Novel functions of Kaposi’s sarcoma herpesvirus viral FLICE inhibitory protein

KSHV viral FLICE inhibitory protein (vFLIP) is a potent activator of NF-κB signaling and an inhibitor of apoptosis and autophagy. Inhibition of vFLIP function and NF-κB signaling promotes lytic reactivation. Here we provide evidence for a novel function of vFLIP in promoting NFκB signaling through inhibition of the DUB activity of the negative regulator, A20. We demonstrate interaction of vFLIP with the Itch/A20 ubiquitin editing complex. We have identified a SUMO interaction motif in vFLIP that is required for NF-κB activation. Mutation of the SIM in BAC16 resulted in increased spontaneous RTA expression and loss of spindle cell morphology. Our results suggest a role for SUMO in mediating vFLIP function and provide evidence for vFLIP modulation of the negative regulation of NF-κB signaling by A20. Our results provide further insight into the function of vFLIP and SUMO in the regulation of NF-κB signaling and the latent lytic transition.

Introduction editing protein with both C-terminal ubiquitin ligase activity and N-terminal deubiquitinase activity.
In one well characterized mechanism, A20 forms a ubiquitin editing complex with Itch, RNF11, and TAX1BP1, and downmodulates NF-κB signaling through removal of K63-linked polyubiquitin chains from RIPK1 followed by addition of K48-linked polyubiquitin chains, resulting in degradation of RIPK1 via the proteasome. A20 is reported to deubiquitinate a number of signaling intermediates within the NF-κB pathway in addition to RIPK1, including IKK , TRAF6, TRAF2 and MALT1 [20][21][22].
Here we report two observations suggesting novel functions of vFLIP. 1. vFLIP inhibition of A20 activity and 2. vFLIP has a SUMO interaction motif (SIM) that functions in maintaining latency.
We previously reported that RTA induces the degradation of vFLIP early in lytic reactivation resulting in the termination of NF-κB signaling, presumably to promote transition from latency to lytic replication [23]. RTA induced degradation of vFLIP is dependent on the activity of the Itch ubiquitin ligase [24]. We identified mutants of vFLIP that are unable to interact with Itch and cannot activate NF-κB [24]. Here we report that vFLIP interacts with the A20/Itch ubiquitin editing complex and this interaction occurs independently of RTA. We propose that vFLIP inhibits A20 DUB activity to modulate NF-κB signaling through interference with negative regulation. We demonstrate reduced A20 activity and increased levels of RIPK1K ubiquitin conjugates following stimulation with TNFA in the presence of vFLIP.
Small Ubiquitin-like Modifier (SUMO) proteins, when covalently conjugated to substrate proteins, can modulate the stability, interaction and activity of proteins. We have identified a SUMO interacting motif (SIM) in vFLIP. vFLIP SIM mutants exhibited reduced SUMO1 and SUMO 2/3 binding and were unable to activate NF-κB signaling. Small molecule inhibition of SUMO conjugation resulted in increased virus production, suggesting a role for SUMO the latent to lytic transition. We generated a BAC16 mutant where the V22 in the vFLIP SIM is changed to E. iSLK cells containing the vFLIP SIM mutant exhibited higher levels of spontaneous reactivation and lost their spindle shaped morphology. Taken together, our findings suggest a novel role for vFLIP in activation of NF-κB signaling via inhibition of the DUB activity of A20 and that a SIM in vFLIP plays a role in maintaining latency. Future studies will need to determine whether the SIM in vFLIP is required for interaction with A20 and whether this interaction is required for maintaining latency.
These observations expand our understanding of vFLIP function and have the potential to increase our understanding of the mechanisms governing the maintenance of latency.

vFLIP interacts with Itch and A20
We previously reported that RTA induces the degradation of vFLIP via the cellular ubiquitin ligase, Itch. We hypothesized that RTA was recruiting Itch to vFLIP to promote ubiquitination and degradation of the viral protein. Upon further characterization of the interactions between vFLIP and Itch as part of the Itch/A20 ubiquitin editing complex, we observed interaction between vFLIP, Itch and A20, even in the absence of RTA (Fig 1a-b). Based on this data we hypothesized that vFLIP interacts with the Itch/A20 ubiquitin editing complex in latency and RTA expression, occurring early in lytic reactivation, results in activation of Itch and/or A20 ubiquitin ligase activity (either directly or indirectly), resulting in the subsequent ubiquitination and degradation of vFLIP.
If this model were correct, we would expect to see degradation of additional targets of Itch and/or A20 in the presence of RTA. In fact, we detected a modest decrease in the levels of A20, a known target of Itch, in cells transfected with RTA compared to empty vector control (Fig. 2a). We evaluated additional Itch and/or A20 substrates, RIPK1 and c-Jun, by immunoblot and observed a modest decrease in protein levels compared to the internal tubulin control (Fig. 2b).
To further examine the effect of RTA on the ubiquitination of Itch and A20 substrates, we conducted an analysis of the ubiquitinated proteome in SILAC labeled RTA transfected 293T cells using the anti-diglycine remnant (K-ε-GG) antibody. We detected two known Itch or A20 substrates that exhibited a significant increase in ubiquitination in the presence of RTA. We observed a 1.5-fold increase in the ubiquitination of BRAT1, a known Itch substrate, and a nearly 3-fold increase in ubiquitinated UBE2N, a known A20 substrate, in RTA transfected cells compared to empty vector transfected controls (Fig. 2c). Taken together, this data suggests that while RTA may play a role in the activation of the Itch ubiquitin ligase or Itch/A20 ubiquitin editing complex, the exact mechanism remains unclear.
vFLIP inhibits the debiquitinase activity of A20 A20 is a well characterized negative regulator of NF-κB signaling. Following stimulation of NF-κB via the TNF receptor (TNFR), A20 downregulates signaling by removal of K63 linked polyubiquitin chains from RIPK1 and in concert with Itch, adds K48 linked polyubiquitin, resulting in RIPK1 degradation via the proteasome. It was previously reported that vFLIP induces the expression of A20. It has been proposed that A20 expression, in the context of latent KSHV infection, is necessary to limit the inflammatory phenotype induced by persistent NF-κB signaling. We hypothesized that A20 activity needs to be tightly regulated, as excessive activity has the potential interfere with latency and cell survival, and inhibition of NF-κB signaling has been shown to promote apoptosis and lytic reactivation. To this end, we assessed the impact of vFLIP on A20 DUB activity. Using purified K63-linked tetraubiquitin, A20 and vFLIP, we evaluated A20 DUB activity via in vitro assay. Addition of purified A20 alone to tetraubiquitin, resulted in cleavage of tetraubiquitin to faster migrating mono and polyubiquitin species (Ub-3, Ub-2, Ub-1), however addition of recombinant vFLIP resulted in a dose dependent decrease in DUB activity (Fig. 3a).
A well characterized target of A20 DUB activity is RIPK1, following TNFR stimulation. Within  minutes of TNFA treatment, followed by deubiquitination 2h post treatment (Fig. 3b). Addition of vFLIP, however, resulted in detection of sustained RIPK1 ubiquitin conjugates, which was accentuated in cells transfected with K63-only ubiquitin (Fig. 3b). These data, taken together, suggest that vFLIP has an inhibitory effect on A20 DUB activity.

vFLIP has a SUMO interaction motif (SIM)
PML nuclear bodies have been implicated in antiviral defense, apoptosis, and the DNA damage response as well as other cellular processes. They are known to be highly modified by the small ubiquitin-like modifier SUMO. In our initial vFLIP studies, we observed vFLIP colocalization with PML via immunofluorescence assay, prompting the question as to whether vFLIP has a SIM (unpublished observation). Analysis of the vFLIP protein sequence resulted in the identification of a putative tandem SIM spanning amino acids V21-VLFLLN-V28 based on the following consensus motif (V/L/I, X, V/L/I, V/L/I) or (V/L/I, V/L/I, X, V/L/I) (Fig. 4a). Indeed, wild-type vFLIP was able to bind recombinant SUMO 1 and SUMO 2/3, albeit to a lesser extent, while mutant vFLIP V22E, was no longer able to bind to either SUMO 1 or 2/3 (Fig. 4b).
We previously reported that this vFLIP mutant was resistant to RTA induced degradation and was unable to activate NF-κB, suggesting that this SIM is important for interaction with components of the NF-κB signaling pathway and RTA [23,24]. vFLIP V22E is also unable to interact with the Itch ubiquitin ligase [24]. To determine whether Itch and A20 are SUMOylated, we immunoprecipitated FLAG tagged Itch or A20 and probed the 500mM NaCl washed immunoprecipitates with antibody against endogenous SUMO 1 or SUMO 2/3. We observed corresponding SUMO 1 and 2/3 bands in the Itch immunoprecipitates (Fig. 4c). Itch is not reported to be SUMOylated in the literature or in the Phosphosite plus database suggesting that either the modification is highly transient or that Itch is interacting with a SUMOylated protein. Taken together, these results support a model where vFLIP may interact with the Itch/A20 ubiquitin editing complex either directly or indirectly via a SUMO dependent mechanism.
Mutation of the SIM in BAC16 results in loss of spindle cell morphology and increased RTA expression.
Our observations suggest that SUMOylation plays a role in regulation of NF-κB signaling induced by vFLIP. Previous reports demonstrated the importance of vFLIP induced NF-κB signaling for maintaining viral latency and cell survival, as inhibition with Bay 11-7082 and transfection with an IKBA dominant negative mutant resulted in increased apoptosis and lytic reactivation [25] [26]. We hypothesized that if SUMOylation was required for maintaining latency, we would see a dose dependent increase in lytic reactivation upon SUMOylation inhibition and in fact we did observe this in Vero rKSHV.294 cells treated with the small molecule SUMOylation inhibitor 2-D08 [27](manuscript in preparation).
To further explore the function of the vFLIP SIM, we constructed a V22E vFLIP mutant in BAC16 and generated iSLK BAC16 vFLIPV22E cells. The BAC16 vFLIPV22E mutant was sequenced and analyzed by digest with Sbf1 followed by pulsed field gel electrophoresis (PFGE). The BAC16 vFLIP V22E sequence was consistent with the BAC16 consensus except for the mutation we introduced into the vFLIP coding sequence (Fig. 5a). Through PFGE analysis, we observed similar banding patterns between wild type and mutant BAC16 except for a shift in the bands above 48kb suggesting a possible alteration in the terminal repeat length (Fig. 5b). We hypothesized that if the SIM in vFLIP is required for vFLIP function we should see an increase in spontaneous lytic reactivation. To evaluate lytic reactivation, we analyzed RTA expression in iSLK cells containing wild-type or mutant BAC16. As expected, doxycycline treatment resulted in a 9fold increase in RTA expression in iSLK cells containing wild-type BAC16, however cells containing BAC16 vFLIPV22E displayed a 12-fold increase in RTA expression in the absence of doxycycline treatment, suggesting an increase in spontaneous lytic reactivation (Fig. 5c).
Doxycycline induction of RTA expression was 2-fold higher in cells containing the SIM mutant compared to wild-type BAC16 (Fig5c, compare WT +DOX to VE +DOX). We observed a corresponding increase in RTA protein levels (Fig. 5d). Interestingly, we observed a striking difference in cell morphology when comparing iSLK cells containing wild-type and SIM mutant BAC16. Cells containing wild-type BAC16 had the characteristic spindle shape morphology associated with vFLIP induced activation of NF-ΚB, whereas in cells containing the SIM mutant virus, vFLIP had more of a cobblestone appearance, similar to iSLK cells lacking BAC16 (Fig 5e).
Taken together these data support a functional role for the SIM in vFLIP, as virus lacking an intact SIM exhibited higher levels of spontaneous lytic reactivation and infected cells lost the characteristic vFLIP induced NF-κB associated spindle cell morphology.

Discussion
We have presented multiple observations supporting a novel mechanism by which vFLIP promotes NF-κB signaling and latency. vFLIP is an established activator of NF-κB signaling and this activity is associated with viral latency. However, activation of NF-κB results in the expression of several negative regulators of the signaling pathway. Expression of one such negative regulator, A20, was shown to be induced by vFLIP. While NF-κB signaling is important for maintaining latency, prolonged NF-κB activation could contribute to an inflammatory phenotype.
In fact, this is what occurs when negative regulators of NF-κB are either naturally or experimentally defective. Deficiencies in Itch ubiquitin ligase expression or function are associated with immune deficiencies and the Itch -/-knock out mouse displays an "itchy" phenotype for which this gene is named. A20 -/-mice also display a phenotype associated with inflammation and autoimmunity, exhibiting hypersensitivity to TNF and premature death. To establish and maintain a latent infection, vFLIP must activate NF-κB and signaling must be sustained without killing the host and to accomplish this, the virus must control negative regulators of NF-κB.
We observed interaction of vFLIP with the Itch and A20 ubiquitin editing complex. We previously reported that in the presence of RTA, Itch targets vFLIP for degradation. These observations suggest that vFLIP may be interacting with the Itch/A20 ubiquitin editing complex in latency and reactivation along with expression of RTA may be modulating the activity of this complex. We evaluated multiple known substrates of Itch via western blot and proteomic analysis and observed modest decreases in protein levels when RTA was expressed suggesting that RTA is altering Itch substrate stability. Our proteomics data revealed identification of 146 proteins with RTA dependent alterations in ubiquitination, however only two were known Itch or A20 substrates, suggesting that while RTA has a demonstrated effect on the cellular ubiquitome, the mechanism(s) governing this observation remains unclear. We reasoned that vFLIP interaction with the Itch A20 ubiquitin editing complex may function to promote NF-κB signaling, and expression of RTA abrogates signaling by inducing the degradation of vFLIP as well as other members of the complex. vFLIP had no effect on Itch/A20 complex assembly, suggesting that vFLIP was not inhibiting protein complex formation as had been described with HTLV Tax [28]. We observed, through in vitro assay and through immunoprecipitation of RIPK1 conjugates, inhibition of A20 DUB activity by vFLIP. Detection of sustained K63 ubiquitinated RIP1 in the presence of vFLIP, suggests that A20 DUB activity is limited thereby allowing for sustained NF-ΚB signaling.
We identified a SIM in vFLIP and observed interaction with SUMO-1 and 2/3. We previously reported that this motif is required for activation of NF-ΚB, degradation of vFLIP by RTA and interaction with Itch. Taken together these data suggest that vFLIP interacts with the Itch/A20 complex via a SUMO dependent mechanism. We observed evidence of Itch SUMOylation and inhibition of global SUMOylation resulted in a dose dependent increase in infectious virus production.

Reagents, Plasmids, and Antibodies
The proteasome inhibitor MG132 (Boston Biochem) was used in this study.

Immunoblot Analysis
Proteins were run on 12% Tris-Glycine or Any kD mini-PROTEAN Precast Gel (Biorad) with Tris-glycine running buffer. The proteins were then transferred to a PVDF membrane using semi-dry transfer system at 20V for 20 minutes. The membranes were blocked in 5% non-fat dry milk in PBS for one hour. Primary antibodies were diluted in with 2.5% non-fat dry milk at 1uL antibody: 1000uL milk and applied to the membranes. The membranes were incubated on a shaker at 4°C overnight and were washed in PBS with 0.1% Tween the following day. Secondary antibodies were applied to the membranes in 2.5% non-fat dry milk at 1uL antibody: 1000uL milk.
The membranes were incubated at room temperature on a shaker for one hour and afterward were washed with PBS and 0.1% Tween. Proteins were visualized with the addition of ECL substrate and the detection of the luminescence on x-ray film or scanned by a Li-COR C-DiGit Blot Scanner.

Immunoprecipitation
Transfected cells with appropriate constructs were harvested 48h post-transfection with PBS and centrifuged at 1500 rpm for 10 min. The PBS was removed and 1mL of lysis buffer with 10µl of a protease inhibitor cocktail kit (Thermo Scientific) were added to each cell pellet. When appropriate 12.5µL of 5 mM NEM was added to each cell pellet. Cell lysates were centrifuged at 10,000 rpm for 5 minutes to remove cell debris. The resulting supernatant was precleared with protein A/G PLUS-agarose (Santa Cruz) for 30 min at 4°C. The lysates were transferred to a new 1.5mL tube and protein concentrations were measured and normalized with a Pierce BCA protein assay kit. Approximately 50 µg of protein was transferred to new 1.5mL tube to serve as control lysate. 1μg of the appropriate primary antibody was added to the remaining cell lysate and incubated on a rotator overnight at 4°C. 25µL of protein A/G-agarose were added the following day for 1hr and washed 4x with RIPA lysis buffer. 50µl of 2X Laemmli Buffer were added and samples were boiled at 100°C for 10 min. Samples were visualized through immunoblot analysis as described above  In vitro deubiquitination assay V5-His tagged vFLIP was expressed in E. coli (BL21) and purified using Ni-NTA resin (ThermoFisher). A20-Flag was purified as previously described. Purified tetra-K63 ubiquitin was purchased from Boston Biochem. The following reagents were added to 20μl reactions where indicated: A20 (2μM), vFLIP (1μM, 5μM, 10μM), tetra-K63 Ub (500nM). Reactions were incubated at 37ºC for 2hrs followed by the addition of 4x Laemmli loading buffer. Reactions were analyzed by SDS PAGE followed by immunoblot.

Construction of BAC16 vFLIP V22E
Mutant BAC16 was constructed using the protocol of Tischer et al using BAC16 in GS1783 generously provided by Jae Jung [29][30][31]Briefly, using PCR (primers in Primers Table), a Kanamycin resistance gene containing an I-SceI restriction enzyme site flanked by regions from the KSHV genome containing the mutation of interest, was amplified. The kanamycin cassette was amplified from pEPkan-S and integrated into BAC16 using Red recombination. An overnight culture of GS1783, containing BAC16 and Red recombinase, was used to inoculate a fresh culture and grown to an OD of 0.5. The culture was transferred to a 42°C shaking water bath for 10 min.
to induce Red recombinase expression followed by 10-20min incubation on ice. Bacteria were pelleted and washed 3x in ice cold water in preparation for electroporation. 100ng of kanamycin cassette were added to 50ul of cells in residual water and electroporated using the following parameters: 1.5kV, 25uF, 200 ohms, time constant=3.5-4.5 ms. 250ul SOC media was added to bacteria and incubated at 30°C for 1h for recovery. Cells were plated on LB agar containing kanamycin (20ug/ml) and incubated at 30°C for 24h. Kanamycin positive colonies were verified for insert by colony PCR using kanamycin specific primers. Following confirmation of kanamycin cassette, cells were cultured overnight at 30°C in LB containing kanamycin. Overnight cultures were used to inoculate LB containing chloramphenicol 15ug/ml (chlor resistance cassette is present in BAC16). To remove the kanamycin cassette, leaving the mutation behind, I-SceI expression was induced by adding arabinose to a final concentration of 1%. Cultured were incubated, shaking for 45 min at 30°C. To complete the removal of the kanamycin cassette, Red recombinase expression was again induced via incubation at 42°C for 10 minutes, followed by incubation at 30°C for 2h. 10x serial dilutions were plated on LB plates containing 15ug/ml chloramphenicol and 1% arabinose. Colonies were replicate plated on kanamycin to identify colonies that no longer contained the kanamycin resistance cassette. Chlor positive/kan negative clones were sequenced using vFLIP specific primers. Clones were further analyzed by sequencing (Illumina for full BAC16 and Sanger for repeats) and pulsed field gel electrophoresis. Quality trimming, filtering, and read mapping. Reads were trimmed by low-quality base pairs (quality limit = 0.05), including 20 nucleotides at the 5′ terminal and 10 nucleotides at the 3' terminal. All reads shorter than 50 nucleotides were filtered out. High -quality, trimmed reads were mapped to the KSHV genome (GenBank accession number NC_009333) using the map-toreference tool on CLC Genomics Workbench v20.0.3 (Qiagen) with default parameters. A length fraction of 0.5 and a similarity fraction of 0.8 were selected and non-specific reads were ignored for better mapping accuracy.

PCR amplification of KSHV gap regions and Consensus sequence generation
Primer pairs tagged with the universal M13 primers were used to amplify the gap regions of the KSHV genome which are not typically covered by NGS technology [32]. PCR conditions used were previously described [32]. PCR products were Sanger Sequenced and mapped to the reference genome (NC_009333) along with the original NGS reads using the parameters RTA and RPS13 (housekeeping gene) were amplified using the primers listed in the primers table.
Relative gene expression was calculated using the ΔCT method using the RPS13 housekeeping gene for normalization. Error bars represent the standard deviation. Statistical analysis was conducted using GraphPad Prism 9 using an unpaired t test comparing WT-DOX to VE -DOX.