Putative E3 Ubiquitin Ligase of Human Rotavirus Inhibits NF-κB Activation by Using Molecular Mimicry To Target β-TrCP

Human and porcine NSP1 proteins conserve a C-terminal phosphodegron-like motif.

NSP1 proteins differentially antagonize the host innate immune response (27). To examine the genetic basis for these activities, 556 RVA NSP1 open reading frames (ORFs) were obtained from GenBank for phylogenetic analysis. These NSP1 sequences formed four distinct clusters that comprised genotypes: (i) A1, A2, and A8; (ii) A3, A11, A12, and A13/A14; (iii) A5, A7, A9, and A6/A10; and (iv) A4/A16 (Fig. 1A). NSP1 genes from human and porcine strains were typically characterized by an A1, A2, or A8 genotype, all of which segregated together according to their phylogenetic history (Fig. 1B). Within this set, the NSP1 genes derived from human strains fell predominantly into the subgroups A1-ST3, A1-Wa, and A2. The NSP1 genes from the remaining subgroups (A1-OSU, A1-R479, and A8) derived from a wider range of host species: predominantly pigs, but occasionally humans, cows, or horses. To date, the RV strains collected from avian hosts carry NSP1 genes, exclusively of the A4/A16 genotypes, that appear only distantly related to those of mammalian strains. The NSP1 genes from the remaining two animal-like genotype clusters—A3, A11, A12, and A13/A14 (UK-like); and A5, A7, A9, and A6/A10 (SA11-4F-like)—derived from a broad range of host species, occasionally including humans. The A12 genotype, which has been observed infrequently and only in humans, appeared to form a cluster phylogenetically distinct from the other UK-like genotypes, but one that bootstrapping could not confidently distinguish. The genetic distance that separated the human/porcine-, animal-, and avian-like NSP1 genes suggested a high level of molecular divergence, which was confirmed by comparing the topology of the phylogenetic tree constructed from the translated sequences (data not shown).

OSU NSP1 (porcine A1) mediates the degradation of β-TrCP (26), and inspection of its sequence identified a C-terminal DSGIS peptide (residues 479 to 483) similar to the β-TrCP recognition motif (DSGΦxS) found in several human (IκBα, IκBβ, β-catenin, NRF2, YAP, and others) and viral (Vpu, LMP1, and A49) proteins (Fig. 1C and D) (42). The site of polyubiquitination on cellular targets of SCF^β-TrCP is a lysine 9 to 14 residues N-terminal to the phosphodegron (12), and a lysine (K465) is also present 14 amino acids N terminal to the DSGIS motif in OSU NSP1. Within the full set of RVA NSP1 genes, a phosphodegron-like motif (DSGΦS) and an N-terminal lysine residue were found in 98% of sequences from human and porcine strains (Fig. 1B and C). The C termini of the NSP1 genes derived from the two animal-like genotype clusters appeared to be conserved among themselves but shared little identity with those of the human/porcine-like strains, suggesting that these sequences have evolved independent strategies of immune antagonism. Studies with the simian RVA strains SA11-4F and SA11-5S (genotype A5)—isogenic except for a 17-residue, C-terminal truncation of SA11-5S NSP1 that arose following serial passage of SA11-4F in cell culture (43)—have demonstrated a critical role for the C terminus in NSP1 function. SA11-5S, unlike SA11-4F, fails to block the induction of IFN-β, and its NSP1 protein cannot mediate the degradation of IRFs (33, 37). The conservation of a C-terminal PDL motif across the majority of human- and porcine-derived RVA strains suggested its importance in NSP1 function, distinct from that described for the other classes of animal RVAs.

OSU NSP1 requires an intact C terminus to mediate its NF-κB antagonist activities.

To explore the role of the PDL motif in OSU NSP1 function, a 13-residue, C-terminal truncation mutant (OSU-ΔC13) was cloned for transient expression. This deletion removed the PDL motif and was analogous by sequence alignment to SA11-5S NSP1 (Fig. 2A). To evaluate the capacity of OSU-ΔC13 NSP1 to block NF-κB activation, HEK293T cells were cotransfected with NSP1 and an NF-κB luciferase reporter. OSU NSP1, but not its ΔC13 mutant, blocked NF-κB activation following stimulation with the cytokine TNF-α (Fig. 2B). OSU-C42A, a point mutant predicted to disrupt the N-terminal RING domain, and SA11-4F NSP1, which mediates the degradation of IRFs but not β-TrCP (27), also failed to block NF-κB activation. NF-κB was nearly twice as active in cells that expressed OSU-C42A versus OSU-ΔC13 NSP1, perhaps indicating that NSP1 mediates the degradation of β-TrCP and another cellular factor(s) to fully block NF-κB activation (Fig. 2B; see Fig. S1A in the supplemental material). Alternatively, disruption of the RING domain may have perturbed the overall structure of NSP1 and affected functions other than its putative E3 ubiquitin ligase activity.

To evaluate the role of the PDL motif in NSP1-mediated turnover of β-TrCP, HEK293T cells were cotransfected with NSP1 and FLAG-β-TrCP. The level of β-TrCP was assayed 24 h posttransfection (p.t.) by quantitative anti-FLAG immunoblotting and, as shown in Fig. 2C, it was 50% lower in cells coexpressing OSU versus OSU-ΔC13 NSP1. This effect was relatively stable over a range of DNA concentrations but could be saturated (see Fig. S2 in the supplemental material). OSU-ΔC13 NSP1 elevated β-TrCP expression versus an empty vector control (see Fig. S1B in the supplemental material), despite being expressed at a slightly lower level than that of the wild type (wt) (see Fig. S3 in the supplemental material), and was therefore used as a normalizing control in degradation assays. OSU-ΔC13 NSP1 had no effect on β-TrCP level compared to SA11-4F or OSU-C42A NSP1, suggesting that a 13-residue truncation of OSU NSP1 is sufficient to abolish all β-TrCP degradation activity. Mutation of the lysine residue N-terminal to the PDL motif (K465R) did not affect the steady-state level of OSU NSP1, nor did it disrupt NSP1 function, likely indicating that β-TrCP was not responsible for polyubiquitinating or otherwise regulating NSP1 turnover (see Fig. S4 in the supplemental material). Likewise, degradation of HIV-1 Vpu occurs independently of β-TrCP activity (44), despite the presence of a lysine 14 residues N-terminal to the Vpu PDL motif (Fig. 1D). Together, these data suggest that OSU NSP1 requires an intact C terminus and N-terminal RING domain to block NF-κB activation and to mediate β-TrCP degradation.

Human and porcine NSP1 proteins conserve NF-κB antagonist activity.

Despite the preponderance of available human and human-like RVA NSP1 sequences (Fig. 1B), most functional studies have focused on NSP1 proteins from animal RVAs. To examine if the NF-κB antagonist and β-TrCP degradation activities of OSU NSP1 extended to related NSP1 proteins, particularly those from human RVA strains, a panel of OSU-like NSP1 sequences and their corresponding ΔC13 mutants (representing genotypes A1, A2, and A8) were cloned for transient expression. These sequences absolutely conserved a C-terminal DSGΦS motif (Φ, I/L/V) (Fig. 3A) and were well conserved over their entire length: they shared 81% identity among themselves but only 38% identity with SA11-4F (see Fig. S5 in the supplemental material). HEK293T cells were first transfected with NSP1 (wt or ΔC13) and an NF-κB luciferase reporter. As shown in Fig. 3B, these OSU-like NSP1 proteins invariably blocked NF-κB activation following TNF-α treatment (Fig. 3B). This block required the NSP1 C terminus, as the corresponding ΔC13 mutants, SA11-4F and SA11-5S NSP1, all failed to inhibit NF-κB activation. DS-1 and KU NSP1 proteins (human A2 and A1-Wa, respectively) both contained a nonconservative mutation at Y478 (histidine and aspartic acid, respectively), immediately N-terminal to the PDL motif (Fig. 3A) and in a position contacted by β-TrCP in other phosphodegrons (12); however, reversion to tyrosine had negligible effect on their NF-κB antagonist activities (data not shown).

HEK293T cells were then cotransfected with NSP1 (wt or ΔC13) and FLAG-β-TrCP, and the level of β-TrCP was assayed 24 h p.t. by quantitative immunoblotting. As shown in Fig. 3C, this group of OSU-like NSP1 proteins did not universally conserve β-TrCP degradation activity, despite all members having a C-terminal PDL motif and potently blocking NF-κB activation. Along with porcine OSU, NSP1 proteins from human D, DS-1, KU, P, and Wa RVA mediated the degradation of β-TrCP. Other human (IAL28, ST3, and WI61) and porcine (Gottfried and YM) RVA NSP1 proteins failed to induce β-TrCP degradation, indicating that this activity was not strictly conserved among NSP1 proteins from either host species. To confirm this finding, transfected cells were treated with the proteasome inhibitor MG132; the level of β-TrCP was rescued in cells that coexpressed OSU NSP1, whereas no effect was observed in cells that coexpressed Gottfried or SA11-4F NSP1 (see Fig. S6 in the supplemental material). Within the panel tested, no residues could be identified that predict the ability of a given NSP1 protein to induce β-TrCP turnover (data not shown). Together, these data suggest that NSP1-mediated degradation of β-TrCP may not be required for NSP1 to inhibit NF-κB activation. This finding is consistent with the action of bovine UK NSP1, which mediates the degradation of IRF3 in a host-cell-specific manner but which can also block IRF3 transcriptional activity without inducing IRF3 degradation (45). Furthermore, EBV LMP1 (39), HIV-1 Vpu (40), and VACV A49 (41) all interact with β-TrCP to block NF-κB activation, without inducing β-TrCP turnover.

The serine residues of the PDL motif are critical for OSU NSP1 function.

Cellular degrons must be phosphorylated on a pair of conserved serine residues to be recognized by β-TrCP (12). All four viral PDL motifs contain this serine pair (Fig. 1D), although only HIV-1 Vpu has been confirmed to undergo phosphorylation, a modification required for its association with β-TrCP (46). The serine residues of the OSU NSP1 PDL motif lie within predicted casein kinase I (CK) I and II sites (Fig. 4A) (47–49). Mutation of the CKII (E486Q), but not CKI (D477N), priming residue almost completely abolished NF-κB antagonism by OSU NSP1 (Fig. 4B). This is consistent with OSU NSP1 being phosphorylated by CKII, which also phosphorylates the Vpu PDL motif (50). Mutation of the serine residues in tandem—to nonphosphorylatable alanine (OSU-AA), phosphomimetic aspartic (OSU-DD) or glutamic (OSU-EE) acid, or phosphorylatable threonine (OSU-TT)—disrupted NF-κB antagonism by OSU NSP1 (Fig. 4C) and indicated a strict requirement for serine (likely, phosphoserine) at these positions.

HEK293T cells were then cotransfected with NSP1 and FLAG-β-TrCP, and the level of β-TrCP was assayed 24 h p.t. by quantitative immunoblotting. As shown in Fig. 4D, OSU and OSU-EE NSP1 were functionally equivalent in this assay, while OSU-AA, -DD, and -TT NSP1 proteins also mediated a modest degree of β-TrCP degradation. This suggested that transient interaction of the NSP1 RING domain with the E2 subunit of the SCF^β-TrCP complex may have been sufficient to stabilize a likely weak interaction between the mutated PDL motif and the β-TrCP subunit. FLAG immunoprecipitation of β-TrCP:NSP1 complexes from cotransfected HEK293T cells demonstrated that the serine residues were critical for this interaction (Fig. 4E). Together, these data support a critical role for the serine residues within the NSP1 PDL motif and suggest that NSP1 function depends on phosphorylation of these residues.

OSU NSP1 targets the phosphodegron-binding pocket of β-TrCP.

β-TrCP is a member of the F-box protein family that provides specificity to SCF-type E3 ubiquitin ligases (51). These proteins contain an N-terminal F-box domain that interfaces with the other subunits of the SCF complex and a C-terminal WD40 domain that mediates recognition of cellular targets. To identify the domain recognized by OSU NSP1, the F-box (β-TrCP-N287) and WD40 (β-TrCP-288C) domains were separately cloned for transient expression (Fig. 5A). HEK293T cells were cotransfected with NSP1 and FLAG-β-TrCP, and the level of β-TrCP was assayed 24 h p.t. by quantitative immunoblotting. As shown in Fig. 5B, NSP1 mediated the degradation of β-TrCP-288C, although less efficiently than it did full-length β-TrCP. This suggested preferential recognition of the full-length protein, perhaps in the context of the SCF^β-TrCP complex. The F-box domain was dispensable for degradation, as the level of β-TrCP-N287 was slightly elevated in the presence of OSU versus OSU-ΔC13 NSP1. Despite this, OSU NSP1 coimmunoprecipitated with both fragments of β-TrCP, albeit less efficiently than with full-length β-TrCP (Fig. 5E). This supports a two-pronged interaction between NSP1 and β-TrCP:RING:E2, bridged by β-TrCP-N287, and PDL motif:WD40, mediated by β-TrCP-288C.

Phosphodegrons bind the WD40 domain of β-TrCP along its top face, by dipping into the central channel of this β-propeller structure (12). Aspartic acid, which almost invariably marks the phosphodegron start, forms the most extensive network of contacts, in particular with residues R510 and Y524 (β-TrCP isoform 1 numbering) (Fig. 5C). Alanine substitution at either position blocks polyubiquitination of IκBα (12), and it was predicted that such a mutation would also protect β-TrCP from NSP1-mediated degradation. HEK293T cells were cotransfected with OSU NSP1 and an R510A mutant of FLAG-β-TrCP or FLAG-β-TrCP-288C, and the level of β-TrCP was assayed 24 h p.t. by quantitative immunoblotting. As shown in Fig. 5D, OSU NSP1 could not mediate the degradation of either β-TrCP-R510A or β-TrCP-288C-R510A, indicating that NSP1 targets the binding pocket used by β-TrCP to interact with cellular phosphodegrons. FLAG-β-TrCP-R510A also failed to coimmunoprecipitate OSU NSP1 (Fig. 5E). Together, these data demonstrate that β-TrCP recognizes cellular phosphodegrons and the NSP1 PDL motif in a similar manner.

The NSP1 PDL motif is a transferrable functional element.

Despite significant sequence variation, all full-length RVA NSP1 proteins are thought to utilize an N-terminal RING domain to interact with a cellular E2 ubiquitin-conjugating enzyme and a C-terminal sequence element to provide binding specificity for a host innate immune target (Fig. 1C) (52). Formation of such a complex is predicted to induce polyubiquitination of the target, followed by its degradation in the proteasome. Sequence variability often masks structural conservation (53), and it was hypothesized that NSP1 conserves its architecture across strains, with one of three C-terminal motifs mediating target specificity (Fig. 1C). To test this possibility, the C terminus of simian SA11-4F NSP1, which specifically targets IRFs (27), was substituted for with that of OSU NSP1. A (D/E)-Φ-(D/E) motif (Φ, aromatic residue) conserved in both proteins was chosen as the site of substitution, in order to keep intact the putative CK sites in OSU NSP1 (Fig. 4A and 6A).

This chimeric NSP1, dubbed 4F-OSU, was cotransfected into HEK293T cells with an NF-κB luciferase reporter and, as shown in Fig. 6B, mediated a 2-fold block in NF-κB activation versus SA11-5S and SA11-4F NSP1. 4F-OSU lacked the potency of OSU NSP1, indicating that regions outside the C-terminal PDL motif likely contribute to NF-κB antagonism by OSU NSP1 and are not fully captured by SA11-4F NSP1. HEK293T cells were then cotransfected with NSP1 and FLAG-β-TrCP, and the level of β-TrCP was assayed 24 h p.t. by quantitative immunoblotting (Fig. 6C). Unlike SA11-4F NSP1, the 4F-OSU chimera was able to induce β-TrCP turnover comparable to that of OSU NSP1. This finding suggests that the PDL motif, when paired with an intact RING domain, is sufficient to target SCF^β-TrCP and that SA11-4F and OSU NSP1 proteins conserve an architecture and mechanism similar enough to permit this transfer of functionality. This was confirmed by a coimmunoprecipitation assay, in which 4F-OSU, but not SA11-4F, NSP1 associated with FLAG-β-TrCP (Fig. 6D). Together, these data suggest that the C-terminal PDL motif is a transferrable functional element, likely due to structural conservation across NSP1 proteins, but that other sequence elements and/or functions of OSU and OSU-like NSP1 proteins contribute to NF-κB antagonism.

Cells and antibodies.

HEK293T cells were grown in Dulbecco’s modified Eagle’s medium (Quality Biological) supplemented with 10% fetal bovine serum (Gibco). Rabbit polyclonal antisera against simian SA11-5S and porcine OSU NSP1 (27) and mouse monoclonal anti-FLAG M2 antibody (F1804; Sigma) were used at a 1:5,000 dilution. Rabbit polyclonal antibody against PCNA (sc-7907; Santa Cruz Biotechnology) was used at a 1:2,000 dilution. IRDye 800CW-conjugated anti-rabbit IgG (926-32211; LI-COR Biosciences) and IRDye 680-conjugated anti-mouse IgG (926-32220; LI-COR Biosciences) goat polyclonal antibodies were used at a 1:20,000 dilution.

Plasmids and sequences.

The vector pCAGGS (58) (a gift from K. W. Boehme) was modified by standard DNA techniques to introduce a T7 promoter and ligation-independent cloning (LIC) site downstream of the CAG promoter (pLIC6). This vector was further modified to introduce an N-terminal FLAG tag (MADYKDDDDKSG) upstream of the LIC site (pLIC6F). To generate LIC sticky ends, pLIC6 or pLIC6F was digested with NsiI and then treated with T4 DNA polymerase in the presence of dTTP. Full-length or truncated β-TrCP or NSP1 ORFs were PCR amplified from previously described vectors (27) with primers having LIC-compatible ends. Mutations were generated by outward PCR of these original plasmids with suitable primers; ORFs were then PCR amplified with LIC-compatible primers. PCR fragments were purified and treated with T4 DNA polymerase in the presence of dATP to generate LIC sticky ends. Treated insert and pLIC6 or pLIC6F were mixed, incubated at 25°C for at least 20 min, and transformed into Escherichia coli strain TOP10. All constructs were verified by sequencing. Endotoxin-free plasmid DNA was prepared using an EndoFree plasmid kit (Qiagen) and diluted to 250 ng/µl in endotoxin-free Tris-EDTA (TE) buffer before use. The β-TrCP (isoform 1) nucleotide sequence is available in GenBank (accession no. NM_033637). Accession numbers for the NSP1 nucleotide sequences used in functional assays are listed in Table S1 in the supplemental material.

Quantitative immunoblotting.

Transfection complexes (94 µl Opti-MEM [Gibco], 500 ng pLIC6-NSP1, 500 ng pLIC6F-β-TrCP, 2 µl Lipofectamine 2000 [Life Technologies]) were prepared according to the manufacturer’s protocols and transferred to wells of a 12-well plate, unless otherwise specified. Nine hundred microliters of a HEK293T cell suspension (1.3 × 10⁶ cells/ml) was added to each well. Cells were collected 24 h p.t., washed once with phosphate-buffered saline (PBS [pH 7.4]), and lysed in a 100-µl solution of 1× radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 50 mM Tris [pH 8.0], 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate), 1× NuPAGE lithium dodecyl sulfate (LDS) sample buffer (Life Technologies), 25 mM dithiothreitol, 1× phosphatase inhibitor cocktail (1 mM each sodium fluoride, sodium orthovanadate, β-glycerophosphate, and sodium pyrophosphate), and 1× Complete EDTA-free protease inhibitor cocktail (Roche). Lysates were sonicated briefly, electrophoresed on 12% bis-Tris gels, and transferred onto nitrocellulose membranes for immunoblotting. Blots were imaged on an Odyssey Classic and quantified using Image Studio Lite v3.1.4 (LI-COR Biosciences). Bands of interest were normalized to the loading control PCNA. Data are presented as means ± standard deviations (SD) from three independent transfections.

NF-κB reporter assay.

Transfection complexes (9.25 µl Opti-MEM, 92 ng pLIC6-NSP1, 30 ng pGL4.32[luc2P/NF-κB-RE/Hygro] [Promega], 3 ng pGL4.74[hRluc/TK] [Promega], 0.25 µl Lipofectamine 2000) were prepared in triplicate according to the manufacturer’s protocols and transferred to wells of a 96-well plate. Ninety microliters of a HEK293T cell suspension (1.1 × 10⁶ cells/ml) was added to each well. At 24 h p.t., cells were stimulated for 4 h with medium ± 25 ng/ml TNF-α (PeproTech). Following stimulation, cells were lysed in 20 µl 1× passive lysis buffer (Promega), and luciferase activity was measured with the Dual-Glo luciferase assay system (Promega), according to the manufacturer’s protocols. Relative luciferase activity (NF-κB stimulation) was calculated by normalizing firefly to Renilla luciferase activity. The data presented (mean ± SD) are from one of three independent experiments.

Immunoprecipitation.

Transfection complexes (1.45 ml Opti-MEM, 7.25 µg pLIC6-NSP1, 7.25 µg pLIC6F-β-TrCP, 29 µl Lipofectamine 2000) were prepared according to the manufacturer’s protocols and transferred to 10-cm dishes of HEK293T cells. At 40 h p.t., cells were treated for 8 h with 20 µM MG132 (Cayman Chemical). Cells were collected, washed twice with PBS, and resuspended in lysis buffer (50 mM Tris [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1× phosphatase inhibitor cocktail). Cell lysates were clarified by centrifugation and incubated overnight with anti-FLAG M2 magnetic beads (M8823; Sigma). Beads were washed three times with wash buffer (50 mM Tris [pH 7.4], 150 mM NaCl), and bound protein was eluted with 0.1 M glycine (pH 3.0).

Phylogenetic analysis.

All available RVA NSP1 ORFs (833 in total) were downloaded from GenBank on 1 February 2013, and partial sequences were discarded. The GenBank to FASTA tool (http://rocaplab.ocean.washington.edu/tools/genbank_to_fasta/) was used to extract the name, country, host species, date of isolation, and sequence for each strain; missing fields were entered manually. Sequences were omitted whose annotation lacked sufficient detail, which left 556 ORFs with lengths ranging from 1,459 to 1,732 bases. Nucleotide sequences were aligned using the ClustalW plugin of Geneious Pro v5.6.5 (Biomatters). To compare evolutionary history, the alignment was translated into the predicted amino acid sequence (length range, 486 to 557 residues) and aligned using a BLOSUM cost matrix. Linear motifs were annotated using the EMBOSS fuzzpro tool (59). MEGA v5.0.5 (60) was used to select the best available substitution model for each alignment, according to the Bayesian information criterion (BIC) ranking. Phylogenetic trees were constructed using the RAxML BlackBox server (61). One thousand rapid bootstrap replicates were performed, with PO-13 (pigeon A4) NSP1 used as the outgroup. Initial trees were visualized using FigTree v1.4.0 (http://tree.bio.ed.ac.uk/software/figtree/). The NSP1 genotype representing each branch was identified based on clustering and confirmed using the RotaC server (62). The A1 branch was further defined into four clearly resolved subgroups, which were supported by bootstrap values greater than 75. Circular cladograms were generated using the EvolView server (63) and were colored according to host species and conservation of a DSGxS motif.

Statistical analysis.

Prism v6.0f (GraphPad Software) was used to perform all statistical analyses. Unless otherwise stated, data were analyzed with one-way analysis of variance (ANOVA) using Dunnett’s multiple comparisons test.