies in ALS suggest a role for arginine methylation in disease progression and in polyGR-related toxicity (Ikenaka et al., 2019;Gittings et al., 2020).
Protein arginine methyltransferases (PRMTs) are a family of enzymes that post-translationally modify proteins by methylating nitrogen atoms of arginine residues. These modifications influence many cellular processes including transcription, RNA processing, signal transduction cascades, DNA damage response, and liquidliquid phase separation (Guccione and Richard, 2019). Specifically, glycine-and arginine-rich (GAR) motifs, typical in histones and RNA binding proteins, are common targets for PRMT mediated modifications that are reported to influence protein localization and gene expression (Thandapani et al., 2013). In the present study we examined whether the cytotoxic effects of exogenously applied polyGR and polyPR would be affected by pharmacological inhibition of PRMT activity.
PRMTs are responsible for the monomethylation (MMe), asymmetric dimethylation (ADMe), and symmetric dimethylation (SDMe) of arginine residues, primarily within a GAR motif (Najbauer et al., 1993;Cheng et al., 2007) and are classified as type I, type II, or type III depending on the type of methylation they catalyze. Type I PRMTs catalyze ADMe with MMe as an intermediate, and include PRMT1, 2, 3, 4, 6 and 8. Type II PRMTs catalyze SDMe with MMe as an intermediate, and include PRMT5 and 9. Type III PRMTs perform MMe only and include PRMT7 (Blanc and Richard, 2017).

NSC-34 Cell Culture
NSC-34 cells (Cedarlane Laboratories, Burlington, ON, CA) were cultured in a complete medium consisting of high glucose Dulbecco's modified eagle medium (DMEM) (Millipore-Sigma, Burlington, MA, USA) supplemented with 10% US-origin fetal bovine serum (Thermo Fisher Scientific, Cambridge, MA, USA), 1% 200 mM L-glutamine solution (Thermo Fisher Scientific, Cambridge, MA, USA), and 1% 10,000 U/mL penicillinstreptomycin solution (Thermo Fisher Scientific, Cambridge, MA, USA). Prior to preparation of NSC-34 complete medium, L-glutamine and penicillin-streptomycin solutions were aliquoted and stored at -20°C, and DMEM/high glucose was stored at 4°C. At each passage, cells were washed once with Dulbecco's phosphate-buffered saline (DPBS) with calcium and magnesium (Thermo Fisher Scientific, Cambridge, MA, USA) and treated with 0.25% Trypsin-EDTA solution (Thermo Fisher Scientific, Cambridge, MA, USA) for 5 min at 37°C and 5% CO 2 for dissociation. Prepared complete medium, DPBS, and Trypsin were always heated in a 37°C water bath before use and stored at 4°C between uses.

Preparation of PRMT Inhibitor Solutions
Small-molecule PRMT inhibitors MS023 and GSK591 (Tocris Bioscience, Briston, UK), MS049 and EPZ020411 (Cayman Chemical, Ann Arbor, MI, USA), GSK3368715 (Medchem Express, Monmouth Junction, NJ, USA), and negative control MS094 (Millipore-Sigma, Burlington, MA, USA) were purchased and stored at -20°C prior to and following reconstitution. After reconstitution using the solvents specified in Table 1, stocks were aliquoted to 15-20 µL and immediately stored at -20°C. UltraPure ™ DNase/RNase-Free Distilled Water (Thermo Fisher Scientific, Cambridge, MA, USA. Plate was read on the LI-COR Odyssey 9120 Infrared Imaging System. Data were expressed as a ratio of 800 channel signal to 700 channel signal (test condition to total protein).

Plating NSC-34 and Dosing With DRPs and PRMT Inhibitors
NSC-34 cells were in clear, flat-bottom, full volume, 96-well tissue culture-treated plates (Thermo Fisher Scientific, Cambridge, MA, USA). One row on the top and bottom of the plate and two columns on either side of the plate were left without cells and contained culture medium only to minimize experimental well volume evaporation. After plating, cells were incubated for 24h at 37°C and 5% CO 2 prior to DRP and/or PRMT inhibitor addition. At time of DRP/PRMT inhibitor addition, desired doses of DRP for challenge and inhibitor for treatment were achieved by diluting aliquots of each stock in warm culture medium. During experiments where both PRMT inhibitors and DRPs were used, PRMT inhibitors were always applied to wells first, followed by DRP application. Vehicle controls were included as wells treated with equivalent DMSO concentrations to those that had been DRPtreated, drug treated, or both. Inhibitor toxicity controls were included as wells treated with the desired doses of drug for the experiment, but no DRP. DRP toxicity controls were included as wells only treated with the doses of DRP used for challenge. Once dosed, plates were incubated for 24 h at 37°C and 5% CO 2 prior to running the WST-1 or LDH assay endpoints. Additional controls needed for each endpoint are specified in the "WST-1 Assay" and "LDH Assay" sections of these methods.

WST-1 Assay
Cells were plated and prepared using the steps described in the "Plating NSC-34 and Dosing with DRPs and PRMT Inhibitors" section of these methods. Other controls for this experiment included wells containing only cells in culture medium, and culture medium only. At time of testing, culture medium was removed from wells and replaced with a warmed, sterile-filtered solution consisting of DPBS with calcium and magnesium and 4.5 g/L of D-glucose (Millipore-Sigma, Burlington, MA, USA).
To wells containing 200 µL of DPBS-glucose solution, 20 µL/well of WST-1 reagent (Millipore-Sigma, Burlington, MA) was applied, and plates were then incubated at 37°C and 5% CO 2 for 1 h before plates were read at 450 nm on a SpectraMax M3 Microplate Reader (Molecular Devices, San Jose, CA, USA). As indicated in figures, WST-1 data was calculated relative to either GR/PR challenge or no challenge conditions using the following formulas. If DMSO was used as a solvent for compounds in test conditions, then same concentrations of DMSO were added to untreated controls.

LDH Assay
Cells were plated and prepared using the steps described in the "Plating NSC-34 and Dosing with DRPs and PRMT Inhibitors" section of these methods. Other controls for this experiment included several sets of wells with only cells in culture medium (one triplicate designated for "untreated," one triplicate designated for "lysed" positive control) and wells with culture medium only. An additional control added only to the transfer plate at time of testing was 5 µL of LDH only.

In Vitro Methylation (IVM) Assay
Individual assays were conducted in 0.5 mL, flat cap PCR tubes (Thermo Fisher Scientific, Cambridge, MA, USA). Systems contained recombinant PRMT1 (Active Motif, Carlsbad, CA, USA), S-(5-adenosyl)-L-methionine iodide (SAM, Sigma-Aldrich), either Histone H4 (Active Motif) or GR 15 (Genic Bio), 10X PBS (Thermo Fisher Scientific), and nuclease-free water (Thermo Fisher Scientific). All reagents were added to achieve a final volume of 30 µl. Up to 0.8 µg of PRMT1, 25 µM of SAM, 3 µM of Histone H4, and 6.7 µM of GR 15 were added to the system. Three microliters of 10X PBS was added to achieve 1X PBS. Once desired ratios of reagents were added to each tube, systems were lightly mixed and incubated for 2 h at 37°C and 5% CO 2 . After the incubation, reactions were stopped using 10 µl of 4x LDS Sample Buffer (Thermo Fisher Scientific).

Statistics
Statistical analyses were performed using Graphpad Prism v.8 and Microsoft Excel. Statistical tests included two-way ANOVAs with Dunnett's, Sidak's multiple comparisons tests, one-way ANOVAs with Dunnett's multiple comparison test, fiveparameter logistical regression models to calculate EC50s, and a four-parameter logistical regression model to calculate IC50s. One 1 µM and 0.2 µM data point for MS049 and one 20 µM data point for EPZ020411 in Figure 1A were excluded for being outside of two standard deviations of their respective means. A full listing of n values that were not represented in main text figures are presented in Supplementary Material Tables  S2-S6. Experiments were performed in technical triplicates or quadruplicates, with n values greater than 3 or 4 indicating combination with biological replicates.

RESULTS
Our lab has reported that exogenous application of GR 15 and PR 15 to NSC-34 cells induces cellular toxicity, as measured by WST-1 metabolism, LDH release, BrdU labeling, and Caspase-3 activity (Gill et al., 2019). In the present study, we use the WST-1 metabolism and LDH release assays to evaluate changes in DRP toxicity in the presence of a PRMT inhibitor.
To test the effects of inhibition of various PRMTs in the presence of polyGR or polyPR, we acquired commercially available, small molecule PRMT inhibitors that were capable of inhibiting either ADMe or SDMe. Because multiple PRMTs are capable of catalyzing asymmetric dimethylation of arginine residues, multiple small molecule inhibitors were selected with varying potencies against the various type I PRMTs. These included MS023, GSK715, EPZ020411, and MS049. Because PRMT5 is the more common and abundant of the two Type II PRMTs across and in various cell types (Stopa et al., 2015), we selected a potent and specific inhibitor of PRMT5, GSK591. As a negative control, we used MS094, a previously described structural analog of MS023, which is known to be inert against all PRMT activity (Eram et al., 2016).
We first determined the potency of the PRMT inhibitors in NSC-34 cells using an In-Cell Western (ICW) assay measuring total ADMe for Type I PRMT inhibitors and total SDMe for the Type II PRMT inhibitor (Table 2 and Figures 1A, B). After establishing a working range of concentrations for each inhibitor, we measured metabolic activity (WST1 metabolism endpoint) and cytotoxicity (LDH release endpoint) in NSC-34 cells challenged by various concentrations of GR 15 or PR 15 with or without co-incubation with PRMT inhibitors.
We found that the Type I PRMT inhibitors were effective at abrogating the decreased metabolic activity and increased cytotoxicity associated with the application of GR 15 or PR 15 , with MS023 in particular demonstrating the lowest EC50s ( Table  2 and Figures 1C-F). At some concentrations, incubation with Type I PRMT inhibitors resulted in complete rescue of GR 15 or PR 15 effects. Most of the inhibitors were inert with regards to LDH and WST1 endpoints in the absence of polyGR and polyPR at concentrations that were effective at abrogating polyGR and polyPR toxicity. The lone exception was EPZ020411. At concentrations at and above 10 µM, MS023, MS049, and EPZ020411 did reduce WST1 metabolism and elevated cytotoxicity, possibly contributing to their bell-shaped doseresponse curves (Supplementary Figures 1A-D and Table  S1). MS094, the reported inert analog of MS023, displayed a negligible effect at inhibiting ADMe at 0.1 and 0.2 µM concentrations and did not abrogate the GR 15 -and PR 15related toxicity at any concentration (Figures 1H, I and Supplementary Figure 2A). The Type II PRMT5 inhibitor, GSK591, did inhibit SDMe but did not abrogate the decreased metabolic activity due to GR 15 and PR 15 challenge ( Figure 1G and Supplementary Figure 3). These results suggest that the activities of Type I PRMTs contribute to the toxicity produced by GR 15 and PR 15 .
One possible interpretation of our data is that the asymmetric dimethylation of the arginine-rich DRPs is essential for arginine-rich DRP toxicity. To evaluate poly-GR as a substrate for Type I PRMT activity we conducted an in vitro methylation assay, using recombinant PRMT1 as the enzyme and S-adenosyl methionine (SAM) as the methyl donor group. We used recombinant Histone H4, a known substrate of PRMT1 26 , as a positive control for asymmetrical and symmetrical dimethylation activity. In one experiment, we used antibodies against total ADMe and Histone H4 asymmetrical dimethylation (H4R3me2a) and revealed that GR 15 is subject to ADMe by PRMT1 in this system and can be increasingly dimethylated when incubated with increasing amounts of PRMT1 (Figure 2A). The H4R3me2a antibody was able to detect the ADMe of GR 15 , possibly due to the antibody having been raised against the H4R3me2a epitope containing a methylated arginine 3, which is preceded by a glycine (Figure 2A).
After determining that GR 15 could be arginine methylated, we had ADMe-GR 15 synthesized. We first compared effects of ADMe-GR 15 challenge to effects of unmethylated GR 15 challenge in our LDH and WST-1 assays in NSC-34 cells. ADMe-GR 15 challenge produced similar levels of cytotoxicity as challenge with unmethylated GR 15 peptide and caused a significant decrease in cellular metabolic function beyond the effects seen with unmethylated GR 15 ( Figures 2B, C). To further elucidate the mechanism behind the protective effects of Type I PRMT inhibitors, we challenged NSC-34 cells with ADMe-GR 15 and dosed with MS023. In contrast to treatment after challenge with unmethylated GR 15 , MS023 was not able to abrogate the toxicity produced by ADMe-GR 15 challenge ( Figures 2D, E). We also had ADMe-PR 15 synthesized and conducted a similar set of experiments as those with ADMe-GR 15 and unmethylated GR 15 . Again, we found that ADMe-PR 15 challenge led to similar levels of cytotoxicity and caused a significantly greater decrease in metabolic activity when compared to unmethylated PR 15 challenge ( Figures 2B, C). Interestingly, MS023 co-incubation led to abrogation of toxicity caused by PR 15 and to, a much lesser extent, toxicity caused by ADMe-PR 15 (Figures 2F, G). Together, these results suggest the importance of asymmetric dimethylation to the toxicity caused by the arginine-rich DRPs however, the post-modification mechanism driving polyGR toxicity could be different than that of polyPR.

DISCUSSION
The present study reveals that Type I PRMT inhibitors can completely abrogate toxicity produced by exogenous polyGR and polyPR challenge in NSC34 cells and suggests that Type I PRMT inhibition is a potential therapeutic strategy for C9orf72associated ALS. We also determined that polyGR is subject to ADMe modification, and the ADMe of exogenous polyGR and polyPR is crucial to the toxicity caused by the arginine-rich dipeptide repeats. The partial rescue of ADMe-PR 15 toxicity by MS023 leaves the question of how the mechanisms of polyGR and polyPR differ. As PRMTs typically react with GAR motifs, methylation and demethylation dynamics could be different between the two DRPs, leading to different responses in the presence of a PRMT inhibitor (Thandapani et al., 2013). Alternatively, it is possible that ADMe-PR 15 interferes downstream with various PRMT substrates, and the addition of an inhibitor partially prevents those interactions. It is noteworthy that our co-incubation of DRP and Type I PRMT inhibitor produced a bell-shaped dose-response curve, possibly indicating a drastic increase in Type I PRMT activity in hormetic response to the inhibitor (Calabrese et al., 2018). Recent work examining dimethylation found in human cortical tissue suggests symmetric dimethylation of DRPs can extend disease duration (Gittings et al., 2020). As symmetric dimethylation prevents  Products of in-vitro methylation assay immunoblotted for asymmetrically dimethylated arginine 3 in Histone 4 and GR dipeptide (left) and total asymmetrical arginine dimethylation and GR dipeptide (right). Also shown is the peptide sequence of Histone 4 with the epitope of the H4R3me2a antibody highlighted. Both blots show ADMe of GR 15 and that it is increasingly dimethylated with increasing amounts of PRMT1. (B) Percent metabolic activity after challenging with ADMe-GR 15 compared to unmethylated GR 15 (****P < 0.0001, ***P = 0.0002, **P = 0.0013, *P = 0.0281) or ADMe-PR 15 compared to unmethylated PR 15 ( #### P < 0.0001, ## P = 0.004; two-way ANOVA with Sidak's multiple comparison; n = 3 for each dose of DRP; mean ± s.e.m.). (C) Percent LDH release after challenging with ADMe-GR 15 compared to unmethylated GR 15 (**P = 0.0083, *P = 0.0117) or ADMe-PR 15 compared to PR 15 ( # P = 0.0239; two-way ANOVA with Sidak's multiple comparison; n = 3 for each dose of DRP; NS P > 0.05 mean ± s.e.m.).

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ACKNOWLEDGMENTS
This work was supported by Augie's Quest. We would like to thank people with ALS and their families and friends who have supported and inspired this work; in particular, we acknowledge those living with C9orf72 repeat expansion-mediated ALS. We would like to thank Beth Levine for her invaluable advice in the drafting of this manuscript. This manuscript has been released as a pre-print at BioRxiv, 2020.05.20.106260.