1. Howard Hughes Medical Institute,
2. Department of Pharmacology,
3. Department of Genome Sciences,
4. Institute for Stem Cell and Regenerative Medicine, University of Washington, School of Medicine, Box 357280, Seattle, Washington 98195, USA
5. †Present address: Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario M5S 3M2, Canada
6. *These authors contributed equally to this work

Correspondence to: Randall T. Moon^1,2,4Ning Zheng² Correspondence and requests for materials should be addressed to N.Z. (Email: nzheng@u.washington.edu) or R.T.M. (Email: rtmoon@u.washington.edu).

Distinct from all known cullin adaptors, DDB1 is a large multidomain protein featuring an independent -propeller (BP) domain (BPB) flexibly connected to a clam-shaped double-propeller fold (BPA–BPC)¹⁷. To reveal how DDB1 incorporates into the CUL4A–ROC1 complex and mediates substrate recruitment, we determined the 3.1 Å crystal structure of a DDB1–CUL4A–ROC1 complex bound to the V protein of simian virus 5 (SV5-V), previously known as the VDC complex (Table 1). By docking the STAT proteins to DDB1, the V protein of the paramyxovirus is able to accelerate the ubiquitination and degradation of STAT1, thereby blocking the antiviral response of the host¹⁶. With all the components in their full-length forms, the 240 kDa quaternary protein complex represents a functionally complete form of the E3 apparatus.

Table 1: X-ray refinement statistics

Full table

The DDB1–CUL4A–ROC1–SV5-V complex has an overall elongated shape that is about 170 Å long and 100 Å wide, resembling SCF (SKP1–cullin–F-box-protein)¹⁸ (Fig. 1). As a close paralogue of CUL1, CUL4A adopts a CUL1-like structure with an arc-shaped helical amino-terminal domain (NTD) and a globular carboxy-terminal domain (CTD). Lacking the SKP1/BTB fold, DDB1 devotes its entire BPB domain for CUL4A binding, anchoring at the tip of the NTD of cullin. The BPA–BPC double-propeller fold of DDB1, however, is positioned in the complex facing towards the E2-binding subunit of the E3 ligase, ROC1. As a result, the viral protein—which sits at the opening of the large pocket of the DDB1 double-propeller fold—is about 45 Å away from the RING finger subunit. By comparing this overall architecture of the virally hijacked DDB1–CUL4A–ROC1 complex with SCF¹⁸, we conclude that the cullin-docking function and the substrate-presenting function of DDB1 are separately performed by the BPB domain and the BPA–BPC double-propeller fold, respectively.

Figure 1: Crystal structure of the DDB1–CUL4A–ROC1 ubiquitin ligase complex hijacked by the V protein of simian virus 5.

a, b, Two orthogonal views of the complex structure are shown in ribbon diagram. DDB1, CUL4A, ROC1 and SV5-V are coloured in blue, green, red and magenta, respectively. The BPB domain of DDB1 is coloured in cyan. The zinc atoms in ROC1 and SV5-V are represented as orange spheres. A previously identified STAT2-binding site¹⁷ of the viral protein, as well as the four domains of DDB1, are indicated.

High resolution image and legend (129K)

The complex formation between DDB1 and CUL4A is predominantly mediated through two separate interfaces, both involving the BPB domain of DDB1 (Fig. 2a). At one interface, the top surface of the propeller fold of DDB1-BPB is recognized by the H2 and H5 helices of CUL4A, which are unexpectedly the same ones used by CUL1 to bind SKP1¹⁸ (compare Fig. 2b, c). At the other interface, CUL4A further embraces the BPB domain of DDB1 using an N-terminal extension sequence, which is unique among all cullin family members¹⁷. Running across two blades of DDB1-BPB, this CUL4A N-terminal sequence wraps around the propeller of DDB1 at the peripheral side via a hydrogen-bond network and hydrophobic packing (Fig. 2a, d). The equal importance of these two separate interfaces in DDB1–CUL4A binding is reflected in the strict conservation of most of the interacting residues among CUL4 and DDB1 orthologues (Supplementary Fig. 1). Furthermore, mutations of the DDB1 or CUL4A residues involved in either interface can significantly weaken the complex formation^{13, 17}. Overall, the bipartite binding mode between DDB1 and CUL4A not only strongly supports a specific role of DDB1 in the CUL4A-based E3 machinery, but also demonstrates a surprising variation in cullin–adaptor interactions.

Figure 2: Binding interfaces and structural flexibility of DDB1 on CUL4A.

a, An overall view of the structural elements mediating the binding of the DDB1 BPB domain (cyan) with the CUL4A NTD (green). b, The interface between the top surface of DDB1-BPB (cyan) and the H2–H5 helices of CUL4A (green) viewed from above with participating side chains shown as sticks (grey in DDB1 and green in CUL4A). c, Superposition of the H2–H5 helices of CUL4A (green) and CUL1 (grey) with CUL1-bound SKP1 (blue) shown in the background. The view is the same as that shown in panel b. CUL4A-bound DDB1 is not shown for clarity. d, The interface between the CUL4A N-terminal extension sequence (green) and the peripheral side of the DDB1 BPB domain (blue). e, Superposition analyses of DDB1 alone (right) and DDB1–SV5-V complex (left) structures in the context of the full DDB1–CUL4A–ROC1 complex. The models are docked through the BPB domain and viewed from the same angles as shown in Fig. 1b.

High resolution image and legend (76K)

The crystal structures of DDB1 alone and in complex with SV5-V have previously revealed two distinct structural states of DDB1¹⁷, which show dramatic differences in the relative orientation between the substrate-presenting BPA–BPC double-propeller fold and the cullin-binding BPB domain. In the quaternary-complex crystal, DDB1 adopts yet a new domain arrangement. Intriguingly, superposition analysis of the three structures indicates that all three domain configurations of DDB1 are in fact geometrically allowed in the context of the full DDB1–CUL4A–ROC1 complex (Fig. 2e). The BPA–BPC double-propeller folds of DDB1 in the resulting three models all similarly point towards the E2-binding RING subunit, even though the orientations of the fold itself are correlated by large degrees of rotation (compare Figs 1b and 2e). Because the BPA–BPC double-propeller fold of DDB1 has a very limited number of contacts with the cullin scaffold (Supplementary Fig. 2), we speculate that on binding to different cellular factors, DDB1 might be able to undergo conformational changes, which can modulate the ubiquitin ligase activity of the E3 machinery.

Although a multidomain structure might allow DDB1 to directly recruit substrates to the E3 platform, ubiquitination of several known DDB1–CUL4A substrates has been shown to require additional cellular factors^{7, 19}. Mimicked by the viral V protein, endogenous substrate-specific receptors might be used by DDB1 to expand its substrate repertoire. To further clarify the substrate targeting mechanisms of the DDB1–CUL4A E3 machinery, we used a proteomic approach optimized for isolating CUL4A and DDB1 protein complexes from mammalian cells by tandem-affinity purification²⁰. In these analyses, more than 30 proteins were consistently identified by mass spectrometry, giving rise to a protein-interaction network shown in Fig. 3a.

Figure 3: Identification of a novel family of WD40 proteins as the potential substrate receptors for the DDB1–CUL4A E3 machinery.

a, Mammalian protein interaction network of the DDB1–CUL4A ubiquitin ligases established by tandem-affinity analyses of CUL4A, DDB1, DDB2, WDTC1 and H326. The novel family of DDB1–CUL4A-associated WD40 domain proteins (DCAF proteins) is shown in red. Arrows represent unidirectional (single arrowhead) and bidirectional (double arrowhead) interactions. b, In vitro GST pull-down assay followed by western blot (WB) analysis showing the direct interactions between DDB1 and two DCAF proteins, WDR23 (DCAF11) and H326 (DCAF8), with STAT1 as a negative control. c, Sequence alignment showing the DxR motifs (highlighted in yellow) conserved in two consecutive -propeller blades of the DCAF proteins identified in our studies and two yeast WD40 proteins known to bind the DDB1(Rik1)–Cul4A complex. The two -strands preceding the DxR motifs are indicated. d, Mutagenesis of the arginine residues in the double DxR box of two representative DCAF proteins impairs DDB1 binding. Interactions between DDB1 and the wild-type and mutant DCAF proteins were assessed by co-affinity purification (AP) using streptavidin (DDB1 pull down) and western blot (WB) analysis using anti-Flag antibodies (DCAF proteins) (middle panel, lanes 1–10). DDB1 purification efficiency and the expression of the DCAF proteins were monitored as shown in the top and bottom panels, respectively. Flag–GFP was used as a negative control. e, Competition between SV5-V and WDR23 (DCAF11) for DDB1 binding. The experiments shown in panel b were repeated with increasing amounts of purified SV5-V protein. A molar ratio of 1:2 (DDB1:SV5-V) was reached in the final lane.

High resolution image and legend (69K)

Notably, besides several known cullin regulators and DDB1-binding proteins, numerous novel protein factors were identified in both the CUL4A and DDB1 complexes. In particular, a group of 16 proteins, including DDB2 and CSA, clearly stands out by sharing a predicted WD40 domain. The consistent presence and abundance of these WD40 proteins in the complexes strongly suggest that they represent bona fide members of the human DDB1–CUL4A E3 machinery (Supplementary Tables 2 and 3). It is also of note that reciprocal tandem-affinity pull-down analyses with DDB2 and two randomly selected WD40 proteins (H326 and WDTC1) confirmed that they are individually associated with the DDB1–CUL4A complex (Fig. 3a). Because many of these WD40 proteins do not have any known function other than DDB1–CUL4A binding, we propose to name them the DCAF (DDB1–CUL4A-associated factor) proteins (Supplementary Table 2).

As DDB2 and CSA have already been demonstrated necessary for DDB1–CUL4A-mediated ubiquitination of certain substrates, we speculated that the DCAF proteins we have identified might represent a novel family of proteins serving as the substrate-recruiting module for the E3 machinery. Using co-affinity purification experiments, we first confirmed our mass spectrometry results by showing the in vivo interactions of DDB1 with a panel of randomly selected DCAF proteins (Supplementary Fig. 3). Given that both DDB2 and CSA have been reported to directly bind DDB1, we next examined the direct interactions between DDB1 and two DCAF proteins. As shown in Fig. 3b, bacterially produced and purified glutathione S-transferase (GST)-tagged WDR23 (DCAF11) and H326 (DCAF8) were indeed able to interact with the purified recombinant DDB1 protein in an in vitro GST pull-down experiment.

If the DCAF proteins represent a family of substrate receptors for DDB1–CUL4A, then they might share a common DDB1-docking motif, analogous to the substrate receptors of other cullin E3 complexes³. As no recognizable common sequence other than the WD40 repeats was found among the DCAF proteins, we postulated that this motif might consist of amino acids exposed on the surface of their WD40 domains. Structure-based sequence analysis of the DCAF proteins revealed the presence of two conserved DxR motifs within the WD40 domain of almost all members (Fig. 3c). These two motifs are separately located at the end of strand c of two consecutive blades of the WD40 fold, and we propose to name these two motifs the 'double DxR box'. On the basis of several known WD40 domain structures^{21, 22}, the conserved Asp and Arg residues in the double DxR box are predicted to be solvent-exposed, lining at the bottom surface of the WD40 propeller fold (Supplementary Fig. 4).

Although other unique and hidden features of the DCAF proteins might contribute to their specific interactions with DDB1, several lines of evidence suggest that the double DxR box is one of the main determinants for their binding to the CUL4A adaptor. First, a naturally occurring mutation (R273H) found in a subset of xeroderma pigmentosum group E (XPE) patients has been mapped to an arginine within the double DxR box motif present in DDB2. As previously reported²³ and confirmed by us (Fig. 3d, lane 2), this mutation abolishes the binding of DDB2 to DDB1. Second, our mutagenesis analysis of two representative DCAF proteins, WDR23 (DCAF11) and H326 (DCAF8), shows that both arginine residues in the double DxR box contribute to DDB1 binding. As shown in Fig. 3d, although single mutation of either of the two arginine residues in the double DxR box of the two DCAF proteins markedly weakens DDB1 binding (lanes 4, 5 and 8, 9), the double mutations completely ablate complex formation (lanes 6 and 10). Third, the double DxR box in two consecutive WD40 blades are also found in two fission yeast WD40 proteins, Cdt2 and Dos1 (also known as Clr8) (Fig. 3c), which are known to function with yeast DDB1–Cul4, and its close relative Rik1–Cul4, E3 complexes^{19, 24, 25}. The double DxR motif, therefore, probably underlies an evolutionarily conserved mechanism for docking the substrate-recruiting module to the DDB1–CUL4A ubiquitin ligase machinery.

The presence of a conserved motif shared among the DCAF proteins predicts a common docking site on DDB1. The overall architecture of the virally hijacked DDB1–CUL4A–ROC1 complex strongly suggests that such a site might be located on the BPA–BPC double-propeller fold of DDB1, probably occupied by SV5-V, even though the viral protein does not have a WD40 domain. In fact, in an in vitro binding assay, association of DDB1 and two representative DCAF proteins, WDR23 (DCAF11) and WDTC1 (DCAF9) (data not shown), can be stoichiometrically competed off by increasing amounts of purified SV5-V protein (Fig. 3e). This in vitro result is consistent with a previous report showing that the same viral protein can compete with DDB2 for binding with DDB1²⁶. It is interesting to note that, in order to anchor itself to DDB1, the viral protein not only binds to the surface of the BPA–BPC double-propeller fold of DDB1, but also inserts an entire helix into the large pocket of the DDB1 double-propeller fold. Although the size and secondary structure elements of a regular WD40 domain do not allow it to do so, it remains possible that some DCAF proteins might use sequence regions outside their WD40 domains to bind to the pocket of the DDB1 double-propeller fold, thereby further securing their associations with the CUL4A adaptor. Such an assembly mode could be used by the few DCAF proteins that lack a perfect double DxR box motif. Exactly how the DCAF proteins are docked to DDB1 awaits future structural studies.

As gene expression varies in different cell types and under different cellular states, the DCAF proteins identified in our studies are not expected to represent the complete family of substrate receptors of the DDB1–CUL4A E3 in human cells. Having established the structural and molecular basis of the assembly of the E3 machinery, our studies pave the road for a future understanding of the functions of the DDB1–CUL4A complex in eukaryotic biology and in human diseases.

References

1. Hershko, A. & Ciechanover, A. The ubiquitin system. Annu. Rev. Biochem. 67, 425–479 (1998) | Article | PubMed | ISI | ChemPort |
2. Pickart, C. M. Mechanisms underlying ubiquitination. Annu. Rev. Biochem. 70, 503–533 (2001) | Article | PubMed | ISI | ChemPort |
3. Petroski, M. D. & Deshaies, R. J. Function and regulation of cullin-RING ubiquitin ligases. Nature Rev. Mol. Cell Biol. 6, 9–20 (2005) | Article |
4. Nag, A., Bondar, T., Shiv, S. & Raychaudhuri, P. The xeroderma pigmentosum group E gene product DDB2 is a specific target of cullin 4A in mammalian cells. Mol. Cell. Biol. 21, 6738–6747 (2001) | Article | PubMed | ISI | ChemPort |
5. Chen, X., Zhang, Y., Douglas, L. & Zhou, P. UV-damaged DNA-binding proteins are targets of CUL-4A-mediated ubiquitination and degradation. J. Biol. Chem. 276, 48175–48182 (2001) | PubMed | ISI | ChemPort |
6. Sugasawa, K. et al. UV-induced ubiquitylation of XPC protein mediated by UV-DDB-ubiquitin ligase complex. Cell 121, 387–400 (2005) | Article | PubMed | ISI | ChemPort |
7. Groisman, R. et al. CSA-dependent degradation of CSB by the ubiquitin–proteasome pathway establishes a link between complementation factors of the Cockayne syndrome. Genes Dev. 20, 1429–1434 (2006) | Article | PubMed | ChemPort |
8. Groisman, R. et al. The ubiquitin ligase activity in the DDB2 and CSA complexes is differentially regulated by the COP9 signalosome in response to DNA damage. Cell 113, 357–367 (2003) | Article | PubMed | ISI | ChemPort |
9. Kapetanaki, M. G. et al. The DDB1–CUL4A^DDB2 ubiquitin ligase is deficient in xeroderma pigmentosum group E and targets histone H2A at UV-damaged DNA sites. Proc. Natl Acad. Sci. USA 103, 2588–2593 (2006) | Article | PubMed | ChemPort |
10. Wang, H. et al. Histone H3 and H4 ubiquitylation by the CUL4–DDB–ROC1 ubiquitin ligase facilitates cellular response to DNA damage. Mol. Cell 22, 383–394 (2006) | Article | PubMed | ChemPort |
11. Liu, C. et al. Cop9/signalosome subunits and Pcu4 regulate ribonucleotide reductase by both checkpoint-dependent and -independent mechanisms. Genes Dev. 17, 1130–1140 (2003) | Article | PubMed | ISI | ChemPort |
12. Higa, L. A., Mihaylov, I. S., Banks, D. P., Zheng, J. & Zhang, H. Radiation-mediated proteolysis of CDT1 by CUL4–ROC1 and CSN complexes constitutes a new checkpoint. Nature Cell Biol. 5, 1008–1015 (2003) | Article |
13. Hu, J., McCall, C. M., Ohta, T. & Xiong, Y. Targeted ubiquitination of CDT1 by the DDB1–CUL4A–ROC1 ligase in response to DNA damage. Nature Cell Biol. 6, 1003–1009 (2004) | Article |
14. Bondar, T., Ponomarev, A. & Raychaudhuri, P. Ddb1 is required for the proteolysis of the Schizosaccharomyces pombe replication inhibitor Spd1 during S phase and after DNA damage. J. Biol. Chem. 279, 9937–9943 (2004) | PubMed | ChemPort |
15. Wertz, I. E. et al. Human De-etiolated-1 regulates c-Jun by assembling a CUL4A ubiquitin ligase. Science 303, 1371–1374 (2004) | Article | PubMed | ISI | ChemPort |
16. Horvath, C. M. Weapons of STAT destruction. Interferon evasion by paramyxovirus V protein. Eur. J. Biochem. 271, 4621–4628 (2004) | Article | PubMed | ChemPort |
17. Li, T., Chen, X., Garbutt, K. C., Zhou, P. & Zheng, N. Structure of DDB1 in complex with a paramyxovirus V protein: viral hijack of a propeller cluster in ubiquitin ligase. Cell 124, 105–117 (2006) | Article | PubMed | ChemPort |
18. Zheng, N. et al. Structure of the Cul1–Rbx1–Skp1–F box^Skp2 SCF ubiquitin ligase complex. Nature 416, 703–709 (2002) | Article | PubMed | ISI | ChemPort |
19. Liu, C. et al. Transactivation of Schizosaccharomyces pombe cdt2⁺ stimulates a Pcu4–Ddb1–CSN ubiquitin ligase. EMBO J. 24, 3940–3951 (2005) | Article | PubMed | ChemPort |
20. Angers, S. et al. The KLHL12–Cullin-3 ubiquitin ligase negatively regulates the Wnt–-catenin pathway by targeting Dishevelled for degradation. Nature Cell Biol. 8, 348–357 (2006) | Article |
21. Lambright, D. G. et al. The 2.0 Å crystal structure of a heterotrimeric G protein. Nature 379, 311–319 (1996) | Article | PubMed | ChemPort |
22. Wall, M. A. et al. The structure of the G protein heterotrimer G_i112. Cell 83, 1047–1058 (1995) | Article | PubMed | ISI | ChemPort |
23. Shiyanov, P. et al. The naturally occurring mutants of DDB are impaired in stimulating nuclear import of the p125 subunit and E2F1-activated transcription. Mol. Cell. Biol. 19, 4935–4943 (1999) | PubMed | ISI | ChemPort |
24. Li, F. et al. Two novel proteins, dos1 and dos2, interact with rik1 to regulate heterochromatic RNA interference and histone modification. Curr. Biol. 15, 1448–1457 (2005) | Article | PubMed | ChemPort |
25. Thon, G. et al. The Clr7 and Clr8 directionality factors and the Pcu4 cullin mediate heterochromatin formation in the fission yeast Schizosaccharomyces pombe. Genetics 171, 1583–1595 (2005) | Article | PubMed | ChemPort |
26. Leupin, O., Bontron, S. & Strubin, M. Hepatitis B virus X protein and simian virus 5 V protein exhibit similar UV-DDB1 binding properties to mediate distinct activities. J. Virol. 77, 6274–6283 (2003) | Article | PubMed | ChemPort |

Supplementary Information

Supplementary information accompanies this paper.

Acknowledgements

We thank the beamline staff of the Advanced Light Source at Berkeley for help with data collection, P. Zhou, W. Xu, J. Beavo, Z. Gao and members of the Zheng and Moon laboratories for discussions and help. We also thank J. W. Harper and Y. Xiong for communicating unpublished results. This work is supported by grants from the National Institutes of Health (to N.Z. and M.J.M.), the Pew Scholar Program (N.Z.), and the Howard Hughes Medical Institute (S.A. and R.T.M.). Author Contributions S.A. and T.L. contributed equally to this work.

Competing interests statement

The authors declare no competing financial interests.

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.