USP39 Serves as a Deubiquitinase to Stabilize STAT1 and Sustains Type I IFN–Induced Antiviral Immunity

Ubiquitination, a posttranslational modification covalently attaching ubiquitin to target proteins, alters target proteins’ biological activity, stability, and subcellular localization (1). Ubiquitin contains seven lysine residues and can form eight different linkages of polyubiquitin chains (K6, K11, K27, K29, K33, K48, K63, and Met1) (2). Distinct linkage types may result in different functions, such as protein degradation and regulating signaling pathways (3). For instance, K48-linked polyubiquitination mainly mediates protein degradation through the ubiquitin proteasome system, whereas K63-linked polyubiquitination mainly contributes to DNA repair, endocytosis, and signal transduction (4).

Deubiquitinating enzymes (DUBs), proteases that reverse-modify protein ubiquitination by removing ubiquitins or ubiquitin-like molecules, have been considered crucial regulators of both ubiquitin-mediated degradation and other biological processes (5). The human genome encodes ∼100 potential DUBs, which can be classified into six families: ubiquitin-specific proteases (USPs), ubiquitin C-terminal hydrolases (UCHs), ovarian tumor proteases (OTUs), Josephins, JAB1/MPN/MOV34 family (JAMMs), and MCP 1–induced protein (MCPIP) (6). Among these families, USPs, UCHs, OTUs, Josephins, and the newly identified MCPIP families belong to thiol proteases, whereas JAMMs, are zinc metalloproteases (7). DUBs regulate multiple signal transduction pathways in all organelles, such as regulating transcription and DNA repair in the nucleus and cell division and apoptosis in the cytoplasm (8). However, the specific mechanisms of DUBs in various signaling pathways have not been fully illuminated.

The type I IFN family plays a critical role in the defense of viral infection and regulating innate immunity (9–11). In the classical type I IFN signaling pathway, IFN-α/β binds to its receptors (IFNAR1 and IFNAR2) and induces tyrosine phosphorylation of the JAK family (JAK1 and Tyk2). Subsequently, activated JAK1 and Tyk2 induce phosphorylation of the signal transducers and activators of transcription (STAT1 and STAT2). Phosphorylated STAT1, STAT2, and IRF9 form a signaling complex known as ISGF3, which then translocates into the nucleus and binds to the IFN-stimulated response elements (ISRE) promoter to induce the expression of IFN-stimulated genes (ISGs). Finally, these ISGs perform multiple biological functions, such as antivirus, antitumor, and immunoregulation (12, 13). To date, clarifying the specific mechanism of type I IFN–induced antiviral signaling remains a major challenge.

Several DUBs have been reported to have antiviral effects (14–19). USP4 interacts with RIG-I and removes K48-linked polyubiquitin chains from RIG-I to stabilize its protein level, thus positively regulating RIG-I–mediated antiviral immune responses (20). In the contrast, USP19 negatively regulates virus-induced type I IFN signaling through removing K63-linked polyubiquitin chains of TRAF3 (21). To systematically explore the roles of DUBs in antiviral immunity, we analyzed the roles of 97 DUBs during vesicular stomatitis virus (VSV) infection. Among these DUBs, USP39 was turned to be a promising candidate against viral infection. We found that USP39 can promote innate antiviral immunity by stabilizing STAT1 in the type I IFN signaling pathway and USP39-mediated regulation of type I IFN signaling is dependent on its deubiquitinase activity. Our findings uncover a novel innate antiviral immunity mechanism of USP39 and may provide a new target for enhancing IFNs antiviral therapeutic efficacy.

293T, HT1080, and BHK21 cells were reserved by the laboratory. The U3A cell was provided by Dr. G. Chen. 293T, BHK21, and U3A cells were cultured in DMEM (HyClone, South Logan, UT). HT1080 cell was cultured in RPMI 1640 medium (HyClone). All media were supplemented with 10% FBS (Life Technologies, Gaithersburg, MD), 100 U/ml penicillin, and 100 μg/ml streptomycin. All cells were maintained at 37°C in a 5% CO₂ laboratory incubator that was routinely cleaned and decontaminated. All transient transfections were carried out using Longtrans (UcallM Biotechnology, Wuxi, China) for DNA plasmids or Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA) for small interfering RNA (siRNA) according to the manufacturer’s instructions.

Human DUBs siRNA library was purchased from RiboBio (STLC002; Guangzhou, China). The library contains 97 human DUBs. There were three different siRNAs targeting to each DUB. A total of 5 × 10⁴ 293T cells seeded in one well of 48-well plate were individually transfected with three mixed siRNAs for one DUB (50 nM) using Lipofectamine RNAiMAX (Invitrogen). Forty-eight hours posttransfection, cells were infected with VSV-GFP (multiplicity of infection [MOI] = 1.0), and the green fluorescence of infected cells was measured by flow cytometry 12 h postinfection.

The USP39 siRNA (siUSP39) were purchased from RiboBio (STB005338B). The human USP39 cDNA was PCR amplified from 293T cell with the following primer pair sequence for USP39: forward (F), 5′-CCGGAATTCCATGTCCGGCCGGTCTAAGC-3′ and reverse (R), 5′-GCTCTAG AGCTCAAGCCCCCTGCTGGTTG-3′. The amplified fragment was cloned into pFlag-CMV₂ vector using EcoRI and XbaI restriction enzyme sites and we named it Flag-USP39. We also used the Site-Directed Mutagenesis Kit (R401; Takara Bio, Tokyo, Japan) to generate C136A and C139A double-mutated USP39 (Flag-USP39 ^C136A/C139A) with the following primer sequences: USP39 ^C136A/C139A: F, 5′-ATCAATGCTTATGCCGCCCT GGTGGCCGGCAAGTACTTTCAAGGCCGGGGTTTG-3′ and USP39 ^C136A/C139A R, 5′-CTTGAAA GTACTTGCCGGCCACCAGGGCGGCATAAGCATTGATGTGTGAGAGGGAG-3′. The mutated sites were underlined in the primer sequence. All plasmids were confirmed by DNA sequencing. Luciferase reporter plasmids (IFN-β–Luc and ISRE-Luc) and other plasmids including HA-Ub, HA-Ub-K6, HA-Ub-K11, HA-Ub-K27, HA-Ub-K29, HA-Ub-K33, HA-Ub-K48, HA-Ub-K63, and Flag-STAT1 were kind gifts from Dr. H. Zheng (Soochow University, Suzhou, China). Recombinant human IFN-α was purchased from PBL InterferonSource (Waltham, MA). IFN-α was used at the concentration of 1000 IU/ml unless stated otherwise. Cycloheximide (CHX) and MG132 were purchased from Sigma-Aldrich (St. Louis, MO).

VSV-GFP was obtained and reserved by the laboratory. Sendai virus (SeV) was a gift from Dr. F. Zhou (Soochow University, Suzhou, China). To assess the antiviral ability of USP39, control siRNA or siUSP39 were transfected into cells. After 48 h of transfection, cells were challenged with VSV-GFP (MOI = 1.0) for 12 h. Cells were then collected, and viral levels were analyzed by flow cytometry. To determine the antiviral effect of IFN-α, cells were transfected with siRNAs against USP39. Forty-eight hours after transfection, cells were treated with 30 IU/ml IFN-α for 15 h. After washing twice, cells were infected with viruses at MOI = 1.0 for 15 h, then the viral levels were analyzed by Western blot using VSV-G Ab or real-time PCR.

293T cells infected with VSV-GFP (MOI = 1.0) were subjected to analysis by flow cytometry. For flow cytometry analysis, cells were collected with cold 1× PBS and were acquired in a FACS Canto II (BD Biosciences, San Jose, CA) equipped with a 488-nm argon laser. FACS data were analyzed with FlowJo software (FlowJo, Ashland, OR).

Total cellular RNA was extracted using RNeasy Total RNA Kit (Takara Bio, Tokyo, Japan), and cDNA synthesis was performed using random primers with 500 ng of total RNA. Real-time PCR was performed using a SYBR Green PCR Master Mix with specific primers (Applied Biosystems, Waltham, MA) and normalized with the human β-actin gene. The primer sequences used are in Table I.

Cells were harvested using lysis buffer containing 150 mM NaCl, 20 mM Tris–HCl (pH 7.4), 1% NP-40, 0.5 mM EDTA, PMSF (50 μg/ml), and protease inhibitors (Roche, Basel, Switzerland). Equivalent protein quantities were subjected to SDS-PAGE and transferred to PVDF membranes. Membranes were then blocked with 5% nonfat milk or 5% BSA for 2 h at room temperature and then probed with the primary Abs, followed by the respective HRP-conjugated goat anti-mouse or goat anti-rabbit secondary Abs (SouthernBiotech, Birmingham, AL). The following Abs were used: Abs against USP39 (1:2000; ab131244; Abcam, Cambridge, MA), IFNR1 (1:1000; DF6571; Affinity, Cincinnati, OH), IFNR2 (1:1000; sc-271105; Santa Cruz Biotechnology, Santa Cruz, CA), STAT1 (1:1000; catalog no. 9172; Cell Signaling Technology [CST], Danvers, MA), pY701-STAT1 (1:1000; catalog no. 9167; CST), STAT2 (1:1000; catalog no. 72704; CST), pY690-STAT2 (1:1000; catalog no. 88410; CST), JAK1 (1:1000; sc-1677; Santa Cruz Biotechnology), p-JAK1 (1:1000; catalog no. 3331s; CST), Tyk2 (1:1000; catalog no. 14193; CST), p-Tyk2 (1:1000; catalog no. 9321; CST), VSV-G (1:5000; sc-66180; Santa Cruz Biotechnology), HA (1:2000; ab9110; Abcam), ubiquitin (1:1000; sc-8017; Santa Cruz Biotechnology), Flag (1:2500; F7425; Sigma-Aldrich), Myc (1:2000; catalog no. 9B11; CST), GAPDH (1:20,000; G9545; Sigma-Aldrich), tubulin (1:5000; 66031-1-Ig; Proteintech, Rosemont, IL). The band intensities were quantified by ImageJ software (Media Cybernetics, Silver Spring, MD).

293T cells were transfected with Flag-USP39 plasmids. Forty-eight hours posttransfection, the cells were lysed with protein lysis buffer (20 mM Tris–HCl [pH 7.5], 150 mM NaCl, 1% Triton X-100, sodium pyrophosphate, β-glycerophosphate, EDTA, Na₃VO₄, and leupeptin). Cell lysates were incubated with rabbit protein A agarose at 4°C for 4 h, then immunoprecipitated with anti-STAT1 Ab at 4°C overnight. The proteins were analyzed by Western blot analysis with anti-Flag Ab.

For detection of IFN-β production, 1 × 10⁵ 293T cells seeded in one well of a 24-well plate were individually transfected with siUSP39 (50 nM), together with 100 ng of IFN-β-Luc luciferase and 25 ng of Renilla luciferase plasmids. Forty-eight hours posttransfection, the cells were infected with virus for 15 h and then harvested. For analysis of IFN-induced transcriptional activity, 1 × 10⁵ 293T cells seeded in one well of 24-well plate were individually transfected with siUSP39 (50 nM) or Flag-USP39 plasmid (200 ng), together with 100 ng of ISRE-Luc and 25 ng of Renilla luciferase plasmids. Forty-eight hours posttransfection, the cells were treated with or without IFN-α (1000 IU/ml) for 12 h and then harvested. The luciferase activities were measured using the Dual-Luciferase Reporter Assay System (catalog no. E1910; Promega, Madison, WI).

The half-life of total STAT1 was determined by CHX (Sigma-Aldrich) chase assay. For analysis of total STAT1 levels, 293T cells were transfected with vector or Flag-USP39. Forty-eight hours after transfection, cells were treated with DMSO or CHX (50 μg/ml) for the indicated time points, then an equal amount of boiled lysates were analyzed by Western blot. The protein level was measured by the gray intensity using ImageJ software. The amount of STAT1 was calibrated by the GAPDH. The STAT1 degradation rate at A time point was calculated as (STAT1_0 h − STAT1_A h)/STAT1_0 h.

A total of 1 × 10⁶ 293T cells seeded in a 6-cm dish were transfected with 1 μg of control vector or HA-K6, together with 1.5 μg of Flag-STAT1 plasmids. Thirty-six hours posttransfection, cells were treated with DMSO or MG132 (10 μM) for 6 and 12 h. Cell lysates were incubated with anti-Flag beads (EZview Red ANTI-FLAG M2 Affinity Gel, Sigma-Aldrich) at 4°C overnight, and the K6-linked ubiquitination of STAT1 was analyzed by Western blot.

293T cells were cotransfected with Flag-STAT1 and HA-Ub. The ubiquitinated Flag-STAT1 proteins were immunoprecipitated using the anti-Flag beads. Similarly, the cell lysates obtained from 293T cells transfected with Flag-USP39 were used to pull down Flag-USP39 proteins by the anti-Flag beads. All immunoprecipitates were eluted with 1× Flag peptide (Sigma-Aldrich). The ubiquitinated Flag-STAT1 and Flag-USP39 proteins were then added to a deubiquitination reaction buffer (50 mM Tris–HCl [pH 7.4], 150 mM NaCl, 5 mM MgCl₂, and 10 mM DTT) for reaction at 37°C for 2 h. The ubiquitination levels of immunoprecipitated STAT1 were analyzed using Western blot.

Six- to eight-week-old C57BL/6 female mice were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Sciences. The mice were housed in a barrier facility with controlled temperature at Soochow University. All animal experiments were approved by the Institutional Animal Care and Use Committees of Soochow University. Polyplexes were prepared at a nitrogen/DNA phosphate ratio = 7. The polymer and 50 μg of DNA were both diluted in 5% dextrose to a volume of 100 μl. DNA and polymer were then mixed, whirled for 30 s, and incubated for 30 min with a volume of 200 μl after mixing. The mice were i.v. injected with 50 μg of polyetherimide (PEI)-packaged Flag-CMV₂ plasmid or Flag-USP39 overexpression plasmid. On day 2 postinjection, the lung tissues were removed from sacrificed mice, and USP39 expression was detected by Western blotting analysis. For other parallel groups on day 1 postinjection, the mice were intranasally infected with 1 × 10⁸ PFUs VSV-GFP. The mRNA levels of IFN-β or ISGs in mice lung tissues were analyzed by real-time PCR 3 d postinfection. The frozen sections of lung tissues were also prepared, and the GFP fluorescence was detected under the fluorescence microscope. For survival rate measurement, each group containing eight infected mice was observed until day 10 of infection.

Comparison between different groups was analyzed by two tailed Student t test. Data were shown as the mean ± SD. All p values <0.05 were considered statistically significant. Kaplan–Meier survival curves were generated and analyzed for mouse survival study performed in GraphPad Prism 6.0.

The mechanism of host antiviral innate immune responses regulated by deubiquitinases is not totally understood. To identify which DUBs have antiviral effects, we conducted the screening of a siRNA library containing 97 human DUBs upon VSV infection. 293T cells were individually transfected with DUB siRNAs and infected with a reporter virus, VSV-GFP (MOI = 1.0) 48 h post-siRNA transfection. The green fluorescence of infected cells was measured by flow cytometry 12 h postinfection (Supplemental Fig. 1). If the DUBs have an inhibitory effect on viral replication, the viral infection should increase with siRNA transfection in 293T cells; it yielded that USP4, SENP7, USP39, USP30, OTUD7A, EIF3H, CYLD, and COPS5 ranked as the most antiviral effect, as shown in Fig. 1A. We excluded USP4 and SENP7 because the role of USP4 in VSV infection has been studied before (20) and the level of SENP7 in 293T cells was low. We then focused on the USP39 gene, the most antiviral gene in the screened genes. We performed the siUSP39 transfection and VSV infection in human fibrosarcoma (HT1080) cells and BHK21 cells, further confirming USP39 antiviral ability (Fig. 1B). The USP39 protein level in infected cells increased and reached a peak 4 h postinfection, suggesting that USP39 may execute an antiviral function during VSV infection (Fig. 1C). To further verify USP39 antiviral effect, we exogenously expressed Flag-USP39 in 293T cells and did the VSV infection 36 h posttransfection. We found that VSV replication was inhibited as measured either by viral G protein or the GFP⁺ infected cells (Fig. 1D), which was consistent with siUSP39 knockdown results.

Type I IFN is the first defense of the host to combat virus, so we then asked whether USP39 affects IFN production or the IFN-mediated signaling pathway. 293T cells were transfected with control siRNA or siUSP39, following stimulation of VSV infection for type I IFN production. We found that siUSP39 knockdown had no effect on the production of type I IFN or on the overexpression of USP39 in 293T cells (Fig. 2A). Meanwhile, the IFN-β reporter assay showed that the luciferase activity was not significantly changed with USP39 knockdown in 293T cells (Fig. 2B). Similar results were observed when cells were stimulated with SeV, which is also a potent type I IFN inducer (data not shown). Given the above finding that USP39 did not affect the type I IFN production, we next wondered whether USP39 affected the IFN-mediated downstream signaling pathway. VSV has been thought to be very sensitive to type I IFN treatment. Pretreatment of cells with IFN-α enhanced cellular antiviral defenses and therefore decreased the levels of viral envelope protein VSV-G (Fig. 2C, lane 2 versus lane 4), and USP39 knockdown remarkably reduced IFN-α–induced antiviral defenses (Fig. 2C, lane 4 versus lane 5), which was consistent with real-time PCR results (Fig. 2D) and suggests that USP39 may play a role in the IFN-mediated signaling pathway. In addition, it seems that IFN-α stimulation increases the USP39 protein level; we measured the change of USP39 in U3A (STAT1 deficient) cells and its parental cells, HT1080, treated with IFN-α (1000 IU/ml) at different time points. As expected, the USP39 protein level was increased upon IFN-α stimulation in HT10180 cells, suggesting it might be an ISG. However, we did not see the change of USP39 protein level in U3A cells upon IFN-α treatment (Supplemental Fig. 2).

It has been well demonstrated that the activated STAT1/2 and IRF9 will form ISGF3 complex, which translocates into nucleus and binds to the ISRE promoter and finally induces ISG expression (22). To further confirm the effects of USP39 on type I IFN signaling, we analyzed the ISRE promoter activation with or without USP39 knockdown. Using an ISRE activity luciferase reporter assay, we found that knockdown of endogenous USP39 significantly suppressed type I IFN–mediated ISRE promoter activity, whereas exogenous expression of USP39 enhanced ISRE activity (Fig. 3A, 3B), suggesting that USP39 promotes ISRE promoter activity and enhances type I IFN signaling.

Because activation of the ISRE promoter will result in ISG expression, we also analyzed whether USP39 has effects on the expression of ISGs, such as IFIT1 and ISG54. Knockdown of endogenous USP39 remarkably inhibited mRNA levels of both IFIT1 and ISG54 in a time-dependent manner with IFN-α treatment (Fig. 3C). On the contrary, exogenous expression of USP39 increased IFN-induced expression of IFIT1 and ISG54 compared with vector control (Fig. 3D). To further investigate the role of USP39 in type I IFN signaling, we used siUSP39 to knockdown endogenous USP39 in HT1080 cells and U3A cells, then infected them with VSV-GFP and measured the infectivity by the percentage of GFP⁺ cells (Fig. 3E). Consistent with results in Fig. 1B, siUSP39 knockdown in HT1080 cells led to significantly increased viral replication. But knockdown of USP39 in the U3A cells did not affect VSV replication, further confirming that USP39 can enhance antiviral activity through type I IFN signaling.

To determine which molecular USP39 regulates in the type I IFN signaling pathway, we first analyzed the influence of USP39 on the mRNA levels of key factors involved in type I IFN signaling, including IFNAR1, IFNAR2, JAK1, Tyk2, STAT1, and STAT2. The mRNA levels of these genes were not affected by the siRNA knockdown of USP39 in 293T cells (Fig. 4A). However, when we examined the protein levels of these factors, we found that silence of endogenous USP39 obviously decreased the level of STAT1 and phosphorylated STAT1 but not the other proteins in type I IFN signaling (Fig. 4B), and we also found that it was dose dependent (Supplemental Fig. 3). Furthermore, when Flag-USP39 was exogenously expressed in 293T cells, the increased STAT1 level was observed (Fig. 4C). Moreover, to examine whether USP39 could interact with STAT1, a coimmunoprecipitation assay was performed. It was shown that Flag-USP39 interacted with STAT1 Ab and was coprecipitated by protein A agarose beads (Fig. 4D).

To address how USP39 regulates the STAT1 protein level, we traced the protein stability of STAT1 in cells treated with the protein synthesis inhibitor CHX with or without exogenous expression of USP39. The endogenous STAT1 protein level was higher in 293T cells with ectopic expression of USP39 at all of the time points post-CHX treatment (Fig. 5A). The STAT1 degradation, as measured by ImageJ software, showed that the STAT1 protein degradation rate in USP39 exogenous expression cells was less compared with that of control cells (0.24 versus 0.58) 12 h post-CHX treatment. We further asked whether USP39 regulates the STAT1 protein level through ubiquitination. Interestingly, USP39 was previously not considered as a deubiquitinase; we found that USP39 knockdown markedly enhanced STAT1 ubiquitination (Fig. 5B). On the contrary, an in vitro deubiquitination assay showed that overexpression of Flag-USP39 reduced the ubiquitination level of STAT1 protein compared with that of the vector (Fig. 5C), demonstrating that USP39, as a deubiquitinase, was able to deubiquitinate STAT1. Additionally, we noticed that the localization of USP39 after type I IFN stimulation did not change, as it was detected predominantly in the nucleus, similar with nonstimulation. Furthermore, it is shown that the zinc finger USP (ZnF-UBP) domain is critical for the enzyme activity (23). We then generated a USP39 mutant (USP39^C136/139A) in which the Cys¹³⁶ and Cys¹³⁹ of the ZnF-UBP were replaced by Ala. Because of the impaired enzyme activity of mutant USP39^C136/139A, overexpression of USP39^C136/139A resulted in the increased cellular ubiquitination level compared with wild-type (WT) USP39. Moreover, it was no longer able to enhance the stability of STAT1 protein (Fig. 5D). When the transfected cells were infected with virus, quantitative PCR showed that there was no difference in IFN-β mRNA level between Flag-Vec, Flag-USP39, and Flag-USP39 ^C136A/C139A. However, the ISGs (IFIT1 and ISG54) increased in Flag-USP39–transfected cells compared with Flag-Vec–transfected cells. The expression of these ISGs was reduced when the Flag-USP39 was mutated, indicating the enzymatic activity is critical for ISG induction. Consequently, the viral replication in Flag-USP39–transfected cells was decreased, as measured by viral mRNA (Fig. 5E).

K48- and K63-linked ubiquitination of substrates have so far been extensively studied. Because USP39 can decrease STAT1 ubiquitination, we tried to determine which type of ubiquitination USP39 can affect STAT1. To this end, HA-Ub-K6, 11, 27, 29, 33, 48, and 63 were used to analyze the types of STAT1 ubiquitination. Our data showed that knockdown of USP39 significantly sustained K6-linked ubiquitination of STAT1 but not the other types of ubiquitination (Fig. 6A). Despite the absence of K6-specific Ab, 293T cells transfected with HA-K6 and Flag-STAT1 plasmids were treated with MG132 36 h posttransfection. The K6-linked ubiquitination was examined 6 and 12 h after MG132 treatment. We found that MG132 treatment significantly upregulated K6-linked ubiquitination levels of STAT1, suggesting that K6-linked STAT1 ubiquitination may undergo proteasome-dependent degradation (Fig. 6B).

Our results above suggest that USP39 can deubiquitinate STAT1 and enhance the stability of STAT1 protein level. To investigate the role and functional importance of USP39 in host antiviral response in vivo, we upregulated the USP39 expression in mice by i.v. injection of PEI-packaged plasmids. Forty-eight hours postinjection, we found that the exogenous USP39 expression was obviously upregulated in mice lung tissue (Fig. 7A) but not in liver, spleen, and kidney (data not shown). We challenged WT- and USP39-upregulated mice with VSV using intranasal infection and found that the IFN-β production was not significantly different between USP39-upregulated mice or control vector-injecting mice (Fig. 7B). Meanwhile, the ISG expression was higher in USP39-upregulated mice (Fig. 7C), and less viral infection was measured in their lung tissue than WT mice 72 h postinfection (Fig. 7D). When viral infection was extended to 10 d, the mice-survival results showed that USP39-upregulated mice were more resistant to VSV infection in overall survival assays (Fig. 7E). Collectively, our data suggest that USP39-upregulated mice possess more potent host defense against VSV by enhancing type I IFNs signaling (Fig. 8).

The human genome encodes ∼100 DUBs that are categorized into six major families. Although USP39 harbors a DUB domain, the three residues (cysteine, histidine, and aspartic acid) important for deubiquitinase activity were not conserved in USP39, as previously reported (23). Thus, this deubiquitinase may not act as an actual deubiquitinase. Several studies reported that USP39 may act as an oncogenic factor in several cancers, such as breast cancer, thyroid carcinoma, and hepatocellular carcinoma (24–27), and they revealed a novel role for USP39-mediated mRNA splicing in the process of cell proliferation (28). However, it is less clear whether or how USP39 plays a role in other biological functions, such as antiviral immunity. In the present study, we characterized USP39 as a positive regulator of the type I IFN signaling pathway by RNA interference–based screening.

It has been well described that IFNs antiviral activity is achieved by activated JAK/STAT signaling pathways. STAT1 is a key molecule in type I IFN–induced antiviral signaling. It becomes activated and phosphorylated at the tyrosine 701 site (pY701-STAT1) and regulates diverse cellular processes, including innate immune responses. In this study, we demonstrated that USP39 is a critical regulator of type I IFN antiviral activity during viral infections. We identified that USP39 actually did not affect the production of type I IFN but significantly promoted type I IFN signaling as evidenced with enhanced ISRE activity and ISGs expression. Recently, several groups provided important evidence for the ubiquitination regulation of total STAT1 (17, 29). As a matter of fact, total STAT1 protein levels are kept relatively stable, even if protein synthesis has been inhibited for over 24 hours. Likewise, short treatment of cells with IFNs for a couple of hours does not obviously change the level of total STAT1 proteins. Recently, two E3 ligases, SLIM and Smurf1, have been reported to catalyze STAT1 ubiquitination and promote its degradation (30, 31). Therefore, these observations suggest that STAT1 seems to have a mechanism to cope with the ubiquitin-mediated degradation. One example is that deubiquitinase USP13 was proved to interact with the nonactivated form of STAT1 and stabilize STAT1 protein level, resulting in an enhanced IFN antiviral response (17). Initially, we thought that USP39 lacked three residues critical for protease-activity and is an inactive deubiquitinase, and we asked whether USP39 regulates STAT1 protein level by regulating Smurf1. Unfortunately, our coimmunoprecipitation assay turned out that there was no interaction between USP39 with Smurf1 (data not shown). Instead, our findings reveal that STAT1 is a substrate of USP39 deubiquitination. First of all, we demonstrated that the STAT1 protein level could be increased by USP39 in a dose-dependent manner and this effect relied on the deubiquitinating activity of USP39. We also showed that USP39 interacted with STAT1 and regulated its ubiquitination to sustain the protein level, thus having an enhanced type I IFN antiviral efficacy. To our knowledge, this is the first report that USP39, as a cellular DUB, possesses the capability of deubiquitinating its substrate protein to positively regulate type I IFN–induced antiviral immunity.

Of the DUBs family, USPs have raised special interest because of the multiple family members described in different eukaryotic organisms. It is thought that not all USPs that contain well-conserved motifs have deubiquitinating activity. Most USPs contain a catalytic domain that includes an N-terminal Cys-box and a C-terminal His-box (32). Other functional ubiquitin-binding domains, such as the ubiquitin-interacting motif, the UBA domain, and the ZnF-UBP domain are also contained (23). The ubiquitin-interacting motif and UBA domains can contribute to the combination of substrates and enzymes, and the ZnF-UBP domain might stimulate the hydrolysis process (33). USP39 contains a ZnF-UBP domain (23), which is important for the catalytic activity of USPs, suggesting that USP39 may have deubiquitinating activity. Although USP39 was predicted to lack three residues critical for protease-activity and to be inactive as a DUB, our findings reveal an interesting mechanism of USP39 deubiquitinating function. We identify USP39 as an important deubiquitinase that interacts with STAT1 directly and regulates STAT1 ubiquitination to sustain its protein level. Nevertheless, we still could not exclude the possibility that USP39 binds to other different DUBs that are mixed in the immunoprecipitation and catalyze STAT1 deubiquitination because it is hard for us to show that the immunoprecipitated USP39 does not contain any other DUBs; we also could not get the recombinant USP39 from the Escherichia coli. Furthermore, we identified that the cysteines 136 and 139 in ZnF-UBP domain were critical for USP39 deubiquitinase activity, as we showed STAT1 ubiquitination level was not significantly changed with the USP39 mutant expression. Overexpression of USP39^C136/139A resulted the increased cellular ubiquitination level compared with WT USP39. Moreover, it was no longer able to enhance the stability of STAT1 protein. Therefore, we believe USP39 most likely acts as another in vivo DUB of STAT1. Similarly, some DUBs, such as USP50 and USP52, which were predicted to be inactive by primary structure alignment, may also have the deubiquitinating functions on one specific substrate (23). Interestingly, the latest report demonstrated that the previously identified pseudodeubiquitinase USP52 is a bona fide deubiquitinase of ASF1A. It removes K48-linked polyubiquitin chains of ASF1A and promotes ASF1A stabilization (34). They also noted that full-length USP52 prefers ASF1A longer ubiquitin chains (K48 linked not K63 linked) and displays higher catalytic activity than USP52/UCH. The discrepancies could be owing to factors, such as protein–protein interactions, intraprotein collaborations, or posttranslational modifications, which could potentially enhance USP39 enzymatic activity or influence its topology preference of polyubiquitin chains. It is essential to investigate which domain of USP39 is interacted with STAT1 directly and could display higher catalytic activity.

As a DUB, which type of ubiquitin chains USP39 removes in its substrates is also interesting. In our studies, we find that USP39 is related to polyubiquitination modification. Furthermore, we show that USP39 can remove K6-linked polyubiquitin chains of STAT1, which again demonstrates that USP39 possesses the ability to remove different types of polyubiquitin chains. The cellular functions of K6-linked ubiquitin chains are currently unclear. K6 linkages have been reported to participate in DNA repair events through the E3 ligase BRCA1 (35, 36). Furthermore, USP8 regulates ubiquitylated mitophagy by removing K6 linkages assembled by the E3 ligase Parkin (37). Few data in the literature suggest that K6 linkages have degradative roles. However, recent research suggests that MID1-mediated K6 ubiquitination promotes proteasomal degradation of BRAF35 (38). Another report recently demonstrated that during viral infection the ubiquitin ligase UBE4A mediates K6-linked ubiquitination at Lys²⁰⁶ of viperin, which promotes degradation of viperin (39). In our study, the treatment of MG132 significantly upregulated K6-linked ubiquitination levels of STAT1, suggesting that STAT1 K6-linked ubiquitination may undergo proteasome-dependent degradation. However, the exact sites in STAT1 for K6-linked ubiquitination will needed to be further investigated.

In summary, we elucidate that USP39 plays an important role in regulating type I IFN–mediated antiviral activity and the type I IFN signaling pathway (Fig. 8). USP39 physically interacts with STAT1 and removes K6-linked polyubiquitin chains of STAT1, which increases the STAT1 protein level and stability in cells. These findings uncover, to our knowledge, a novel biological function of USP39, which may deepen the understanding for innate antiviral defense and provide a novel strategy for improving IFN-based antiviral therapy.

We thank Dr. Wei Xu for helpful comments about this paper. We also greatly appreciate Dr. Guoqiang Chen for providing U3A cells and Dr. Fangfang Zhou for providing SeV.

The authors have no financial conflicts of interest.