The SARS coronavirus papain like protease can inhibit IRF3 at a post activation step that requires deubiquitination activity

Firefly luciferase plasmids containing the IFN-β or NF-κB promoter and the GFP- and HA-tagged SARS-CoV PLpro expression plasmids were described previously [16]. The SARS-CoV PLpro mutant used contains a double mutation in the active site (C1651A and D1826A) as described previously [17]. Flag-tagged IRF3(5D) was a gift from John Hiscott (described in[34]). Ha-tagged Ubiquitin was previously described [16]. HEK293T cells were purchased from ATCC (Catalog #CRL-3216) (Manassas, VA), grown in DMEM (Invitrogen, Carlsbad, CA) with 10% FBS and 1% penicillin/streptomycin.

To analyze the induction of IFNβ induced genes, a luciferase reporter assay was used in HEK293T cells. Briefly, an expression construct containing the luciferase ORF and the IFNβ promoter (IFNβ/luciferase) was co-transfected with either a GFP control plasmid or the designated PLpro plasmid. Transfections of reporter plasmids into HEK293T cells were performed with the Lipofectamine LTX (Invitrogen) transfection reagent as directed by the manufacturer. For all transfections, 10 ng of Renilla luciferase, 200 ng of luciferase plasmid, 200 ng of viral expression plasmid, 200 ng of inducer plasmid (total 600 ng/well) was used in each well of a 48 well plate with 1ul of Lipofectamine LTX. For control wells that contained only the inducer and luciferase plasmid and did not contain 600 ng plasmids in total, the remaining DNA was filled with an empty GFP plasmid so we could assess transfection efficiency and so that all transfections contained the same amount of plasmid DNA. All transfections were performed in triplicate and the average of 3 experiments is shown in figures. At 18 hours post transfection cells were lysed and assayed for luciferase expression using the Dual Luciferase Reporter assay (Promega Inc). The ratio of experimental treatment to control inducer after normalization to Renilla luciferase was graphed in each figure.

Plasmids vectors containing HA and FLAG tags ligated to the C or N termini of the indicated ORFs were transfected into HEK293T cells as described below. After 18 hours of expression, cells were treated with lysis buffer (20 mM Tris/HCl pH 7.5, 150 mM NaCl, 1% NP-40), the extract was centrifuged for 10 minutes at 4°C, and the supernatant was removed. 25ul of washed EZ View Red Anti-FLAG M2 Affinity Gel beads (Sigma, St Louis MO, #F2426) were added to each extract and rotated overnight at 4°C. Extract was then washed 3 times with lysis buffer and re-suspended in SDS PAGE loading buffer before boiling and electrophoresis.

For ubiquitination or deubiquitination experiments, the protocol above was followed with the addition of HA-tagged ubiquitin (HA-Ub) or mutant ubiquitin plasmids. Mutant ubiquitin plasmids that are only able to be ubiquitinated at either K48 or K63 were added to the transfection experiments as described above. The mutant and wildtype ubiquitin plasmids were previously described [35].

Expression plasmids were assayed for protein expression by Western blotting. Lysates were then run on SDS-PAGE gels (NuPage, Invitrogen) and blotted to PVDF membrane (Invitrogen). Proteins were visualized using anti-GFP antibody (G1544, Sigma Aldrich), anti-HA (Sigma H3663), anti-Flag (Sigma F7425), HRP-conjugated secondary anti-rabbit antibody (NA934, GE Life Sciences) and HRP-conjugated secondary anti-rabbit antibody (NA931, GE Life Sciences).

IRF3 dimer gels were performed as described in Iwamura et al. [36] with the following alteractions. All buffers were made 24 hours in advance and chilled overnight at 4°C. Tris/Glycine running buffer was supplemented with 0.3 g Sodium Deoxycholate. Native gels (Biorad Ready gel) was pre-run for 30 minutes at 150 V before loading samples. Native PAGE gels were run at 10 V for 2.5 hours before soaking gel in running buffer for 30 minutes and transferring to PVDF for 90 minutes. An IRF3 and IRF3(5D) sampled was boiled in SDS/PAGE loading dye for 5 minutes before loading as a monomer control. Proteins were visualized with anti-Flag antibody (Sigma F7425) and anti-HA antibody (Sigma H3663).

Binding assays were performed with ten micrograms of whole cell extracts incubated in a total volume of 15 μL with buffer containing 50 mM HEPES (pH 7.9), 10% glycerol, 200 mM KCl, 5 mM EDTA (pH 8), 1 mM MgCl₂, 5 mM DTT, and 1 μg of poly (deoxyinosine-deoxycytidylic) acid sodium salt (Sigma-Aldrich) to eliminate non-specific binding. Samples were incubated on ice for 10 min, followed by the addition of 150,000 CPU of ³²P-labeled DNA probe and 20 min of incubation at room temperature. Oligonucleotide probes corresponding to the ISREs of ISG15 and OAS were annealed to their complementary oligonucleoides using annealing buffer containing 100 mM NaCl and 50 mM HEPES (pH 7.6). Forward sequences for probes used:Oas1b ISRE: TTCCCGGGAAATGGAAACTGAAAGTCCCAT,ISG15 ISRE: GATCGGAAAGGGAAACCGAAACTGAAGCC. T4 PNK (New England Biolabs) was used to end-label annealed probes with γ32ATP. Samples were electrophoresed at 180 Volts in 0.5% Trisborate-EDTA buffer on a 5% native polyacrylamide gel composed of 49:1 acrylamide to bis-acrylamide. Gels were dried on Whatman paper at 80°C for 1 h and exposed by autoradiogram.

Chromatin immunoprecipitation (ChIP) analysis was performed by using the Chromatin Immunoprecipitation (ChIP) assay kit (Millipore, 17–295) and previously described [25]. Briefly, HEK293T cells were transfected with 200 ng of each plasmid expressing either HA tagged PLpro, Flag tagged IRF3 or Flag tagged IRF3(5D) singly or in combination. At 24 hours post transfection, cells were fixed, pelleted and immediately frozen at −80°C. For immunoprecipitation the pellets were lysed and sonicated, with sonication conditions chosen to produce the desired size distribution of chromatin, between 300 and 1,200 bp. Lysates were immunoprecipitated with anti-IRF3 (Active Motif, 39033), anti-Flag (Sigma, F7425), anti-acetyl-Histone H3 (Millipore, 06–599) as a positive control, and anti-IgG (Jackson Labs, 315-005-003) as a negative control. To affirm the presence or absence of specific IRF3-binding to the IFNβ promoter following ChIP, PCR was performed. Response-specific IFNβ promoter regions were amplified by using the following primers: IFN-f 5’- GAATCCACGGATACAGAACCT-3’, IFN-r 5’-TTGACAACACGAACAGTGTCG-3’. The amplification of GAPDH served as an input control (forward 5’-CATGGGGAAGTTGAAGGTCG-3’, reverse 5’-TTGATGGTACATGACAAGGTGC-3’). PCR products were run on a 1.5% agarose gel for visualization.

GFP tagged PLpro, Flag tagged IRF3 and IRF3(5D) and GFP plasmids were transfected into HeLa cells as described above and cells fixed with 4% PFA. Fixed cells were incubated with mouse anti-Flag M2 antibody (Sigma-Aldrich, St. Louis MO). Cover-slips were incubated with secondary antibodies; Alexa Fluor 546 conjugated goat anti-mouse (Invitrogen, Carlsbad, CA). Fluorescence imaging was performed using a Zeiss Axioskop Microscope. The IRF3 and IRF3(5D) localization assays were performed in triplicate. Several fields of view were imaged for each transfection experiment and representative images consistent across each experiment displaying the localization of IRF3 and IRF3(5D).

SARS-CoV PLpro has been shown to inhibit IRF3 phosphorylation, nuclear translocation and subsequent IFNβ gene induction [16],[23]. We sought to identify whether PLpro acts at additional downstream targets to inhibit IFNβ induction. To bypass the IRF3 phosphorylation step in the signaling pathway, we used a constitutively active phosphomimetic mutant of IRF3, called IRF3(5D). IRF3(5D) has 5 of the C-terminal serine residues mutated to aspartic acid (D), mimicking the C-terminal phosphorylated serine of active IRF3, bypassing the need for its phosphorylation [34]. We hypothesized that, as PLpro inhibits IRF3 phosphorylation, IRF3(5D) would not have any effect on the downstream activation of STING/IKK/IRF3 and allow us to analyze the effects of PLpro on IRF3 signaling after its phosphorylation step.

To investigate this hypothesis, HEK293T cells were transfected with the IFNβ/luciferase plasmid alone, or in combination with a plasmid encoding IRF3(5D) (a potent inducer of the IFNβ promoter) and either an empty expression plasmid or a plasmid expressing a HA-tagged PLpro of SARS-CoV (Figure 1). We find that IRF3(5D) induces ~60-fold higher levels of luciferase compared to mock transfected cells and an empty GFP plasmid has no effect on luciferase expression, as expected. However, expression of HA-tagged PLpro caused a strong inhibition of IRF3(5D) dependent luciferase production. These data demonstrates that PLpro is able to inhibit IRF3 induced IFNβ gene induction at a post-phosphorylation step.

Upon phosphorylation, IRF3 homodimerizes before entering the nucleus to induce IFN production. We hypothesized that SARS-CoV PLpro may be blocking IRF3(5D) dimerization to inhibit IFNβ induction. To this end, HEK293T cells were transfected with Flag-tagged IRF3 with and without SARS-CoV PLpro, and then infected with Sendai Virus (SeV), a potent inducer of IRF3 activation and IFNβ induction, for six hours before all cells were lysed and dimer formation assessed by SDS-PAGE (Figure 2A) and native PAGE gel electrophoresis (Figure 2B). As expected, in mock-infected cells, IRF3 is found as a dimer, which is not changed during SeV infection (Figure 2B) [37]. This is expected due to the previous finding that overexpression of IRF3 alone can induce IFNβ gene induction. However, when HA-tagged PLpro is co-transfected with IRF3, dimer formation is inhibited (Figure 2B). This confirms the previous finding that PLpro inhibits IRF3 phosphorylation, and thus its dimerization [23].

We performed the same assay on IRF3(5D)-transfected cells, with and without HA-tagged PLpro. We performed native PAGE on transfected HEK293T cells with a plasmid expressing Flag-tagged IRF3(5D) alone or in combination with HA-tagged PLpro (Figure 2C). We found that IRF3(5D) readily dimerizes when over-expressed alone, and PLpro has no effect on IRF3(5D) dimerization. These data demonstrate that the inhibition of IRF3(5D)-dependent IFNβ gene induction by PLpro is not due to PLpro blocking IRF3(5D) dimerization.

The next step after dimerization in IRF3 signaling is nuclear import. We, therefore, hypothesized that PLpro is blocking nuclear import of IRF3(5D). To test this, we transfected Vero E6 cells with either Flag-tagged IRF3 and IRF3(5D) with or without HA-tagged PLpro. As a positive control, Flag-tagged IRF3 was transfected alone and at 24 hours post transfection cells, were infected with SeV for six hours. As expected, we found that prior to SeV infection, IRF3 is localized primarily in the cytoplasm, and migrates to the nucleus upon SeV infection (Figure 3A). When IRF3 and PLpro were co-transfected in the presence of SeV, we found that IRF3 is only in the cytoplasm. This demonstrates that PLpro most likely blocks nuclear import of activated IRF3 (Figure 3A).

IRF3(5D) localized to both the nucleus and the cytoplasm in transfected cells (Figure 3B), as previously reported [34], reflecting the continuous cytoplasmic translation and nuclear import IRF3(5D). When PLpro is over-expressed, IRF3(5D) is found in both the nucleus and the cytoplasm, identical to what is found when IRF3(5D) is expressed alone (Figure 3B). Therefore, we do not see a block of nuclear import of IRF3(5D) when PLpro is expressed. These data demonstrate that PLpro is not blocking IRF3(5D) nuclear import to inhibit IFNβ gene induction.

After activated IRF3 dimers are imported into the nucleus, they bind to specific response elements in the promoter region of IRF3-induced genes. Since PLpro does not block IRF3(5D) nuclear import, we hypothesized that it may block the ability of IRF3(5D) to bind to DNA promoters. We hypothesized that PLpro may be inhibiting the ability of IRF3 to bind to the DNA directly. Because PLpro is not found as a soluble protein in infected cells, rather it is a domain of the larger NSP3 protein, we would not hypothesize that PLpro would be binding DNA natively. However, it is used as a comparator for IRF3 binding to the promoter regions. In these experiments, we assayed the ability of PLpro to affect IRF3 binding to DNA using two assays. First, we tested for the ability of IRF3 to bind to the IFNβ promoter by performing chromatin immuno-precipitation (ChIP) and second, we assayed for the ability of co-transfected PLpro to inhibit IRF3 binding to other IFN-inducible promoters by gel-mobility shift assay.

First, we performed ChIP of IRF3(5D) to identify whether PLpro was altering the ability of IRF3(5D) to bind the IFNβ promoter. HEK293T cells were transfected with Flag-tagged IRF3 or Flag-tagged IRF3(5D) with or without HA-tagged PLpro. At 24 hours post transfection, cells were treated with paraformaldehyde (PFA) to cross-link protein bound to DNA, lysed and sonicated to shear protein bound chromatin for ChIP experiments. Immunoprecipitation was performed against either IRF3, Flag or IgG (negative control). The pre-ChIP input was scored for GAPDH to control for the amount of template in the reactions (Figure 4A). After immunoprecipitating the samples, the crosslink was reversed and used for PCR with IFNβ promoter primers spanning the IRF3 binding site (Figure 4B). Negative controls of either untransfected protein or HA-tagged PLpro transfected alone, showed no amplification of the IFNβ promoter in immunoprecipitated samples, demonstrating no background binding of PLpro to the IFNβ promoter (Figure 4B). As expected, IFNβ promoter binding is observed in cells transfected with either IRF3 or IRF3(5D) (Figure 4B, at arrow). To assess the effect of PLpro on IRF3 binding activity, HA-tagged PLpro was co-transfected with either Flag-tagged IRF3 or Flag-tagged IRF3(5D). We found strong amplification of the IFNβ promoter in each case (Figure 4C), suggesting that PLpro had no effect on IRF3 or IRF3(5D) binding to the IFNβ promoter. These data demonstrate that the inhibition of IRF3(5D) by PLpro is not due to inhibition of IRF3(5D) binding to the IFNβ promoter.

Electro-mobility Shift Assays (EMSA) were used to test whether IRF3(5D) from cells containing both IRF3(5D) and PLpro was able to bind to the ISG15 or OAS1b promoter, both of which are bound by IRF3 during infection. In this assay, HEK293T cells were transfected with either empty plasmid, Flag-tagged IRF3(5D), HA-tagged PLpro or Flag tagged IRF3(5D) and HA-tagged PLpro together. Protein lysates were incubated with radiolabeled oligonucleotide probes that correspond to the ISG15 and OAS1 promoters. The negative control lysate from HA tagged PLpro transfected cells did not bind to either probe. However, supershifts, designating a protein-DNA complex, were visible for both ISG15 and OAS1 promoter elements in lysates containing IRF3(5D) (Figure 4D). Interestingly, lysate from cells containing both HA-tagged PLpro and Flag-tagged IRF3(5D) still bound to the promoter elements of both genes (Figure 4D). These data suggest that PLpro does not affect the ability of activated IRF3 to bind its DNA targets.

IRF3 has not been conclusively shown to be ubiquitinated. Other proteins in the IRF3 signaling pathway, such as RIGI-I [38], are ubiquitinated and can effect IRF3 activation. We examined whether ubiquitination of IRF3 is necessary for IRF3- dependent induction of IFNβ. We then sought to identify whether the ubiquitination of IRF3(5D) was affected by PLpro. PLpro contains de-ubiquitination (DUB) enzymatic activity and we tested whether IRF3(5D) was de-ubiquitinated by PLpro. We transfected HEK293T cells with either wildtype ubiquitin (Ub) tagged with HA (HA-Ub), K48-linked Ub (K48-Ub), or K63-linked Ub (K63-Ub) with and without Flag-tagged IRF3(5D) (Figure 5A). As expected, we observed a ladder of ubiquitinated proteins when probed with anti-HA in the lanes containing IRF3(5D) with either of the HA-tagged Ub variants or HA-tagged Ub variants alone. When Flag-tagged IRF3 and HA-tagged Ub were co-transfected with GFP-tagged PLpro we found that overall ubiquitinated protein levels were reduced (Figure 5A). Interestingly, K48-Ub and K63-Ub plasmid transfections displayed higher levels of ubiquitinated proteins compared to wildtype HA-Ub. For this reason, K48-Ub and K63-Ub were solely used in immunoprecipitation experiments in Figure 5B. However, wildtype HA-Ub does show reduced ubiquitinated protein levels when PLpro is co-expressed, identical to the results with K48-Ub and K63-Ub plasmid transfections. When anti-Flag coated beads were used to pull-down Flag-tagged IRF3(5D) from selected lysate shown in Figure 5A, we found that IRF3(5D) is ubiquitinated. Interestingly, when co-transfected with PLpro and K48 and K63 ubiquitin, there is only minor ubiquitination of IRF3(5D) (Figure 5B). These data demonstrate that IRF3(5D) is ubiquitinated and that PLpro is able to deubiquitinate IRF3(5D) in vitro.

We hypothesized that the ubiquitination of IRF3(5D) was necessary for its activity, as has been shown for IRF3 [39], so we mutated the active site of PLpro to remove its deubiquitination activity [28]. To confirm the inhibition of the DUB activity, HEK293T cells were transfected with either a GFP-only plasmid, HA-tagged Ub alone or HA-tagged Ub in combination with either wild type PLpro (PLpro/wt) or active site mutant PLpro (PLpro/mt) (Figure 6A). As before, both PLpro plasmids are expressed as C-terminal GFP fusion proteins. We observe that when HA-tagged Ub is transfected alone, a large smear of ubiquitinated proteins are seen by western blot. When GFP-tagged WT PLpro was co-transfected with HA-tagged Ub, we observe a marked reduction in the amount of ubiquitinated proteins. When the mutant GFP-tagged PLpro is co-transfected with HA-tagged Ub the smear of ubiquitinated proteins reappeared (Figure 6A), demonstrating that the mutant PLpro no longer contains deubiquitinase activity.

We used this mutant PLpro to assay for its ability to inhibit IRF3(5D)-dependent induction of IFNβ/luciferase reporter plasmid. As seen previously, when either IRF3 or IRF3(5D) are transfected into HEK293T cells with the IFNβ/lux reporter we observe strong induction of luciferase (32-fold and 21-fold, respectively) (Figure 6B). Also, as observed previously, when wildtype PLpro is co-transfected with both we find that IFNβ/lux induction is significantly inhibited. However, when the deubiquitination- deficient mutant PLpro is transfected with either IRF3 or IRF3(5D) we found that there is no longer an inhibition of IFNβ/lux induction. These data demonstrate that the deubiquitination activity of PLpro is responsible for limiting gene induction of IRF3(5D) on the IFNβ promoter.