Regulation of the EBNA1 Epstein-Barr Virus Protein by Serine Phosphorylation and Arginine Methylation

Identification of phosphorylated serine in EBNA1.Thirty milligrams of EBNA1 protein, purified from Sf9 insect cells as previously described (23), was treated with 200 ng of trypsin overnight at 37°C and analyzed on an Applied Biosystems/MDS Sciex QTrap LC/MS/MS mass spectrometer in the Proteomic and Mass Spectrometry Centre at the University of Toronto, Faculty of Medicine. The digested protein was directly infused to the mass spectrometer. A neutral loss scan of H₃PO₄ (98 Da) was performed first to specifically identify the phosphorylated peptides, and then an enhanced product ion scan was performed to sequence the selected phosphorylated peptide.

Construction of plasmids expressing EBNA1 serine mutants.EBNA1 serine mutants, in which the four serines in the EBNA1 325-376 region are converted to alanine or aspartic acid individually or in combination, were generated in the mammalian expression plasmid pc3oriP (56). Mutants are named according to the position of the mutated serine(s), where SSSS refers to Ser334, Ser 342, Ser350, and Ser365, respectively. pc3oriP expressing ASSS or SAAS was created using pc3oriPE (56) as a template for QuikChange (Stratagene) mutagenesis. pc3oriP expressing ASSA, SSAS, SASS, AASS, SDDS, or DDDD was constructed using the following cloning strategy. The EBNA1 coding sequence was PCR amplified from pc3oriPE using primers containing NdeI and BamHI sites and inserted into NdeI and BamHI sites of the pUC19 plasmid. The resulting plasmid was digested with BmsFI, which released an EBNA1 fragment containing residues 325 to 376. The BmsFI-digested plasmid was ligated to two sets of complementary oligonucleotide cassettes containing BmsFI ends. These cassettes together encompassed the residues contained in the released EBNA1 fragment and contained the relevant serine mutation(s). The EBNA1 serine mutants obtained in this manner were then excised from pUC19 using AccI and BamHI and ligated to pc3oriPE digested with the same enzymes. All clones in pc3oriPE were sequenced to confirm the mutations.

For expression in Saccharomyces cerevisiae, the EBNA1 serine mutants were cloned in p416MET25. This involved PCR amplification of the EBNA1 serine mutants in pc3oriP, phosphorylation of the ends of the PCR fragments by T4 polynucleotide kinase, and ligation into the SmaI site of p416MET25.

S. cerevisiae plasmid loss assays.Yeast strain KY320 (MATaleu2-PET56 ura3-52 trp1-Δ1 lys2-801am ade2-101oc his3-Δ200GAL⁺) was transformed with the segregation test plasmid YRp7FR, the hEBP2-expressing plasmid p425PGK.hEBP2, and p416MET25 expressing EBNA1, an EBNA1 serine mutant, or no EBNA1. The plasmids and plasmid loss assays were as described by Kapoor et al. (29). Briefly, positive transformants were grown in selective medium and then diluted into medium nonselective for the segregation test plasmid and grown for 11 generations. Ten-fold serial dilutions of the resulting cultures were then spotted on selective and nonselective plates to assess the stability of the segregation plasmid. The percentage of cells that retained the segregation plasmid was quantified by counting colonies grown on selective and nonselective plates from equal amounts of culture.

Mammalian transient-replication and long-term maintenance assays.HeLa cells were plated at 1 × 10⁶ cells per 6-cm plate 24 h before transfection with the ESCORT II reagent (Sigma) as per the manufacturer's directions. Cells were transfected with pc3oriP (negative control), pc3oriPE expressing EBNA1 (56), or a pc3oriP plasmid expressing an EBNA1 serine mutant. Transfected cells were grown for 3 days without selection (transient replication) or 2 weeks in G418 (300 μg/ml) to select for cells containing the plasmid (long-term maintenance). Plasmids from equal cell numbers (5 × 10⁶ to 1 × 10⁷) were isolated, linearized with StuI, DpnI digested, and Southern blotted for quantification as described by Shire et al. (56).

Transcription activation assays.HeLa (Tet-On) cells were plated at 2.5 × 10⁵ cells per 6-cm plate 24 h prior to transfection using ESCORT II reagent. Cells were cotransfected with 2 μg of the pFRTKCAT reporter construct and 5 μg pc3oriP expressing EBNA1 or EBNA1 mutants or lacking the EBNA1 cassette. At 48 h posttransfection, cells were lysed and 50 μg of lysate was assayed for chloramphenicol acetyltransferase (CAT) activity as previously described (8, 67). The amount of acetylated product generated by each lysate was quantified for several reaction times and used to determine the acetylation rate.

Two-hybrid assays.Coding sequences for EBNA1 serine mutants were PCR amplified from the p416MET25 expression constructs and cloned between NdeI and BamHI sites in pAS2. The mutants were then subcloned into pLexA-kan (a kind gift from Igor Stagljar) containing Nde1 and BamH1 sites for expression as LexA DNA binding domain fusions. S. cerevisiae strain L40a (MATatrp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacZ) (60), which contains integrated HIS3 and lacZ reporter genes under the control of LexA binding sites, was transformed with one of the pLexA.EBNA1 constructs and with pACTII expressing hEBP2 (fused to the GAL4 activation domain) or with empty pACTII (negative control). Yeast were grown overnight in medium lacking Trp and Leu. The cultures were rinsed with sterile water and diluted to an optical density at 600 nm of 0.1, and then 10-μl aliquots of 10-fold serial dilutions were spotted on plates lacking Trp and Leu and on plates lacking Trp, Leu, and His containing 0, 2, and 5 mM 3-aminotriazole (AT).

EBNA1 proteins for affinity columns and methylation assays.The highly purified EBNA1 proteins were used as substrates in the methylation reactions. The version of EBNA1 used lacks most of the Gly/Ala repeat region, as do the EBNA1 internal deletion mutants. EBNA1 (23), EBNA1GA (with full Gly/Ala repeat [5]), and Δ41-376 (EBNA1 lacking amino acids 41-376 [1]) were purified from insect cells as previously described. Δ61-83 (EBNA1 lacking amino acids 61 to 83 [67]) was purified from insect cells on a metal chelating column by virtue of an N-terminal six-histidine tag. The EBNA1 truncation mutants containing the indicated amino acids, 351 to 641, 363 to 641, 377 to 641, and 452 to 641, were purified from Escherichia coli as previously described (3, 17), as was Δ325-376 (23). The EBNA1 mutant 320-355/459-607, in which amino acids 320 to 355 are fused at the N terminus to amino acids 459 to 607, was purified from E. coli on a metal chelating column by virtue of an N-terminal six-histidine tag.

EBNA1 affinity columns.Affinity columns containing purified EBNA1 and whole-cell extracts from HeLa cells were generated as described by Holowaty et al. (23). A 400-μl aliquot of the HeLa lysate (at 14 mg/ml) was applied under physiological salt conditions to a 40-μl column containing EBNA1 or deletion mutants coupled at 1 mg/ml. After washing, bound proteins were eluted in high salt followed by 1% sodium dodecyl sulfate (SDS) and identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry as previously described (23). SDS eluates were also analyzed by Western blotting using rabbit anti-PRMT1 (Upstate Biochemicals) or mouse anti-PRTM5 (Transduction Laboratories) antibodies.

Purification of PRMT enzymes and GAR substrate.PRMT1, PRMT3, PRMT5, and PRMT6 and GAR substrate were expressed as glutathione S-transferase (GST)-fusion proteins from pGEX plasmids in E. coli strain BL21(DE3)plysS. Plasmids were kindly provided by Harvey Herschman, Steven Clarke, and Sidney Pestka and have been described previously (15, 16, 37, 59). Expression of GST-fusion protein was induced with 0.5 mM isopropyl-1-thio-β-d-galactopyranoside (IPTG) for 3 h at 30°C for GST-GAR and with 0.2 mM IPTG overnight at 15°C for GST-PRMT1, GST-PRMT3, GST-PRMT5, and GST-PRMT6. After washing with phosphate-buffered saline (PBS), the cell pellet was resuspended in PBS containing 1 mM phenylmethylsulfonyl fluoride (PMSF), 10 mM dithiothreitol (DTT), and 0.5% Triton X-100 and lysed by sonication. The lysate was clarified by centrifugation and then applied to a glutathione-Sepharose 4B column (Amersham Pharmacia Biotech). Bound proteins were eluted from the column with 20 mM reduced glutathione in 50 mM Tris-HCl pH 8.0 and stored in aliquots at −70°C.

In vitro methylation assays.Purified GST-PRMT proteins were incubated in a 40-μl reaction mixture containing 0.2 μCi of S-[methyl-¹⁴C]adenosyl methionine (60.0 mCi/mmol; Amersham Biosciences), 2 μg of EBNA1 protein or GST-GAR, 25 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 mM EGTA,1 mM PMSF, 1 mM benzamidine. After 2 h at 37°C, the reaction was stopped with addition of an equal volume of 2× SDS-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (50 mM Tris-HCl pH 6.8, 100 mM dithiothreitol, 2% SDS, 10% glycerol, and 0.1% bromophenol blue) and then boiled for 10 min prior to SDS-PAGE analysis. Following electrophoresis, the gel was fixed in 45% (vol/vol) methanol and 10% (vol/vol) acetic acid for 30 min, soaked in Amplify solution (Amersham Biosciences) for 30 min, and then dried. Radioactive bands were detected by exposing the gel to Kodak BioMax XAR X-ray film at −70°C for 2 to 14 days. The reaction kinetics of PRMT1 with EBNA1 and Δ325-376 substrates was determined as described above, except that aliquots were removed from the reaction mixture every 5 min. Incorporation of ¹⁴C into the EBNA1 proteins was quantified by phosphorimager analysis of SDS-polyacrylamide gels.

In vivo methylation.U2OS cells were grown in H21 medium to 90% confluence in a 15-cm dish and then were transfected with 10 μg of pc3oriP plasmid expressing EBNA1 or EBNA1 mutants Δ32-54, Δ325-376, or Δ40-376 (described in references 56 and 67) using Lipofectamine 2000 (Invitrogen). After a 4-h incubation, cells were split into two 15-cm dishes. Three days posttransfection, the medium was replaced with 8 ml of H21 containing 100 μg/ml cycloheximide and 10 μg/ml chloramphenicol. After 30 min at 37°C, the cells were rinsed twice with methionine-free Dulbecco's modified Eagle's medium and then incubated for 3 h at 37°C in 4 ml methionine-free Dulbecco's modified Eagle's medium containing 100 μg/ml cycloheximide, 10 μg/ml chloramphenicol, and 40 μCi l-[methyl-³H]methionine. Cells were then harvested, rinsed with PBS, and incubated in 700 μl lysis buffer (100 mM Tris-HCl pH 8, 250 mM NaCl, 1% NP-40, 1 mM EDTA, 1 mM PMSF, Roche protease inhibitors) for 30 min on ice. After centrifugation for 15 min in a microcentrifuge, a 20-μl bed volume of protein A-Sepharose and 2 μl of R4 anti-EBNA1 rabbit serum were added to the supernatant and incubated overnight at 4°C with rocking. The Sepharose beads were then washed three times with lysis buffer, and bound proteins were eluted with 40 μl 2% SDS. All of the eluate was loaded onto a 12% polyacrylamide gel and transferred to polyvinylidene difluoride membrane. EBNA1 proteins were identified by Western blotting using anti-EBNA1 mouse monoclonal antibody OTIx (supplied by J. Middledorp). The membrane was then stripped and exposed to film for 1 month to detect protein bands containing tritium.

MTA treatment and immunofluorescence microscopy.HeLa cells were plated at 1 × 10⁶ cells per 6-cm plate 24 h before transfection with 2 μg of pc3oriPE by using the ESCORT II reagent. Four hours posttransfection, the transfection mixture was removed and cells were thoroughly washed with PBS to terminate the transfection process. Cells were replated and incubated in medium containing 0.3 mM 5′methyl-thioadenosine (MTA) (dissolved in dimethylformamide [DMF]) or an equivalent amount of DMF alone (negative control) for 48 h before processing for immunofluorescence microscopy. For microscopy, log-phase cells were adhered to coverslips, fixed, and permeabilized as described by Wu et al. (66). Cells were stained with mouse anti-EBNA1 monoclonal antibody OT1x and either rabbit anti-hEBP2 (66), rabbit anti-USP7 (23), or goat anti-B23 antibody (Santa Cruz). For samples that were costained for EBNA1 and either hEBP2 or USP7, the secondary antibodies used were goat anti-mouse conjugated to Texas Red and goat anti-rabbit conjugated to fluorescein isothiocyanate (FITC; Santa Cruz). For cells stained for both EBNA1 and B23, the secondary antibodies used were goat anti-mouse conjugated to FITC (Santa Cruz) and donkey anti-goat conjugated to Cy3 (Chemicon). All cells were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI; 25 ng/μl) and visualized at 400× magnification using a Leica DMR microscope and OpenLab software.

PRMT1 silencing.U2OS cells were seeded at 50% confluence in six-well plates containing coverslips, 24 h prior to Lipofectamine 2000 transfection with 3 μg of pSilencer3.1-U1hygro (Ambion) expressing PRMT1 hairpin RNA ([AA]GGAGGTGGACATCTATAC). The cells were grown for 4 days in the presence of 100 μg/ml hygromycin B and then transfected as above with 3 μg pC3oriPE expressing EBNA1. Two days later, the cells were prepared for immunofluorescence microscopy as described above. PRMT1 was detected with rabbit antibody (07-404; Upstate Biochemicals) followed by FITC-conjugated goat anti-rabbit antibody (Invitrogen). EBNA1 was detected with the mouse monoclonal antibody OT1x followed by Texas Red-conjugated goat anti-mouse antibody (Molecular Probes). In all cases, DNA was visualized by DAPI staining and microscopy as described above.

In vivo phosphorylation of the EBNA1 325-376 region.EBNA1 is phosphorylated on multiple serine residues when expressed in both human and insect cells (18, 21, 49), and the functionally important 325-376 region contains multiple serines that are highly predicted to be phosphorylated. To investigate the possibility that the functionality of the 325-376 region is modulated by phosphorylation, we first asked whether phosphorylation of these sequences could be detected in EBNA1 purified from insect cells after baculovirus expression. To this end, highly purified EBNA1 was completely digested with trypsin and phosphorylated peptides were examined using a QTrap LC/MS/MS mass spectrometer. A phosphorylated tryptic peptide of the expected size for the 325-376 region was selected, sequenced, and shown to be comprised of GGSGGR. This sequence is found in triplicate in the 325-376 region (Fig. 1B), indicating that one or more copies of this sequences are phosphorylated in the context of the functional EBNA1 protein.

Contribution of serine phosphorylation to the EBNA1 segregation function.To test the possible functional importance of phosphorylation of the serine residues in the EBNA1 325-376 region, each of the serines in the GGSGGR repeated sequence (serines 334, 342, and 350) was mutated to alanine in the context of functional EBNA1, and combinations of serine mutations were also generated. Mutants were named according to the four serines (or alanine mutations) in the 325-376 region, where SSSS refers to serines 334, 342, 350, and 365. EBNA1 proteins containing these mutations were first tested for their ability to segregate plasmids containing the EBV segregation element FR in a reconstituted segregation system in Saccharomyces cerevisiae (29). In this system, plasmids replicate from an ARS element using the yeast replication machinery but can only segregate during cell division by virtue of EBNA1 and the FR element. Segregation occurs through the attachment of EBNA1 to the yeast chromosomes, which requires the human EBP2 protein exogenously expressed in the yeast cells (27), a mechanism that appears to be the same as for EBNA1-mediated plasmid segregation in human cells. The ability of the EBNA1 serine mutants to maintain the segregation test plasmids was determined after 11 cell generations in the absence of selection for these plasmids, and the results were compared to wild-type EBNA1 and no EBNA1 by spotting dilutions of the cultures on plates selective for the segregation plasmid (Fig. 2A) and by comparing the numbers of colonies grown on selective and nonselective plates (Table 1). In all cases expression of the EBNA1 proteins at similar levels was confirmed by Western blotting (data not shown). While mutations of individual serines had no significant effect on the EBNA1 segregation function, mutation of two serines in adjacent repeated sequences consistently caused a twofold reduction in plasmid segregation (AASS and SAAS mutants). This segregation defect was restored by converting the alanines to aspartic acid (mutant SDDS) to mimic the charge of the phospho-serine. Plasmid segregation efficiency was further reduced (threefold relative to wild type) by mutating all four serines in the 325-376 region to alanines (mutant AAAA) and was again restored by conversion of all four alanines to aspartic acid (mutant DDDD), indicating that phosphorylation of these serines plays a role in segregation.

• Open in new tab
• Download powerpoint

FIG. 2.

Effects of serine mutations on EBNA1 segregation function and hEBP2 binding assayed in Saccharomyces cerevisiae. A. Assay of the loss of FR-containing plasmids after 11 generations in the absence of selection for the FR plasmid. Assays were performed in the presence of EBNA1 (positive control) or an EBNA1 mutant with the indicated serine mutations or in the absence of EBNA1 (negative control). Serial dilutions of each culture were then grown on nonselective or selective plates with respect to the FR-containing plasmid to determine the number of total yeast cells present (no selection) and the fraction of the cells that contained the plasmid (selection). B. Two-hybrid assay showing the interaction of EBNA1 and EBNA1 serine mutants with hEBP2, as determined by activation of a HIS3 reporter gene. EBNA1 with empty pACT2 (-EBP2) is shown as a negative control (second row). Ten-fold serial dilutions of the cultures were grown on plates containing His or lacking His and containing 0, 2, or 5 mM AT as indicated.

To determine if the serine phosphorylation effect observed for EBNA1-mediated segregation in the yeast system also occurs in human cells, we expressed the EBNA1 serine mutants in human cells from plasmids containing the EBV oriP. We then tested the ability of these plasmids to be stably maintained in dividing cells for a 2-week period, as determined by the level of recovered plasmid relative to that in the presence of wild-type EBNA1 (Fig. 3B and Table 1). Stable maintenance of oriP plasmids is known to require the segregation function of EBNA1 in addition to replication. While all EBNA1 proteins were expressed at the same level (Fig. 3A), the SAAS, AASS, and AAAA EBNA1 mutants were consistently less able to maintain oriP plasmids than wild-type EBNA1. As in the yeast system, this defect was rescued by conversion of the alanines to aspartic acid, indicating that phosphorylation of these serines affects the plasmid maintenance ability of EBNA1. To ensure that these effects were due to the ability of EBNA1 to mediate the segregation of the oriP plasmids as opposed to their replication, we also assessed the transient replication of the oriP plasmids in the presence of each of the EBNA1 mutants. None of the serine mutations was found to significantly affect the replication of the oriP plasmids (Table 1). Therefore, the phosphorylation of multiple serines in the EBNA1 325-376 region positively regulates the ability of EBNA1 to segregate plasmids in both human cells and the reconstituted yeast system.

• Open in new tab
• Download powerpoint

FIG. 3.

Plasmid maintenance ability of EBNA1 serine mutants in human cells. A. Western blot showing the expression of EBNA1 and EBNA1 serine mutants in HeLa cells 3 days posttransfection. B. Southern blot showing the linearized oriP plasmids that were recovered from equal numbers of HeLa cells 2 weeks posttransfection with oriP plasmids expressing EBNA1, no EBNA1 (none), or the EBNA1 serine mutants indicated.

Serine phosphorylation affects the EBNA1-hEBP2 interaction.One possibility of how phosphorylation of serines in the 325-376 region affects the segregation function of EBNA1 is by its altering EBNA1 binding to hEBP2. hEBP2 interacts with EBNA1 through sequences in the 325-376 region (56). This interaction is absolutely required for EBNA1-mediated segregation in yeast and is also important for oriP segregation in human cells, in both cases serving to tether EBNA1 to the cellular mitotic chromosomes (27, 28). To determine if the decreased segregation function associated with the EBNA1 SAAS and AAAA mutants is due to less efficient binding to hEBP2, we compared the hEBP2 binding ability of these EBNA1 mutants to wild-type EBNA1 in a LexA-based yeast two-hybrid assay, where binding stringency can be assessed by comparing activation of the His3 gene in the presence of increasing amounts of AT (Fig. 2B). We found that, while the SAAS mutant appeared to bind hEBP2 at similar levels as wild-type EBNA1, the SDDS mutant had increased hEBP2 binding under stringent conditions (His 5AT panel). We also found that the AAAA mutant was clearly impaired for binding hEBP2 (His and His 2AT panels) and that the binding was restored by converting the four alanines to aspartic acids, resulting in higher binding activity than wild-type EBNA1 (His 2AT and His 5AT panels). Therefore, the ability of EBNA1 to bind hEBP2 is positively affected by phosphorylation of multiple serines in the 325-376 region, which likely affects the segregation function.

Contribution of serine phosphorylation to the EBNA1 transcription activation function.In addition to its role in segregation, the 325-376 region of EBNA1 is important for transcriptional activation from the FR element. Therefore, we tested the EBNA1 serine mutants for possible effects on activating the expression of a CAT reporter gene under the control of the FR element. To this end, HeLa cells were cotransfected with an FR-CAT reporter construct and an oriP plasmid expressing an EBNA1 mutant, and the level of CAT activity was determined 2 days posttransfection in relation to cell lysates expressing wild-type EBNA1. A representative experiment is shown in Fig. 4, and quantification from multiple experiments is shown in Table 1. Some of the single and double alanine substitutions were found to decrease transcriptional activity to various degrees, as did the quadruple alanine substitution. However, conversion of the four alanines to aspartic acid did not restore transcription function, but rather further decreased it. The results are more consistent with effects on transcription due to sequence alteration rather than phosphorylation effects.

• Open in new tab
• Download powerpoint

FIG. 4.

Transcriptional effects of EBNA1 serine mutants. HeLa cells were transfected with an FR-CAT reporter plasmid and pc3oriP expressing EBNA1, no EBNA1, or the EBNA1 serine mutants as indicated. Two days later, equal amounts of cell lysates were then assayed for CAT activity by following the acetylation of chloramphenicol in 5-, 15-, and 30-min reactions.

EBNA1 interactions with protein arginine methyltransferases.We have previously examined cellular protein interactions with EBNA1 using an in vivo TAP-tagging approach as well as an EBNA1 affinity column approach (23). The TAP-tagging experiments showed that EBNA1 stably interacted with the protein arginine methyltransferase PRMT5, in addition to other proteins. Strangely, PRMT5 was not identified as an EBNA1 binding protein using the affinity column approach, where cell lysates were applied to EBNA1 microcolumns under physiological salt conditions and retained proteins were then eluted in high salt. We have recently revisited the EBNA1 affinity column method by adding an SDS elution step following the high-salt elution, in order to elute any proteins that may bind EBNA1 largely through hydrophobic interactions and hence not be efficiently eluted with salt. In doing so, we found that PRMT5 was retained on the EBNA1 column after the high-salt elution and subsequently eluted with SDS (Fig. 5A). PRMT5 was also found to bind EBNA1 mutants lacking amino acids 325 to 376 or 61 to 83, indicating that these sequences are not required for the PRMT interaction. In the same experiments, a second protein arginine methyltransferase, PRMT1, was found to bind EBNA1 and the Δ325-376 and Δ61-83 mutants, eluting from the columns with SDS (the most efficient retention was seen with the Δ61-83 affinity column). Western blot assays performed on the column eluates in Fig. 5A confirmed that both PRMT1 and PRMT5 were specifically retained on the EBNA1 columns (but not on the control column) (Fig. 5B).

• Open in new tab
• Download powerpoint

FIG. 5.

Retention of PRMT1 and PRMT5 on EBNA1 affinity columns. Equal amounts of HeLa cell lysates were applied to a 40-μl Affi-gel column resin alone (control) or coupled to purified EBNA1 or EBNA1 deletion mutants (Δ325-376 and Δ61-83). Columns were then washed, and bound proteins were eluted with high salt followed by SDS. A. SDS elutions were analyzed by SDS-PAGE and silver staining, and the indicated bands were identified by MALDI-TOF mass spectrometry. Note that some of the EBNA1 proteins themselves elute from the column with SDS. B. The column eluates in panel A were analyzed by Western blotting using antibodies specific for PRMT5 and PRMT1.

In vitro methylation of EBNA1.The interaction of EBNA1 with PRMT1 and PRMT5 may indicate that EBNA1 is a substrate for methylation by these enzymes. To investigate this possibility, we expressed and purified PRMT1 and PRMT5 as GST fusions in E. coli and tested their ability to methylate EBNA1 in vitro in comparison to equivalent amounts of a positive control substrate (GST-GAR). Both PRMT1 and PRMT5 were found to methylate EBNA1, containing either the small or full-length Gly-Ala repeat (EBNA1 and EBNA1GA, respectively) with similar efficiency as they methylated GST-GAR (Fig. 6A and C), although PRMT5 was less active on both GST-GAR and EBNA1 substrate than PRMT1. We then repeated the methylation assays using a variety of EBNA1 deletion and truncation mutants in order to map the EBNA1 regions that were methylated by these enzymes (Fig. 6A and C). For both PRMT1 and -5, methylation of EBNA1 was decreased but not abolished by deleting amino acids 325 to 376, indicating that this region is methylated but is not the only methylated region. The decreased methylation of Δ325-376 with PRMT1 is most obvious in the time course in Fig. 6B, conducted using equal amounts of EBNA1 substrates and PRMT1 enzyme. Methylation by both PRMT1 and PRMT5 was abrogated by the 41-376 deletion, which deletes multiple GRG motifs between amino acids 41 and 50, in addition to the methylation sites in the 325-376 region (note that amino acids 101 to 324 of the Gly-Ala repeat are missing in all EBNA1 proteins except EBNA1GA). This suggests that sequences in both the 41-50 and the 325-376 regions of EBNA1 are targets for methylation by both PRMT1 and PRMT5.

• Open in new tab
• Download powerpoint

FIG. 6.

In vitro methylation of EBNA1 and EBNA1 mutants. Equal amounts of the purified EBNA1 proteins indicated or the GST-GAR positive control were incubated with S-[methyl-¹⁴C]adenosyl methionine and either PRMT1 (A and B), PRMT5 (C), PRMT3 (D), or PRMT6 (E) as described in Materials and Methods. Reaction mixtures were then analyzed by SDS-PAGE and autoradiography. In each case the major labeled band corresponds to the position of the full-length protein. Note that EBNA1 362-641 has a small fusion at the C terminus, making it run larger than 351-641 (17). In panel B, methylation reactions were conducted for various times ranging from 5 to 40 min and the amount of labeled EBNA1 was quantified. A direct comparison of methylation of GST-GAR and EBNA1 substrates by PRMT1, -3, and -6 is shown in panel F and used amounts of enzyme that gave equal methylation of GST-GAR.

Further information about which sequences in the 325-376 region are methylated was gained from methylation assays performed with a series of EBNA1 truncations mutants (Fig. 6A and C). EBNA1 N-terminal truncation mutants 351-641 and 363-641 were methylated by both PRMT1 and PRMT5, while neither enzyme methylated the 377-641 mutant. This indicated that one or more methylation sites exist between residues 364 and 377. We also found that an EBNA1 mutant containing residues 320 to 355 fused to the DNA binding domain (459 to 607) was methylated by both PRMT1 and PRMT5, indicating that one or more methylation sites exist in the 320-355 region. Therefore, multiple sites in the 325-355 region can be methylated by PRMT1 and PRMT5.

To further investigate the specificity of EBNA1 methylation, we asked whether other PRMTs that have not been found to bind EBNA1 also methylated EBNA1. To this end, we expressed and purified PRMT3 and PRMT6 as GST fusions in E. coli and compared their abilities to methylate EBNA1 and GST-GAR (positive control) substrates (Fig. 6D and E). While both PRMT3 and -6 resulted in some labeling of EBNA1 and EBNA1GA, this was much less than that seen with an equivalent amount of GST-GAR substrate, indicating that both of these enzymes have very low specificity for EBNA1. The methylation of EBNA1 that was observed occurred within the 41-376 region, since removal of these sequences abrogated labeling. Methylation at levels equivalent to GST-GAR was only seen with the 320-355 fragment when fused to the 459-607 DNA binding domain, indicating that this sequence can be methylated by PRMT3 and -6 when removed from its normal context. In Fig. 6F, we directly compared the relative specificities of PRMT3 and PRMT6 to PRMT1, using amounts of each enzyme that gave equivalent methylation of GST-GAR. This clearly showed that the level of EBNA1 methylation by PRMT3 and -6 was much less than for PRMT1. Together, the results suggest that PRMT1 and -5, but not PRMT3 and -6, are likely to methylate EBNA1 in vivo.

In vivo methylation of EBNA1.To determine whether EBNA1 is methylated in human cells, we expressed EBNA1 and several deletion mutants in U2OS cells. After blocking protein production with cycloheximide, the cells were grown in the presence of [³H]methyl-methionine, and incorporation of the tritiated methyl groups into EBNA1 was determined by autoradiography of SDS-polyacrylamide gels containing immunoprecipitated EBNA1. As shown in Fig. 7A, EBNA1 and all of the deletion mutants were recovered at similar levels by immunoprecipitation. Exposure to film showed that EBNA1 and Δ34-52, lacking the N-terminal Gly-Arg-rich region, had been methylated, whereas methylation was not detected in the Δ325-376 and Δ41-376 mutants. This indicates that the 325-376 region is a major site of methylation in vivo. The amount of label incorporated in Δ34-52 was consistently about 60% of that for EBNA1, in keeping with the possibility that some methylation might also occur in the N-terminal Gly-Arg-rich region in vivo.

• Open in new tab
• Download powerpoint

FIG. 7.

In vivo methylation of EBNA1. U2OS cells expressing EBNA1, Δ34-52, Δ325-376, or Δ41-376 were incubated with cycloheximide and then with [³H]methyl-methionine. EBNA1 proteins were then immunoprecipitated with anti-EBNA1 rabbit serum and separated by SDS-PAGE. A. Western blot of immunoprecipitated EBNA1 probed with mouse anti-EBNA1 monoclonal antibody. B. Autoradiograph of the blot in panel A, showing incorporation of [³H]methyl-methionine.

Effect of methylation on EBNA1 localization.To determine what effect methylation might have on EBNA1 functions, we looked at the localization and transcriptional activation of EBNA1 in human cells after treating the EBNA1-expressing cells with the methylation inhibitor MTA. MTA concentrations were used that have been previously known to affect protein methylation (43) but which do not inhibit cell proliferation (data not shown). We did not observe an effect on the transcriptional activity of EBNA1, compared to mock-treated cells, nor did we observe an obvious effect on EBNA1 attachment to mitotic chromosomes by immunofluorescence (data not shown). We did, however, consistently see effects of MTA treatment on the localization of EBNA1 in interphase cells. After treatment of EBNA1-expressing cells with MTA (at concentrations ranging from 0.1 to 0.3 mM), EBNA1 was observed to localize to the outer region of the nucleolus in approximately 25% of the cells (values ranged from 20 to 30% in nine separate experiments), resulting in the formation of nucleolar rings that were not seen in mock-treated cells (Fig. 8). This effect appeared to be specific for EBNA1, because the localization of the nucleolar proteins B23 (Fig. 8A) and hEBP2 (Fig. 8B) was not affected by the MTA treatment, nor was the localization of the nuclear protein USP7 (Fig. 8C). The results suggest that the methylation state of EBNA1 affects its localization.

• Open in new tab
• Download powerpoint

FIG. 8.

Effect of MTA treatment on EBNA1 localization. (A to C) Log-phase HeLa cells expressing EBNA1 were either treated with MTA (+) or mock treated (-) and then were stained with anti-EBNA1 rabbit antiserum and anti-B23 goat antibodies (A), with mouse anti-EBNA1 monoclonal antibody and anti-hEBP2 rabbit serum (B), or with mouse anti-EBNA1 monoclonal antibody and anti-USP7 rabbit serum (C). Secondary antibodies conjugated to Texas Red, FITC, and Cy3 were used to detect mouse, rabbit, and goat primary antibodies, respectively. All cells were counterstained with DAPI and visualized at 400× magnification using similar exposure times. (D) U2OS cells were transfected with a PRMT1 silencing plasmid and then with an EBNA1 expression plasmid. Cells were stained with anti-EBNA1 and anti-PRMT1 antibodies, followed by Texas Red- and FITC-conjugated secondary antibodies, respectively. All cells were counterstained with DAPI and visualized at 400× magnification using similar exposure times. EBNA1 localization was compared in cells with little or no PRMT1 staining (-) and with more obvious PRMT1 staining (+).

We also examined the effect of PRMT silencing on EBNA1 localization. To this end, U2OS cells were transfected with a plasmid expressing hairpin RNA for PRMT1 and, 2 days later, transfected with an EBNA1 expression plasmid. Cells were then stained for EBNA1 and PRMT1 and visualized by immunofluorescence microscopy (Fig. 8D). EBNA1 was observed to form rings around the nucleoli in approximately 90% of the cells with little or no PRMT1 expression. In contrast, only 10% of cells in the same experiment that had higher PRMT1 staining levels (and expressed EBNA1) showed the nucleolar ringing of EBNA1. This effect was not seen upon transfection of the same cells with a PRMT5 silencing construct; however, efficient silencing of PRMT5 was not apparent by microscopy (data not shown). Therefore, PRMT1 has a major effect on the nuclear distribution of EBNA1.