Targeted ubiquitination of CDT1 by the DDB1–CUL4A–ROC1 ligase in response to DNA damage

L E T T E R S

NATURE CELL BIOLOGY VOLUME 6 | NUMBER 10 | OCTOBER 2004 1003

Targeted ubiquitination of CDT1 by theDDB1–CUL4A–ROC1 ligase in response to DNA damageJian Hu1, Chad M. McCall1, Tomohiko Ohta1,2 and Yue Xiong1,3

Cullins assemble a potentially large number of ubiquitinligases by binding to the RING protein ROC1 to catalysepolyubiquitination, as well as binding to various specificityfactors to recruit substrates1–4. The Cul4A gene is amplified inhuman breast and liver cancers, and loss-of-function of Cul4results in the accumulation of the replication licensing factorCDT1 in Caenorhabditis elegans embryos and ultraviolet (UV)-irradiated human cells. Here, we report that human UV-damaged DNA-binding protein DDB1 associatesstoichiometrically with CUL4A in vivo, and binds to an amino-terminal region in CUL4A in a manner analogous to SKP1,SOCS and BTB binding to CUL1, CUL2 and CUL3, respectively.As with SKP1–CUL1, the DDB1–CUL4A association isnegatively regulated by the cullin-associated and neddylation-dissociated protein, CAND1. Recombinant DDB1 and CDT1bind directly to each other in vitro, and ectopically expressedDDB1 bridges CDT1 to CUL4A in vivo. Silencing DDB1prevented UV-induced rapid CDT1 degradation in vivo andCUL4A-mediated CDT1 ubiquitination in vitro. We suggest thatDDB1 targets CDT1 for ubiquitination by a CUL4A-dependentubiquitin ligase, CDL4ADDB1, in response to UV irradiation.

E3 ubiquitin ligases contain two distinct activities: catalysis of isopep-tide bond formation and recruitment of substrates to this catalyticactivity1. The cullins are a family of evolutionarily conserved proteinsthat assemble a large family of cullin-dependent E3 ligases (CDL). Thehuman cullin family includes six closely related proteins (CUL1,CUL2, CUL3, CUL4A, CUL4B and CUL5) and three distantly relatedproteins (CUL7, PARC and APC2). All cullins contain a conserved car-boxy-terminal domain of approximately 100 amino acids, which bindsto a small RING finger protein: ROC1 (RING of Cullins, also knownas Hrt1 and Rbx1), ROC2 or APC11 (refs 2, 3). ROC proteins activatean E2 ubiquitin-conjugating enzyme to catalyse polyubiquitinationthrough their RING finger, and bind with cullins through an N-termi-nal sequence flanking the RING domain to recruit the catalytic func-tion of the E2 to cullins4. A unique feature of CDLs is that the cullins,through a conserved N-terminal domain, interact with a specificity

factor — either directly or through an adaptor protein or adaptor com-plex — to recruit specific substrates, rather than binding to substratesdirectly as most other ligases do. The SKP1 adaptor bridges an F-boxprotein to CUL1-dependent ligases5–8, a heterodimeric elongins B andC complex brings SOCS (suppressor of cytokine signalling) proteins toCUL2-dependent, and possibly CUL5-dependent, ligases9–12, whereasCUL3 binds directly to the BTB domain (Broad-Complex C (BR-C),Tramtrack (Ttk) and Bric-a-brac)13–16. The presence of numerous sub-strate specificity factors — mammals express more than 60 F-box, 40SOCS and 200 BTB proteins — suggests that individual cullins mayassemble into multiple E3 ligase complexes.

CUL4 has been highly conserved during evolution, with closelyrelated orthologues present in fission yeast, plants, worms, flies andmammals. Loss-of-function of the Cul4 gene results in elongatedcells with decondensed chromosomes in fission yeast17 and massiveDNA re-replication in C. elegans embryos18, and deletion of Cul4Aresults in early embryonic lethality in mice19. Together, these resultsindicate an essential function for CUL4 (or CUL4A) in cell-cyclecontrol, genomic stability and development. Human Cul4A isgenomically amplified or overexpressed in a portion of breast andliver tumours20,21, suggesting an oncogenic function of CUL4A.Degradation of several substrates has been linked to the function ofCUL4, including CDT1 (refs 18 and 22), HOXA9 (ref. 23), STAT1and STAT3 (refs 24–26), and c-Jun27. The substrate-targeting mecha-nism of CUL4-dependent ligases is not well understood. It wasrecently proposed that DDB1 participates in targeting the substratec-Jun to the CUL4–ROC1 ligase by interacting with an undefinedmotif present in the human De-etiolated-1 (DET1) protein27.

Our initial investigation into the function of DDB1 in CUL4–ROC1ligases was inspired by the co-purification of DDB1 with ROC1 fromHeLa cells (Fig. 1a). As previously reported, transfection-immunopre-cipitation assays demonstrated that when overexpressed, DDB1 read-ily associates with both CUL4A and CUL4B (ref. 28 and data notshown). To view quantitatively the DDB1–CUL4A association in vivo,we immunopurified an endogenous CUL4A complex from BT474breast cancer cells, which express a high level of CUL4A, and visual-ized the CUL4A complex by silver staining. CUL4A associated nearly

1Department of Biochemistry and Biophysics, Lineberger Comprehensive Cancer Center, and Program in Molecular Biology and Biotechnology, University of NorthCarolina at Chapel Hill, NC 27599-7295, USA. 2Present address: Division of Breast and Endocrine Surgery, St Marianna University School of Medicine, Kawasaki216-8511, Japan. 3Correspondence should be addressed to Y.X. (e-mail: [email protected])

Published online: 26 September 2004, DOI: 10.1038/ncb1172

print ncb1172 14/9/04 3:22 PM Page 1003

© 2004 Nature Publishing Group


L E T T E R S

1004 NATURE CELL BIOLOGY VOLUME 6 | NUMBER 10 | OCTOBER 2004

stoichiometrically with DDB1, and associated abundantly with sub-units of the COP9 signalosome (Fig. 1b), indicating that DDB1 is amajor partner of CUL4A. In the same purification, as in ROC1immunopurification from HeLa cells, we did not detect any DDB2.This indicates that DDB2, if associated with DDB1–CUL4A in BT474cells, is present as a minor component. To gain mechanistic insightinto the DDB1–CUL4A interaction, we took advantage of the definedfunctions of different domains in cullins, and determined whatsequence in CUL4A is involved in binding with DDB1. Deletion of theN-terminal 97 residues in CUL4A (∆N97) completely abolished DDB1binding. Conversely, a fragment of CUL4A retaining the N-terminal406 residues (N406) exhibited a DDB1-binding activity that is similarto that of wild-type CUL4A (Fig. 1c). Hence, an N-terminal sequencein CUL4A is both necessary and sufficient for binding to DDB1. Twohydrophobic helical surfaces in the N-terminal tip of CUL1, H2 andH5 pack with hydrophobic and polar residues from SKP1 to form alarge interface. The N-terminal regions of other cullins form similar

H2 and H5 helices, which contain residues that are invariably con-served in orthologues, but are different in paralogues8. The substitu-tion of residues in the H2 (from Leu 86-Tyr 87-Gln 88-Ala 89-Val 90to Ala 86-Ala 87-Ala 88-Ala 89-Ala 90) or H5 (from Trp 139-Gln 140-Asp 141-His 142 to Ala 139-Ala 140-Asp 141-Ala 142) helices ofCUL4A substantially reduced DDB1–CUL4A binding (Fig. 1d).Together, these results demonstrate that DDB1 binds to CUL4A in amanner similar to SKP1–CUL1 binding.

Mutations in either the H2 or H5 helices also abolished theCAND1–CUL4A association (Fig. 1d), suggesting that CAND1 andDDB1 may competitively bind to an overlapping sequence in CUL4A.Supporting this idea, no CAND1 was detected in the DDB1 immuno-complex; and conversely no DDB1 was detected in the CAND1immunocomplex, under conditions where CUL4A was readilydetected in both DDB1 and CAND1 complexes (Fig. 2a). Notably,DDB1 associated with both unneddylated and neddylated forms ofCUL4A, whereas CAND1 selectively associated with unneddylated

c

Myc3−Cul4A

IP: Anti-Myc

N96

N20

2

N40

6

N53

1

N64

0

∆N97

∆N20

3

∆N40

7

∆N53

2

∆N64

1

WT

97

66

45

30

97

DDB1

Myc3−Cul4A

IgG(H)D

DB

1M

ycD

DB

1

Western

IgG(L)

DDB1

b

97DDB1

CUL4A

CSN1

CSN2CSN4CSN3

67

43

CU

L4A

MW

M

Coomassieblue staining

DDB1

CAND1

Myc3−Cul4A

ROC1

97

97

66

14

Myc3−Cul4A

IP:87654321

DD

B1

CA

ND

1M

ycR

OC

1

Anti-Myc

Wes

tern

Myc3−Cul4ANedd8

H2−

H5

H5

H2

WT

H2−

H5

H5

H2

WT

d

a

CUL1,2,3,4

−

ROC1

97

66

45

30

DDB1

ROC1

+Peptide

20

14

Silver staining

1 2 3 4 5 6 7 8 9 10 11

Mr(K)

Mr(K)

Mr(K)

Mr(K)

Figure 1 DDB1 binds to an N-terminal region in CUL4A. (a) DDB1associates with ROC1 in vivo. Clarified lysates derived from HeLa cells wereimmunoprecipitated with an anti-ROC1 antibody and resolved on anSDS–PAGE gel before silver staining. Specific ROC1-interacting proteinswere identified by being competed off after addition of a molar excess ofantigen peptide. The identities of proteins determined by massspectrometric analysis are indicated. (b) An anti-CUL4A immunoprecipitatederived from 150 mg of clarified lysate from logarithmically growing BT474cells was resolved by SDS–PAGE before Coomassie blue staining. Specific

bands were excised and subjected to mass spectrometric analysis; theidentified proteins are indicated. (c, d) DDB1 binds to an N-terminal regionin CUL4A. Various deletion mutants and site-specific mutants in the H2, H5or both H2 and H5 helices of CUL4A were ectopically expressed in 293Tcells by transient transfection, and assayed for binding with DDB1 byimmunoprecipitation and western blot analysis. Asterisks indicate individualdeletion mutants of myc-tagged CUL4A (c, middle). The concentration ofendogenous DDB1 in each transfected cell population was determined bydirect immunoblotting (c, bottom).

print ncb1172 14/9/04 3:23 PM Page 1004



L E T T E R S


CUL4A. Suppression of DDB1 expression by RNA interference (RNAi)considerably increased the CAND1–CUL4A association, whereas ithad little effect on the steady state level of either protein (Fig. 2b).Conversely, knocking down CAND1 by RNAi did not affect the steadystate levels of CUL4A and DDB1, but did increase the CUL4A–DDB1association (Fig. 2c). The competitive interaction between CAND1and DDB1 for binding with CUL4A is very similar to the competitiveinteraction between CAND1 and SKP1 for binding with CUL1(refs 29, 30), providing additional evidence that DDB1 binds toCUL4A in a similar manner to SKP1–CUL1 binding.

To test whether DDB1 might function as a substrate-recruiting fac-tor for CUL4A–ROC1 ligases, we examined the possibility that DDB1may target CDT1 for CUL4-mediated ubiquitination. TheDDB1–CDT1 association can be detected readily and reciprocally bycoupled transfection and co-immunoprecipitation assays (Fig. 3a).Although DDB1 binds to both fast- and slow-migrating forms of

CDT1, it preferentially associated with a slow-migrating form ofCDT1 (Fig. 3a), which may correspond to phosphorylated CDT1(ref. 31). To determine whether CDT1 and DDB1 interact with eachother directly, or whether this interaction requires an additional factor,and knowing that budding yeast cells do not have apparent homo-logues of CUL4 and DDB1, we examined the interactions of CDT1with these two proteins by a directed yeast two-hybrid assay. WhereasDDB1 positively interacted with both CUL4A and CDT1, noCDT1–CUL4A interaction was detected (Fig. 3b). To confirm thedirect binding between CDT1 and DDB1, the CDT1 immunocomplexfrom cultured mammalian cells ectopically overexpressing CDT1 andDDB1 was resolved on an SDS–PAGE gel and visualized by silver stain-ing. DDB1 co-immunoprecipitated with CDT1 nearly stoichiometri-cally, whereas no other protein shown in the gel was abundant enoughto bridge CDT1–DDB1 binding (Fig. 3e). In repeated assays, anendogenous CDT1–DDB1 complex could not be detected. This maybe, in part, because CDT1 is physically separated from DDB1, orbecause the DDB1–CUL4–ROC1 ligase rapidly destroys CDT1 once itbinds to DDB1. Finally, in an in vitro binding assay using purified pro-teins, DDB1 stoichiometrically bound to glutathione S-transferase(GST)-tagged CDT1, but not with GST alone (Fig. 3f). We thereforeconclude that DDB1 directly binds to CDT1.

When overexpressed in cultured human cells, CDT1 and CUL4Adid not appreciably associate with each other, but a CDT1–CUL4Acomplex became readily detectable when these two proteins were co-expressed with DDB1 (Fig. 3c). We postulate that the CDT1–CUL4Aassociation bridged by endogenous DDB1 may be rapidly dissociatedafter ubiquitination and thus escape detection. However, the high con-centrations of the CDT1–DDB1–CUL4A ternary complex assembledin cells overexpressing all three proteins might saturate endogenouscomponents (for example, ROC1) required for rapid ubiquitinationand degradation of CDT1. Deletion analysis indicated that bothCUL4A and CDT1 bind to a central region in DDB1, between residues380 and 956 (Fig. 3d). The N-terminal 768 residues (N768) are suffi-cient for binding with CUL4A, but almost completely abrogate bind-ing with CDT1, indicating that CUL4A and CDT1 may bind todifferent sequences in DDB1. Together, these results indicate thatDDB1 can directly interact with both CUL4A and CDT1 and that it isa rate-limiting factor to bridge CDT1 to CUL4A in vivo.

Consistent with a recent report that CDT1 is degraded by theCUL4–ROC1 ligases after DNA damage22, UV irradiation caused rapiddegradation of CDT1 in a proteasome-dependent manner (Fig. 4a).Knocking down expression of both CUL4A and CUL4B, but not eitheralone (see Supplementary Information, Fig. S1a) or two other cullins,CUL1 and CUL3 (see Supplementary Information, Fig. S1b), preventedUV-induced CDT1 degradation, but had no detectable effect on CDT1degradation in the absence of DNA damage (Fig. 4b). Remarkably,knocking down DDB1 effectively prevented UV-induced CDT1 degra-dation, but had no discernible effect on the concentration of CDT1 innon-irradiated cells (Fig. 4b; also see Supplementary Information,Figs S1a and c). In addition, knocking down DDB2 expression had noappreciable effect on the concentration of CDT1 in the presence orabsence of DNA damage (Fig. 4b). We noted that DDB2 itself wasdecreased to a nearly undetectable level after UV irradiation (Fig. 4b);however, the significance and mechanism underlying the UV-inducedDDB2 decrease remains to be determined.

The CUL4A immunocomplex efficiently catalysed polyubiquitina-tion of CDT1 in vitro in an E1- and E2-dependent manner (Fig. 4c).Deletion of either the N-terminal 203 or 97 residues from CUL4A,which disrupted DDB1–CUL4A association, abolished the ability of theCUL4A immunocomplex to ubiquitinate CDT1. Knocking down

DDB1

CAND1

Cul4ANedd8

Cul4A

CAND1 RNAi

Anti-Cul4A− + − +

IP: − −

DD

B1

CA

ND

1C

ul4A

Wes

tern

66

97

97

a

b

c

DDB1 RNAi Anti-Cul4A

1

+

DD

B1

CA

ND

1C

ul4A

Wes

tern

97

97

DDB1

CAND1

Cul4A

2 3 4

1 2 3 4

− + −IP: − −

Cul4ANedd8

T7−DDB1

Myc−CAND1IP: Anti-T7 Anti-Myc − −

++

++−

− −−−

−−−− −

DDB1

Myc−CAND1CAND1

Cul4ANedd8

Cul4A

97

97

66

Wes

tern

CAND1

Cul4A

DDB1

1 2 3 4 5 6 7 Mr(K)

Mr(K)

Mr(K)

Figure 2 DDB1 and CAND1 bind to CUL4A in a mutually exclusive manner.(a) Total cell lysates were prepared from 293T cells transiently transfectedwith a plasmid expressing either T7–DDB1 or Myc–CAND1 and subjected toeither direct immunoblotting or immunoprecipitation and western blotanalysis with the indicated antibodies. (b, c) Total cell lysates were preparedfrom cells transiently transfected with siRNA silencing the expression ofeither DDB1 or CAND1. The level of individual proteins was determined bydirect immunoblotting and CUL4A–DDB1 and CUL4A–CAND1 complexeswere examined by immunoprecipitation and western blot analysis.

print ncb1172 14/9/04 3:23 PM Page 1005



L E T T E R S


DDB1, but not DDB2, by RNAi substantially reduced the efficiencyof CUL4A-catalysed CDT1 polyubiquitination (Fig. 4d). Together,these results indicate that DDB1 is required for UV-induced CDT1degradation in vivo and CUL4A-mediated CDT1 ubiquitination invitro. DDB2 has long been linked with DDB1 functionally, andforms a complex with DDB1 (ref. 32). The exact role of DDB2 inconnection with CUL4 remains unclear, but it has been suggested

to be a substrate of, or to recruit substrates to, the DDB1–CUL4 lig-ase33,34. Our results also suggest that DDB2 has only a minor, if any,role in mediating CDT1 ubiquitination by the DDB1–CUL4A–ROC1ligase in response to UV irradiation. This notion is supported by thefact that DDB2 is not present in non-mammals, whereas the other four components (DDB1, CUL4, ROC1 and CDT1) are evolutionarily conserved.

Cul4A+ROC1

Cul4A+CDT1 CDT1+Cul4A

DDB1+vector

DDB1+Cul4A DDB1+CDT1

a

b

T7−DDB1

2IP: T7 Myc

97

CDT1

DDB1

66

4

Myc−CDT1+ ++ + + +

+ −+−

1Myc T7

45

HA−DDB2T7 Myc

++

+ +− − − −

DDB2

5 6

dWT

N115N768N956

∆N220∆N381∆N770∆N959

+CUL4ABinding with:

1 1140

−+++++−

+−−++++−

CDT1

Myc−CDT1 + Myc3−Cul4A

Myc3−CUL4A

Myc3−CUL4A

T7-DDB1

1 3 5 6 8IP:

N11

5

N76

8

N95

6

∆N38

1

∆N77

0

WT

97

66

66

45

30

Western: T7 +

Myc

∆N20

0

Myc−CDT1

Myc−CDT1

∆N95

9

IP:

Myc−CDT1Myc3−Cul4A +

T7−DDB1

97

66

Myc3−CUL4A

T7−DDB1

Myc−CDT166

Myc−CDT1

+ ++ +

++−−GST−CDT1

Cul4A Cul4A

c

GST−CDT1

T7−DDB1 + + −− + +

DDB1

CDT1

IgG(H)

IgG(L)

1 3

Silver staining

175

80

62

47

32

25166

e

IP: CDT1T7 Myc−CDT1

GST−CDT1(µg)

GST−CDT1

GST

DDB1(µg)

10

−

−−

−15

11

−−

1 3

5

Silver staining

DDB1175

80

62

47

32

25166

f

GST (µg)

−LW −LWH Key

3

2 41 3

+ ++

2 4 7T7

2

2 4

Mr(K)

Mr(K)

Mr(K)

Mr(K)

Mr(K)

Figure 3 DDB1 binds to CDT1 and bridges CDT1 to CUL4A. (a) Total 293T-cell lysates prepared from cells co-transfected with plasmids expressingboth T7–DDB1 and Myc–CDT1 were analysed for DDB1–CDT1 associationby immunoprecipitation and western blot analysis. Note that DDB1preferentially associates with the slower migrating (most probablyphosphorylated) form of CDT1. (b) The interaction between CUL4A, CDT1and DDB1 was examined by a directed yeast two-hybrid assay in buddingyeast cells that lack apparent homologues of CUL4A and DDB1. (c) Theassociation of CUL4A and CDT1 was examined by immunoprecipitation andwestern blot analysis in 293T cells ectopically transfected with plasmidsexpressing both proteins with or without additional co-expression of DDB1.The levels of ectopically expressed CDT1 in each transfected cellpopulation were determined by direct immunoblotting of the same celllysates (bottom). (d) CDT1 and CUL4A bind to separate regions of DDB1.

Various deletion mutants of DDB1 were co-expressed with CDT1 andCUL4A in 293T cells, and immunoprecipitated DDB1–CDT1 andDDB1–CUL4A bindings were determined by immunoprecipitation andwestern blot analysis. Arrows indicate individual wild-type and deletionmutants of T7-tagged DDB1 (top). The levels of ectopically expressedCUL4A and CDT1 in each transfected cell population were determined bydirect immunoblotting of the same cell lysates (bottom). (e) Cell lysateswere prepared from 293T cells transfected with plasmids expressingT7–DDB1 and/or Myc–CDT1, immunoprecipitated with an anti-T7 or anti-CDT1 antibody and resolved by SDS–PAGE and then silver stained. (f) GST–CDT1 and GST were purified from bacteria using glutathioneagarose beads and incubated with DDB1 (purified from insect cells) in ahypotonic buffer. The glutathione agarose beads were washed, and boundproteins were resolved by SDS–PAGE and then silver stained.

print ncb1172 14/9/04 3:23 PM Page 1006



L E T T E R S


The mechanism by which DDB1 contributes to DNA repair hasremained elusive since its discovery as a UV-damaged-DNA-bindingprotein32,35. Four lines of evidence support the conclusion that DDB1targets CDT1 for ubiquitination by the CUL4–ROC1 ligases inresponse to UV irradiation: first, DDB1 associates nearly stoichiomet-rically with CUL4A, and binds directly to CUL4A in a manner analo-gous to SKP1 and BTB binding to CUL1 and CUL3 respectively;second, the DDB1–CUL4A association is regulated by CAND1 simi-larly to the SKP1–CUL1 association; third, DDB1 binds directly andstoichiometrically to CDT1 and bridges CDT1 to CUL4A; and finally,suppression of DDB1 function inhibited UV-induced CDT1 degrada-tion in vivo and CUL4A-mediated CDT1 ubiquitination in vitro.CUL4–ROC1-mediated CDT1 degradation is consistent with recentreports that loss of Cul4 function results in the accumulation of CDT1in C. elegans embryos18. We suggest that by targeting CDT1 forCUL4–ROC-mediated rapid degradation after UV irradiation, DDB1could prevent the replication of damaged DNA, contributing to DNArepair and the maintenance of genomic integrity. Three major compo-nents of this pathway — CUL4, DDB1 and CDT1 — are evolutionarilyconserved from fission yeast to mammals, suggesting both an early

origin and conservation of the regulation of CDT1 by theDDB1–CUL4–ROC1 ligase in the cellular response to UV irradiation.

One remarkable feature of CUL1-, CUL2- and CUL3-dependent lig-ases is the assembly of various distinct complexes through interactionwith a common motif present in multiple adaptor proteins. DDB1-mediated binding of the substrate CDT1 to CUL4A does not seem torequire an additional adaptor, which is different from SKP1-mediatedCUL1-dependent ligases but more similar to BTB-mediated CUL3-dependent ligases. DDB1 has also been reported to target STAT1 ubiq-uitination involving simian parainfluenza virus 5 (SV5)-encoded Vprotein24,25,36 and c-Jun involving DET1–COP1 (ref. 27). It thus seemspossible that DDB1 may recruit different substrates to CUL4–ROC1ligases, through both directly binding to a substrate or through anadaptor or an adaptor complex, to CUL4–ROC1 ligases.

METHODS

Plasmids, cell culture and cell transfection. Plasmids expressing, and antibod-ies recognizing, human CUL4A, ROC1 and CAND1, and procedures for yeasttwo-hybrid assay, immunoprecipitation and immunoblotting were asdescribed13,29,37. Plasmids expressing DDB1 and CDT1 were obtained from

aMG132

DDB1

CUL4A

CDT1

UV +

1 2 3

++−−

−

CUL4ANedd8

DDB1 RNAi

DDB1

DDB2 RNAi

IP: CUL4A

DDB2

ROC1

CUL4A

b

1UV − +− −

RNAi DD

B1

Cul

4AC

ul4B

−

2 5 7

DD

B2

Cul

4AC

ul4B

DD

B2

DD

B1

−+ +

8

DDB1

CUL4A

CDT1

DDB2

CSN5

CUL4Anedd8

d

Myc-CDT1

+

DDB1 RNAi

1

66

2 3 6

DDB2 RNAi

IP:E1 & E2HA−Ub

Myc−CDT1 + + + + +− + + + + +

− ++ + + +

− − + −− − + −

CUL4A

Myc

−CD

T1−(

Ub

) n

−

c

Myc

−CD

T1−(

Ub

) n

Myc3−Cul4AIP:

Myc−CDT1E1+E2+HA−Ub

1 2 3 4

+ + −Myc

WT ∆N97 ∆N203

+ + ++

Myc-CDT1

1 2 3

+

Left

on

bea

ds

Myc3−Cul4AMyc3−Cul4A ∆N97

Myc3−Cul4A ∆N203

Elu

tionMyc3−Cul4A

Myc3−Cul4A ∆N97

Myc3−Cul4A ∆N203

WT

3 4 6− +

−−

−

− − +− −− − −− +

−

4 5

1 2 3 4 5

Mr(K)

Mr(K)

Mr(K)

∗

Figure 4 DDB1 targets CDT1 for CUL4A-mediated ubiquitination. (a) UV irradiation induced a rapid and proteasome-dependent degradationof CDT1. HeLa cells were irradiated by UV (50 J/m2) with or withoutMG132 treatment. The level of CDT1 protein was determined by directimmunoblotting of total-cell lysate prepared 30 min after UV treatment. (b) DDB1 is required for the UV-induced rapid degradation of CDT1. HeLacells were transfected with siRNA oligonucleotides silencing DDB1, DDB2or a combination of both CUL4A and CUL4B. Cells were UV-irradiated(50 J/m2) 48 h after transfection and then lysed for 30 min. The steadystate levels of CDT1, as well as DDB1, DDB2 and CUL4A, were determinedby direct immunoblotting. (c) The N-terminal sequence of CUL4A isrequired for in vitro CDT1 ubiquitination. Myc-tagged wild-type and mutantCUL4A were ectopically expressed in 293T cells, immunoprecipitatedusing an anti-Myc antibody and incubated with separately precipitatedMyc–CDT1 from transfected 293T cells in the presence or absence ofvarious components as indicated. The Myc antigen peptide was added to

the mixtures to elute the ligases and the substrates. The ubiquitinationreaction mixtures, as well as the eluate (middle) and proteins remaining onthe protein G agarose beads (bottom), were resolved on an SDS-PAGE gel,before immunoblotting with an anti-CDT1 or anti-Myc antibody. (d) DDB1is required for in vitro CUL4A-mediated ubiquitination of CDT1. TheCUL4A immunocomplex was precipitated from untransfected HeLa cells orHeLa cells transfected with siRNA oligonucleotides silencing DDB1 orDDB2. The targeted reduction of DDB1 or DDB2 expression was verified bydirect immunoblotting (bottom). The substrate, Myc–CDT1, wasimmunoprecipitated from separately transfected 293T cells and incubatedwith the CUL4A immunocomplex in the presence or absence of variouscomponents as indicated. A CUL4A peptide and the Myc peptide wereadded to the mixtures to elute the ligases and the substrates, respectively.After an in vitro ubiquitination reaction, the mixtures were resolved on anSDS–PAGE gel and CDT1 ubiquitination was examined by immunoblottingwith an anti-CDT1 antibody (top).

print ncb1172 14/9/04 3:23 PM Page 1007



L E T T E R S


collaborators (see Acknowledgments section). Mutations were introduced bysite-directed mutagenesis using the Quick-Change Kit (Stratagene, La Jolla,CA) and verified by DNA sequencing. All human cells were cultured in DMEMcontaining 10% FBS in a 37 °C incubator with 5% CO2. Cell transfections wereperformed using a calcium-phosphate buffer.

Antibodies, proteins, immunological procedures and mass spectrometricanalysis. Antibodies to haemagglutinin (HA) (12CA5; Boehringer-Mannheim,Mannheim, Germany), to Myc (9E10; NeoMarker, Fremont, CA), to T7(Novagen, Madison, WI), to FLAG (M2; Sigma, St Louis, MO) and to CSN5(JAB1-2A10.8; GeneTex, San Antonio, TX) were purchased. Rabbit polyclonalantibody to ROC1 has been described37, and rabbit polyclonal antibodies toCUL4A (N-MADEAPRKGSFSALVGRTNG-C), CDT1 (N-ADLAHITARLAHQ-TRAEEGL-C) and DDB1 (N-REKEFNKGPWKQENVE-H) were raised againsta synthetic peptide derived from the respective human proteins. The rabbit poly-clonal antibody to GST–CDT1 (amino acids 267–546) was a gift from X. Wu(Scripps Research Institute, CA). DDB1 protein purified from insect cells was agift from N. Zheng (University of Washington, WA). Procedures for proteinpurification, immunoprecipitation and immunoblotting have beendescribed13,37. To purify the endogenous ROC1 complex, HeLa cells were lysedwith a 0.5% NP-40 lysis buffer and lysates were pooled. To purify the CUL4Acomplex, 24× 150-mm plates of BT474 breast cancer cells were lysed with a 0.5%NP-40 lysis buffer and lysates were pooled (150 mg total). Lysates were incubatedwith affinity purified anti-ROC1 (18 µg) or anti-CUL4A (3 µg) antibodies.Immunocomplexes were precipitated by Protein-A or Protein-G agarose beadsand then eluted by incubating with a molar excess of antigen peptide. Elutedimmunocomplexes were resolved on an SDS–PAGE gel, stained with Coomassieblue and protein bands digested with trypsin before mass spectrometric analysisat the University of North Carolina Proteomics Core Facility.

RNA interference. All siRNA oligonucleotides were synthesised with 3′ dTdToverhangs by Dharmacon (Lafayette, CO) in a purified and annealed duplexform. The sequences targeting each human gene were as follows:Ddb1, 5′-CCUGUUGAUUGCCAAAAAC-3′; Ddb2, 5′-GAGCGAGAUCC-GAGUUUAC-3′; Cul4A, 5′-GAACUUCCGAGACAGACCU-3′; Cul4B,5′-AAGCCUAAAUUACCAGAAA-3′; Cand1, 5′-AATGATTTGATGACG-GAACTG-3′. OPTI-MEM medium (500 µl) was mixed with Lipofectamine2000 reagent (10 µl) for 5 min and then incubated with 10 µl (20 mM) of siRNAfor 20 min at room temperature. DMEM–10%-FBS medium (1.5 ml) was addedto the mixture, and the entire 2.5 ml was added to HeLa cells cultured on a 60-mmplate at 30–40% confluency. The cells were transfected once with Cul4A, Ddb1 andDdb2 siRNA and analysed 48–72 h after transfection, and transfected three timeswith Cand1 siRNA every 24 h and analysed 24 h after the last transfection.

In vitro ubiquitin ligation assays. The procedures for ubiquitin labelling wereas described13,29. Briefly, to purify substrate, Myc-tagged CDT1 was ectopicallyexpressed in 293T cells, extracted in RIPA lysis buffer (50 mM Tris-HCl at pH8.0, 150 mM NaCl, 1% NP-40, 0.5% DOC and 0.1% SDS), and immunopre-cipitated using an anti-Myc antibody. To purify CUL4A ligases from HeLacells, CUL4A immunocomplexes were immunoprecipitated using a CUL4Aantibody from untreated cells or HeLa cells transfected with Ddb1 or Ddb2siRNA oligonucleotides, immobilised on protein A–agarose beads and washedthree times with an NP-40 buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl,0.5% NaCl and 50 mM NaF) and twice with a ligase assay buffer (25 mM Tris-HCl at pH 7.5, 50 mM sodium chloride, 1 mM EDTA, 0.01% NP-40 and 10%glycerol). For the in vitro CDT1 ubiquitination, the CUL4A immunocomplexwas mixed with Myc–CDT1 substrate and the mixtures were added to a ubiq-uitin ligation reaction (final volume 30 µl) containing the following: 50 mMTris-HCl at pH 7.4, 5 mM MgCl2, 2 mM sodium fluoride, 10 nM okadaic acid,2 mM ATP, 0.6 mM dithiothreitol, 12 µg bovine ubiquitin, 1 µg HA–ubiquitin(Sigma), 60 ng E1, 300 ng E2 (hUbc5c), 5 µg Myc peptide and 5 µg CUL4Apeptide. Reactions were incubated at 37 °C for 60 min, terminated by boilingfor 5 min in an SDS sample buffer containing 0.1 M dithiothreitol andresolved on an SDS–PAGE gel before immunoblotting with the anti-CDT1antibody to examine ubiquitin ladder formation.

BIND identifiers. Four BIND identifiers (www.bind.ca) are associated with thismanuscript: 153957, 153958, 153959 and 153960.

Note: Supplementary Information is available on the Nature Cell Biology website.

ACKNOWLEDGEMENTSWe thank P. Raychaudhuri for providing the DDB1 expression vector; X. Wu forproviding a CDT1 expression vector and a GST–CDT1 antibody; N. Zheng forproviding purified DDB1 protein; S. Jackson for generating the CUL4A antibody;and M. Furukawa and members of the Xiong laboratory for the discussion andhelp throughout the course of this study. C.M.M. is a recipient of the George H.Hitchings New Investigator Award in Health Research and Training of the TriangleCommunity Foundation. This work is supported by National Institutes of Healthgrant GM067113 to Y.X.

COMPETING FINANCIAL INTERESTSThe authors declare that they have no competing financial interests.

Received 19 August 2004; accepted 25 August 2004Published online at http://www.nature.com/naturecellbiology.

1. Hershko, A. & Ciechanover, A. The ubiquitin system. Annu. Rev. Biochem. 67,425–479 (1998).

2. Deshaies, R. J. SCF and cullin/RING H2-based ubiquitin ligases. Annu. Rev. Cell Dev.Biol. 15, 435–467 (1999).

3. Jackson, P. et al. The lore of the RINGs: substrate recognition and catalysis by ubiq-uitin ligases. Trends Cell Biol. 10, 429–439 (2000).

4. Furukawa, M., Ohta, T. & Xiong, Y. Activation of UBC5 ubiquitin-conjugating enzymeby the RING finger of ROC1 and assembly of active ubiquitin ligases by all cullins. J.Biol. Chem. 277, 15758–15765 (2002).

5. Bai, C. et al. SKP1 connects cell cycle regulators to the ubiquitin proteolysis machin-ery through a novel motif, the F-box. Cell 86, 263–274 (1996).

6. Skowyra, D., Craig, K., Tyers, M., Elledge, S. J. & Harper, J. W. F-box proteins arereceptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex.Cell 91, 209–219 (1997).

7. Feldman, R. M. R., Correll, C. C., Kaplan, K. B. & Deshaies, R. J. A complex ofCdc4p, Skp1p, and Cdc53p/Cullin catalyzes ubiquitination of the phosphorylatedCDK inhibitor Sic1p. Cell 91, 221–230 (1997).

8. Zheng, N. et al. Structure of the Cul1–Rbx1–Skp1–F-box–Skp2 SCF ubiquitin ligasecomplex. Nature 416, 703–709 (2002).

9. Kamura, T. et al. The Elongin BC complex interacts with the conserved SOCS-boxmotif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat fami-lies. Genes Dev. 12, 3872–3881 (1998).

10. Zhang, J. G. et al. The conserved SOCS box motif in suppressors of cytokine signalingbinds to elongins B and C and may couple bound proteins to proteasomal degrada-tion. Proc. Natl Acad. Sci. USA 96, 2071–2076 (1999).

11. Kamura, T. et al. Muf1, a novel Elongin BC-interacting leucine-rich repeat proteinthat can assemble with Cul5 and Rbx1 to reconstitute a ubiquitin ligase. J. Biol.Chem. 276, 29748–29753 (2001).

12. Stebbins, C. E., Kaelin, W. G., Jr. & Pavletich, N. P. Structure of theVHL–ElonginC–ElonginB complex: implications for VHL tumor suppressor function.Science 284, 455–461 (1999).

13. Furukawa, M., He, Y. J., Borchers, C. & Xiong, Y. Targeting of protein ubiquitination byBTB–Cullin 3–Roc1 ubiquitin ligases. Nature Cell Biol. 5, 1001–1007 (2003).

14. Geyer, R., Wee, S., Anderson, S., Yates, J. & Wolf, D. A. BTB/POZ domain proteins areputative substrate adaptors for cullin 3 ubiquitin ligases. Mol. Cell 12, 783–790(2003).

15. Pintard, L. et al. The BTB protein MEL-26 is a substrate-specific adaptor of the CUL-3 ubiquitin-ligase. Nature 425, 311–316 (2003).

16. Xu, L. et al. BTB proteins are substrate-specific adaptors in an SCF-like modularubiquitin ligase containing CUL-3. Nature 425, 316–321 (2003).

17. Osaka, F. et al. Covalent modifier NEDD8 is essential for SCF ubiquitin-ligase in fis-sion yeast. EMBO J. 19, 3475–3484 (2000).

18. Zhong, W., Feng, H., Santiago, F. E. & Kipreos, E. T. CUL-4 ubiquitin ligase maintainsgenome stability by restraining DNA-replication licensing. Nature 423, 885–889(2003).

19. Li, B., Ruiz, J. C. & Chun, K. T. CUL-4A is critical for early embryonic development.Mol. Cell. Biol. 22, 4997–5005 (2002).

20. Chen, L.-C. et al. The human homologue for the Caenorhabditis elegans cul-4 gene isamplified and overexpressed in primary breast cancers. Cancer Res. 58, 3677–3683(1998).

21. Yasui, K. et al. TFDP1, CUL4A, and CDC16 identified as targets for amplification at13q34 in hepatocellular carcinomas. Hepatology 35, 1476–1484 (2002).

22. Higa, L. A., Mihaylov, I. S., Banks, D. P., Zheng, J. & Zhang, H. Radiation-mediatedproteolysis of CDT1 by CUL4–ROC1 and CSN complexes constitutes a new check-point. Nature Cell Biol. 5, 1008–1015 (2003).

23. Zhang, Y. et al. CUL-4A stimulates ubiquitylation and degradation of the HOXA9homeodomain protein. EMBO J. 22, 6057–6067 (2003).

24. Andrejeva, J., Poole, E., Young, D. F., Goodbourn, S. & Randall, R. E. The p127 sub-unit (DDB1) of the UV-DNA damage repair binding protein is essential for the targeteddegradation of STAT1 by the V protein of the paramyxovirus simian virus 5. J. Virol.76, 11379–11386 (2002).

25. Ulane, C. M. & Horvath, C. M. Paramyxoviruses SV5 and HPIV2 assemble STAT pro-tein ubiquitin ligase complexes from cellular components. Virology 304, 160–166(2002).

print ncb1172 14/9/04 3:23 PM Page 1008



L E T T E R S


26. Ulane, C. M., Rodriguez, J. J., Parisien, J. P. & Horvath, C. M. STAT3 ubiquitylationand degradation by mumps virus suppress cytokine and oncogene signaling. J. Virol.77, 6385–6393 (2003).

27. Wertz, I. E. et al. Human De-etiolated-1 regulates c-Jun by assembling a CUL4Aubiquitin ligase. Science 303, 1371–1374 (2004).

28. Shiyanov, P., Nag, A. & Raychaudhuri, P. Cullin 4A associates with the UV-damagedDNA-binding protein DDB. J. Biol. Chem. 274, 35309–35312 (1999).

29. Liu, J., Furukawa, M., Matsumoto, T. & Xiong, Y. NEDD8 modification of CUL1 disso-ciates p120(CAND1), an inhibitor of CUL1–SKP1 binding and SCF ligases. Mol. Cell10, 1511–1518 (2002).

30. Zheng, J. et al. CAND1 binds to unneddylated CUL1 and regulates the formation ofSCF ubiquitin E3 ligase complex. Mol. Cell 10, 1519–1526 (2002).

31. Li, X., Zhao, Q., Liao, R., Sun, P. & Wu, X. The SCF(Skp2) ubiquitin ligase complexinteracts with the human replication licensing factor Cdt1 and regulates Cdt1 degra-dation. J. Biol. Chem. 278, 30854–30858 (2003).

32. Keeney, S., Chang, G. J. & Linn, S. Characterization of a human DNA damage binding

protein implicated in xeroderma pigmentosum. J. Biol. Chem. 268, 21293–21300(1993).

33. Groisman, R. et al. The ubiquitin ligase activity in the DDB2 and CSA complexes isdifferentially regulated by the COP9 signalosome in response to DNA damage. Cell113, 357–367 (2003).

34. Nag, A., Bondar, T., Shiv, S. & Raychaudhuri, P. The xeroderma pigmentosum group Egene product DDB2 is a specific target of cullin 4A in mammalian cells. Mol. CellBiol. 21, 6738–6747 (2001).

35. Chu, G. & Chang, E. Xeroderma pigmentosum group E cells lack a nuclear factor thatbinds to damaged DNA. Science 242, 564–567 (1988).

36. Lin, G. Y., Paterson, R. G., Richardson, C. D. & Lamb, R. A. The V protein of theparamyxovirus SV5 interacts with damage-specific DNA binding protein. Virology249, 189–200 (1998).

37. Ohta, T., Michel, J. J., Schottelius, A. J. & Xiong, Y. ROC1, a homolog of APC11, rep-resents a family of cullin partners with an associated ubiquitin ligase activity. Mol.Cell 3, 535–541 (1999).

print ncb1172 14/9/04 3:23 PM Page 1009



S U P P L E M E N TA RY I N F O R M AT I O N

WWW.NATURE.COM/NATURECELLBIOLOGY 1

Figure S1 CUL1, CUL3, and CUL4A alone are not required for the UV-induced rapid degradation of CDT1. (a) HeLa cells were transfected with siRNA oligonucleotides silencing DDB1, CUL4A or mismatched siRNA oligonucleotides. (b) HeLa cells were transfected with siRNA oligonucleotides silencing CUL1 or CUL3. (c) HeLa cells were transfected with chemically synthesized siRNA oligonucleotide targeting nucleotides 199-207 of DDB1 mRNA (NM_001923), or siRNA oligonucleotides

generated by recombinant Dicer protein using the template double strand RNA corresponding to nucleotides 2966-3519 of DDB1 mRNA or nucleotides 103-705 relative to the translation start of GFP gene. 48 hours after transfection, cells were UV-irradiated (50 J/m2) and lysed 30 minutes after UV irradiation. The steady state levels of CDT1, as well as DDB1, CUL4A, CUL1, CUL3, ROC1 and tubulin were determined by direct immunoblotting.


Documents

Targeted ubiquitination of CDT1 by the DDB1–CUL4A–ROC1 ligase in response to DNA damage