Toll-like receptor signaling controls reactivation of KSHV from latency

Results

We first screened latently infected PEL for TLR gene expression. BCBL-1 and BC-3 cells were harvested and expression of TLRs 1 through 10 were analyzed by qPCR with specific TLR primers. TLR profiling revealed that these cells express multiple TLRs, including TLR7 and TLR8 (Fig. 1A).

To determine whether TLR activation could induce reactivation of KSHV, we stimulated BCBL-1 and BC-3 cells with individual cognate TLR ligands for 24 h, as well as a positive control, TPA (Fig. 1B). Intracellular DNA viral load was determined by real-time PCR and normalized for input using GAPDH copy number. Fold activation for each sample was determined relative to the vehicle control using the ΔΔCT method. TLR2 stimulation with zymosan or TLR4 stimulation with LPS derived from Escherichia coli strain 0111:B4 showed no significant difference in lytic gene expression. Treatment of BCBL-1 cells with FSL-1 (TLR2/6 ligand) also had no significant effect on replication compared with vehicle control and neither did TLR9 stimulation with CpG DNA. We observed a decrease in viral DNA copy number in response to 1 μg/mL recombinant flagellin (TLR5 ligand) or 10 μg/mL polyI:C (TLR3 ligand). This was in contrast to stimulation with single-stranded RNA, a molecular pattern associated with viral infection that is recognized by TLR7 and TLR8 (9). ssPoly-U is a synthetic homolog of single-stranded RNA that specifically activates TLR8 (10). Stimulation with 10 μg/mL ssPoly-U resulted in a 4-fold increase in viral genome copy number compared to control (Fig. 1B). Moreover, ssPoly-U reactivated KSHV to a greater extent than the positive control, TPA. In summary, ssPoly-U was the only TLR agonist that up-regulated KSHV lytic gene expression, suggesting TLR8 activation reactivates KSHV from latency.

Since KSHV does not replicate to high titers in PEL, quantification of DNA replication is not very sensitive. We therefore tested TLR8-mediated reactivation of KSHV transcription in BCBL-1 cells. BCBL-1 cells were treated with ssPoly-U for various timepoints. ORF57 and ORF49 mRNA levels were subsequently analyzed by qPCR as a marker of viral reactivation. ssPoly-U treatment resulted in >10-fold reactivation compared to the vehicle control. In response to ssPoly-U, the levels of KSHV replication peaked at 48 h and declined after 72 h (Fig. 1C). For this experiment, we used a combination of both TPA and sodium butyrate (NaB) as a more potent positive control for KSHV reactivation, due to the variability in responses of individual PEL lines to TPA alone. This combination gives considerably higher reactivation levels than either agent alone. To establish that TLR8-mediated reactivation was common to PEL in general, we screened additional PEL cells using ssPoly-U. BCBL-1, BC-3, BCP-1, or VG-1 cells were treated with ssPoly-U for 48 h. This resulted in lytic reaction of KSHV in all samples (Fig. 1D). This demonstrates that the activation of specific TLR signaling pathways can lead to KSHV reactivation.

We next queried the specificity of TLR7 and TLR8 for mediating KSHV lytic replication since both receptors recognize single-stranded RNA. Both TLR7 and TLR8 reside within the endosome, have similar structures, and are often functionally redundant (10, 11). We used several validated TLR7 and TLR8 ligands to test for KSHV reactivation in BCBL-1 cells. CL-075 (3M-002) predominantly activates TLR8, and TLR7 to a lesser degree (10). Imiquimod has been shown to be specific for TLR7 (12). We stimulated BCBL-1 cells with increasing concentrations of CL-075, Imiquimod, and CpG DNA (TLR9 ligand) and measured lytic gene transcription. Increasing concentrations of CL-075 reactivated KSHV approximately 2.5- and 4.5-fold over mock (Fig. 2A). Interestingly, the TLR7 ligand, Imiquimod, reactivated KSHV to similar levels as CL-075. In contrast, increasing concentrations of TLR9 agonist (CpG DNA) had no effect. In summary, stimulation of TLR7 and TLR8 led to reactivation of latent KSHV. We used the TLR4 agonist, LPS, as an additional negative control. We treated BCBL-1, BC-3, BCP-1 and VG-1 PEL cells with 10 and 50 μg/mL LPS derived from E. coli strain 0111:B4. Increasing concentrations of LPS did not result in KSHV reactivation from latency (Fig. 2B), whereas the positive control (TPA and NaB) did reactivate the virus. Another TLR4 agonist, LPS from E. coli strain K12, also failed to induce KSHV lytic replication.

TLRs 7 and 8 are known to signal through the NF-κB and IRF7 pathways, leading to the up-regulation of pro-inflammatory cytokines and type I IFN, respectively. First, we looked at NF-κB activation following ssPoly-U stimulation by probing for IκB-α, an inhibitor of NF-κB nuclear translocation and transactivation. Upon activation of TLR7/8, IκB-α is ubiquitinated and targeted for proteasome-mediated degradation. Treatment of BCBL-1 cells with ssPoly-U for 5 h induced IκB-α degradation (Fig. 3A), indicating that the canonical NF-κB pathway was activated by TLR7/8 stimulation in PEL. Next, we tested whether the NF-κB promoter was activated in response to ssPoly-U stimulation. BCBL-1 cells were nucleofected with 200 ng NF-κB-luciferase plasmid. Cells were pooled after 48 h, and either treated with ssPoly-U or vehicle control. Sixteen hours post-treatment, cells were harvested and analyzed for luciferase activity. ssPoly-U treatment induced an approximate 7-fold activation of the NF-κB reporter compared to vehicle treated cells (Fig. 3B). Therefore, ssPoly-U potently activates NF-κB in BCBL-1 cells. ssPoly-U stimulation also induced the up-regulation of IFN-α, IFN-β and IFN-γ, as well as NF-κB transcription in PEL (Fig. 3C). Induction of NF-κB and IFN-α, β, and γ transcription serve as validated markers for the activation of TLR signaling in many different cell types (4). Moreover, activation of IFN-α/β induces IRFs (including IRF-7) in a positive-feedback loop necessary for amplification of the initial innate immune response. Therefore, we analyzed IRF-7 protein expression 24 h after ssPoly-U stimulation of BCBL-1 cells and observed IRF-7 up-regulation as expected (Fig. 3D). To determine whether IRF-7 was critical for TLR7/8 induced lytic replication, we transfected an IRF-7 dominant-negative (IRF-7DN) construct (13) into BCBL-1 cells (Fig. 3E, Left). Forty-eight hours later, the cells were split into 3 samples and treated with equivalent amounts of either ssPoly-U, vehicle, or TPA. Twenty-four hours post-treatment, cells were harvested and lysates were analyzed for vIL-6 as a marker for lytic replication by western blot. We found that IRF-7DN expression resulted in a marked reduction of KSHV lytic reactivation compared to vector control (Fig. 3E, Right). Taken together, our data demonstrate that the TLR7/8 pathway is activated in response to ssPoly-U stimulation of PEL and imply that activation of this pathway can lead to KSHV reactivation from latency in B lymphocytes. IRF-7 is a critical component of this signaling pathway, whose activation by TLR7/8 is required for viral reactivation.

RTA/ORF50 is the key lytic switch protein of KSHV. It is necessary and sufficient to activate KSHV replication (14). We therefore queried whether TLR7/8 signaling in PEL activated the KSHV RTA/ORF50 promoter. BCBL-1 cells were nucleofected with 5 μg ORF50 promoter-luciferase reporter plasmid, and 48 h post-transfection, cells were pooled and treated with ssPoly-U RNA or vehicle. After 16 h, luciferase activity was quantified as previously described (7). ssPoly-U treatment resulted in a 7-fold increase in luciferase activity compared with vehicle control (Fig. 4A). Hence, ssPoly-U-induced TLR7/8 signaling can activate the RTA promoter.

Next, we measured KSHV viral DNA in the ssPoly-U treated BCBL-1 cells (Fig. 4B), and were able to detect a 4-fold increase in viral DNA levels. To corroborate these results, KSHV early lytic protein (ORF59) expression was analyzed by immunofluoresence assays (IFA). KSHV ORF59 was detected in ssPoly-U treated BCBL-1 cells as well as the positive control, but not in mock-treated cells (Fig. S1). Similar results were obtained when we analyzed the expression of KSHV vIL-6 in ssPoly-U-treated cells by western blot analysis (Fig. 4C). In summary, our data demonstrate that ssPoly-U stimulates the KSHV RTA promoter and induces KSHV early lytic gene transcription, replication, and the expression of early lytic proteins, vIL-6 and ORF59.

Although the above data suggested that KSHV lytic gene transcription and protein expression was activated by ssPoly-U treatment of PEL, it remained possible that lytic replication was incomplete. Therefore, we profiled genome-wide KSHV transcription using our KSHV qPCR array (15, 16). We treated BCBL-1 and BC-3 cells with either ssPoly-U or TPA (positive control) for 48 h and analyzed KSHV transcription. Two independent cell lines, BCBL-1 and BC-3, showed up-regulation of KSHV viral gene expression 48 h after ssPoly-U treatment (Fig. 5A). Their expression profile resembled that of the positive control, TPA. As a negative control, we treated BCBL-1 and BC-3 cells with 50 μg/mL LPS for 48 h, but did not observe any KSHV reactivation (Fig. 5A). LPS treated BCBL-1 cells displayed the same transcription profile as the vehicle control-treated cells.

We also performed a time course experiment. BCBL-1 and BC-3 cells treated for 0 h, 48 h, and 72 h with ssPoly-U or TPA displayed lytic replication at the 48 h and 72 h time points, with peak activation at 48 h (Fig. 5B). These results indicate that complete KSHV lytic gene transcription occurs in response to TLR7/8 stimulation of PEL.

To further verify that infectious virus was produced by ssPoly-U mediated reactivation of PEL, we treated BCBL-1 and BC-3 cells for 48 h with 50 μg/mL ssPoly-U. Supernatants were subsequently harvested and used to infect uninfected Vero cells. Ninety-six hours postinfection, viral DNA was isolated and KSHV viral loads were analyzed using qPCR with ORF57 primers. As shown in Fig. 5C, treatment of BCBL-1 and BC-3 cells resulted in approximate 5- and 20-fold increases in production of infectious KSHV virions, respectively. Of note, the endogenous levels of TLR8 in these 2 cell lines may correlate with differences in lytic replication between PEL cells. BCBL-1 cells have reduced levels of TLR8 expression compared to BC-3 (Fig. 1A), which may explain observed differences in infectious virus produced upon ssPoly-U stimulation of different PEL.

To demonstrate that TLR7/8 stimulation was necessary for ssPoly-U induced KSHV lytic replication, we used TLR7- and TLR8-specific shRNA vectors to create stable BCBL-1 cells containing knock-down of either TLR7 or TLR8. shRNA directed against luciferase was used as a control. We achieved moderate knock-down of TLR7 and complete loss of TLR8 in our knock-down stable cells (Fig. 6A).

BCBL-1 cells knocked-down for TLR7 or for TLR8 were used in ssPoly-U reactivation assays. Viral reactivation was measured by western blot analysis for the lytic protein, vIL-6. The control cell line, psiLuciferase-BCBL-1, showed considerable reactivation after treatment with ssPoly-U (Fig. 6B). Treatment of psiTLR7-BCBL-1 with ssPoly-U resulted in slightly reduced vIL-6 expression, compared to the psiLuciferase-BCBL-1 treated cells, demonstrating a potential role for TLR7 in KSHV reactivation. However, ssPoly-U treatment of the psiTLR8-BCBL-1 cells displayed substantially reduced vIL-6 expression compared to the psiLuciferase-BCBL-1 treated cells. Importantly, all of the knock-down cell lines remained similarly responsive to lytic reactivation by TPA, and showed considerable reactivation (Fig. 6B). In summary, while TLR7 may play a role in KSHV reactivation, TLR8 activation is more important for viral reactivation since the TLR8 knock-down cells showed a significant reduction in ssPoly-U-stimulated viral reactivation.

To translate our results to a physiologically relevant setting, we used the single-stranded RNA virus, vesicular stomatitis virus (VSV), for infection of PEL. VSV is endocytosed in a clathrin-dependent manner (17, 18) and is known to activate TLR7/8 in mouse and human cells (19). BCBL-1 cells were infected with VSV at an MOI of 1, or treated with TPA as a positive control. KSHV vIL-6 protein expression was determined 24 h and 48 h postinfection. Infection with VSV resulted in KSHV reactivation at both time points (Fig. 6C). Similar results were seen with VSV infection at 0.1 MOI. VSV-induced reactivation was also accompanied by increased NF-κB, IFN-α, β, and γ transcription (Fig. S2) and cell death was visually observed. Thus, single-stranded RNA, specifically viral RNA, is capable of inducing KSHV reactivation.

Discussion

In summary, we report evidence for KSHV reactivation through TLR signaling. TLR7/8 mediated reactivation from latency offers a paradigm for how KSHV may initiate lytic replication in vivo. Periodic reactivation of herpesviruses is important for the establishment of new latent reservoirs of virus, viral transmission, and persistence in the host. Since human herpesviruses establish life-long latency in their hosts, it is very plausible that a latently infected cell is exposed to infection by a secondary pathogen, be it a virus, bacterium, or fungus. We report that secondary viral infection that is dependent on endosomal entry and stimulation of TLR7/8 signaling can lead to reactivation of latent KSHV from B lymphocytes. It is well-established that VSV activates TLR7/8. Although VSV is a rhabdovirus and a natural pathogen of livestock, human infections have been documented (20, 21), and VSV can infect human B lymphocytes (22). Other RNA viruses that activate TLR7/8, such as dengue virus, also infect B lymphocytes (23). Of note, it was previously reported that infection of PEL with VSV-pseudotyped HIV particles reactivated KSHV (24), although the exact mechanism was not deduced. Since HIV RNA is known to activate TLR7 and TLR8 (9, 25), the observed outcome in this earlier study may be attributable to HIV RNA-induced activation of TLR7/8 in PEL.

In immunosuppressed KSHV-positive individuals, KSHV reactivation rates may be increased by secondary infections that trigger TLR7/8 activation. Since lytic replication of KSHV is linked to an increased risk of KSHV-associated cancers (26), one can speculate that KSHV-positive individuals exposed to TLR7/8-activating pathogens are at a higher risk for developing KSHV-associated malignancies. Endogenous RNAs from necrotic cells also activates TLR7/8 (27). Hence, it is equally plausible that KSHV reactivation is triggered by endosomal trafficking of endogenous RNA from necrotic cells. A recent publication indicates that MHV-68 can also be reactivated in vivo by TLR signaling (28), although it was shown that multiple TLR ligands could reactivate MHV-68 from B lymphocytes. This may reflect a biological difference between the two viruses, or the difference in TLR signaling between normal and malignant B lymphocytes.

Lastly, our data have implications for oncolytic therapy of viral cancers. There is a growing interest in using VSV oncolysis to treat tumors (29, 30). Our data suggests that VSV infection of KSHV-infected lymphomas results in oncolysis, but also reactivates KSHV. This may be detrimental to KSHV-positive cancer patients since VSV oncolysis would inadvertently aid viral reactivation and dissemination. Thus, for some viral cancers it may be advisable to use VSV oncolytic treatment in combination with viral replication inhibitors, such as ganciclovir, to prevent reactivation and spread of the virus to naïve cells in the host.

Defining the physiological stimuli of reactivation from latency is a fundamental question in herpesvirus biology. We report evidence of innate immune receptors mediating KSHV reactivation. Moreover, TLR7/8 signaling mediated by secondary infection of KSHV-infected B lymphocytes not only induces KSHV reactivation and lytic replication, but also allows the virus to propagate itself and escape a cell that is fated to die. This ensures KSHV survival and persistence for the lifetime of the infected host.

Materials and Methods

Supplementary Material