Automethylation of Protein Arginine Methyltransferase 8 (PRMT8) Regulates Activity by Impeding S-Adenosylmethionine Sensitivity*

Background: Protein arginine methyltransferase (PRMT) 8 is negatively regulated by its N terminus, which contains two automethylation sites. Results: Preventing automethylation increased sensitivity to AdoMet. Mimicking methylation had the opposite effect. Conclusion: Automethylation of PRMT8 negatively regulates activity by decreasing affinity for AdoMet. Significance: Selective reduction of AdoMet affinity for PRMT8 by automethylation is a unique regulatory mechanism among protein arginine methyltransferases. Protein arginine methyltransferase (PRMT) 8 is unique among the PRMTs, as it has a highly restricted tissue expression pattern and an N terminus that contains two automethylation sites and a myristoylation site. PRMTs catalyze the transfer of a methyl group from S-adenosylmethionine (AdoMet) to a peptidylarginine on a protein substrate. Currently, the physiological roles, regulation, and cellular substrates of PRMT8 are poorly understood. However, a thorough understanding of PRMT8 kinetics should provide insights into each of these areas, thereby enhancing our understanding of this unique enzyme. In this study, we determined how automethylation regulates the enzymatic activity of PRMT8. We found that preventing automethylation with lysine mutations (preserving the positive charge of the residue) increased the turnover rate and decreased the Km of AdoMet but did not affect the Km of the protein substrate. In contrast, mimicking automethylation with phenylalanine (i.e. mimicking the increased hydrophobicity) decreased the turnover rate. The inhibitory effect of the PRMT8 N terminus could be transferred to PRMT1 by creating a chimeric protein containing the N terminus of PRMT8 fused to PRMT1. Thus, automethylation of the N terminus likely regulates PRMT8 activity by decreasing the affinity of the enzyme for AdoMet.

Protein arginine methyltransferase (PRMT) 8 is unique among the PRMTs, as it has a highly restricted tissue expression pattern and an N terminus that contains two automethylation sites and a myristoylation site. PRMTs catalyze the transfer of a methyl group from S-adenosylmethionine (AdoMet) to a peptidylarginine on a protein substrate. Currently, the physiological roles, regulation, and cellular substrates of PRMT8 are poorly understood. However, a thorough understanding of PRMT8 kinetics should provide insights into each of these areas, thereby enhancing our understanding of this unique enzyme. In this study, we determined how automethylation regulates the enzymatic activity of PRMT8. We found that preventing automethylation with lysine mutations (preserving the positive charge of the residue) increased the turnover rate and decreased the K m of AdoMet but did not affect the K m of the protein substrate. In contrast, mimicking automethylation with phenylalanine (i.e. mimicking the increased hydrophobicity) decreased the turnover rate. The inhibitory effect of the PRMT8 N terminus could be transferred to PRMT1 by creating a chimeric protein containing the N terminus of PRMT8 fused to PRMT1. Thus, automethylation of the N terminus likely regulates PRMT8 activity by decreasing the affinity of the enzyme for AdoMet.
Arginine methylation is a post-translational modification catalyzed by the protein arginine methyltransferase (PRMT) 2 family (1)(2)(3). PRMTs catalyze the transfer of a methyl group from S-adenosylmethionine (AdoMet) to a peptidylarginine on a target protein substrate (1)(2)(3). The PRMTs methylate a large variety of substrates and are involved in many biological processes from DNA repair to signal transduction (1)(2)(3). PRMT1, the predominant member, is responsible for 85% of the arginine methylation in the cell, is ubiquitously expressed, and is embryonically lethal when genetically deleted (4). In contrast, much less is known about its closely related family member, PRMT8. PRMT8 is Ͼ84% identical to PRMT1 in the methyltransferase core but has a unique N terminus that is known to inhibit its activity (5,6). The N terminus of PRMT8 contains two automethylation sites and a myristoylation site, which localizes PRMT8 to membranes within the cell, and is the only PRMT known to have this localization (5,6). Additionally, in contrast to all other PRMT family members, PRMT8 is the only family member with a highly restricted expression pattern; PRMT8 is expressed only in the central nervous system (5,7,8).
For PRMTs, enzyme kinetics can reveal many particular aspects of catalysis, such as substrate recognition motifs and the order of substrate binding and product release (25)(26)(27)(28)(29)(30)(31). The kinetics of PRMT1-catalyzed methylation are well character-ized, and it was determined that the kinetic mechanism proceeds in a random sequential fashion, where AdoMet and the protein substrate bind in a random fashion (26).
A wide range of factors are known to regulate the activity of PRMTs (9,(32)(33)(34)(35). The formation of homodimers is essential for the function of PRMTs, and heterodimers have been found to also regulate activity (18, 20, 21, 33, 36 -41). Most PRMTs also have distinct regions on their N terminus that confer substrate specificity and regulate activity (1,3,32,34,35,42). For example, PRMT3 contains a zinc finger domain that is required for methylation of RNA-associated substrates (36). Among post-translational modifying enzymes, self-modification often alters the activity and/or substrate specificity of the enzyme (43,44). For example, automethylation of the lysine methyltransferase G9a, while not affecting its methyltransferase activity in vitro, allows it to bind to other proteins, and mutating this site prevents its specific localization within the cell (43). Similarly, mutation of an automethylation site on CARM1 impairs CARM1-activated transcription and mRNA splicing (45). Automethylation of PRMT6 increases stability of the enzyme and regulates its anti-HIV-1 activity (46). Currently, the mechanism by which the N terminus inhibits the enzymatic activity of PRMT8 is unknown. In this study, we made specific mutants of the two automethylation sites, either to a lysine to preserve charge but prevent automethylation or to a phenylalanine to mimic automethylation, and examined the effect of self-modification on PRMT8 activity. Herein, we report that automethylation regulates AdoMet binding to PRMT8, thereby inhibiting its methyltransferase activity.
Immunofluorescence-HeLa cells were plated onto glass coverslips and transfected using Lipofectamine LTX (Invitrogen) following the manufacturer's instructions. The cells were washed with PBS, stained for 15 min with Alexa Fluor 488conjugated wheat germ agglutinin (Invitrogen) at 0.5 g/ml, washed with PBS, fixed in 4% paraformaldehyde, and stained with DAPI. The coverslips were mounted onto slides and imaged by fluorescence microscopy. Images were analyzed using ImageJ and analyzed with JACoP (47).
Site-directed Mutagenesis-Mutagenesis was performed with the QuikChange Lightning site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions.
Recombinant Protein Expression and Purification-Over-Express C41(DE3)pLysS chemically competent cells (Lucigen) were transformed with the above constructs and grown on LB medium ϩ antibiotic agar plates. Selected colonies were grown in 2ϫ YT medium to A 600 ϭ 0.8 and induced with 0.25 mM isopropyl ␤-D-thiogalactopyranoside for 4 h at 37°C. Cell pellets were resuspended in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, and 10 mM imidazole) supplemented with 1 mg/ml lysozyme (Sigma), sonicated, and centrifuged at 17,000 ϫ g for 30 min. The clarified lysates were collected and incubated overnight with nickel-nitrilotriacetic acid slurry (Qiagen) at 4°C. The nickel-nitrilotriacetic acid beads were collected at 1000 ϫ g for 5 min and washed twice with lysis buffer ϩ 30 mM imidazole for 15 min. Protein was eluted three times with lysis buffer ϩ 300 mM imidazole for 1 h at 4°C. PD-10 columns (GE Healthcare) were used to size-exclude contaminants and exchange the buffer to PBS, and protein-containing fractions were concentrated using an Amicon centrifugal filter (Millipore). Protein concentration was determined using the BCA kit (Pierce). The purification yielded Ͼ95% purity (by SDS-PAGE). GST-human fibrillarin (residues 1-148) fusion protein (designated GST-GAR) was purified in a similar fashion using PBS as a lysis buffer, GST slurry (Thermo Scientific) to capture the protein, and 20 mM glutathione to elute.
In Vitro Methylation Assays-Recombinant protein (0.8 M) was incubated overnight at room temperature in standard methylation buffer (20 mM Tris-HCl (pH 8.0), 200 mM NaCl, and 0.4 mM EDTA) with 1 g of substrate protein (histone H4 or GST-GAR) and 2 l of [ 3 H]AdoMet (2 Ci). Reactions were quenched with 6ϫ SDS loading buffer, boiled for 5 min, and separated by SDS-PAGE. Gels were stained with Coomassie stain (0.25% Coomassie Brilliant Blue, 50% MeOH, and 10% acetic acid) and destained with 50% MeOH and 10% acetic acid. After a 30-min incubation in Amplify (GE Healthcare), gels were dried and exposed to HyBlot CL film at Ϫ80°C.
Steady-state Kinetics Assays-Reactions were performed in methylation buffer for kinetic studies (50 mM HEPES (pH 8.0), 1 mM EDTA, 50 mM NaCl, and 0.5 mM dithiothreitol) with 30 M [ 14 C]AdoMet and varying concentrations of histone H4 or vice versa. Mixtures were preincubated at 37°C for 10 min. Enzyme was then added (0.4 M for PRMT1, ⌬N-PRMT8, and Ch8-1 or 0.8 M for PRMT8 and point mutants), and reactions were incubated for 15 min (within steady state for PRMT1, ⌬N-PRMT8, and Ch8-1) (data not shown) or 60 min (within steady state for PRMT8 and mutants) (data not shown) before being quenched with 6ϫ SDS-PAGE loading buffer and boiled at 95°C for 5 min. Samples were separated by SDS-PAGE (Bio-Rad), stained with Coomassie stain, and destained overnight. After rinsing in H 2 O, gels were incubated in Amplify for 30 min, and dried. Incorporated radioactivity was measured using a PhosphorImager, and values were fit to Equation 1 using GraFit v7.
Co-immunoprecipitation Assays-Samples were prepared in standard methylation buffer. GST-tagged PRMT8 or PRMT8-E194Q (0.1 M) was incubated overnight with either His-tagged PRMT8 or PRMT8-E194Q (0.1 M) with 3 l of anti-GST antibody (Cell Signaling 2624) and 20 l of protein G slurry (Amersham Biosciences) at 4°C. Beads were washed three times with PBS, separated on a SDS-polyacrylamide gel, transferred to PVDF membrane, and probed with anti-PRMT1 antibody (Upstate 07-404), which cross-reacts with PRMT8 via immunoblotting.
Inhibitor Kinetic Assays-Prior to the 15-min incubation at 37°C, PRMT1 or ⌬N-PRMT8 was incubated with the isolated N-terminal domain (amino acids 1-69) of PRMT8 for 1 h at room temperature. The subsequent steps were performed as described above. Values for incorporated radioactivity were fit to competitive inhibition (Equation 2), mixed inhibition (Equation 3), noncompetitive inhibition (Equation 4), and uncompetitive inhibition (Equation 5). Appropriate fit was determined by reduced 2 analysis.

Removal of the PRMT8 N Terminus Results in Increased
Enzymatic Activity-The activity and/or substrate specificity of the PRMTs is often regulated by the N terminus of the enzyme (1,3,32,34,35,42). Previously, Clarke and co-workers (6) showed that the N terminus of PRMT8 negatively regulates its activity because recombinant PRMT8 lacking the first 60 residues (⌬N-PRMT8) has significantly more activity compared with the full-length enzyme. Indeed, PRMT1 and ⌬N-PRMT8 can be regarded as very similar enzymes because they are Ͼ84% identical. Because this work was done using an N-terminal GST tag, we determined the kinetic activity of ⌬N-PRMT8 with a C-terminal His tag compared with C-terminal His-tagged versions of PRMT1 and full-length PRMT8 under steady-state conditions. As shown in previously published studies (5, 26), PRMT1 has much higher activity than PRMT8. For example, the k cat /K m value for AdoMet is 210-fold higher than that obtained with PRMT8 ( Fig. 1 and Table 1). Consistent with the increase in activity seen by Clarke and co-workers, both k cat(AdoMet) and k cat(H4) were dramatically increased (between 10-and 20-fold) upon deletion of the N terminus of PRMT8 (compare the values for ⌬N-PRMT8 versus full-length PRMT8) ( Fig. 1 and Table 1). Thus, our C-terminal His-tagged versions of PRMT8 behaved as expected. Furthermore, the lack of the first 60 residues of the N terminus did not result in a change in K m(H4) . In contrast, the K m(AdoMet) was decreased in the truncated version of PRMT8 compared with full-length PRMT8 ( Table 1), suggesting that the N-terminal domain of PRMT8 regulates the affinity of PRMT8 for the AdoMet cofactor but not the protein substrate.
Automethylation Occurs in cis-Although the enzyme core amino acid sequence of PRMT8 is Ͼ84% identical to PRMT1, the 80-residue N terminus of PRMT8 is unique among PRMTs (5,6). It contains a myristoylation site at Gly-2 and two automethylation sites, i.e. Arg-58 and Arg-73 (5,6). Because dimer formation is necessary for PRMT activity (18, 20, 37) and mutations in the dimerization arms of multiple PRMTs drastically  Table 1. decrease enzymatic activity (18,20), we sought to determine whether automethylation of PRMT8, which is also known to dimerize (6), occurs in cis or trans. To distinguish between these two possibilities, we utilized GST-or His-tagged recombinant versions of PRMT8 to track automethylation. In addition, we created a catalytically impaired PRMT8 mutant, i.e. PRMT8-E194Q ( Fig. 2A), and confirmed that this mutant lacked significant activity (Fig. 2C). The corresponding mutation in PRMT1 decreased activity by 30 -100-fold (20,48). Additionally, we confirmed that heterodimers formed between differently tagged enzymes by pulldown assays (Fig.  2B). For these experiments, PRMT8-His and GST-PRMT8-E194Q were incubated together, as well as the reverse combination, i.e. GST-PRMT8 and PRMT8-E194Q-His, for 1 h with [ 3 H]AdoMet with no additional methyl-accepting substrate present in the reaction. Automethylation was detected only on the wild-type PRMT8 subunit and never on PRMT8-E194Q (Fig. 2D). Thus, automethylation occurs in cis on the PRMT8 homodimer.
Automethylation of Arg-73 Is Dependent on Arg-58-To determine whether automethylation of PRMT8 regulates methyltransferase activity, we used site-directed mutagenesis to create point mutants that either mimicked or prevented automethylation. The use of these mutants allowed for the control of the methylation status of the bacterially purified wildtype enzyme, which is likely a heterogeneously methylated population. Initially, Arg-58 and Arg-73 were individually mutated to lysine to generate the PRMT8-R58K and PRMT8-R73K mutants; lysine was used because it preserves the positive charge while preventing methylation (Fig. 3A). PRMT8-R58F and PRMT8-R73F were also generated because phenylalanines are often used to mimic the increased hydrophobicity of a methylated arginine (Fig. 3A) (49).
To establish whether the arginine methylation states of Arg-58 and Arg-73 are interdependent, we examined the automethylation activity of the PRMT8-R58K and PRMT8-R73K mutants using [ 3 H]AdoMet as the methyl donor. Fig. 3B shows that the R58K mutant did not automethylate, whereas the R73K mutant did (6). Interestingly, PRMT8-R73K showed a higher level of Arg-58 automethylation compared with wildtype PRMT8 (Fig. 3C). To confirm that the automethylation detected with the PRMT8-R73K mutant is located on Arg-58, we examined the automethylation of the double lysine mutant, PRMT8-R58K/R73K. In this case, no automethylation was detected (Fig. 2B), strongly suggesting that Arg-58 is the site of automethylation on the PRMT8-R73K mutant. There are at least two interpretations of the above results, including the possibility that Arg-58 is the major site of automethylation and that its modification requires a positively charged residue at position 73. Alternately, automethylation of Arg-58 may occur  independently of Arg-73 methylation, but Arg-58 methylation is a prerequisite for methylation of Arg-73. In contrast to the results obtained for the lysine mutants, neither of the phenylalanine mutants, i.e. PRMT8-R58F and PRMT8-R73F, automethylated (Fig. 1, B and C). This result is consistent with the notion that Arg-58 is the preferred site of automethylation and that its methylation requires a positively charged residue, either lysine or arginine, at position 73.
Automethylation Does Not Alter Membrane Targeting of PRMT8-In eukaryotic cells, PRMT8 is modified by myristoylation at Gly-2, which promotes membrane localization of the enzyme (5). Because the myristoylation modification occurs cotranslationally, modification of PRMT8 by myristoylation must occur prior to automethylation (50,51). However, it is possible that automethylation regulates the association of myristoylated PRMT8 with the cell membrane. To determine whether any of the automethylation mutants affect the subcellular localization of PRMT8, we transiently transfected HeLa cells with GFP-tagged PRMT8 variants. Transfected cells were also stained with fluorescently labeled wheat germ agglutinin (Alexa Fluor 555-labeled) to identify the cell membrane and with DAPI as a nuclear stain (Fig. 4A). Cells with GFP-tagged PRMT8 were imaged by microscopy, and colocalization was analyzed by determining the Pearson's coefficient (47). Additionally, colocalization was visualized by determining fluorescence intensity across the diameter of the cell. As expected, wild-type PRMT8 and point mutants all displayed a similar level of colocalization (Pearson's coefficient of 0.91-0.98) compared with GFP alone (Pearson's coefficient of 0.76) (Fig. 4, A  and B). In total, these results indicate that the membrane localization of PRMT8 is independent from its automethylation activity.
Blocking Automethylation Increases Sensitivity to AdoMet-To determine whether automethylation regulates enzymatic activity, we next investigated whether prevention of automethylation using the PRMT8-R58K or PRMT8-R73K mutant changes the kinetic parameters of the PRMT8 enzyme. As wild-type PRMT8 purified from bacteria likely consists of a heterogeneous population of automethylated and unautomethylated enzymes due to the presence of AdoMet within the bacterial cells, the use of the lysine and phenylalanine mutants allowed for the control of the automethylation status of PRMT8. Because the biological substrates of PRMT8 are currently unknown, we used a histone H4 peptide containing a single arginine (Arg-3), which can be methylated by PRMT1 and PRMT8 in vitro (6). As depicted in Table 1, when either automethylation site was blocked by lysine mutations (R58K or R73K), in kinetic assays with constant H4 and varying AdoMet concentrations, the turnover rate (k cat(AdoMet) ) was increased by 1.5-3-fold compared with wild-type PRMT8, and the K m(AdoMet) was decreased by 2-fold. In contrast, when the AdoMet concentration was held constant and the H4 concentration was varied, the turnover rate also increased, but there was no corresponding decrease in K m(H4) ( Table 1).
These data led us to postulate that a further increase in catalytic activity could be achieved by generating the R58K/R73K double mutant. As expected, the k cat /K m(AdoMet) for the R58K/ R73K enzyme was increased to a level that was comparable to the enzyme lacking the entire N terminus (⌬N-PRMT8) ( Table  1). Because the increase in k cat /K m(AdoMet) was largely driven by a decrease in K m(AdoMet) with little to no effect on the kinetic parameters for histone H4, these data suggest that automethylation likely regulates either the conformation of the AdoMetbinding pocket or the ability of AdoMet to access this site.
Mimicking Automethylation Decreases Sensitivity to AdoMet-Because blocking methylation increased both turnover rates and decreased K m(AdoMet) , we predicted that a mimic of methylarginine, phenylalanine, would have the opposite effect, i.e. decrease the turnover rate and increase the K m(AdoMet) . As expected, the k cat(AdoMet) was reduced for both of the phenylalanine single mutants, corresponding to a decrease in k cat / K m(AdoMet) ( Table 1). Importantly, although a similar decrease in activity was seen in the k cat(H4) , no corresponding decrease was detected in the K m(H4) ( Table 1). Thus, automethylation of PRMT8 most likely affects R58K/R73K sensitivity but not the affinity of the enzyme for H4.
Likewise, we sought to determine whether the double methylation mimic would also enhance the differences seen with the individual phenylalanine mutants. Interestingly, in contrast to the single mimic mutants PRMT8-R58F and PRMT8-R73F, the activity of the double phenylalanine mutant, PRMT8-R58F/ R73F, was dramatically increased relative to wild-type PRMT8 (Table 1). Although the reason for this discrepancy is unclear, it likely relates to the fact that phenylalanine is an imperfect mimic of a methylarginine because it lacks positive charge and is therefore incapable of forming the idealized cationinteractions that are characteristic of methylarginine-and methyllysine-binding domains (52). Regardless, the data for the single mutants provide compelling evidence for the hypothesis that methylarginines bind to a site on the enzyme that alters the affinity of the AdoMet-binding pocket for AdoMet.
Isolated PRMT8 N-terminal Domain Competitively Inhibits AdoMet-Because these data suggested that the N terminus of PRMT8 inhibits AdoMet binding, we sought to determine the mechanism by which the N-terminal domain of PRMT8 acts as an inhibitor. For these studies, we expressed and purified the N-terminal domain (residues 1-69) as a His-tagged fusion protein and used it in kinetic studies with ⌬N-PRMT8.
To determine whether the N-terminal domain can inhibit ⌬N-PRMT8 activity in trans, we also determined the steadystate kinetic parameters for these enzymes in the presence of increasing concentrations of this domain. Based on visual inspection of the double-reciprocal plots and a comparison of the reduced 2 values, the N-terminal domain acted as a competitive inhibitor of AdoMet for ⌬N-PRMT8 (K i(AdoMet) ϭ 6.0 Ϯ 1.3 M) (Fig. 5A). This suggested that the isolated N-terminal domain of PRMT8 would act as a substrate for ⌬N-PRMT8. Unfortunately, methylation of the N-terminal domain by ⌬N-PRMT8 was below the limit of detection (data not shown). Thus, we isolated point mutants of the N-terminal domain of PRMT8 to mimic the methylated or unmethylated states. The R58F point mutant of the N-terminal domain of PRMT8 acted as a competitive inhibitor of AdoMet for ⌬N-PRMT8 (K i(AdoMet) ϭ 2.2 Ϯ 0.5 M) (Fig. 5B), supporting the hypothesis that automethylation of the N-terminal domain inhibits AdoMet binding. In contrast, the R58K point mutant of the N-terminal domain acted as a noncompetitive inhibitor of AdoMet for ⌬N-PRMT8 (K i(AdoMet) ϭ 5.4 Ϯ 1.3 M) (Fig. 5C), suggesting that there are additional mechanisms by which the N-terminal domain regulates PRMT8 activity that do not involve AdoMet binding. This conclusion seems likely because the N terminus acts as both a substrate and an inhibitor of AdoMet binding.
The results obtained for AdoMet were in contrast to the effect of the N-terminal domain of PRMT8 on H4 kinetics, where this domain acted as a noncompetitive inhibitor of H4 (K i(H4) ϭ 8.2 Ϯ 1.4 M) (Fig. 5D). In total, these data, combined with our above studies, suggest that the N terminus of PRMT8 impairs the AdoMet/PRMT8 interaction.
Chimeric PRMT1 with the N-terminal Domain of PRMT8 Has Decreased AdoMet Sensitivity-Finally, to confirm that inhibition is an intrinsic property of the N-terminal domain of PRMT8, we generated a chimeric enzyme containing the PRMT1 core and the N-terminal domain of PRMT8 (referred to as Ch8-1). In steady-state kinetic studies, the K m(AdoMet) of Ch8-1 was increased to a similar level compared with fulllength PRMT8, whereas the K m(H4) remained similar to that of PRMT1 (Table 1). The k cat(AdoMet) and k cat(H4) of Ch8-1 were also decreased by 2-4-fold compared with PRMT1 (Table 1). Additionally, automethylation was detected on Ch8-1 at similar levels compared with PRMT8, whereas PRMT1 and ⌬N-PRMT8 did not have detectable levels of automethylation (Fig. 6). Thus, automethylation of the N-terminal domain of PRMT8 reduces the activity of the enzyme by increasing the concentration of AdoMet necessary for the catalytic reaction to proceed but does not greatly affect the access and binding of the protein substrate.

DISCUSSION
Regulation of enzymatic activity is critical to maintain cellular homeostasis and to allow for the proper functioning of signaling networks. Given that the PRMTs are generally thought to be constitutively active when purified from orthogonal sources, there must be mechanisms to negatively regulate and "turn off" their activity. For example, phosphorylation of CARM1 is known to decrease the activity of this isozyme (53). With respect to PRMT8, the unique N terminus of this enzyme is known to be myristoylated and automethylated, and these modifications appear to be important for regulating the subcellular localization and activity of this enzyme. In fact, deletion of the first 60 residues of the protein results in a large increase in activity (6). Here, we have shown that automethylation within the N-terminal domain of PRMT8 regulates its activity by altering the affinity of PRMT8 for its cofactor AdoMet.
On the basis of our data, we propose that the methyltransferase activity is regulated by the following mechanism (Fig. 7). PRMT8 is unmethylated in its open and active state. As PRMT8 progresses and methylates its endogenous protein substrate, this decreases the availability of unmethylated substrate. The lack of unmethylated substrate results in automethylation of PRMT8, causing the N terminus to bind to the enzyme core and blocking AdoMet access to the catalytic site, leading to downregulation of PRMT8 activity. The N terminus of PRMT8 extends from the conserved ␣-helix of the PRMTs, which is required for enzymatic activity and AdoMet binding (18 -21). Automethylation of the N-terminal domain of PRMT8 may induce a conformational shift in the ␣-helix, preventing its ability to properly bind AdoMet, thus lowering activity.
In support of this model, we used two alternative mutations of the arginine automethylation sites, Arg-58 and Arg-73: lysine, which preserves the positive charge of the residue but is not methylated by PRMT8, and phenylalanine, which mimics the increased hydrophobicity of a methylated arginine (49). Interestingly, we found that neither of the Arg-to-Phe mutants, i.e. PRMT8-R58F and PRMT8-R73F, had detectable levels of automethylation, nor did the methylation-null mutant PRMT8-R58K. However, PRMT8-R73K had increased levels of automethylation, and the double lysine mutant PRMT8-R58K/R73K did not have detectable levels of automethylation. The suggests that Arg-58 is the automethylation site of the PRMT8-R73K mutant. It appears that Arg-58 is the "gateway" automethylation site, without which Arg-73 is not methylated. In addition, the presence of increased automethylation in PRMT8-R73K indicates that Arg-73 is the automethylation site that more strongly affects the methylation activity of PRMT8.
These observations played out when we examined the activity of PRMT8 methylating the substrate histone H4, which is methylated at Arg-3. In both cases of preventing automethylation, PRMT8-R58K and PRMT8-R73K, there was an increase in turnover rates and a decrease in K m(AdoMet) . The opposite was true for PRMT8-R58F and PRMT8-R73F, the methylation mimics; both had decreased methylation activity. Intriguingly, these effects could be transferred to a similar enzyme, PRMT1, by fusing on the N terminus of PRMT8, thus verifying that the N-terminal domain of PRMT8 directly decreases the ability of PRMT8 to utilize AdoMet.
In this study, we determined that automethylation of PRMT8 negatively regulates its methyltransferase activity. Because the biological function of PRMT8 is currently unknown, the involvement of these post-translational modifications in cellular pathways has yet to be determined. However, because automethylation appears to function as a negative regulator by lowering AdoMet affinity of PRMT8, myristoylation of the endogenous enzyme and subsequent membrane association may affect as a positive regulator of catalytic activity, potentially increasing enzyme velocity by increasing the affinity of AdoMet or the protein substrate for PRMT8. Future crystallography studies on the structure of PRMT8 will help provide insight as to how these residues interact with the rest of the protein to regulate activity.