K-RTA is a Ser and Thr-rich protein (122 out of 691 amino acids, 17.7%) but the sites and roles of post-translational modifications (PTMs) on these residues are largely unexplored. In this study, by using mass spectrometric analysis we demonstrated that K-RTA is an O-GlcNAcylated protein and that Thr-366/Thr-367 is an O-GlcNAcylation motif. Thr to Ala mutations at this motif increased K-RTA's transactivation capability in luciferase reporter assays and KSHV reactivation in 293/rKSHV.219 cells, indicating that O-GlcNAcylation may impose a suppressive effect on K-RTA. This notion was further supported by increased K-bZIP expression and viral particle production in KSHV infected cells depleted with O-GlcNAc transferase (OGT). The inhibitory effect of O-GlcNAcylation on K-RTA was attributed to an increasing affinity between glycosylated K-RTA and PARP1. Noteworthy, PARylation of K-RTA by PARP1 was previously shown to negatively modulate the activity of K-RTA . Thereby, our results established a link between these two PTMs in regulating K-RTA. Furthermore, given that O-GlcNAcylation is a dynamic process keenly responding to glucose fluctuation, we speculate that the activity of K-RTA is closely controlled by the metabolic state of the host cell. For example, Delgado et al. recently demonstrated that induction of Warburg effect in KSHV-latently infected endothelial cells is a required process for tumor cell survival . As increased glucose uptake will produce more UDP-GlcNAc, in theory Warburg effect would be coupled with elevated O-GlcNAcylation. Thus, Warburg effect may create an environment suppressive to K-RTA's full functionality that leads to a crippled KSHV reactivation. This could explain why most KSHV remain latent in the KS biopsies.
In Figure 5, the M2-FLAG resin-precipitated T366A/T367A mutant is still reactive to α-O-GlcNAc antibody (RL2), indicating that Thr-366/Thr-367 is not the only site modified by O-GlcNAc. This partial de-O-GlcNAcylation in Thr-366 and Thr-367 mutants may explain why the enhancement effects in our functional assays were small, although statistically significant (Figure 3, 4). We reasoned that those unidentified O-GlcNAcylation sites most likely are located at larger tryptic peptides that have been excluded in our mass spectrometry (e.g., amino acids 531-633, approximately 10.6 kDa, contain 25 Ser/Thr). Alternatively, the fragile O-GlcNAc moieties might have been lost during protein purification. Thus, the use of other proteases such as chymotrypsin to supplement the trypsin digestion, or employing more advanced methods including "QUICK-Tag" and electron transfer dissociation mass spectrometry [37, 38] should disclose additional O-GlcNAcylation sites in K-RTA and provide a more comprehensive biological relevance.
Suppression of transactivation activity by O- GlcNAcylation is best known in Sp1. An O-GlcNAcylated activation domain of Sp1 repelled its hydrophobic interaction with TAF110 in vitro and removal of the O-GlcNAc moiety in Sp1 elicited a signal for activation [29, 39]. Hyper-O- GlcNAcylation of Sp1 severely impaired the transactivation of p21/WAF1 promoter in HeLa cells , and, the LTR promoter activity of HIV-1 in T cells . Other than Sp1, O-GlcNAcylation of C/EBPβ, YY1 and NF-κB p65 also brought suppressive transcriptional effects on their target genes [30, 41, 42]. Here, we showed that forced de-O-GlcNAcylation of Thr-366 and Thr-367 in K-RTA resulted in moderate enhancement in activity (Figure 3, 4). This motif is located in the middle of the 691 amino acids consisting of K-RTA, a region previously mapped to interact with positive regulator RBPJκ . Thus, it is likely that unmodified Thr-366/Thr-367 in this region may provide a better interaction domain for RBPJκ or factors in the transcription machinery, as has been described in the case of Sp1. In addition, we found that O-GlcNAcylated K-RTA "attracts" more PARP1, another kind of PTM enzyme that transfers large and negatively charged polymer (PAR) onto numerous nuclear factors. Frequently, the attachment of PAR led to altered activity and function of target proteins through both steric and charge inhibition . In agreement, a previous report demonstrated that PARP1 PARylated K-RTA in vitro and suppressed its activity in a co-expression assay . Combined, we speculate that O-GlcNAcylated K-RTA repels more of its positive regulators, associates with more PARP1 thus being PARylated, and ultimately its full functionality is restricted. It would be of great interest to investigate whether O-GlcNAcylation and PARylation exert synergistic or feedback effects in modulating K-RTA.
O-GlcNAcylation-mediated transcriptional suppression could also take place at the chromatin level. First discovered by Yang et al., OGT was targeted by mSin3A and participated in gene silencing . Drosophila OGT was found to be encoded by Polycomb group gene, super sex combs (sxc) , and directly participated in epigenetic gene silencing on polytene chromosomes [14, 15]. In mammals, murine OGT stability was regulated by polycomb repressive complex 2 and the cellular O-GlcNAc level was crucial to the transcription repertoire in embryonic stem cells . Together, these findings provide a new perspective that OGT could be an authentic chromatin remodeller. Indeed, O-GlcNAcylation of H2B facilitates its subsequent monoubiquitination  and in vivo O-GlcNAcylation sites of other histone members have been revealed . Intriguingly, to add one more layer of complexity, the role of PARP1 in chromatin structure modification is also increasingly evident . Thereby, it is tempting to propose that modification of K-RTA by OGT and interaction of K-RTA with PARP1 may be only the secondary events. Targeting by K-RTA, OGT and PARP1 may actively modify the viral genome structures at epigenetic level before decorate and oppress K-RTA function. With this regard, future experiments are required to delineate the spatiotemporal occupancies of OGT, PARP1 and K-RTA on the viral genome and their relevance to K-RTA mediated latent-lytic switch.
We began our study by computational analysis of K-RTA primary amino acid sequences, realized that 11 of the 20 potential O-GlcNAcylation sites could also be O-phosphorylated (Figure 1C). Thr-367 is one of them. Interestingly, although our mass spectrometry did not resolve whether Thr-366 or Thr-367 is morel likely to be the real acceptor for O-GlcNAc, in Figure 5, substitution of Thr to Ala at Thr-367 seemed to cause a more distinct phenotype than substitution introduced at Thr-366. Thus, these hypothetical sites could serve as a blueprint for further studies in understanding alternate modifications of O-GlcNAcylation and O- phosphorylation in K-RTA.