PLK1‐dependent phosphorylation restrains EBNA2 activity and lymphomagenesis in EBV‐infected mice

Abstract While Epstein–Barr virus (EBV) establishes a life‐long latent infection in apparently healthy human immunocompetent hosts, immunodeficient individuals are at particular risk to develop lymphoproliferative B‐cell malignancies caused by EBV. A key EBV protein is the transcription factor EBV nuclear antigen 2 (EBNA2), which initiates B‐cell proliferation. Here, we combine biochemical, cellular, and in vivo experiments demonstrating that the mitotic polo‐like kinase 1 (PLK1) binds to EBNA2, phosphorylates its transactivation domain, and thereby inhibits its biological activity. EBNA2 mutants that impair PLK1 binding or prevent EBNA2 phosphorylation are gain‐of‐function mutants. They exhibit enhanced transactivation capacities, accelerate the proliferation of infected B cells, and promote the development of monoclonal B‐cell lymphomas in infected mice. Thus, PLK1 coordinates the activity of EBNA2 to attenuate the risk of tumor incidences in favor of the establishment of latency in the infected but healthy host.


Introduction
Epstein-Barr virus (EBV) is associated with multiple malignancies including B-cell lymphomas and B-cell lymphoproliferative diseases (Longnecker et al, 2013;Farrell, 2019;Shannon-Lowe & Rickinson, 2019). More than 90% of the world population are infected with EBV. Based on an intimate virus-host interaction, the virus establishes a latent infection in resting memory B cells (Thorley-Lawson & Gross, 2004). Infected B cells are activated, enter the cell cycle, and proliferate and maintain the viral circular genomes in a process that is strictly coordinated with the cell cycle of the host cell. While entering the memory B-cell compartment, viral gene expression is gradually silenced to evade host immune surveillance. Reactivation of viral replication and virus production occurs occasionally (Munz, 2019).
EBV nuclear antigen 2 (EBNA2) is a key transcription factor that initiates and maintains the expression of viral and cellular target genes which are critical for the growth transformation of B cells by EBV West, 2017). EBNA2 can activate the transcription of the viral latent membrane protein 1 (LMP1) and the cellular Myc gene in the absence of de novo protein synthesis. Thus, both are direct target genes of EBNA2 (Kaiser et al, 1999). Both proteins can induce an avalanche of secondary EBNA2 target genes. LMP1 promotes proliferation and cell survival. This multifunctional membrane oncoprotein activates the cellular NFjB, JNK, and STAT signaling cues and thereby mimics CD40 signaling (Thorley-Lawson, 2001). The proto-oncoprotein and transcription factor MYC stimulate global RNA and protein synthesis to foster cell growth of quiescent lymphocytes and cell cycle progression in normal nonmalignant cells. High-level MYC expression is associated with lymphomagenesis, but in the absence of co-stimulatory signals, it may also induce apoptosis (Baluapuri et al, 2020).
In the nucleus, EBNA2 associates with cellular proteins to execute its function. The B-cell-specific transcription factor EBF1 (Friberg et al, 2015;Lu et al, 2016;Glaser et al, 2017) and CBF1/ RBPJ, the major downstream effector of NOTCH signaling, serve as DNA anchors for EBNA2 (Henkel et al, 1994;Hsieh & Hayward, 1995;Ling & Hayward, 1995;Farrell et al, 2004). The C-terminal acidic transactivation domain (TAD) of EBNA2 recruits components of the pre-initiation complex and histone acetylases and also contributes to ATP-dependent chromatin remodeling by interacting with hSNF5/Ini Cohen, 1992;Tong et al, 1995aTong et al, , 1995bTong et al, , 1995cWang et al, 2000;Kwiatkowski et al, 2004). The TAD of EBNA2 is intrinsically unstructured. In complex with the histone acetylase CBP or the TFIIH subunit Tfb1, a 9-residue amphipathic a-helix within this TAD is formed (Chabot et al, 2014). Cellular proteins may control EBNA2 transcriptional activities by protein-protein interactions or modification. The MYND domain of the repressor BS69 can bind as a dimer to the EBNA2 TAD and its flanking regions and attenuates EBNA2 activity as well as transformation efficiency (Harter et al, 2016;Ponnusamy et al, 2019).
In search for additional cellular factors controlling EBNA2 function, we have performed a label-free mass spectrometry-based quantification of cellular proteins in EBNA2 immunoprecipitates and found polo-like kinase 1 (PLK1). PLK1 is a serine/threonine-protein kinase that controls G2/M transition and progress through mitosis and cytokinesis in a tightly controlled order to secure genomic stability of the dividing cell. Through phosphorylation of specific substrates, PLK1 promotes activation of the mitotic driver Cyclin B1/ CDK1 in the late G2-phase, triggering prophase onset (Nakajima et al, 2003;Watanabe et al, 2005;Gheghiani et al, 2017). PLK1 recognizes its substrates by a conserved C-terminal polo-box domain (PBD). Frequently, PBD binds to pre-phosphorylated epitopes generated by the mitotic CDK1 kinase or other proline-directed kinases such as MAPK (Elia et al, 2003a(Elia et al, , 2003bLowery et al, 2005). This process is referred to as non-self-priming (Barr et al, 2004;Lee et al, 2014). Since primary tumors from various tissues express elevated levels of PLK1, and this high-level expression frequently correlates with a poor prognosis (Wolf et al, 1997), PLK1 is a potential oncotarget for molecular cancer therapy and a prognostic marker (Yuan et al, 1997;Rodel et al, 2020;Rosenblum et al, 2020). Preclinical and clinical studies currently test potential further clinical indications for small molecule or siRNA-based PLK1 inhibitors (Liu et al, 2017). However, there is now increasing evidence that PLK1 can also act as a tumor suppressor when expressed in the context of specific tumor types. Thus, there is a strong need to characterize and define the exact molecular features of tumors that may be treated with PLK1 inhibitors (de Carcer, 2019).
Here, we show that PLK1 directly binds to EBNA2. EBNA2/PLK1 complex formation is strongly enforced by EBNA2 residue S379 phosphorylation catalyzed by the mitotic Cyclin B/CDK1 complex. PLK1 phosphorylates the C-terminal transactivation domain of EBNA2 and attenuates its activity. EBNA2 mutants that lack either the main PLK1 docking site or the two phosphorylation sites are gain-of-function mutants that promote lymphoma incidence in EBVinfected humanized mice. This indicates that PLK1 acts as a tumor suppressor in EBV-driven carcinogenesis.

Results
Cyclin B/CDK1 primes EBNA2 residue S379 for polo-like kinase 1 (PLK1) binding To identify potential EBNA2 interacting cellular proteins, we transfected EBV-negative DG75 B cells with HA-tagged EBNA2 expression constructs or the corresponding HA-expression vector and performed immunoprecipitations with HA-specific antibodies. Tryptic peptides of these immunoprecipitates were analyzed and quantified by label-free mass spectrometry. Polo-like kinase 1 (PLK1) was one of 19 proteins that were significantly enriched in EBNA2 coimmunoprecipitates (Table EV1). As expected, the EBNA2 DNA anchor protein CBF1/RBPJ was one of these proteins, demonstrating that the experimental approach was valid. Importantly, PLK1 and EBNA2 specifically co-immunoprecipitated from whole cell extracts of EBV-infected B cells in which both proteins are expressed at endogenous physiological levels (Fig 1).
To identify PLK1 docking sites within EBNA2, EBNA2 deletion fragments were generated and tested for binding to PLK1 by transfection and co-immunoprecipitation (Fig 2A-C (Elia et al, 2003a). Crystal structures of the PLK1 PBD in complex with peptides demonstrate how the positively charged groove of PBD docks onto negatively charged phosphopeptides of diverse substrates. EBNA2 exhibits three potential Cyclin B/ CDK1 phosphorylation and PBD docking sites located at residue T267, S379, and S470 ( Fig 2D). Mutants involving the respective residues were tested for PLK1 binding by transfection and coprecipitation. The mutation TSS377VAA impaired PLK1 binding dramatically while all other EBNA2 mutations did not affect PLK1 binding ( Fig 2E). To specify the contribution of EBNA2 residue S379 phosphorylation, a heptapeptide (PNTSSPS) and phospho-A B Figure 1. EBNA2 and PLK1 complexes in EBV-infected B-cell lines.
Source data are available online for this figure. heptapeptide (PNTSpSPS) were tested for PBD (PLK1 residue 345-603) binding by isothermal titration calorimetry (ITC). The phosphorylated peptide bound PBD in a molar ratio of 1:1 whereas no interaction was detected using the unmodified peptide. The PNTSpSPS/PBD interaction is micromolar with a dissociation constant of K D = 8.19 µM ( Fig 2F). GST-EBNA2 342-422 expressed and purified from bacterial extracts did hardly pulldown any PLK1 from cellular extracts. Phosphorylation of the EBNA2 protein produced in bacteria by Cyclin B/CDK1, an enzyme not present in bacteria, strongly enhanced PLK1 binding. EBNA2 (342-422) S379A mutant did not bind PLK1, and phosphorylation by Cyclin B/CDK1 could not reconstitute binding ( Fig 2G). Also, as shown by transfection, EBNA2 S379A binding to PLK1 was severely impaired ( Fig 2H). We conclude that EBNA2 residue S379 is a PLK1 docking site primed by Cyclin B/CDK1 phosphorylation.
The C-terminus of EBNA2 is the substrate of PLK1 In order to test whether EBNA2 is phosphorylated by PLK1, EBNA2/ PLK1 complexes were co-precipitated from cellular extracts of DG75 cells induced with doxycycline to express HA-EBNA2 (DG75 Dox HA-EBNA2 ) ( Fig 3A). The precipitates were submitted to kinase assays either in the presence or absence of exogenous recombinant PLK1. Phosphorylation of EBNA2 was readily detected in the coprecipitates and was enhanced by the addition of exogenous recombinant PLK1 to the co-precipitates ( Fig 3B). EBNA2 phosphorylation was prevented upon addition of Volasertib, a PLK1-specific inhibitor, to the co-precipitates of EBNA2 and endogenous PLK1 (Fig 3C), corroborating the evidence that the endogenous PLK1 trapped in the co-precipitate is the active kinase.
To identify the amino acid residues that are phosphorylated by PLK1, bacterially expressed EBNA2 was phosphorylated by active recombinant PLK1 or left untreated. The EBNA2 protein was digested by trypsin and endoproteinase GluC (V8 Protease) in parallel. Tryptic or V8 protease-derived peptides and phosphopeptides were identified by mass spectrometry (MS). Since neither trypticnor V8-derived peptide covered the C-terminus of EBNA2 sufficiently, a subfragment (453-474) of EBNA2 flanked by arginine residues and expressed as a GST fusion protein was used for further tryptic digest and phosphopeptide mapping ( Fig EV1). Initially, 11 potential phosphorylation sites were found. Of these, five phosphorylation sites (S184, 258, 457, 479, and T465) were confidently localized ( Fig EV2) and six additional sites (T175, 178, 263, 267, 464, and S266) were mapped ambiguously with less confidence. The alignment (Fig EV3) of the protein sequence of different EBV type 1 or type 2 strains or related non-human primate lymphocryptoviruses (LCV) revealed that residues S184 and S258 were conserved in type 1 EBV strains, and S379 and S479 were identical across species while S457 was conserved across human sequences. T465 was identical across human sequences and replaced by the similar T residue in LCV.
To test, if the five confidently localized phosphorylation sites identified in vitro are relevant also in cells, EBNA2 mutants with a singular or combined mutations were generated and expressed in DG75 B cells. Immunoprecipitations and subsequent kinase assays based on endogenous PLK1 trapped in the precipitates were performed ( Fig 3D). None of the mutations severely impaired PLK1 binding but Volasertib prevented phosphorylation of all EBNA2 proteins. In the absence of Volasertib, phosphorylation of S184A was slightly reduced, while phosphorylation of S258A and S479A was not reduced. Phosphorylation of the EBNA2 mutants S457A and T465V was severely impaired, and the combination of both mutations (S457A/T465V) abolished phosphorylation ( Fig 3E). To corroborate this finding, GST-EBNA2 246-487 either in wild-type or as S457A/T465V mutant was submitted to kinase assays using recombinant PLK1 in vitro. This experiment resulted in efficient phosphorylation of wild-type but not mutant EBNA2 S457A/T465V ( Fig 3F). We conclude that the PLK1 docking and the PLK1 phosphorylation site are two independent EBNA2 modules that can be dissected and analyzed by specific mutations (Fig 3G).
To test the phosphorylation mutants for their biological activity, we used a luciferase reporter construct driven by an EBNA2 responsive artificial promoter that harbors a multimerized CBF1 binding site to recruit EBNA2 (Minoguchi et al, 1997). All mutants that carried either the S457A, the T465V mutation, or both mutations showed enhanced transactivation potential, suggesting that PLK1 might negatively regulate EBNA2 activity ( Fig 3H). Since the C-TAD of EBNA2 is known to bind the histone acetylase and co-activator p300 (Chabot et al, 2014), p300 binding to EBNA2 phosphorylation mutants was tested by GST-pulldown experiments. While p300 binding of the single EBNA2 phosphorylation mutants was strongly enhanced by approximately 2-to 3-fold, binding by the doublemutant S457A/T465V was increased even 8-to 9-fold. Thus, enhanced p300 binding of EBNA2 phosphorylation mutants ◀ Figure 2. EBNA2/PLK1 phosphorylation-dependent complex formation.
A-C (A) Schematic outline of EBNA2, its dimerization domains (END, DIM), the region used by CBF1 to recruit EBNA2 to DNA (WW), the C-terminal transactivation domain (TAD), the nuclear localization signal (N), and the EBNA2 fragments used to map the PLK1 docking site (UniProt ID: P12978.1). The panel on the right summarizes the results of the co-immunoprecipitations in (B) and (C). Transfection of DG75 B cells and immunoprecipitation of (B) HA-tagged or for smaller fragments (C) GFP-tagged EBNA2 fragments to co-precipitate endogenous PLK1. Total protein lysates and immunoprecipitates (IP) were analyzed by Western blotting (WB). correlates well with improved transactivation activity. This finding suggests that PLK1 phosphorylation hinders p300 recruitment to EBNA2, thereby inhibiting EBNA2 transactivation ( Fig 3I).
To directly study whether PLK1 inhibits EBNA2 functions, PLK1 and EBNA2 were co-expressed and luciferase reporter assays were performed ( Fig 4A). In these transient transfection experiments, all EBNA2 mutants and all PLK1 mutants were expressed well ( Fig 4B).
EBNA2 wt activity was significantly reduced by co-expression of PLK1 but not affected by the kinase-dead K82M PLK1 mutant. The activity of the docking site mutant S379A was weakly impaired by co-expression of PLK1, suggesting that residual binding activity that is evident in Fig 2H ( demonstrating that EBNA2 S457A/T465V is resistant to PLK1 treatment. Since the kinase-dead PLK1 mutant did not impair the transactivation capacity of any EBNA2 protein, we conclude that PLK1 binding is necessary but not sufficient to inhibit EBNA2 and the phosphorylation of EBNA2 by PLK1 is required to inhibit EBNA2. In summary, PLK1 phosphorylates S457 and T465 within the TAD of EBNA2 to attenuate its activity and to impair p300 binding. PLK1 uses S379A as a phosphorylation-dependent docking site that can be primed by Cyclin B/CDK1. EBV strains expressing EBNA2 mutants deficient for PLK1 docking or phosphorylation are gain-of-function mutants in B-cell immortalization assays To test EBNA2 mutants defective for PLK1 binding or phosphorylation by PLK1 for their B-cell immortalization potential, three new viral mutants based on the viral backbone of the recombinant EBV strain p6008 (Mrozek-Gorska et al, 2019; Pich et al, 2019) were generated. Carboxyl-terminally HA-tagged EBNA2 (EBV wt), HAtagged EBNA2 S379A (EBV S379A), and HA-tagged EBNA2 S457A/ T468V (EBV S457A/T465V) were inserted into the backbone of the recombinant EBV strain p6008 by homologous recombination using a two-step positive-negative selection protocol, also called recombineering (Wang et al, 2009). The integrity of the viral genomes was controlled by restriction digest and sequencing (Fig EV4A-D). Infectious viruses were produced in HEK293 cells. Primary B cells from three unrelated donors were infected. During the first 6 days postinfection, the proliferation rates of EBV wt, EBV S457A/T465V, or EBV S379A-infected B cells were determined. The phosphorylation mutant S457A/T465V proliferated the fastest, followed by S379Aand wt-infected cells (Figs 5A and EV4E and F). Also, long-term cultures infected with EBV wt and both mutants could be established. The proliferation of these lymphoblastoid cell lines (LCLs) was characterized by counting the cells daily ( Fig 5B). The proliferation rates of both long-term cultures infected with mutant viruses were higher than EBV wt-infected cells. EBNA2 expression levels in these long-term cultures were similar for wt and S379A but less pronounced for the S457A/T465V mutant. LMP1 expression varied between the cell lines obtained from different donors but was consistently elevated in EBV S457A/T465V-infected cell lines compared with S379A and EBV wt-infected B cells. MYC expression was seen in all infected cultures, but MYC levels were not affected by individual virus variants indicating that MYC was not the driving force for the accelerated proliferation of the established LCLs ( Fig 5C). In summary, both mutants were gain-of-function mutants with respect to proliferative capacities of the infected cell cultures and LMP1 activation. In comparison, these features were more pronounced in EBV S457A/T465V than in S379A-infected B cells.
Enhanced frequencies of B-cell lymphomas in humanized mice infected with EBV S379A or EBV S457A/T465V Since EBV does not infect murine B cells, tumor development and immune control were studied in humanized mice. Immunodeficient NSG (NOD scid c c À/À ) mice can be engrafted with CD34 + human hematopoietic progenitor cells to develop human immune system components that can be readily infected with human lymphotropic pathogens, including EBV. These chimeric humanized mice were used previously to study EBV-induced tumor formation after EBV infection and anti-viral immune mechanisms against EBV in vivo (Strowig et al, 2009). Hence, we infected three cohorts (designated experiments 1, 2, and 3) of humanized mice intraperitoneally with 10 5 Green Raji Units (GRU) of EBV and analyzed viral blood loads, CD8 + and CD4 + T-cell composition as well as activation in their blood in weekly intervals. Five weeks post-infection, all EBV-positive mice were sacrificed to analyze tumor development ( Fig 6A). In two out of three independent experiments, infection with EBV S457A/T465V caused increased mortality rates compared with EBV wt-or S379A-infected animals (66% survival in the second and 20% in the third experiment; Fig 6B). In addition, mice that were infected with either of the two EBV mutants presented with higher incidence of tumor development in spleen, peritoneal cavity, or lymph nodes (23% for WT, 36% for S379A, and 44% for S457A/T465V; Fig 6C). Viral loads in the spleen of mice infected with mutants were slightly elevated 5 weeks post-infection when mice were sacrificed (Fig 6D). In blood, viral loads of EBV wt-and EBV S379A-infected mice gradually increased until week 3 and remained at similar levels until the end of the experiment. In contrast, viral loads in mice within the EBV S457A/T465V mutant group reached higher levels already 2 weeks after infection and started to decrease at 4 weeks postinfection ( Fig 6E).
A Doxycycline (Dox) induction of EBNA2 in DG75 Dox HA-EBNA2 cells treated for 24 h. Co-immunoprecipitates of HA-EBNA2 and endogenous PLK1 are visualized by Western blotting. B EBNA2/PLK1 co-precipitates were submitted to kinase reactions using [c-32 P] ATP in the absence (control) or presence of 50 ng recombinant active PLK1. C EBNA2 co-precipitates were submitted to kinase reactions in the absence (control) or the presence of the PLK1 inhibitor Volasertib (40 nM). D EBNA2 candidate phosphorylation mutants were expressed in DG75 B cells and tested for PLK1 binding by co-immunoprecipitations followed by Western blotting. E Immunoprecipitates were submitted to kinase reactions as in B, but samples were treated with Volasertib (40 nM) (+) or treated with solvent only (À). F GST-EBNA2 fragment 246-487 wt and mutant S457A/T465V were treated with recombinant active PLK1 (+) in the presence of [c-32 P] ATP in vitro or left untreated (À). CRS, an artificial PLK1 test substrate (Yuan et al, 2002), was used as a positive control. G Schematic presentation of EBNA2 phosphorylation sites by CDK1 and PLK1. H HA-tagged EBNA2 candidate phosphorylation mutants were expressed in DG75 B cells and tested for activation of an EBNA2/CBF1 responsive promoter reporter luciferase plasmid. Activation of the reporter gene is shown as relative response ratio normalized to Renilla luciferase activity and shown relative to wt activity. Data were presented as the mean AE SEM of n = 3 biological replicates. Statistical significance was tested by one-way ANOVA followed by Tukey's multiple comparison test (ns: not significant, ***P < 0.001, ****P < 0.0001, vs WT). I GST-pulldown assay using GST-EBNA2 446-474 as a bait to purify cellular proteins from DG75 cells followed by Western blotting and quantification of signals obtained by GST-and p300-specific antibodies. Relative binding affinities (rel. p300 bindg.) are normalized to the wt signal. Since 3 lanes of the original blot were excised a broken line marks the borders between two parts in the figure.
Source data are available online for this figure.
6 of 17 The Authors EBV infection of B cells is well known to elicit a strong T-cell response. In particular, extensive CD8 + T-cell expansion and activation in blood is a common trait marking EBV infection in humanized mice. It usually follows, with a delay of about a week, rising viral loads (Strowig et al, 2009;Chatterjee et al, 2019) and correlates with these (Caduff et al, 2020;Zdimerova et al, 2021). As CD8 + T cells expand more strongly than CD4 + T cells in response to EBV infection, a rising CD8 + to CD4 + T-cell ratio indicates the extensive proliferation of CD8 + T cells.
CD8 + and CD4 + T-cell populations of mice infected with mutant virus started to expand already 2 weeks post-infection (Fig EV5A  and B). Also, the relative fraction of CD8 + T cells expressing the activation marker HLA-DR increased from week 2 post-infection, thus earlier than the control EBV wt-infected group (Fig EV5C and  D). A similar trend could be observed for mice infected with the EBNA2 S379A mutant. Interestingly, the EBV S457A/T465V mutant more strongly induced the activation of CD4 + T cells in the blood of mice within 5 weeks of infection compared with the EBV wtinfected group (Fig EV5D). In contrast to blood, the fractions of CD8 + and CD4 + T cells in the spleen of infected animals did not differ between the groups ( Fig EV4E) but a higher fraction of CD8 + and a significantly increased fraction of CD4 + T cells were HLA-DR-positive and thus activated in EBV S457A/T465V-infected animals (Fig EV4F and G). The development of memory T cells, i.e., effector memory (Tem), central memory (Tcm), or terminally differentiated subsets that re-express CD45RA (Temra), however, was not influenced by EBNA2 mutations (Fig EV5H and I). In summary, EBV S379A and EBV S457A/T465V infections developed more rapidly in humanized mice, resulting in earlier immune responses, increased lymphomagenesis and more frequent demise of the mice. Figure 4. PLK1 inhibits EBNA2 activity by phosphorylating residues S457/T465.

A B
A Flag-tagged active (PLK1) and kinase inactive PLK1 (K82 M) were co-expressed with HA-EBNA2 wt, docking site mutant (S379A) and phosphorylation mutant (S457A/ T465V) and tested for activation of an EBNA2/CBF1 responsive promoter reporter in DG75 cells. Activation of the reporter firefly luciferase gene is shown as relative response ratio normalized to Renilla luciferase activity and shown relative to EBNA2 wt activity. Data were presented as the mean AE SEM of n = 3 biological replicates. Statistical significance was tested by two-way ANOVA followed by Tukey's multiple comparison test (ns: not significant, *P < 0.05, **P < 0.01). B 30 µg of cellular extracts that were produced for luciferase assays were tested for protein expression using Flag-, HA-and GAPDH-specific antibodies.
Source data are available online for this figure.
ª 2021 The Authors EMBO reports 22: e53007 | 2021 Tumors of EBV S379A-or EBV S457A/T465V-infected mice frequently present as monoclonal B-cell lymphomas The diversity of immunoglobulin gene rearrangements and sequences can be used to define the clonal composition of a given B-cell population (Kuppers et al, 1993;Rajewsky, 1996). In experiment 2, 1 out of 5 EBV wt, 2 out of 4 EBV S379A, and 2 out of 4 EBV S457A/T465V-infected mice developed tumors. The clonality of macroscopically visible tumors detected in experiment 2 was assessed by PCR of rearranged IgH V genes, using subgroup-specific A B C Figure  IgH V primers together with an IgH J primer mix in separate reactions, followed by Sanger sequencing of amplificates obtained (Table EV2). Polyclonal B-cell expansions were found in the splenic tumors of EBV wt (#13) and in one splenic tumor of EBV S379A (#23). In all of the four analyzed tumor-bearing mice infected with mutant EBV, monoclonal B-cell populations were identified, sometimes on a background of remaining polyclonal B cells. In mouse #23, the spleen harbored a polyclonal B-cell population, while the peritoneum showed a monoclonal B-cell expansion. This analysis validates that the tumors developing in EBNA2 mutant mice are indeed monoclonal, and hence bona fide malignant B-cell lymphomas. In all five monoclonal lymphomas, productive in-frame IgH V genes were obtained. Three of the samples showed additional out-of-frame rearrangements, which likely represent the second IgH alleles of the monoclonal lymphomas defined by the in-frame rearrangements. Since humanized NOD scid c c À/À mice in the experimental setup of our study cannot generate germinal centers, we did not expect somatic hypermutation to occur. Indeed, somatic mutations were not observed in any specimen.

Discussion
EBNA2 mutations, which impair PLK1 binding or prevent phosphorylation by PLK1, accelerate cellular proliferation and tumor formation Our study demonstrates that PLK1 directly binds to the phosphorylated EBNA2 residue S379 and phosphorylates S457 and T465 located in the transactivation domain of EBNA2. S457A/T465V missense mutants can still bind to PLK1 but are not phosphorylated. The EBNA2 docking site S379A and EBNA2 S457A/T465V phosphorylation mutants, both show enhanced transactivation capacities with S457A/T465V being stronger than S379A. This enhanced potential might be caused in part by the elevated levels of histone acetylase and co-activator p300 that binds to the S457A/T465V mutant. However, p300 is not the only candidate molecule that binds to the EBNA2 N-terminal TAD. Since several cellular factors that trigger transcription initiation and chromatin remodeling interact with the TAD of EBNA2, these might directly or indirectly contribute to the phenotype of S457A/T465V. The alignment of EBV and LCV protein sequences revealed a high conservation of all three residues across EBV strains. S379 and T465 were conserved even across species. Considering that the respective mutants are gain of function, i.e., activated proteins, we think that PLK1 contributes to the establishment of latency by attenuating EBNA2 activity. Both EBV mutants, which expressed EBNA2 S379A or S457A/ T465T, were fully immortalization competent and initiated longterm proliferating B-cell cultures with accelerated cell division rates and elevated induction of the EBNA2 target gene LMP1. The impact of both mutations was further studied in humanized mice susceptible to EBV infection. Blood samples of infected mice were analyzed weekly for viral loads and immune responses of CD4 + and CD8 + Tcell populations. Elevated viral loads and CD4 + and CD8 + T-cell activation in the blood of EBV S457A/T465V of infected mice preceded CD8 + T-cell expansion. On the day of sacrifice, splenic EBV loads were slightly elevated in both EBNA2 mutants compared with EBV wt-infected animals. Based on earlier publications, increasing viral loads were expected to precede expansion and activation of EBVspecific CD8 + T cells since antigen abundance stimulates CD8 + T-cell responses (Strowig et al, 2009;Shultz et al, 2010). Irrespective of the early adaptive immune response to infection, more EBV mutantinfected mice developed lymphomas when compared to EBV wt infections. The accelerated proliferation rates that we observed in cell culture might contribute to tumor size in mice. The enhanced transcriptional activities of the EBNA2 mutants induce higher LMP1 levels, which is a critical viral oncogene in B-cell transformation and thus also might promote tumor formation in vivo. In addition, both EBNA2 mutants might also induce distinct viral and cellular RNAs, which promote tumor progression but still are to be identified.
Importantly, tumors detected in mice infected with mutant virus were predominantly monoclonal and were detected in the spleen as well as in the peritoneum and in lymph nodes. The elevated viral loads seen in mice infected with mutant EBV might increase tumor incidence by amplifying the number of infected B cells in these animals and thereby might initiate the lymphoproliferative disease at multiple sites but they do not explain disease progression to monoclonal B-cell lymphomas. It will be interesting to study whether lymphomas in lymph nodes or peritoneum are metastatic descendants of splenic tumors. The genetic or epigenetic processes driving the clonal evolution of single cells and giving rise to monoclonal tumors still need to be explored.

PLK1 restrains the growth transformation potential of EBNA2 to establish tumor-free survival of latently infected hosts
Both mutants independently target highly conserved amino acid residues (Fig EV3) but surprisingly are both gain-of-function ◀ Figure 5. EBV strains expressing EBNA2 mutants deficient for PLK1 docking or phosphorylation accelerate B-cell proliferation and express elevated levels of the viral LMP1 protein.
A Primary B cells were stained with cell trace violet before they were infected, and cell cultures were started. Loss of the fluorescent cell tracking dye was recorded by flow cytometry to monitor proliferation on days 0, 4 and 6 post-infection. Mean values for three biological replicates, standard deviation, and significance of changes in signal loss of mutants compared with wild-type controls Data were presented as the mean AE SEM of n = 3 biological replicates. Statistical significance was tested by two-way ANOVA followed by Tukey's multiple comparison test (*P < 0.05, ***P < 0.001, ****P < 0.0001, vs WT). B Long-term growth-transformed B-cell cultures infected with recombinant EBV mutants were seeded at a starting concentration of 2 × 10 5 cells per ml. Data were presented as the mean AE SEM of n = 4 biological replicates. Statistical significance was tested by two-way ANOVA followed by Tukey's multiple comparison test (*P < 0.05, **P < 0.01, ****P < 0.0001, vs WT). C Expression of EBNA2, LMP1, PLK1, MYC, and GAPDH was analyzed by Western blotting and immunostaining of whole cell extracts of cell lines established from 3 individual donors.
Source data are available online for this figure.
ª 2021 The Authors EMBO reports 22: e53007 | 2021 mutants, which exhibit improved transformation capacities. We conclude that the control of EBNA2 activity is evolutionary conserved and might be beneficial for host and virus to protect against cancerogenesis and establish and stabilize long-term latency and thus the viral reservoir in the host. Since PLK1 activity peaks in mitosis, PLK1 might restrain EBNA2 activities preferentially during A B E F C D Figure 6. EBV strains carrying EBNA2 mutants with diminished PLK1 binding or resistant to PLK1 phosphorylation cause lymphomas in humanized mice more frequently than EBV wt.
Data information: The shape of data points indicates to which cohort (experiment 1, 2, or 3) the respective animal belongs while the color indicates the EBV EBNA2 mutant with which the animal was infected. Statistical significance was tested using the Mann-Whitney U-test with Holm-Sidak correction for multiple comparisons if not indicated differently.
this time window. Unfortunately, since during mitosis the global cellular transcription of the condensed chromosomes and the viral episome is silenced, EBNA2 activity cannot be reliably tested in mitotic cells. There appears to be a strict biological necessity to dynamically control EBNA2 activity by multiple viral and cellular factors. Transient transfections and reporter gene studies have suggested that Cyclin B/CDK1 or the viral lytic kinase BGLF4 can control EBNA2 activity, but these studies have not linked their findings to PLK1 activity (Yue et al, 2004(Yue et al, , 2005(Yue et al, , 2006. PLK1 is a key control element of multiple cell cycle stages including G2/M transition, M-phase progression, and cytokinesis. In addition, it has been shown before that PLK1 can affect the activity of cellular transcription factors through several mechanisms. Phosphorylation of the tumor suppressor p53 and the related p73 protein impair the transactivation activity of both transcription factors. In p73, the substrate site of PLK1 has been mapped to the TAD (Ando et al, 2004;Martin & Strebhardt, 2006;Koida et al, 2008;de Carcer, 2019). Phosphorylation of the transcription factor FOXO1 by PLK1 causes its nuclear exclusion and thereby prevents its action (Yuan et al, 2014). On the other hand, PLK1 phosphorylation of the transcription factor FoxM1, a critical transactivator of mitotic gene expression, induces FoxM1 activity (Marceau et al, 2019). It is tempting to speculate that PLK1 might contribute to the general mitotic transcriptional shut down by inhibiting distinct transcription factor activities. In parallel, PLK1 might guide the transcriptional activity of the cell to processes relevant to mitotic progression and maintenance of genomic integrity.
In healthy immunocompetent hosts, EBNA2 is expressed in a short time window immediately post-infection before either EBNA2 expression is silenced or the EBNA2 expressing cell is eliminated by the immune system. Immunodeficient patients, suffering from profound T-cell suppression, can develop aggressive EBNA2positive B-cell lymphoproliferative diseases, and EBNA2 is a driving force for these tumor entities. The "primary goal" of latent viruses such as EBV is to establish a balanced equilibrium of its latent and lytic phase while tumor development is an accident in its life cycle with no benefit for virus dissemination (Shannon-Lowe & Rickinson, 2019). EBNA2 obeys the intracellular signaling cues that silence its activity during a short time window of the cell cycle in order to establish latency and maintain the viral life cycle rather than promoting tumorigenesis. It is well established that high-level expression of PLK1 promotes carcinogenesis in multiple tissues (Strebhardt & Ullrich, 2006;Strebhardt, 2010;Liu et al, 2017). PLK1 is also considered as an oncotarget for aggressive B-cell lymphomas since it stabilizes MYC (Ren et al, 2018). Currently, clinical trials evaluate the safety and efficacy of PLK1 inhibitors for patient treatment. However, in some tumor types high-level PLK1 expression can suppress cancerogenesis (Raab et al, 2018;de Carcer, 2019). Whether it exerts oncogenic or tumor suppressor functions depends on the cellular environment, defined by tissue origin, genetic lesions and potential additional factors that still need to be defined. Our study makes an important contribution to elucidate the bi-functional nature of PLK1 in its specific context.
Here, we show that PLK1 is an important cellular control factor that restrains the proliferation and transformation of latently EBVinfected B cells driven by a growth program that depends on EBNA2. Since two distinct EBNA2 mutants that both target independent PLK1related functions of the EBNA2/PLK1 complex promote cancerogenesis, we conclude that PLK1 might act as a tumor suppressor in EBNA2-driven lymphomagenesis. Until to date, the interaction of EBNA2 with PLK1 was unknown, and thus, latent EBV was not considered a risk factor in PLK1 inhibitor-based therapeutic strategies. Knowing that latent EBV is ubiquitously found in human hosts, PLK1 inhibitors could cause adverse effects with detrimental consequences for immune-deficient patients, which are prone to EBV-related cancerogenesis. Humanized mice infected with EBV or specific EBV mutants will serve as pre-clinical animal models to study the action of PLK1 inhibitors in the context of latent EBV infection.

Purification of human primary B cell and LCL establishment
Human primary B cells isolated from adenoid from routine adenoidectomy were obtained from the Department of Otorhinolaryngology, Klinikum Grosshadern, Ludwig Maximilians University of Munich, and Dritter Orden Clinic, Munich-Nymphenburg, Germany. All clinical samples were made fully anonymous. To isolate human primary B cells, T cells were depleted by erythrocyte rosetting using sheep red blood cells and B cells were separated by Ficoll density gradient centrifugation as recommended by the manufacturer (GE Healthcare). The remaining erythrocytes were lysed in red blood cell lysis buffer (155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA). Cells were co-stained by anti-CD3 + (UCHT1; BD Pharmingen) and anti-CD19 + (HIB19; BD Pharmingen) antibodies and analyzed by flow cytometry.
To generate LCLs, primary human B cells were infected with recombinant EBV at a ratio of 1 green Raji unit (GRU) to 10 cells for 48 h and cultivated in medium containing 0.5 µg/ml cyclosporine A for 2 weeks before routine cell culture conditions were applied.

Construction of plasmids
All plasmids used in the study were constructed by conventional PCR, restriction digestion, and ligation. Mutated alleles were generated by overlap PCR adapted from the previous protocol (reviewed in Francis et al, 2017). In essence, overlap PCR is based on four strategically designed primers. Internally positioned primers must contain complementary sequences to each other, and both of them must contain the mutation of interest, like a substitution, a deletion, or an insertion. The flanking primers might contain restriction enzyme recognition sites to facilitate the cloning of the amplified fragment. Two steps were performed, in the first round of PCRs using the forward primer of the flanking primers with the reverse primer of the internally positioned primers and vice versa, respectively. The resulting amplified short fragments worked as templates when mixed with the flanking primer pairs, which results in amplification of the final long fragment with the desired mutation in the second round of PCR. The second PCR product is digested and inserted into a corresponding vector. For each plasmid used in the study, detailed information about primer pairs, template, and vector is shown in Table EV2.

Expression and purification of His-or GST-tagged proteins
His 6 -tagged and GST-tagged proteins were expressed in E. coli Rosetta (DE3) cells and purified according to manufacturer's instructions using Ni-chelate agarose (Quiagen) or glutathione coupled Sepharose 4B beads (GE Healthcare).

Kinase assay in vitro
Purified protein or protein coupled to beads were incubated for 30 min at 37°C with recombinant active Cyclin B/CDK1 (100 ng) or PLK1 (50 ng) in the presence of 1mM normal ATP or plus 0.25 mCi/ml c-32 P labeled ATP in PK buffer (50 mM Tris-HCl, 10 mM MgCl 2 , 0.1 mM EDTA, 2 mM DTT, 0.01% Brij 35, pH 7.5) in a total reaction volume of 20 µl.

EBV BAC recombineering
All recombinant EBV strains used in this study were generated by a two-step selection protocol using the k-prophage-based heatinducible For the first step, a Kan/rpsL expression cassette was flanked by 50 nt EBV sequences of the respective EBNA2 gene locus by PCR using the template p6012. The resulting PCR product was used to insert the Kan/rpsL cassette by homologous recombination into the specific EBV/EBNA2 target site by transformation and kanamycin (30 µg/ml) and chloramphenicol (12.5 µg/ml) selection of SW105 pre-transformed with the recombinant target EBV plasmid. As a second step, a synthetic DNA fragment or PCR product carrying the desired mutation flanked by˜300 nt of the genomic viral sequence was used to replace the Kan/rpsL cassette by homologous recombination to generate the final mutant EBV plasmid by streptomycin (1 mg/ml) and chloramphenicol (12.5 µg/ml) selection. For each BACmid used in the study, detailed information about primer pairs, template, and vector is shown in Table EV2.

Production of recombinant virus
HEK293 transfectants carrying the recombinant virus plasmid were induced for virus production by co-transfection of 0.5 µg of the plasmids p509 encoding BZLF1 and p2670 encoding BALF4 per one 6well in 3 ml cell cultures. The supernatants of the transfectants were harvested 3 days post-transfection and passed through a 0.8-µm filter. For quantification of viral titers, 1 × 10 5 Raji cells were infected with serial dilutions of viral supernatants in 2 ml cultures and the percentage of GFP-positive cells was determined by FACS analysis 3 days post-infection. The concentration of viral stocks was expressed as the number of green Raji units (GRU) ( Table 1).

Isothermal titration calorimetry
Experiments were performed on a ITC200 instrument in triplicates and analyzed with the Malvern software. 100 µM PBD was provided in the cell and titrated with 1 mM concentration of wild type (PNTSSPS) or phosphopeptide (PNTSpSPS) with 25 times 1.5 µl injections at 25°C.

Generation and infection of humanized mice
NOD scid c c À/À (NSG) mice obtained from the Jackson Laboratories were bred and maintained under specific pathogen-free conditions at the Institute of Experimental Immunology, University of Zurich. CD34 + human hematopoietic progenitor cells were isolated from human fetal liver tissue (obtained from Advanced Bioscience Resources) using the CD34 MicroBead Kit (Miltenyi Biotec) following the protocol provided by the manufacturer. Newborn NSG mice (age: 1-5 days) were irradiated with 1 Gy by use of an X-ray source. 1-3 × 10 5 CD34 + human hematopoietic progenitor cells were injected intrahepatically 5-7 h after irradiation. Reconstitution of mice with human immune system components was investigated 10-12 weeks after engraftment by flow cytometry for the cell surface expression of huCD45 (BV605 or Pacific Blue, clone HI30; BioLegend), huCD3 (PE, clone UCHT1; BV785, clone OKT3; BioLegend), huCD19 (PE-Cy7, clone HIB19; BioLegend) and (PE-Texas Red, Clone SJ25-C1), huCD4 (APC-Cy7, clone RPA-T4; BioLegend), huCD8 (PerCP, clone SK1; BioLegend), huNKp46 (APC, clone 9E2; BD), and HLA-DR (FITC or PE-Cy7 clone L243; BioLegend) on PBMCs. 12-16 weeks after engraftment, humanized mice were infected intraperitoneally with 1x10 5 GRU of EBV wt, EBV S379A, or EBV S457A/T465V. For each experiment, a different cohort of mice reconstituted with CD34 + cells derived from one donor was generated. The animals were ascribed to a distinct experimental group ensuring similar ratios of males to females and similar reconstitution levels and immune cell activation in the peripheral blood. 5 weeks after infection, mice were sacrificed if not necessitated earlier by the regulations of our experimental animal license as a consequence of general health conditions or weight loss over 20%. For analysis of the experiments, only those mice that showed two of the following signs of infection were regarded as infected and included in the analysis: (i) viral loads in spleen, (ii) viral loads in blood at one time-point during the experiment, and (iii) EBNA2 + cells in spleen as evaluated by histology. The respective animal protocol (ZH159-17/ZH008-20) was approved by the veterinary office of the canton of Zurich, Switzerland.

Whole blood and spleen preparations for immune phenotyping
Whole blood of mice was collected from the tail vein and prepared for immunophenotyping by lysing erythrocytes with NH 4 Cl. Spleens of mice were mashed, subsequently filtered with a 70-µm cell strainer, and afterward mononuclear cells were separated using Ficoll-Paque gradients. Total cell counts were determined from purified mononuclear cell suspensions using a DxH500 Hematology Analyzer (Beckman Coulter). Purified cell suspensions were stained for 30-40 min at 4°C in the dark with a master mix of the respective antibodies followed by a washing step in PBS. The stained cells were analyzed in an LSRFortessa cytometer (BD Biosciences). Flow cytometry data were analyzed using the FlowJo software.

EBV DNA quantification in tissue
Total DNA from splenic tissue and whole blood was extracted using DNeasy Blood & Tissue Kit (QIAGEN) and NucliSENS (bioM erieux), respectively, according to manufacturer's instructions. Quantitative analysis of EBV DNA in humanized mouse spleens and blood was performed by a TaqMan (Applied Biosystems) real-time PCR as described previously (Berger et al, 2001) with modified primers (5 0 -CTTCTCAGTCCAGCGCGTTT-3 0 and 5 0 -CAGTGGTCCCCCTCCCTAG A-3 0 ) and the fluorogenic probe (5 0 -(FAM)-CGTAAGCCAGACAGCA GCCAATTGTCAG-(TAMRA)-3 0 ) for the amplification of a 70-base pair sequence in the conserved BamHI W fragment of EBV. Realtime PCR was run on a ViiA 7 Real-time PCR system or a CFX384 Touch Real-Time PCR Detection System, and samples were analyzed in duplicates.

IgV gene rearrangement sequence analysis
For PCR analysis of rearranged immunoglobulin genes, macroscopically visible tumors were dissected and cryofixated in Tissue Tekâ O.C.T. TM (VWR, Cat# SAKU4583) on dry ice. DNA was isolated from 10 to 20 8-µm frozen tissue sections of biopsies from humanized mice using the PureGene DNA Isolation Kit (Qiagen, Hilden, Germany). PCR was performed with six framework region 1 subgroup-specific primers and a JH primer mix (3 0 JH mix) for 35 cycles in six separate reactions, using 300 ng of DNA per reaction. Primer sequences have been published before (Kuppers et al, 2013). PCR products were purified from agarose gels and Sanger sequenced with the IGHV primers used for PCR and the BigDye Sequencing Kit (ABI, Heidelberg, Germany) on an Applied Biosystems 3130 Genetic Analyzer (ABI). Sequences were evaluated with the IMGT/V-Quest software (http://www.imgt.org/IMGT_vquest/input).
Mass spectrometry of EBNA2 co-immunoprecipitates DG75 cells were transfected with a C-terminal HA-tagged EBNA2 (plasmid: pAG155) using electroporation. After 48 h, cells were lysed in NP-40 lysis buffer (1% NP-40, 10 mM Tris-HCl pH 7.4, 3% glycerol, 150 mM NaCl, 1 mM EDTA supplemental with cOmplete protease and PhoStop phosphatase inhibitor). Cell debris was depleted by centrifugation. The cell extract was incubated with protein G beads covalently coupled with an HA-specific antibody. Immunoprecipitates were eluted with Laemmeli buffer. After reduction and alkylation using DTT and IAA, the proteins were centrifuged on a 30 kDa cutoff filter device and washed thrice with UA buffer (8 M urea in 0.1 M Tris/HCl pH 8.5) and twice with 50 mM ammonium bicarbonate. The proteins were digested for 16 h at 37°C using 1 µg trypsin. After centrifugation (10 min at 14,000 ×g), the eluted peptides were acidified with 0.5% trifluoroacetic acid and stored at À20°C.
LC-MS/MS analysis was performed on a LTQ Orbitrap XL mass spectrometer (Thermo Scientific, Waltham, MA, USA) online coupled to an Ultimate 3000 nano-RSLC (Thermo Scientific). Tryptic peptides were automatically loaded on a C18 trap column (300 µm inner diameter (ID) × 5 mm, Acclaim PepMap100 C18, 5 µm, 100 A, LC Packings) prior to C18 reversed-phase chromatography on the analytical column (Acclaim PepMap C18, 50 µm ID × 250 mm, 2 µm, 100 A, LC Packings) at 300 nl/min flow rate in a 140-min acetonitrile gradient from 4 to 30% in 0.1% formic acid. Profile precursor spectra from 300 to 1,500 m/z were recorded in the Orbitrap with a maximum injection time of 500 ms. TOP10 fragment spectra of charges 2-7 were recorded in the linear ion trap with a maximum injection time of 100 ms, an isolation window of 2.0 m/z, a normalized collision energy of 35, and a dynamic exclusion of 60 s.
Raw files were analyzed using Progenesis QI for proteomics (version 2.0, Nonlinear Dynamics, part of Waters). Features of charges 2-7 were used, and all MS/MS spectra were exported as mgf file. Peptide searches were performed using Mascot search engine (version 2.5.1) against the Ensembl Human protein database (100158 sequences, 37824871 residues). Search settings were as follows: 10 ppm precursor tolerance, 0.6 Da fragment tolerance, one missed cleavage allowed. Carbamidomethyl on cysteine was set as a fixed modification, deamidation of glutamine and asparagine allowed as variable modification, as well as oxidation of methionine. Applying the percolator algorithm with a cutoff of 13 and P < 0.05 resulted in a peptide false discovery rate (FDR) of 1.54%. Search results were reimported in the Progenesis QI software. Proteins were quantified by summing up the abundances of all unique peptides per protein.
Resulting protein abundances were used for the calculation of fold changes and statistical values.
Proteins were separated by SDS-PAGE and stained with Coomassie Colloidal Blue. Bands corresponding to EBNA2 full length (86 kD) or GST-EBNA2 C-terminal fragment (28 kD) were sliced out from the gel lane, and proteins were then reduced, alkylated, and digested with either trypsin or GluC (Roche), as previously described (Shevchenko et al, 1996).
Dried peptides were reconstituted in 0.1% FA/2% ACN and subjected to MS analysis using a Dionex Ultimate 3000 UHPLC + system coupled to a Fusion Lumos Tribrid mass spectrometer (Thermo Fisher). Peptides were delivered to a trap column (75 lm × 2 cm, packed in-house with 5 lm Reprosil C18 resin; Dr. Maisch) and washed using 0.1% FA at a flow rate of 5 ll/min for 10 min. Subsequently, peptides were transferred to an analytical column (75 lm × 45 cm, packed in-house with 3 lm Reprosil C18 resin, Dr. Maisch) applying a flow rate of 300 nl/min. Peptides were chromatographically separated using a 50-min linear gradient from 4 to 32% solvent B (0.1% FA, 5% DMSO in ACN) in solvent A (0.1% FA in 5% DMSO). The mass spectrometer was operated in data-dependent mode, automatically switching between MS and MS/MS. Full-scan MS spectra (from m/z 360-1,500) were acquired in the Orbitrap with a resolution of 60,000 at m/z 200, using an automatic gain control (AGC) target value of 5e5 charges and maximum injection time (maxIT) of 10 ms. The 10 most intense ions within the survey scan were selected for HCD fragmentation with normalized collision energy set to 28%.
Isolation window was set to 1.7 Th, and MS/MS spectra were acquired in the Orbitrap with a resolution of 15,000 at m/z 200, using an AGC target value of 2e5, and a maxIT of 75 ms. Dynamic exclusion was set to 20 s.
Peptide and protein identification was performed using MaxQuant (version 1.5.3.30) with its built in search engine Andromeda (Cox & Mann, 2008). Spectra were searched against a SwissProt database, either the Spodoptera frugiperda (OX 7108 -26,502 sequences) or Escherichia coli (UP000002032 -4,156 sequences), supplemented with the EBNA2 protein sequence. Enzyme specificity was set to Trypsin/P or GluC accordingly, and the search included cysteine carbamidomethylation as a fixed modification, protein N-term acetylation, oxidation of methionine, and phosphorylation of serine, threonine, tyrosine residue (STY) as variable modifications. Up to two and three missed cleavage sites were allowed for trypsin and GluC, respectively. Precursor tolerance was set to 4.5 ppm (after MS1 feature re-calibration) and fragment ion tolerance to 20 ppm. The match between runs feature was enabled. Peptide identification was further filtered for a minimum Andromeda score of 20 or 40, for unmodified and modified (phosphorylated) sequences, respectively. A site localization probability of at least 0.75 was used as the threshold for confident localization (Vizcaíno et al, 2013(Vizcaíno et al, , 2016.

Cell proliferation assay
Human adenoid primary B cells were stained with CellTrace Violet according to the manufacturer's instructions (Thermo Fisher Scientific). Proliferation of CD19 + B cells was monitored by flow cytometry using BD Fortessa, and the data were analyzed using the FlowJo software (Version 10.5.3).
Dual-luciferase assay 5 × 10 6 DG75 cells were electroporated with 1.5 lg EBNA2 expression plasmids and the luciferase construct 1.5 lg pGa981.6 (Minoguchi et al, 1997) carrying a multimerized CBF1 binding site to measure EBNA2 activity and 0.2 lg Renilla luciferase expression plasmid. 24 h post electroporation, cells were washed, pelleted, and lysed in 100 ll lysis buffer for 30 min on ice. Cell extracts were tested by the dual-luciferase assay according to the manufacturer's instructions (Promega).

Data availability
Phosphorylation of EBNA2 by PLK1: The mass spectrometry proteomic data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository http:// proteomecentral.proteomexchange.org/cgi/GetDataset with the dataset identifier PXD022970 (Vizcaíno et al, 2013(Vizcaíno et al, , 2016.
Expanded View for this article is available online.