Ovarian tumor domain-containing viral proteases evade ubiquitin- and ISG15-dependent innate immune responses

Abstract

Ubiquitin (Ub) and interferon-stimulated gene product 15 (ISG15) reversibly conjugate to proteins and mediate important innate antiviral responses. The ovarian tumor (OTU) domain represents a superfamily of predicted proteases found in eukaryotic, bacterial, and viral proteins, some of which have Ub-deconjugating activity. We show that the OTU domain-containing proteases from nairoviruses and arteriviruses, two unrelated groups of RNA viruses, hydrolyze Ub and ISG15 from cellular target proteins. This broad activity contrasts with the target specificity of known mammalian OTU domain-containing proteins. Expression of a viral OTU domain-containing protein antagonizes the antiviral effects of ISG15 and enhances susceptibility to Sindbis virus infection in vivo. We also show that viral OTU domain-containing proteases inhibit NF-kappaB-dependent signaling. Thus, the deconjugating activity of viral OTU proteases represents a unique viral strategy to inhibit Ub- and ISG15-dependent antiviral pathways.

Figure 1

The OTU Domain Sequence Is Conserved across Viral and Mammalian Proteins Multiple alignment of the OTU domains present in the proteins used in this study. In the consensus (Makarova et al., 2000), “h” indicates hydrophobic residues (A, C, F, L, I, M, V, W, Y, T, S, G), “s” indicates small residues (A, C, S, T, D, V, G, P), “+” indicates positively charged residues (R, K), “a” indicates aromatic residues (W, Y, F, H), and “t” indicates residues with high β-turn-forming propensity (A, C, S, T, D, E, N, V, G, P). Highly conserved residues are shaded in black. Numbers at the beginning and end of each sequence indicate the positions of the first and last aligned residue in the respective protein sequences; the numbers between aligned blocks indicate the numbers of residues that are not shown. CCHFV, Crimean Congo hemorrhagic fever virus; DUGV, Dugbe virus; EAV, equine arteritis virus; PRRSV, porcine respiratory and reproductive syndrome virus; VCIP, VCIP135; Cez, Cezanne; OTUB1 and OTUB2, Otubains 1 and 2, respectively.

Figure 2

Levels of Ubiquitinated and ISGylated Proteins in Cells Expressing CCHFV-L and CCHFV-L Mutants (A) Schematic representation of the CCHFV-L constructs utilized in these studies. Predicted protein domains within the protein: OTU domain; ZF, zinc finger; LZ, leucine zipper; RdRp, RNA-dependent RNA polymerase conserved motifs. White stars represent mutations in the OTU domain: 1A (C40A) or 2A (C40A and H151A). All constructs were HA tagged. (B) 293T cells were transfected with either HA-Ub (panel A) or His-ISG15, HA-mUBE1L, and Flag-UbcM8 (panels B–D) along with HA-tagged CCHFV-L constructs, Flag-UBP43, or empty plasmid. Total protein ubiquitination was visualized by immunoblotting with anti-HA (panel A), and protein ISGylation was visualized by anti-ISG15 immunoblot (panel B). ISG15-transfected samples were also probed with anti-HA or anti-Flag (panels C and D) for detection of the CCHFV-L constructs (left arrows), mUbE1L, UBP43 (inset), and UbcM8. Asterisk indicates a nonspecific band. (C) L and L 1A were analyzed for their effect on total ubiquitination (panel A) or ISGylation (panel B) as described in (B). Expression of HA-tagged UBE1L (panel C), L and L 1A (panel D), and L(1–169) and L(1–169)2A (panel E) is shown.

Figure 3

In Vitro Ub and ISG15 Deconjugation Activities of the CCHFV-L OTU Domain (A) Coomassie-stained gel of GST-L(1–169), L(1–169), GST-L(1–169)1A, and L(1–169)1A recombinant proteins. L(1–169) and L(1–169)1A proteins were used for the in vitro experiments. (B) K⁴⁸ (top panel)- or K⁶³ (bottom panel)-linked poly-Ub chains were incubated with reaction buffer (lane 1) or 10-fold dilutions of L(1–169) or L(1–169)1A recombinant proteins, subjected to SDS-PAGE and visualized by Coomassie staining. Isopeptidase T (IsoT) was used as a positive control. Black arrows indicate L(1–169) and L(1–169)1A proteins. (C) Lysates of Ubp43^−/− MEFs (top panel) or ISG15 conjugates purified from ISG15, HA-mUBE1L, and Flag-UbcM8 transfected 293T cells (bottom panel) were incubated with reaction buffer (lane 1) or 10-fold dilutions of L(1–169) or L(1–169)1A protein. ISG15 conjugates were visualized by anti-ISG15 immunoblot. (D) SUMO-2 (top panel) or SUMO-3 (bottom panel) chains were incubated with reaction buffer (lane 1) or 10-fold dilutions of L(1–169) or L(1–169)1A and visualized by Coomassie staining. His₆-SENP2_CD was used as a positive control. Black arrows indicate L(1–169) and L(1–169)1A proteins. S_2–8 indicates number of SUMO-2 or SUMO-3 molecules. (E and F) ProISG15 (panel A), proNedd8 (panel B), proSUMO-1 (panel C) or K⁴⁸-linked Ub (F) chains were incubated with reaction buffer, L(1–169), L(1–169)1A, or A20 catalytic domain (A20_CD) and visualized by Coomassie staining. Positive controls indicate incubation with UBP43 (panel A), NEDP1 (panel B), SENP2_CD (panel C), or IsoT (F). Black arrows indicate L(1–169) and L(1–169)1A proteins, and white arrowhead indicates A20_CD. “Pro” indicates the pro-UBL molecule form, and “Mat” indicates the mature protein.

Figure 4

Ub and ISG15 Deconjugation Activity of OTU Domain-Containing Polypeptides of Viral and Mammalian Origin (A) CCHFV-L(1–169), DUGV-L(1–169), PPRSV-nsp2, EAV-nsp2N, or EAV-nsp2 were cotransfected into 293T cells with either HA-Ub (panel 1) or ISG15, HA-mUBE1L, Flag-UbcM8, and Herc5 plasmids (panels 2 and 3). Samples were immunoblotted for HA (panel 1) or ISG15 (panel 2). ISG15-transfected samples were also probed with anti-HA (panel 3) or anti-Flag plus anti-HA (panel 4) to show expression of HA-mUBE1L, PPRSV-nsp2, CCHFV-L(1–169), DUGV-L(1–169), EAV-nsp2N, or EAV-nsp2 (inset, open triangle). Asterisk indicates a nonspecific band. (B) Otubain 1, Otubain 2, Cezanne, VCIP135, A20, or UBP43 were analyzed for their effect on total ubiquitination (panel 1) or ISGylation (panels 2–4) as described in (A). Expression of HA-tagged Cezanne and VCIP135 (panel 3) and Flag-tagged Otubain 1 and 2 and UBP43 (panel 4) is shown.

Figure 5

Expression of L(1–1325) Transgene Correlates with Increased Susceptibility to Sindbis Virus Infection (A and B) Expression of L(1–1325) transgene and actin in MEFs (A) and brain lysates (B). + indicates a transgene-positive mouse, and − indicates a C57/BL6 mouse. The arrows indicate L(1–1325) protein and open triangles denote actin. (C) Survival of L(1–1325) transgenic mice following infection with Sindbis virus AR86. Transgene-negative littermates from 1836, 1854, and 2929 served as C57/BL6 controls. Numbers of mice in each group are indicated in parenthesis. Comparison by statistical analysis was made between 1836+ and C57/BL6 (p = 0.0011).

Figure 6

Sindbis Viruses Expressing the CCHFV-L OTU Domain Deconjugate Ub and ISG15 and Inhibit ISG15-Mediated Antiviral Effects in Mice (A) Schematic diagram representing the CCHFV OTU domain-expressing Sindbis viruses utilized in these studies. G, genomic promoter; SG, subgenomic promoter. (B) BHK-21 cells were transfected with UBE1L, UbcM8, and Herc5 (left panel), UBE1L, UbcM8, Herc5, and ISG15 (middle panel), or HA-Ub (right panel) and subsequently infected with recombinant Sindbis viruses as indicated. Cells lysates were immunoblotted with anti-ISG15 (left and middle panels) or anti-HA (right panel) antibodies. − indicates untransfected cells. (C) IFNαβR^−/− mice were infected with recombinant Sindbis viruses as indicated and monitored for survival. Data are pooled from four independent experiments, and numbers of mice in each group are indicated in parenthesis. Differences in survival were analyzed by the log-rank test: 169GG and 169 (p < 0.0001), 169GG and MT (p = 0.0032), 169GG and MTGG (p = 0.0015), MTGG and 169 (p < 0.0001), and MTGG and MT (p < 0.0001).

Figure 7

CCHFV-L and EAV-nsp2 OTU Domains Inhibit TNFα-Mediated NF-κB Activation (A) NF-κB reporter assay in 293T cells transfected with OTU domains and treated with TNFα. Results shown are an average of three independent experiments. The western blot indicates expression of viral OTU proteins as detected with anti-HA [CCHFV-L(1–169) and CCHFV-L(1–169)2A] or anti-Flag antibodies (EAV-nsp2N). E, empty plasmid. (B) A549 cells were transfected with indicated plasmids, stimulated with TNFα, and stained for p65 (red) and L(1–169) or L(1–169)2A (green). Nuclei were stained with DAPI (blue). The result shown is a representative of three independent experiments. (C) L(1–169) or L(1–169)2A transfected cells in (B) were scored according to subcellular distribution of p65. Differences in p65 nuclear accumulation in TNFα-treated cells were analyzed by Student's t test: E and L(1–169) (p < 0.0001), E and L(1–169)2A (p = 0.0007), and L(1–169) and L(1–169)2A (p = 0.0045). E, empty plasmid. The result shown is a representative of three independent experiments. Data in (A) and (C) are presented as mean ± SD (n = 3).