RSK1 SUMOylation is required for KSHV lytic replication

RSK1, a downstream kinase of the MAPK pathway, has been shown to regulate multiple cellular processes and is essential for lytic replication of a variety of viruses, including Kaposi’s sarcoma-associated herpesvirus (KSHV). Besides phosphorylation, it is not known whether other post-translational modifications play an important role in regulating RSK1 function. We demonstrate that RSK1 undergoes robust SUMOylation during KSHV lytic replication at lysine residues K110, K335, and K421. SUMO modification does not alter RSK1 activation and kinase activity upon KSHV ORF45 co-expression, but affects RSK1 downstream substrate phosphorylation. Compared to wild-type RSK1, the overall phosphorylation level of RxRxxS*/T* motif is significantly declined in RSK1K110/335/421R expressing cells. Specifically, SUMOylation deficient RSK1 cannot efficiently phosphorylate eIF4B. Sequence analysis showed that eIF4B has one SUMO-interacting motif (SIM) between the amino acid position 166 and 170 (166IRVDV170), which mediates the association between eIF4B and RSK1 through SUMO-SIM interaction. These results indicate that SUMOylation regulates the phosphorylation of RSK1 downstream substrates, which is required for efficient KSHV lytic replication.

SUMOylation is a post-translational modification process, during which small ubiquitinlike modifiers (SUMOs) are covalently conjugated to lysine residues of the target protein substrates. Like ubiquitination, SUMOylation is a three-step enzymatic process involving activating enzyme E1 (SAE1/UBA2), conjugating enzyme E2 (Ubc9), and various E3 ligases. Sometimes, SUMO E3 ligases are dispensable for the target modifications [14]. Mammals encode at least three SUMO isoforms (SUMO1-3), with SUMO2 and SUMO3 sharing 96% identify. SUMO conjugation typically mediates selected protein-protein interaction through SUMO-SIM (SUMO-interaction motif) interaction, leading to subcellular localization changes or transcriptional repression of target proteins [14]. Multiple cellular kinases, such as AKT and PKC undergo SUMOylation modification, which impacts their kinase activity, stability, or substrate specificity [15][16][17]. Here, we showed that RSK1 is SUMOylated primarily at Lys 110 , Lys 335 , and Lys 421 and that SUMOylation of these sites appears to be required for efficient KSHV lytic replication and progeny virus production. We demonstrated that SUMOylation does not affect RSK1 activation or kinase activity, but appearantly modulates its substrate phosphorylation. Specifically, we found that RSK1 SUMOylation is required for ORF45-induced eIF4B phosphorylation and revealed a SIM site in eIF4B that engages recruitment of RSK1 for its phophorylation. Thus, our findings revealed a molecular mechanism by which RSK1 SUMOylation regulates KSHV lytic replication.

RSK1 SUMOylation is enhanced during KSHV lytic replication
SUMOylation of MEK (also known as MAPK kinase or MAP2K) has been shown to interfere with the specific docking interaction between MEK and ERK and thus negatively regulate ERK activation [18]. However the roles of SUMOylation in regulating other components of MAPK cascades remained unexplored. To determine whether SUMOylation regulates RSKs, we first determined whether RSKs undergo SUMOylation in cells. We expressed each Flag-tagged RSKs together with HA-tagged SUMO1, SUMO2, or their G/A mutants (SUMO1 G96A/G97A and SUMO2 G92A/G93A ) in HEK293T cells. SUMO G/A mutants cannot be ligated to lysine residues and served as a negative control for SUMOylation modification of RSKs. Flag-RSKs were immunoprecipitated under denature condition, resolved on SDS-PAGE, and immunoblotted with anti-Flag or anti-HA antibodies. The signal of SUMOylated RSKs migrated about 20 kDa from the unSUMOylated forms, and was abolished under SUMO G/A mutant conditions (lane 3 and lane 5) (Fig 1A). As shown in Fig 1A, RSK1 was strongly modified by both SUMO1 (lane 2) and SUMO2 (lane 4), the SUMOylation of RSK2 and RSK3 were much weaker than RSK1, while RSK4 SUMOylation was undetected in the same condition (Fig 1A), indicating different RSK isoforms are selectively modified by SUMOs. SUMO modification of endogenous RSK1 was also observed in BJAB, A549, and HeLa cells (Fig 1B). In addition, recombinant RSK1 is SUMOylated in vitro when mixed with SAE1/UBA2 (E1), Ubc9 (E2), SUMO1, and ATP/Mg (Fig 1C). These results showed that RSK1 is SUMOylated both in vitro and in vivo.
Next, we examined RSK1 SUMOylation levels over time during KSHV lytic replication. We modified BAC16 by replacing the GFP with TET3G under EF1α promoter and inserted seven

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication tet response elements in the promoter of RTA to generate iBAC (GenBank accession number: OK358814) (S1 Fig), in which RTA expression is controlled by a Tet-On promoter on BACmid [19]. To aid detection of RSK1 SUMOylation, we stably expressed Flag-tagged RSK1 in SLK-iBAC cells with both RSK1 and RSK2 knockout by CRISPR-Cas9 (SLK RSK1/2 DKO -iBAC). SLK-iBAC cell entered lytic replication cycle upon doxycycline and sodium butyrate treatment as shown by immunoblots of several lytic genes in the time course (Fig 1D and 1E). RSK1 SUMOylation was detected in SLK RSK1/2 DKO -iBAC-Flag-RSK1 cells, and increased over time after induction of KSHV lytic replication (Fig 1D). The endogenous RSK1 SUMOylation was HEK293T cells were co-transfected with indicated plasmids and cell lysates were subjected to IP and IB with indicated antibodies. (B) Endogenous RSK1 is SUMOylated. Endogenous RSK1 was immunoprecipitated with anti-RSK1 or IgG (control) antibodies under denature condition from lysates of BJAB, A549, and HeLa cells and the coimmunoprecipitated proteins were analyzed by IB with indicated antibodies. (C) RSK1 is SUMOylated in vitro. Recombinant RSK1 was incubated with SAE1/UBA2 (E1), Ubc9 (E2), SUMO1, and ATP/Mg in SUMO reaction buffer. The SUMOylation of RSK1 were detected by indicated antibodies. (D) RSK1 SUMOylation is increased during KSHV lytic replication. The SLK RSK1/2 DKO -vector, or SLK RSK1/2 DKO -Flag-RSK1, SLK-iBAC RSK1/2 DKO -vector, or SLK-iBAC RSK1/2 DKO -Flag-RSK1 cells were treated with doxycycline (2 μg/ml) and sodium butyrate (1 mM) to induce KSHV lytic replication. Cell lysates were harvested at indicated time points and subjected to IP under denature condition and IB with indicated antibodies. (E) The endogenous RSK1 SUMOylation is elevated during KSHV lytic replication. SLK-iBAC cells were treated with doxycycline (2 μg/ml) and sodium butyrate (1 mM) to induce KSHV lytic replication. Cell lysates were harvested at indicated time points and subject to IP with anti-RSK1 or control IgG under denature condition and IB with indicated antibodies. also elevated during KSHV lytic replication in SLK-iBAC cells (Fig 1E). These results demonstrated that RSK1 is SUMOylated during KSHV lytic replication, suggesting its potential role in regulating KSHV replication.

Identification of RSK SUMOylation sites
Next, we sought to determine which sites of RSK1 are primarily modified by SUMOs. SUMO1 modification of RSK1 resulted in three prominent shifted bands (110~170 kDa) above the unmodified RSK1 (~90 kDa), suggesting at least three different lysine residues are primarily modified by SUMOs (Fig 1A). To locate the SUMOylation sites on RSK1, we truncated it into two fragments, RSK1 NT [amino acid (aa) 1-418] and RSK1 CT (aa 321-735). When co-expressed with SUMO1, RSK1 NT generated two additional bands above the unmodified form, indicating two major lysines are SUMOylated. Mutagenesis analysis revealed that Lys 110 and Lys 335 were two primary SUMOylation sites and mutating both of them to arginine abolished RSK1 NT SUMOylation (Fig 2A). Consistently, K110/335R double mutation in the full-length RSK1 reduced SUMOylation bands from three to only one, comparing lane 4 to 2 ( Fig 2B). This result not only confirms that K110 and K335 are primary SUMOylation sites in aa1-418 region but also suggests an additional SUMO site located in the aa 418-735 region. Therefore, we generated more lysine to arginine mutations on the RSK1 K110/335R backbone to locate the third SUMOylation site. As shown in Fig 2C,  , and Lys 667 , had little or no effect ( Fig 2C). These data demonstrated that Lys 110 , Lys 335 , and Lys 421 of RSK1 are the three major sites that are modified by SUMO1.

RSK1 SUMOylation is required for KSHV lytic replication and progeny virus production
The activation of RSKs by ORF45 has been shown to be crucial for KSHV lytic replication [3,4], we next determined whether RSK1 SUMOylation plays a role in this process. RSK1 and RSK2 but not RSK3 are highly expressed in SLK-iBAC cells ( Fig 3A) [19]. Genetic knockout of

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication RSK1 and RSK2 by CRISPR-Cas9 dramatically reduced KSHV lytic gene expression at both the mRNA and protein levels, as well as viral DNA copy number, and progeny virus production (Fig 3B-3E). To determine whether SUMOylation of RSK1 is required for KSHV lytic replication, we complemented SLK RSK1/2 DKO -iBAC cells with RSK1 K110/335/421R or wild-type RSK1 (RSK1 WT ) and then induced lytic replication with doxycycline and sodium butyrate. We first examined viral protein expression by immunoblot. In SLK-iBAC RSK1/2 DKO cells, the protein expression levels of immediate-early genes (RTA, K8, and ORF45), early genes (K3 and ORF52), and late genes (ORF55 and ORF65) were significantly rescued by RSK1 WT at 24, 48, and 72 h post-induction, to similar levels as wild-type SLK-iBAC cells ( Fig 3B). However, RSK1 K110/335/421R SUMOylation deficient mutant only partially rescued these viral protein expressions (Fig 3B), indicating that RSK1 SUMOylation is required for efficient KSHV lytic replication. Next, to systematically compare KSHV viral gene expression profile between

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication potent activator), ORF45 F66A (a mutant that cannot bind to or activate RSKs) [3], or vector control, we evaluated the phosphorylation level of RSK1 at Thr359, Ser363, and Ser380, which indicate the activation level of RSK1 by the upstream signaling [9,13]. Consistent with our previous findings [3], KSHV ORF45 WT expression led to significantly increased phosphorylation of RSK1 at Thr359, Ser363, and Ser380, as compared to vector control, while ORF45 F66A did not (Fig 4B). Similar to RSK1 WT , ORF45 WT -mediated phosphorylation levels of Thr359, Ser363, or Ser380 were not affected by mutations on RSK1 SUMOylation sites (Fig 4B), indicating that SUMOylation has little or no effect on RSK1 activation by KSHV ORF45.
To further examine whether SUMOylation affects RSK1 kinase activity, we performed an in vitro kinase assay with RSK1 (purified from HEK293A cells) and recombinant GST-S6 peptide (purified from E.coli) containing the phosphorylation site by RSKs. S6 peptide (KEAKEKRQE-QIAKRRRLSSLRASTSKSESSQK) [4] was cloned into pGEX-4T vector and recombinant GST-S6 peptide was purified from E.coli after isopropyl-β-D-thiogalactopyranoside (IPTG) induction. RSK1 WT or RSK1 K110/335/421R were immunoprecipitated by anti-Flag affinity beads from HEK293A cells transiently co-transfected with or without KSHV ORF45 WT , ORF45 F66A (a mutant cannot activate RSK), or empty vector control. A specific phosphorylation antibody against RxRxxS � /T � motif was used to detect the phosphorylation of GST-S6 peptide by RSK1 [19]. As shown in Fig 4C, both wild-type and SUMOylation deficient RSK1 caused similar phosphorylation levels of GST-S6 peptide substrate, and ORF45 WT but not ORF45 F66A dramatically enhanced this phosphorylation (Fig 4C), indicating that SUMOylation does not affect RSK1 kinase activity in vitro. These results demonstrated that SUMOylation modification does not affect RSK1 activation or kinase activity by KSHV ORF45.

SUMO-SIM interaction is required for eIF4B phosphorylation by RSK1
RSK1 SUMOylation mediates eIF4B phosphorylation, which is required for efficient KSHV lytic replication [20]. We next investigated how SUMOylation is required for eIF4B phosphorylation. Due to the marginal effect of RSK1 kinase activation or activity by SUMOylation, we speculated that RSK1 SUMOylation may affect the kinase-substrate interaction. Indeed, RSK1 WT was associated with eIF4B, while RSK1 K110/335/421R interacted with eIF4B with less binding affinity than RSK1 WT (Fig 6A), suggesting SUMOylation of RSK1 is required for efficient interaction between RSK1 and eIF4B. SUMOylation modification could mediate protein-protein interaction through SUMO and SIM association, which is proposed as the Lys-Xaa 3-5 -[Val/Ile]-[Ile/Leu] 2 -Xaa 3 -[Asp/Glu/Gln/Asn]-[Asp/Glu] 2 hydrophobic core motif, sometimes surrounded by Ser-Xaa-Ser motif [21]. Inspection the sequence of eIF4B revealed one potential SIM between the amino acid positions 166 and 170 ( 166 IRVDV 170 ) (Fig 6B). SIM eIF4B fusion with GST tag readily bound to endogenous SUMO1 and RSK1, while mutations on GST-SIM eIF4B abolished these interactions (Fig 6C). SIM mutation within eIF4B reduced the interaction between RSK1 and eIF4B (Fig 6D), suggesting that SUMO-SIM association contributes to RSK1-eIF4B interaction. These results demonstrated that RSK1 SUMOylation facilitates the interaction between RSK1 and eIF4B, leading to the efficient phosphorylation of eIF4B by the RSK1-ORF45 axis.
In conclusion, our results characterized the post-translational modification of RSK1 by SUMO, primarily at Lys 110 , Lys 335 , and Lys 421 , resulting in efficient phosphorylation of eIF4B, which is required for KSHV lytic replication.

Discussion
KSHV modulates multiple cellular signaling pathways to favor virus replication [22]. SUMOylation has emerged as a major posttranslational modification of cellular and viral proteins,

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication affecting a variety of cellular processes upon KSHV infection [23,24]. Multiple KSHV encoded proteins contain SIMs or are shown to be modified by SUMO to modulate the cellular SUMO network. For example, RTA contains tandem SIMs (SIM1: 273 MVDL 277 ; SIM2: 285 AVIL 288 ; and SIM3: 309 VSII 312 ) and functions as a SUMO-targeting ubiquitin E3 ligase (STUbL) that prefers to bind and degrade the targets with poly-SUMO2/3 modifications [25]. The STUbL activity is required for RTA transactivation and KSHV lytic replication [25]. K8 is primarily SUMOylated at Lys 158 and this SUMO modification is essential for the transcriptional repressive activity of K8 [26]. In addition to its role as a substrate for SUMO modification, K8 functions as a SIM-containing SUMO E3 ligase (SIM: 72 VIDL 75 ) that shows higher specificity toward SUMO2/3 than SUMO1 [27]. LANA contains two SIMs (SIM1: 244 IYVG 247 ; SIM2: 264 ISIG 267 ) within the N-terminal domain that preferentially bind to SUMO2 and mediate the recruitment of chromatin remodeling proteins and transcriptional factors, such as TRIM28 and HIF-1α [28]. SUMOylation of Lys 1140 of LANA is important for its function to maintain the KSHV episome and silence RTA activation during latency [28]. vIRF3 (also known as LANA2) has one SIM at the C-terminus and can be SUMOylated at multiple lysine residues. vIRF3 could inhibit SUMOylation of pRb, p107, and p130, leading to blockage of pRb-mediated cell growth arrest [29]. In addition, vIRF3 also disperses PML nuclear body by facilitating the SUMOylation and ubiquitination of PML in a SIM-dependent manner, which potentially contribute to KSHV-mediated cell transformation [30]. The SUMO modification or SUMO

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication binding properties of these viral proteins dynastically control the establishment of latency or switch between latency and lytic replication, contributing to the production of progeny virus and pathogenesis of KSHV. We have shown here that RSK1 SUMOylation is dramatically increased during KSHV lytic replication, which in turn facilitates KSHV lytic replication with high efficiency. RTA, LANA, or K8 cannot promote RSK1 SUMOylation, suggesting they are not the SUMO E3 ligase for RSK1 (S2 Fig). How RSK1 is SUMOylated during KSHV lytic replication is still undetermined, and our group is actively working on this topic to identify the viral or cellular factors that control this process.
Posttranslational modifications modulate protein functions. Besides phosphorylation, other posttranslational modifications, such as ubiquitination and SUMOylation, also delicately regulate the activation and activity of AGC kinases. For example, AKT undergoes K63-linked ubiquitination by TRAF6, which is essential for the membrane recruitment and phosphorylation of AKT upon growth-factor stimulation [31]. AKT is also modified by SUMO at multiple lysine residues and mutations on these SUMOylation sites abolish AKT kinase activity, leading to impaired cell proliferation and tumorigenesis [17,32,33]. SUMOylation potentially changes the subcellular localization, kinase activity, protein stability, or binding partners of the target proteins [14]. Unlike AKT, SUMO modifications at Lys 110 , Lys 335 , and Lys 421 do not alter RSK1 activation by upstream signals or kinase activity but affect the phosphorylation of its downstream substrates. Similar to phosphotyrosine-Src homology 2 (SH2) domain interaction-mediated selective protein-protein interaction (PPI) with phosphorylated tyrosine, SIM--SUMO interaction mediates selective PPI with SUMO modified proteins [14]. The SIM of eIF4B could serve as a docking site for SUMO association and mediates the interaction between eIF4B and RSK1, leading to efficient eIF4B phosphorylation by RSK1, which is required for KSHV lytic replication [20]. Although RSK isoforms (RSK1-RSK4) share 70-85% identity, only RSK1 and RSK2 undergo SUMOylation in normal condition, and SUMOylation of RSK1 is much robust than RSK2. Sequence alignment indicates that Lys 110 is conserved among all RSK isoforms, Lys 335 is only conserved between RSK1 and RSK3, and Lys 421 is unique for RSK1, suggesting a unique role of SUMO modification of RSK1. Although RSK2-RSK4 have less or no modification by SUMO in normal conditions, their SUMOylation upon specific stimulations still needs to be further explored. Our results demonstrate that RSK1 SUMOylation is required for efficient KSHV lytic replication by at least regulating the phosphorylation of eIF4B. Because of its regulatory roles of eIF4B in regulating translation of a subset of mRNAs, RSK1 SUMOylation is expected to delicately modulate the cellular translation processes under certain conditions. In KSHV-infected cells, ORF45 contributes to sustained activation of RSKs by forming high molecular mass protein complexes with RSK and ERK to prevent their dephosphorylation, leading to efficient lytic replication of KSHV. Further studies are required to explore the physiological role of RSK1 SUMOylation in the absence of viral infection.

Generation of stable cell lines
For stable expression, lentiviral plasmids harboring the desired genes were co-transfected into HEK293T cells with the packing plasmids pSPAX and pMD2G at a ratio of 5:3:2. The supernatants containing lentiviruses were collected at 48 h post-transfection and used to infect indicated cells in the presence of polybrene (8 μg/ml). Stable expression cells were selected by puromycin (2 μg/ml), hygromycin (400 μg/ml), or blasticidin (10 μg/ml) at 48 h post-infection until all the control cells were killed. For generation of knockout cells by the CRISPR/Cas9 system, single clones were isolated and assayed by western blot analysis for RSK1/RSK2 protein expression.

Immunoprecipitation and immunoblotting
For co-immunoprecipitation, 2×10 6 HEK293T cells were transfected with 20 μg of plasmid at a confluency of 90% with Lipofectamine 3000 (Thermo Fisher Scientific, #3000015). The cells were washed twice with cold phosphate-buffered saline (PBS) and lysed in a whole cell lysis buffer (WCL) containing (50 mM Tris�HCl [pH 7.4], 150 mM NaCl, 1% NP-40, 1 mM EDTA, 10% glycerol, protease inhibitor cocktail [Roche]) for 20 min on ice at 48 h post-transfection. The cell lysates were then centrifuged at 15,000 rpm for 15 min and the clear supernatants were subjected to immunoprecipitation with anti-Flag M2 agarose resin (Sigma, #F2426) following the manufacturer's instruction. After 4h incubation at 4˚C, the beads were washed for three times with WCL and twice with PBS, and then boiled with the 2 x SDS loading buffer for 10 min. The immunoprecipitants were applied to standard immunoblotting analyses with specific antibodies.
Immunoprecipitation in denaturing conditions for detecting SUMOylated protein was described previously [39]. Briefly, cells were lysed in SDS lysis buffer made by 1:3 ratio of Buffer I (5% SDS, 0.15 M Tris-HCl pH 6.8, 30% glycerol) and Buffer II (25 mM Tris-HCl pH 8.3, 50 mM NaCl, 0.5% NP-40, 0.5% deoxycolate, 0.1% SDS, 1 mM EDTA) supplemented with protease inhibitors cocktail (Roche), 1 mM DTT and 5 mM N-Ethylmaleimide (NEM, Sigma) and denatured 5 min at 95˚C. Cell lysates were then centrifuged at maximum speed for 10 min. Supernatants were either directly resolved by SDS-PAGE or diluted 1:5 in E1A buffer (50 mM Hepes pH 7.5, 250 mM NaCl, 0.1% NP-40, 1 mM EDTA, supplemented with protease inhibitors cocktail, 1 mM DTT and 5 mM NEM) and then immunoprecipitated using anti-Flag M2 agarose resin (Sigma, #F2426). After 4 h incubation at 4˚C, the beads were washed for three times with WCL and twice with PBS, and then boiled with the 2 x SDS loading buffer for 10 min. The immunoprecipitants were applied to standard immunoblotting analyses with specific antibodies.

Recombinant protein expression and purification
S6 peptide, eIF4B SIM WT peptide and eIF4B SIM Mut peptide were cloned into pGEX-4T vector with a N-terminal GST tag. The transformed E. coli were cultured in LB medium with appropriate antibiotic until OD600 reached 0.7. Protein expressions were induced 4 hours at 37˚C with 1 mM isopropyl b-D-1 thiogalactopyranoside (IPTG). Cells were washed and resuspended in PBS buffer, followed by sonication, and then the cell debris was removed by centrifugation (15,000 rpm, 4˚C, 15 min). The proteins with GST tag in the supernatant were affinity-purified by Glutathione Sepharose 4B (GE Healthcare, #17-0756-01).

RNA purification and RT-qPCR
Total RNA was extracted from cells with TRIzol reagent (Sigma) according to the manufacturer's protocol. Afterwards 0.5 μg of total RNA was reverse transcribed by HiScript II Q RT

PLOS PATHOGENS
SUMOylation of RSK1 regulates KSHV lytic replication SuperMix for qPCR (+gDNA wiper) (Vazyme Biotech, #R223-01) and the cDNA was quantified by SYBR green based qPCR using gene specific primers. The relative level of gene expression was calculated by the 2 -ΔCt and the ΔΔCt methods, where GAPDH was used for normalization. The RT-qPCR graphs represent the average of at least three independent experiments. The sequences of the primers used in RT-qPCR have been listed in S1 Table. Measurement of viral DNA copy number during KSHV lytic replication SLK-iBAC-vector, SLK-iBAC RSK1/2 DKO -vector, SLK-iBAC RSK1/2 DKO -RSK1 and SLK-iBAC RSK1/2 DKO -RSK1 K110/335/421R cells were uninduced or induced with doxycycline (2 μg/ ml) and sodium butyrate (1 mM), and lysed in RIPA buffer followed by sonication and then centrifugation to remove cell debris. Total DNA was purified from the supernatant by phenol-chloroform extraction and 10 ng of total DNA was analyzed in qPCR. The viral DNA was measured by qPCR using primers for ORF11 (Fw-GGCACCATACAGCTTCTACGA and Rev-CGTTTACTACTGCACACTGCA). The amount of viral DNA was normalized for the cellular DNA input, which was measured by qPCR specific for the β-actin genomic region (Fw-CGGGAAATCGTGCGTGACATT; Rev-CAGGAAGGAAGGCTGGAA GAGTG).

Quantification of extracellular virion genomic DNA by real-time qPCR
SLK-iBAC-vector, SLK-iBAC RSK1/2 DKO -vector, SLK-iBAC RSK1/2 DKO -RSK1 and SLK-iBAC RSK1/2 DKO -RSK1 K110/335/421R cells were uninduced or induced with doxycycline (2 μg/ ml) and sodium butyrate (1 mM), and viral DNA was isolated from supernatant medium as previously described [40]. Briefly, medium from the infected cells was centrifuged to remove any cellular debris and treated with TurboDNase (Ambion) to remove any unprotected DNA. The viral particles were lysed with buffer AL (Qiagen), and the proteins were degraded with protease K (Qiagen). The DNA was then extracted using phenol-chloroform extraction and analyzed by SYBR green real-time PCRs with KSHV-specific primers ORF11 described above. Viral DNA copy numbers were calculated with external standards of known concentrations of serially diluted BAC16 DNA ranging from 1 to 10 7 genome copies per reaction.

In Vitro kinase assay
In vitro kinase assay with GST-S6 peptide ( 218 KEAKEKRQEQIAKRRRLSSLRASTSK-SESSQK 249 ) and RSK1 was described previously [20]. Briefly, HEK293A cells were cotransfected with RSK or RSK K110/335/421R and ORF45 or ORF45 F66A as indicated. At 24 h post-transfection, the cells were serum-starved for an additional 24 h. WCL were collected and subjected to immunoprecipitation with 50 μl of anti-Flag affinity beads to purify RSK1 kinase complexes at 48 h post-transfection. After two washes with the lysis buffer and three washes with TBS buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl), the immunoprecipitated beads were resuspended in 100 μl of TBS plus 1 mM PMSF, 1 mM Na 3 VO 4 , 1×protease inhibitor mixture (Roche). The kinase reaction was performed by incubation of 5 μl of the precipitated RSK1 complexes with 2.5 μg of GST-S6 substrate in 25 μl of 1×kinase assay buffer (25 mM HEPES, pH 7.5, 50 mM NaCl, 20 mM -glycerophosphate, 1 mM DTT, 20 mM MgCl 2 ,1 mM Na 3 VO 4 , 1 μg/ml BSA, 20 μM ATP). The reactions were kept at 30˚C for 30 min and stopped by addition of 2 × SDS-loading buffer. The samples were analyzed by immunoblotting with the RxRxxS � /T � antibody to evaluate the phosphorylation level of S6 peptide.

GST pull-down assay
HEK293T cells were transfected with pKH3-SUMO1 or pKH3-SUMO2. At 48 h post-transfection, cells were harvested and lysed in a buffer containing 20 mM Tris-HCl at pH 7.5, 0.5% NP-40, 150 mM NaCl and protease inhibitors. The GST-tagged wild-type or mutant eIF4B SIM or GST control proteins were purified by Glutathione Sepharose 4B (GE Healthcare, #17-0756-01), followed by incubation of WCL from cells transiently expressing SUMO1 or SUMO2 for 2h at 4˚C. After washing three times with lysis buffer and twice with PBS, the beads were heated in 2 × SDS loading sample buffer at 95˚C for 10 minutes and the coprecipitated proteins were analyzed by immunoblotting with indicated antibodies.

Quantification and statistical analysis
All data were expressed as mean ± SD, unless otherwise noted. For parametric analysis, the F test was used to determine the equality of variances between the groups compared; statistical significance across two groups was tested by Student's t-test; one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc test were used to determine statistically significant differences between multiple groups. P-values of less than 0.05 were considered significant.