Inhibition of O-Linked N-Acetylglucosamine Transferase Reduces Replication of Herpes Simplex Virus and Human Cytomegalovirus

Cell culture and viruses.

Human foreskin fibroblasts (HFFs; ATCC catalog number CRL-1634) and Vero, HEp-2, HeLa, and HEK-293 cells were obtained from the ATCC (Manassas, VA). HFFs were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS). Vero and HEp-2 cells were cultured in DMEM with 5% FBS and 5% heat-inactivated bovine calf serum (BCS). HSV-1 KOS and HSV-2 186 strains were propagated in Vero cells, and both viruses were titrated on Vero cells. K26GFP recombinant virus (20) was a gift from Prashant Desai (Johns Hopkins School of Medicine). The human cytomegalovirus (HCMV) laboratory strain AD169 was kindly provided by Donald Coen (Harvard Medical School).

Viral infections and inhibitor treatment.

For infections, viruses were diluted in phosphate-buffered saline (PBS) containing 1% BCS and 0.1% glucose and incubated with cells for 1 h on a 37°C shaker. After viral adsorption, the cells were washed with PBS and overlaid with DMEM supplemented with 1% BCS and containing dimethyl sulfoxide (DMSO) (0.5%) or OSMI-1 at the desired concentrations. Progeny virus titers were determined by a plaque assay as described previously (21). To assess HCMV replication in the presence of OSMI-1, HFFs were infected with HCMV at the desired multiplicities of infection (MOIs). After incubation for 1 h at 37°C, the viral inoculum was removed and replaced with DMEM (10% FBS) containing DMSO or OSMI-1. The cells were incubated for a total of 96 h, and the inhibitor was replenished at 48 h postinfection (hpi). To determine viral titers, cells and supernatants were collected in an equal volume of sterile milk and stored at −80°C. Serial dilutions of each sample were titrated on HFF monolayers and overlaid with methylcellulose and DMEM at 1:1 ratio. Viral plaques were counted at 14 days postinfection. The OGT inhibitor OMSI-1 and the control compound PG34 were identified and characterized in a recent publication (19).

siRNA transfections.

HFFs were transfected with an siRNA pool targeting OGT (Santa Cruz) or with a nontarget control siRNA pool (Dharmacon) using DarmaFECT 2 transfection reagent (Dharmacon) at a final siRNA concentration of 50 nM according to the manufacturer's instructions. At 72 h after transfection, the knockdown efficiency was determined by quantitative reverse transcription-PCR (qRT-PCR) and Western blotting, and the cells were used for HSV-1 infections.

Isolation of nuclear, supernatant, and cell-associated DNA.

Nuclei were isolated from HSV-1-infected, siRNA-transfected HFFs using the NE-PER Nuclear and Cytoplasmic Extraction kit (Thermo Scientific). The nuclei were resuspended in PBS, and DNA was isolated using the DNA blood and tissue kit (Qiagen) according to the manufacturer's instructions. The purity of the nuclear and cytosolic fractions was assessed by an immunoblot detecting glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or histone H3. Relative gene copy numbers were determined by qRT-PCR.

qRT-PCR analysis.

To measure transcript levels, total RNA was harvested from cells using TRIzol reagent (Invitrogen). Samples were DNase treated with the DNA-free kit (Ambion), and 1 μg of RNA was reverse transcribed using the high-capacity cDNA reverse transcription kit (Applied Biosystems). Gene copy numbers and transcripts were quantified using the Fast SYBR green PCR master mix (Applied Biosystems) on an ABI 7900 detection system (ABI). Relative copy numbers of genes and transcripts were determined using the standard curve method. Transcript levels were normalized to 18S rRNA. Viral and cellular DNA copy numbers were normalized to levels of a GAPDH pseudogene (22).

SDS-PAGE and immunoblotting.

HFFs were infected with HSV-1 KOS at the desired MOI and treated with OSMI-1 or a DMSO vehicle control. Cell lysates were collected at the times postinfection indicated below and processed as previously described (23). The antibodies used were mouse anti-O-GlcNAc (RL2; Santa Cruz Biotechnology; 1:1,000), mouse anti-ICP27 (ab31631; Abcam; 1:1,000), rabbit anti-gC (R46; 1:1,000), rabbit anti-actin (ab8227; Abcam; 1:2,000), rabbit anti-ICP8 (3-83; 1:2,000), mouse anti-gD (ab6507; Abcam; 1:10,000), mouse anti-GAPDH (ab9484; Abcam; 1:4,000), and rabbit anti-histone H3 (ab1791; Abcam, 1:3,000). Rabbit and mouse horseradish peroxidase (HRP)-conjugated antibodies (Santa Cruz Biotechnology) were used for secondary detection at 1:5,000.

Immunofluorescence.

HFFs grown on glass coverslips were mock or HSV-1 infected and overlaid with medium containing DMSO or 50 μM OSMI-1-containing medium. At the desired times postinfection, cells were fixed with 2% formaldehyde and processed as described previously (24). The coverslips were incubated with mouse anti-VP5 antibody (HA018; EastCoast Bio; 1:500) followed by a secondary anti-mouse Alexa Fluor 488 antibody (1:1,000; Jackson ImmunoResearch). Cells were imaged using a Nikon TE2000 w/C1 point scanning confocal microscope at a ×60 magnification.

Electron microscopy.

HFFs were grown to confluent monolayers in wells of a 12-well plate, infected with HSV-1 KOS at an MOI of 0.1, and maintained in the presence of DMSO or 50 μM OSMI-1. At 18 hpi, the infected cells were fixed with 2.5% paraformaldehyde, 5% glutaraldehyde, and 0.1 M cacodylate buffer (pH 7.2) for 1 h at room temperature. Fixed cells were embedded in resin and sectioned for imaging. For imaging of supernatant-purified virions, the samples were diluted in PBS and adsorbed onto a hydrophilic carbon-coated grid, followed by a negative staining with 0.75% uranyl acetate. Sections were examined by Tecnai G² Spirit Bio Twin electron microscope.

Inhibition of OGT activity in HFFs.

The small molecule inhibitor of OGT, OSMI-1 (Fig. 1A), was previously developed and optimized from a high-throughput screen hit (19). The compound was evaluated for OGT inhibition in an in vitro assay and in several mammalian cell lines (19). In this study, we first evaluated whether OSMI-1 can inhibit OGT activity in human foreskin fibroblasts (HFFs) by examining the change in global O-GlcNAc modification in the presence of the compound. Treatment of HFFs with increasing concentrations of OSMI-1 for 24 h led to a dose-dependent decrease in global O-GlcNAc levels, as measured by immunoblotting (Fig. 1B), without significant toxic effects on the cells (Fig. 1C).

Effect of OGT inhibition on HSV-1 yields.

To test whether OGT inhibition affects HSV-1 replication, we infected HFFs with HSV-1 strain KOS at a multiplicity of infection (MOI) of 0.1 PFU per cell, and immediately following viral adsorption, we treated the cells with either a vehicle control (DMSO) or increasing concentrations of OSMI-1. At 48 hpi, we determined the production of progeny virus by a plaque assay on Vero cells. OGT inhibition by OSMI-1 reduced viral yields over 1,000-fold at the highest concentration of inhibitor (50 μM) (Fig. 2A). We further tested whether the OSMI-1 effect was MOI dependent by infecting HFFs with HSV-1 KOS virus at low or high MOIs and treating them with OSMI-1 at a 50 μM concentration, which previously resulted in the highest yield reduction. We observed a decrease in viral yield in OSMI-1-treated HFFs at both low (950-fold) and high (890-fold) MOIs (Fig. 2B), arguing that OSMI-1 efficiently inhibits a single cycle of replication as well as multiple cycles. OGT inhibition also caused a dose-dependent reduction in HSV-1 yields in HeLa (Fig. 2C), HEp-2 carcinoma cells (Fig. 2D), and HEK-293 cells (Fig. 2E), demonstrating that the effect of the inhibitor was independent of the cell type. A cell viability assay after OSMI-1 treatment showed that the inhibitor did not significantly affect cell viability, as 91% (HeLa), 89% (HEp-2), and 92% (HEK-293) of the cells remained viable after 24 h of treatment (results not shown). We found the concentration of OMSI-1 that resulted in a 50% decrease in HSV-1 yield (EC₅₀) in HFFs to be 6.34 μM (Fig. 2F).

To determine whether the antiviral effect of OSMI-1 was due to OGT inhibition and not due to off-target effects, we tested a structural analog of OSMI-1 in an HSV-1 inhibition assay. The analog PG-34 is structurally similar to OSMI-1, but it contains a phenylalanine in place of the 2-methoxyphenylglycine of OSMI-1 (19). PG34 has similar cell permeability but demonstrates poor in vitro inhibitory activity against OGT and does not reduce global O-GlcNAcylation in cells (19). HSV-1 viral yields remained relatively unaffected by treatment with a 50 μM concentration of the control compound (Fig. 2G). These results were consistent with the interpretation that OSMI-1's effects on viral yields are due to inhibition of OGT.

Effect of OGT inhibition on replication of HSV-2 wild-type and acyclovir-resistant viruses.

To determine if OSMI-1 is effective at inhibition of HSV-2, we infected HFFs with HSV-2 186 virus at 0.1 or 5 PFU/cell and treated the cells with 50 μM OSMI-1 for the duration of the infection. At 24 hpi, the amount of infectious virus was determined by plaque assay. OSMI-1 decreased viral titers by 1,700-fold at the low MOI, whereas there was a lower (60-fold) yield reduction at the high MOI (Fig. 3A). We also tested an HSV-2 mutant lacking thymidine kinase (TK⁻) that is resistant to acyclovir (25). We infected HFFs with wild-type (WT) HSV-2 186 or the TK⁻ mutant virus at 0.1 PFU/cell and measured viral yields. The TK⁻ virus was resistant to acyclovir treatment, while OSMI-1 reduced replication of both WT HSV-2 and the TK⁻ mutant (Fig. 3B). In total, the results argued that OGT is necessary for efficient replication of HSV.

Effect of OGT inhibition on replication of HCMV.

In addition, we tested the effect of OSMI-1 on replication of human cytomegalovirus (HCMV), another human herpesvirus. We infected HFFs with the HCMV lab-adapted AD169 virus strain at MOIs of 0.1 and 1, and we treated infected cells with OSMI-1 or DMSO. Viral yields measured at 96 h after infection showed that the inhibitor reduced HCMV yields in a dose-dependent manner at both low (Fig. 4A) and high (Fig. 4B) MOIs. Treatment with OSMI-1 did not significantly affect cell viability, as over 89% of the cells remained viable during the duration of the treatment (results not shown). These results argue that OGT is involved in the replication cycle of other herpesviruses.

Effect of OGT inhibition on HSV-1 gene expression.

To determine the stage in the HSV-1 replication cycle that is targeted by the compound, we measured the levels of IE, E, and L transcripts in the presence of OSMI-1. We infected HFFs with HSV-1 at an MOI of 5 and measured IE ICP27, E ICP8, and L gC gene transcripts by qRT-PCR at 3, 6, and 9 hpi. Levels of transcripts for all gene classes were not affected by OSMI-1 over the course of 9 h, as shown by equal levels of IE ICP27 (Fig. 5A), E ICP8 (Fig. 5B), and L gC (Fig. 5C) transcripts. Treatment of the cells with OSMI-1 for 16 h prior to HSV-1 infection also had no effect on the expression of viral genes (results not shown). Similarly, the levels of the IE ICP27 and E ICP8 proteins did not decrease throughout the course of infection (Fig. 5D), and a slight increase in protein levels was even observed at 6 and 9 hpi. However, the levels of the L gC protein were markedly reduced in the presence of the inhibitor. To confirm that this decrease was not specific for gC, we measured the expression of two other late proteins, glycoprotein D (gD) and the major capsid protein VP5, and we saw a similar decrease in both proteins at 6 and 9 hpi (Fig. 5E). The decreases in gD and VP5 levels were not observed at the mRNA level (results not shown), arguing that the inhibitor either interferes with the synthesis of late viral proteins or increases their turnover. To confirm that OSMI-1 acts at late times of the viral replication cycle, we performed a time-of-addition assay (Fig. 5F). We added OSMI-1 to HSV-1-infected HFFs at different times after infection and analyzed the viral yields at 24 hpi. We observed that antiviral activity of OSMI-1 was preserved when OSMI-1 was added to infected cells up to 4 h after infection but declined thereafter. Together, these results argued that OSMI-1 acts late in the viral cycle.

Effect of OGT inhibition on viral capsid formation.

Late viral genes encode structural proteins that are mainly involved in the formation of progeny virions (1). Of those, VP5 is the major constituent of the viral capsid and, together with VP26, VP19C, and VP23, forms the immature capsid shell (26). Because we saw lower VP5 protein levels in the presence of the inhibitor, we speculated that capsid assembly and subsequent virion maturation would be negatively affected by OSMI-1. We first examined the effect of OSMI-1 on capsid assembly by evaluating the localization of VP5 and the minor capsid protein VP26 by immunofluorescence. We infected HFFs with the HSV-1 K26GFP recombinant virus that expresses the VP26 small capsid protein fused to green fluorescent protein (GFP), followed by treatment with DMSO control or 50 μM OSMI-1. At 8 hpi, we examined the localization of GFP-VP26 and VP5 by immunofluorescence. Control-treated cells showed GFP-VP26 and VP5 colocalization in a punctate pattern within the cell nuclei (Fig. 6A). In the presence of the inhibitor, we observed fewer and less distinct GFP puncta, and while VP5 appeared to be concentrated in the nucleus, it did not form distinct foci that colocalized with GFP-VP26. Moreover, in many infected cells, we found more cytoplasmic VP5 staining. To confirm that capsid assembly was impaired in OSMI-1-treated cells, we examined infected cells by transmission electron microscopy (TEM). The vehicle control-treated cell nuclei contained a mixed population of capsids in different stages of assembly: empty A capsids, scaffold-containing B capsids, and DNA-filled, mature C capsids (Fig. 6B, panel a). In contrast, the nuclei of OSMI-1-treated cells (Fig. 6B, panel b) contained fewer capsids than those of the control-treated cells. Moreover, particles budding out of inhibitor-treated infected cells lacked the characteristic virion shape (Fig. 6B, panels c and d). EM images of virions isolated from the medium of OSMI-1-treated cells revealed an abnormal shape compared to control cells (Fig. 6, panel e). To quantify the effect of OSMI-1 on capsid formation, we determined the number and the types of viral capsids in the DMSO- and OMSI-1-treated cells (Fig. 7C). In the presence of the inhibitor, we observed an approximately 10-fold decrease in the number of A, B, and C capsids. Therefore, we concluded that inhibition of OGT by OMSI-1 disrupts the assembly of viral capsids and the production of mature viral particles.

Depletion of OGT reduces viral replication and gene expression.

To validate the OSMI-1 phenotype using an siRNA approach, we transfected HFFs with a pool of siRNAs targeting OGT (siOGT) or with a control nonsilencing RNA pool (siControl). At 72 h after siRNA transfection, the levels of OGT mRNA and protein in the siOGT-transfected samples were notably reduced (Fig. 7A). To assess the effect of OGT knockdown on viral yields, we treated cells with siRNAs, infected them with HSV-1 at an MOI of 0.1, and measured viral yields by a plaque assay at 24 hpi. We observed a 10-fold decrease in the production of infectious virus in the OGT knockdown cells (Fig. 7B) that was not due to nonspecific effects such as cell toxicity due to depletion of OGT (Fig. 7C). When we assessed viral mRNA, we observed a 10-fold downregulation in IE (ICP4, ICP27), early (ICP8), and late (gC) transcripts (Fig. 7D); therefore, the reduction of viral yield correlated with the reduction of viral transcripts. Similarly, viral IE (ICP4), E (ICP8), and L (gC) proteins were decreased in the OGT-depleted cells (Fig. 7E). This phenotype was specific for viral genes because OGT depletion appeared to have no effect on the levels of cellular TATA binding protein (TBP) or 18S rRNA transcripts (results not shown). O-GlcNAc modification is the most common posttranslational modification of nuclear pore complex (NPC) proteins (27). Depletion of OGT has been shown to reduce the expression of several members of the NPC (27), and inhibition of O-GlcNAc modification can inhibit protein transport into the nucleus (28). Because the NPC facilitates attachment of viral capsids and delivery of viral DNA into the nucleus (29), we tested whether the OGT-mediated decrease in viral gene expression was due to fewer viral genomes entering the nuclei of OGT-depleted cells. We measured the amount of viral DNA in the nuclei of OGT-depleted and control cells by qRT-PCR and observed equal numbers of viral genomes (Fig. 7F). The absence of GAPDH protein in the Western blot analysis confirmed the purity of the nuclear fractions (results not shown). Thus, the decrease in viral yield and viral gene expression upon OGT knockdown, together with the OSMI-1 phenotype, confirmed that OGT plays a role in the replication cycle of HSV.

Role of OGT as a scaffold.

The catalytic site of OGT resides in its C terminus, while the N terminus contains multiple tetratricopeptide (TRP) repeats, which form an elongated, flexible scaffold that provides a large electrostatic surface for multiple protein-protein interactions (41). The TRP scaffold allows for protein binding in many different binding orientations and modes, thus explaining the absence of a conserved sequence for OGT binding (42). One study shows that a catalytically inactive mutant OGT appears to be unaffected in substrate binding (43), arguing that OGT can serve as a scaffolding platform for the assembly of multiprotein complexes independent of its role as a glycosyltransferase. Moreover, OGT remains associated with its substrates even after glycosylation (41), further suggesting that the function of OGT is not solely modification of proteins. Thus, OGT whose catalytic function is inhibited by OSMI-1 can still retain its ability to interact with proteins and mediate formation of protein complexes. Depletion of OGT with an siRNA pool resulted in a 10-fold reduction in viral yields, which is a more modest phenotype than the 1,000-fold reduction we observed with the inhibitor. Even though the control compound PG34 confirmed OSMI-1's specificity, the reduced effect of OGT knockdown compared to OSMI-1 treatment may be due to off-target effects by the inhibitor. This smaller magnitude of inhibition may result from incomplete depletion of OGT by the siRNA, as we see that 15 to 20% of OGT transcripts remain after knockdown, which could provide enough enzymatic activity for viral replication. When we examined viral gene expression, we noted that unlike OSMI-1, which did not affect the levels of viral transcripts, OGT knockdown decreased expression of viral IE, E, and L genes by ∼10-fold. This difference in phenotypes between the OGT inhibitor and the knockdown most likely reflects the multiple functions of OGT in the cell. We speculate that OGT may serve as a scaffold to facilitate the binding of HCF-1 and other transcription factors to the IE multiprotein complex, and its depletion disrupts this interaction and negatively affects viral gene expression. Moreover, unpublished results from our lab show that OGT is a part of the HCF-1–VP16–OCT transactivation complex (44).

The role of OGT in replication of other viruses.

Several studies implicate O-GlcNAcylation as an important modification that can regulate the replication cycle of viruses. O-GlcNAc modification has been found on many viral proteins, including HCMV UL32 tegument protein (45), adenovirus fiber protein (46), baculovirus tegument protein gp41 (47), rotavirus NS26 protein (48), and several KSHV proteins involved in DNA replication (34, 49). A few studies suggest that high O-GlcNAc levels have an inhibitory effect on viral replication. Increased O-GlcNAcylation of the cellular protein Sp1, which is a critical transcription factor for HIV-1 gene expression, was shown to inhibit transcription from the HIV-1 long terminal repeat (LTR) promoter (35). Also, increasing O-GlcNAcylation by overexpression of OGT reduced KSHV replication (49). In another study, increased O-GlcNAcylation of the KSHV major replication and transcriptional activator (RTA) reduced its ability to transactivate viral genes (34). Consistent with this study, we observed that OGT inhibition did not prevent reactivation of latent KSHV, as measured by the levels of lytic gene expression (results not shown). Unlike the negative effect of O-GlcNAc modification for KSHV and HIV-1, we demonstrate that OGT is needed for the efficient replication of HSV and HCMV, as the OGT inhibitor decreased replication over 1,000-fold.

The results from this study demonstrate that OGT is involved in multiple steps of the HSV replication cycle. The OGT depletion results argue that OGT is needed for efficient formation of the viral IE transactivation complex, while the enzymatic activity of OGT, as demonstrated by the OSMI-1 studies, may be required late in infection, for translation of late viral genes and capsid assembly. In summary, the results described here implicate OGT as a novel factor involved in HSV-1 replication, and further studies are needed to define its precise role in the viral cycle.