Insight into Viral Hijacking of CRL4 Ubiquitin Ligase through Structural Analysis of the pUL145-DDB1 Complex

Abstract

Viruses evolve mechanisms to exploit cellular pathways that increase viral fitness, e.g., enhance viral replication or evade the host cell immune response. The ubiquitin-proteosome system, a fundamental pathway-regulating protein fate in eukaryotes, is hijacked by all seven classes of viruses. Members of the Cullin-RING family of ubiquitin (Ub) ligases are frequently co-opted by divergent viruses because they can target a broad array of substrates by forming multisubunit assemblies comprised of a variety of adapters and substrate receptors. For example, the linker subunit DDB1 in the cullin 4-RING (CRL4)-DDB1 Ub ligase (CRL4^DDB1) interacts with an H-box motif found in several unrelated viral proteins, including the V protein of simian virus 5 (SV5-V), the HBx protein of hepatitis B virus (HBV), and the recently identified pUL145 protein of human cytomegalovirus (HCMV). In HCMV-infected cells, pUL145 repurposes CRL4^DDB1 to target STAT2, a protein vital to the antiviral immune response. However, the details of how these divergent viral sequences hijack DDB1 is not well understood. Here, we use a combination of binding assays, X-ray crystallography, alanine scanning, cell-based assays, and computational analysis to reveal that viral H-box motifs appear to bind to DDB1 with a higher affinity than the H-box motifs from host proteins DCAF1 and DDB2. This analysis reveals that viruses maintain native hot-spot residues in the H-box motif of host DCAFs and also acquire favorable interactions at neighboring residues within the H-box. Overall, these studies reveal how viruses evolve strategies to produce high-affinity binding and quality interactions with DDB1 to repurpose its Ub ligase machinery. IMPORTANCE Many different viruses modulate the protein machinery required for ubiquitination to enhance viral fitness. Specifically, several viruses hijack the cullin-RING ligase CRL4^DDB1 to degrade host resistance factors. Human cytomegalovirus (HCMV) encodes pUL145 that redirects CRL4^DDB1 to evade the immune system through the targeted degradation of the antiviral immune response protein STAT2. However, it is unclear why several viruses bind specific surfaces on ubiquitin ligases to repurpose their activity. We demonstrate that viruses have optimized H-box motifs that bind DDB1 with higher affinity than the H-box of native binders. For viral H-boxes, native interactions are maintained, but additional interactions that are absent in host cell H-boxes are formed, indicating that rewiring CRL4^DDB1 creates a selective advantage for the virus. The DDB1-pUL145 peptide structure reveals that water-mediated interactions are critical to the higher affinity. Together, our data present an interesting example of how viral evolution can exploit a weakness in the ubiquitination machinery.

Keywords: cullin-RING; human cytomegalovirus; innate immunity; ubiquitin ligase; viral hijacking.

FIG 1

Viral H-boxes bind tighter than the human H-boxes used in this study to DDB1. (A) Structure of DDB1 bound to the H-box peptide (green) from HBx (Protein Data Bank code 3I7H) (31). DDB1 consists of three β-propellers (multicolored), but the H-box peptide predominantly interacts with β-propeller C (BPC). (B) The H-box consensus sequence was created using MegAlign from Lasergene. The amino acids are highlighted according to their physicochemical properties. (C) Representative fluorescence polarization (FP) data indicating the binding of H-box peptides from pUL145 or DCAF1 to DDB1. The experiments were performed in at least triplicate and normalized to wild type (WT) as 100%. (D) Table of the affinities (K_D) and standard deviations for the tested H-boxes against DDB1 based on FP data in panel C. BPA, β-propeller A; BPB, β-propeller B; BPC, β-propeller C; CTD, C-terminal domain.

FIG 2

Sequence requirements for H-box binding affinity for DDB1. (A) Table of K_D values and standard deviations for a panel of pUL145 H-box peptides in which the indicated residues were substituted to Ala. The colors of the labels correspond to the magnitude of the fold reduction in binding affinity when the native residue is replaced with Ala (red indicates >100-fold, orange indicates >10-fold, and gray indicates <10-fold). (B) Bar graph of the data from panel A comparing the binding affinities for the alanine-substituted pUL145 peptides. The y axis represents the reduction (fold change) of binding compared to the wild-type peptide. The colors are the same as in panel A. The error bars represent standard deviations. (C) Sequence alignment of known H-boxes that bind to DDB1 with conserved residues shaded. (D) WebLogos of the H-box sequence based on the sequence alignment from panel C colored by their physicochemical properties.

FIG 3

Crystal structures of the H-box peptide derived from pUL145 bound to DDB1. (A) Close-up view of the pUL145 H-box peptide (yellow) in complex with DDB1 (gray). The 2F_o − F_c electron density map is contoured at 1σ. (B) Superposition of the H-box peptides of HBx (black, PDB ID 3I7H), DDB2 (blue, PDB ID 3I7L), DCAF1 (orange, PDB ID 5JK7), and pUL145 (yellow, PDB ID 7UKN) bound to DDB1 (30, 31). (C to K) Close-up views of the key interactions between pUL145 (yellow), DDB1 (gray), and water (gray sphere) based on the binding affinities of the alanine-substituted pUL145 H-box peptides. The 2F_o − F_c electron density map is contoured at 1σ.

FIG 4

Functional epitope of DDB1 for binding H-box sequences. (A) Bar graph of the data shown in Table 2 comparing the binding affinities for the alanine-substituted DDB1 variants. The y-axis represents the reduction (fold change) in binding compared to wild-type DDB1. The colors of the bars correspond to the extent of the fold change in binding affinity because of the alanine substitution. The error bars represent the standard deviations. (B) Representative fits of FP data to determine the binding affinity of DDB1 wild-type and the alanine-substituted variants. Experiments were performed in at least triplicate and normalized to WT as 100%. (C) Surface view representation of DDB1 bound to pUL145 (atoms shown as spheres). DDB1 residues mutated in this study are colored as a gradient to indicate the increasing reduction in binding affinity compared to wild-type DDB1. (D, E) DDB1 and pUL145 variants reduce the interaction and disturb STAT2 degradation, monitored by coimmunoprecipitation (Co-IP) and Western blotting. HEK293T cells were transiently transfected with FLAG-tagged DDB1 and MYC-tagged pUL145 or GFP as a negative control. Cell extracts were prepared, and the pUL145-DDB1 complex was coimmunoprecipitated using an antibody to either the FLAG (D) or MYC (E) tag (n ≥ 3). The STAT2 levels were also monitored by immunoblotting the inputs of the co-IP (n = 7). (F) STAT2 levels in HEK293T cells transfected with DDB1 and pUL145 wild-type and variants from panels D and E. STAT2 levels were calculated as ratios compared to the loading control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and normalized to the levels of STAT2 in control cells. The error bars represent standard error. Statistical significance was assessed by one-way analysis of variance (ANOVA). **, P ≤ 0.01; ****, P ≤ 0.0001 (n = 7 biological replicates). N.D., not determined.

FIG 5

Viral sequences tend to form additional interactions at the C terminus of the H-box sequence to facilitate tighter binding. (A) Table of H-box-DDB1 Rosetta energies of the lowest scoring structure. The total energy is the overall Rosetta energy of the structure, and ΔΔG represents the computational binding energy. It is calculated by subtracting the Rosetta energy for the complex by the energy of the isolated DDB1 and H-box structures. The affinities (K_D) were determined experimentally from Fig. 1 (B to E) Rosetta pair energies of the H-box peptide residue interactions with DDB1 using pUL145 (B), HBx (C), DCAF1 (D), and DDB2 (E). (F, G) Structural comparison of the Rosetta models (green) and crystal structure of DDB1 (gray) bound to the pUL145 (yellow) or DCAF1 (orange) H-box peptide demonstrate that Rosetta formed alternative favorable interactions for R35 and D36 where bridging water molecules were identified in the crystal structure.