PCNA-mediated stabilization of E3 ligase RFWD3 at thereplication fork is essential for DNA replicationYo-Chuen Lina,1, Yating Wanga,1, Rosaline Hsua, Sumanprava Giria, Susan Wopata, Mariam K. Arifa,Arindam Chakrabortya, Kannanganattu V. Prasantha, and Supriya G. Prasantha,2

aDepartment of Cell and Developmental Biology, University of Illinois at Urbana–Champaign, Urbana, IL 61801

Edited by Bruce Stillman, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, and approved November 5, 2018 (received for review August 23, 2018)

RING finger and WD repeat domain-containing protein 3 (RFWD3)is an E3 ligase known to facilitate homologous recombination byremoving replication protein A (RPA) and RAD51 from DNA damagesites. Further, RPA-mediated recruitment of RFWD3 to stalled rep-lication forks is essential for interstrand cross-link repair. Here, wereport that in unperturbed human cells, RFWD3 localizes at replica-tion forks and associates with proliferating cell nuclear antigen(PCNA) via its PCNA-interacting protein (PIP) motif. PCNA associationis critical for the stability of RFWD3 and for DNA replication. Cellslacking RFWD3 show slower fork progression, a prolonged S phase,and an increase in the loading of several replication-fork componentson the chromatin. These findings all point to increased frequencyof stalled forks in the absence of RFWD3. The S-phase defect isrescued by WT RFWD3, but not by the PIP mutant, suggesting thatthe interaction of RFWD3 with PCNA is critical for DNA replica-tion. Finally, we observe reduced ubiquitination of RPA in cellslacking RFWD3. We conclude that the stabilization of RFWD3 byPCNA at the replication fork enables the polyubiquitination of RPAand its subsequent degradation for proper DNA replication.

DNA replication | PCNA | RFWD3 | RPA | ubiquitination

Maintenance of genome integrity is key to cell survival.Defects in DNA replication and errors in the DNA dam-

age response contribute to genome instability and are key con-tributing factors in many diseases, including cancer (1). Replicationstress is a leading cause of genome instability and occurs whenreplication forks progress slowly or stall. A huge repertoire of cel-lular factors can mediate replication stress-induced DNA damage,including deprivation of dNTP pools, defects in DNA replicationproteins, and decreased firing of origins due to defects in replicationinitiation (2). Intricate checkpoint pathways operate to ensure thatthe entry into or progression through S phase is blocked when thecells encounter DNA damage.Cells encounter many assaults to their genome that are repaired

accurately and efficiently to maintain genome integrity. Ubiquitina-tion is emerging as an important player in DNA replication, repair,and damage-signaling pathways. Nondegradative ubiquitin signalinginvolving either monoubiquitination or polyubiquitination hasbeen implicated in the maintenance of genome integrity, includingin processing of DNA double-strand breaks, repair of interstrandcross-link lesions (ICLs), and bypass of lesion during DNA replication(for review, see ref. 3). The signaling events are triggered by ubiq-uitin protein ligases that can initiate mono- or polyubiquitinationthrough nonstandard linkage.DNA ICLs are links between the two strands of DNA with a

covalent bond, and many pathways including nucleotide excisionrepair, structure-specific endonucleases, translesion DNA synthesis(TLS), and homologous recombination (HR) have been impli-cated in resolving such errors (4). ICLs inhibit DNA replicationand transcription, and the dominant mode of ICL repair is believedto happen during S phase and requires converging replicationforks (5, 6). Mutations in ICL repair are associated with Fanconianemia (FA), a rare heritable disorder, and ∼21 FA genesreported thus far have been implicated in ICL repair (7, 8).The role of novel factors in ICL repair and the importance of

ubiquitination in the damage response are becoming intense areasof research.An E3 ubiquitin ligase, RING finger and WD repeat domain-

containing protein 3 (RFWD3), was initially identified in a pro-teomic study as a substrate of ATM/ATR (9). Biallelic mutationsin RFWD3 were reported in patients with FA (10). RFWD3 isemerging as an important component in the FA/BRCA pathwayand has been assigned the alias FANCW. Previous studies haveshown that RFWD3 functions synergistically with Mdm2 toregulate the ubiquitination of tumor suppressor protein p53 inresponse to DNA damage (11). RFWD3 associates with replica-tion protein A (RPA), the single-stranded binding protein; facili-tates the RPA-mediated DNA damage response; and affects HRat stalled replication forks (12–14). Recent studies have shownthat RPA-mediated recruitment of RFWD3 is essential for ICLrepair (15) and that RFWD3-mediated ubiquitination promotesthe removal of RPA and RAD51 from damage sites to allowHR (16). All these findings support the role of RFWD3 in DNA-damage repair.While RFWD3 is emerging as a new component of the FA

pathway, up until now, this role had been thought to be mediatedprimarily by its interactions with the single-stranded DNA-bindingprotein RPA that are important for repair of ICL DNA damage.In the present manuscript, we report that the role of RFWD3 inthe FA pathway may be related to its role in normal DNA repli-cation via its direct interaction with proliferating cell nuclearantigen (PCNA) at the replication fork. We propose that the as-sociation of RFWD3 at the replication fork mediates the ubiquiti-nation of key replication fork components that is essential forefficient progression of DNA replication. This seems to be a fun-damentally different way of considering how RFWD3 functions in

Significance

Ubiquitination of several DNA replication–repair proteins is acritical mechanism by which cellular DNA replication and DNAdamage repair pathways are controlled. The E3 ligase RING fingerandWD repeat domain-containing protein 3 (RFWD3) has emergedas an important regulator of genome stability. We propose thatthe association of RFWD3 to proliferating cell nuclear antigenstabilizes RFWD3 at the fork, enabling ubiquitination of repli-cation protein A and its subsequent removal to facilitate DNAreplication.

Author contributions: Y.-C.L., Y.W., S.G., S.W., K.V.P., and S.G.P. designed research; Y.-C.L.,Y.W., R.H., S.G., S.W., M.K.A., and S.G.P. performed research; A.C. contributed new re-agents/analytic tools; Y.-C.L., Y.W., S.G., K.V.P., and S.G.P. analyzed data; and Y.-C.L.,Y.W., and S.G.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1Y.-C.L. and Y.W. contributed equally to this work.2To whom correspondence should be addressed. Email: supriyap@illinois.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1814521115/-/DCSupplemental.

Published online December 10, 2018.

13282–13287 | PNAS | December 26, 2018 | vol. 115 | no. 52 www.pnas.org/cgi/doi/10.1073/pnas.1814521115

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

DNA metabolism, as it raises the possibility that a major role ofRFWD3 in genome stability occurs via its role in normal, unstressedDNA replication at the replication fork as opposed to solely func-tioning in DNA repair processes.

ResultsRFWD3 Localizes to the Replication Fork and Interacts with PCNA.RFWD3 is an E3 ligase that is known to play important roles inDNA damage response. A recent nascent chromatin capture pro-teomics study showed the enrichment of RFWD3 at the replicationfork (17), supporting the notion that RFWD3 might play importantroles in unperturbed DNA replication. To test this model, weexamined the presence of RFWD3 at the fork and monitored thespatiotemporal dynamics of RFWD3 in a quantitative manner byperforming the isolation of proteins on nascent DNA (iPOND)assay (18). RFWD3 showed enrichment at the unperturbed re-plication fork, but not after thymidine chase (matured DNA)(Fig. 1A). PCNA, the DNA clamp, was also found to be enrichedon the nascent DNA, but not on the mature DNA, confirmingthe specificity of the iPOND assay (Fig. 1A). Histone H4 was foundon nascent as well as mature DNA. Further, immunoprecipita-tion (IP) experiments demonstrated that RFWD3 associatedwith the fork protein PCNA in unperturbed human cells as wellas in DNA-damaged cells (Fig. 1B and SI Appendix, Fig. S1A). Inaddition, by using purified proteins, we demonstrated direct in-teraction between RFWD3 and PCNA (Fig. 1C). Lastly, exam-ination of the localization of RFWD3 in asynchronously grownhuman cells revealed colocalization of RFWD3 with the replica-tion protein PCNA in a subset of cells (Fig. 1D and SI Appendix,Fig. S1B). These data all strongly support the existence of RFWD3at the replication fork.

PCNA is known to be the master regulator of events at thereplication fork (19). PCNA provides a molecular platform thatenables protein–protein and protein–DNA interactions at thefork. A large repertoire of proteins associate with PCNA via ageneral motif, the PCNA-interacting protein (PIP) box sequence[Q-XX-(L/I/M)-XX-(HF/DF/Y)] (20). A close inspection of thesequence of RFWD3 revealed a PIP box consensus toward the Cterminus of the protein (Fig. 1E). IP experiments using the HAantibody in cell lines stably expressing HA-RFWD3-WT, HA-RFWD3-PIPm, or HA-RFWD3-I639K (the latter a mutant knownto abolish RFWD3 interaction with RPA) showed that WT aswell as the I639K mutant (albeit weaker than WT) interactedwith PCNA (Fig. 1F). On the other hand, the PIP box mutant failedto interact with PCNA (Fig. 1F). RFWD3 has previously beenreported to associate with the single-stranded binding proteinRPA. We found that the RPA interaction to RFWD3 was abro-gated in the I639K mutant, as expected (Fig. 1G). RPA interactedwith WT RFWD3 and with the PIP mutant (but to a lesser extentcompared with the WT) (Fig. 1G). To further investigate thecomplex assembly of RFWD3, PCNA, and RPA, purified pro-teins or nuclear extract were fractionated on a Superdex 200 gelfiltration column. A significant portion of RFWD3 was found infraction 7 (∼669 kDa), cofractionating with RPA and PCNA,indicating that RFWD3 is in a high-molecular-weight complexcontaining RPA and PCNA (SI Appendix, Fig. S1 C and D), inaddition to subcomplexes of RFWD3–PCNA and RFWD3–RPA.Further, GST pull-down assays demonstrated that RFWD3, PCNA,and RPA could exist in one single complex, as RPA can bindto RFWD3 without displacing PCNA (SI Appendix, Fig. S1E).Our results suggest that RFWD3 associates to the replicationfork at S phase and interacts with the fork components PCNAand RPA.

RFWD3 Is Required for S-Phase Progression. Because RFWD3 lo-calized to the replication fork in unperturbed cells, we addressedthe role of RFWD3 during the normal cell cycle. Despite nu-merous efforts, our attempts to knock out RFWD3 in humanU2OS cells utilizing the CRISPR/Cas9-mediated genome-editingmethod were unsuccessful, as we failed to get clones with com-plete loss of the RFWD3 protein. This is consistent with resultsfrom other groups, suggesting that RFWD3 is an essential gene(15, 16). We used two independent siRNAs (RFWD3 si-1 andRFWD3 si-2) to deplete RFWD3 in multiple human cell lines,including U2OS (osteosarcoma), HCT116 (colorectal cancer),WT, p53−/−, and p21−/− cells. Depletion of RFWD3 in mostcancer cells, including U2OS and HCT116, showed an increasein the 4C DNA content and accumulation of cells in S phase(Fig. 2 A and B and SI Appendix, Fig. S2 A and B).To address the requirement of RFWD3 in DNA replication, we

depleted RFWD3 in serum-starved WI-38 (human diploid fibro-blast) cells. The control and RFWD3-depleted cells were releasedfrom serum starvation and evaluated by flow cytometry at 0 and24 h after serum release. While the control cells cycled back effi-ciently (as observed by increased G2/M population at 24 h release),cells treated with either of the RFWD3 siRNAs showed defects inprogressing into S phase (SI Appendix, Fig. S2C).Because we observed RFWD3 localized to the fork, we ex-

amined fork progression in RFWD3-depleted cells using a DNAfiber assay. We observed a consistent reduction in replication-fork speed in cells depleted of RFWD3 using either of the siRNAs(Fig. 2C). We hypothesize that the reduced fork speed leads toprolonged S phase; hence, more cells are accumulated in S phase.To address the length of S phase, we synchronized control andRFWD3-depleted cells at the G1/S boundary using double thymi-dine (dT) block and release. At different time points after releasefrom dT block (7, 8, and 9 h), cells were harvested and immu-nostained for PCNA to evaluate the number of cells in S phase.By 9 h after release, <30% of the control cells showed PCNA

PCNA

(HA)RFWD3

HA IgG1% input 30% IP

GST(RFWD3)

PCNAIgG PCNA

1% input 30% IP

B

C SSQ RING Coiled-coil WD401 134 286 333 345 415 488 774

620 624QKMDFQ--L--HF I--DF M--YAKADA

RFWD3-WTPIP consensus

RFWD3-PIPm

E

F

Thy chase 0 60 0 60

input iPOND

RFWD3

PCNA

H4

A D

100150

(min)

37

10

37100150

37

150

mid-S

late-S

YFP-RFWD3 PCNA Merge

WT

PIPmI63

9K- WT

PIPmI63

9K-

1% input 33% HA-IP

100

HA-RFWD3

RFWD3

PCNA37

WT

PIPmI63

9K- WT

PIPmI63

9K-

0.5% input 33% HA-IP

HA-RFWD3

100

37

RFWD3

RPA2

G

Fig. 1. RFWD3 is at the replication fork and associates with PCNA. (A) Cellslabeled with 5-ethynyl-2′-deoxyuridine (EdU), processed by iPOND [0 and 60 minthymidine (Thy) chase], and immunoblotted using RFWD3, PCNA, and H4antibodies. (B) IP using HA antibody from U2OS cells stably expressing HA-RFWD3. RFWD3 and PCNA were analyzed by immunoblotting. (C) Directinteraction of RFWD3 and PCNA using purified proteins. (D) Localization ofYFP-RFWD3 with PCNA. (Scale bar: 15 μm.) (E) Schematic representation ofRFWD3 protein with different domains. Note the PIP motif at amino acids620 to 624 within the WD40 domain. (F) IP in U2OS cells expressing variousHA-RFWD3-WT, HA-RFWD3-PIPm, and HA-RFWD3-I639K mutants using HAantibody and analysis by RFWD3 and PCNA immunoblotting. (G) IP in U2OScells expressing various HA-RFWD3-WT, HA-RFWD3-PIPm, and HA-RFWD3-I639K mutants using HA antibody and analysis by RFWD3 and RPA2immunoblotting.

Lin et al. PNAS | December 26, 2018 | vol. 115 | no. 52 | 13283

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

staining, indicating that most of them had already exited S phase.On the other hand, >50% of RFWD3-depleted cells displayedPCNA-positive signals (compare between 7 and 9 h of release)(Fig. 2D). Further, live-cell imaging of YFP-PCNA in control aswell as in RFWD3-depleted cells demonstrated that cells lackingRFWD3 displayed an average S-phase length of 16 h, while thatof the control was <14 h (Fig. 2E, SI Appendix, Fig. S2D, and MovieS1). These results reveal that the depletion of RFWD3 causeda prolonged S phase.We next addressed whether the association of RFWD3 with

PCNA and/or RPA was required for proper DNA replication.We performed RFWD3 depletion in HA-RFWD3-WT, HA-RFWD3-PIPm, HA-RFWD3-C315A (catalytically inactive), andHA-RFWD3-I639K mutants and tested for cell cycle distributionusing propidium iodide (PI) and BrdU-PI flow analyses. Depletionof RFWD3 showed increased population in S phase (Fig. 2F andSI Appendix, Fig. S2E) as well as reduced incorporation of BrdU,with a population of cells accumulating between 2C and 4CDNA content (SI Appendix, Fig. S2F). The siRNA-resistant version

of WT RFWD3 rescued the S-phase accumulation defects inendogenous RFWD3-depleted cells; however, the other mutantsdid not (Fig. 2F and SI Appendix, Fig. S2 E and F). Similarly, WTRFWD3 was able to rescue the defects in DNA fiber length, but thePIP mutant failed to do so (Fig. 2G). These results suggest thatthe association of RFWD3 and PCNA is important for efficientDNA replication.

Loss of RFWD3 Causes Fork Stalling and Sister Chromatid CohesionDefects. To gain molecular insight into the cause of DNA repli-cation defects in cells lacking RFWD3, we evaluated the status ofthe replisome components at the fork. We performed chromatinfractionation to determine the loading of the replisome componentsin control and RFWD3-depleted cells that had been synchro-nized in S phase. Immunoblotting using PCNA antibodies demon-strated that RFWD3 knockdown resulted in increased PCNAmonoubiquitination, suggesting stalling of the DNA replicationfork (Fig. 3A). This modification is known to mediate the switchfrom replicative DNA polymerases to TLS polymerases (21). Weobserved an increase in the total as well as chromatin-associated

control siRFWD3 si-1RFWD3 si-2

****

********

********

CldU, 30min IdU, 30min

A

C D

U2O

S

control-si

RFWD3 si-1

RFWD3 si-2

(24 hours)1st thymidine

block

(24 hours)2nd thymidine

block(12 hours)Release

(9 hours)Release

RFWD3 RNAi PCNA immuno-stainingat 7h, 8h, 9h

0

20

40

60

80

100

7h 8h 9h

S p

hase

leng

th (h

)

Per

cent

age

of

PC

NA

posi

tive

cells

(%)

control si(N=34)

RFWD3 si(N=24)

control si RFWD3 si-1 RFWD3 si-2G1: 57.1S: 16.83G2/M: 26.02

G1: 36.87S: 23.9G2/M: 38.92

G1: 35.4S: 30.03G2/M: 34.52

B

0

20

40

60

80

contr

ol si

(N=1

58)

RFWD3 si-1

(N=1

58)

RFWD3 si-2

(N=1

46)

Fibe

r len

gth

(μm

)

E

********

0

201612

24

84

0

30

40

20

10perc

enta

ge%

control si

RFWD3 si-1

RFWD3 si-2

**

S phase

13.87516

**

F

G

control si RFWD3 siHA-RFWD3

WT

PIPm

-

C315A

I639K

**** *

RFWD3 si - + - + - +

WT PIPm-HA-RFWD3

30

20

10

0

Fibe

r len

gth

(μm

)

(N=28) (N=42) (N=28) (N=33) (N=35) (N=31)

40

G1: 51.78S: 18.75G2/M: 29.34

G1: 32.93S: 27.7G2/M: 39.05

G1: 47.17S: 21.52G2/M: 31.15

G1: 49.08S: 22.2G2/M: 28.57

G1: 45.18S: 21.45G2/M: 33.15

G1: 35.4S: 28.22G2/M: 36.14

G1: 51.87S: 18.44G2/M: 29.6

G1: 40.49S: 26.03G2/M: 33.12

G1: 51.28S: 18.54G2/M: 30

G1: 42.97S: 25.09G2/M: 31.76

live cell image

Fig. 2. RFWD3 is required for S-phase progression.(A) Cell cycle profile (PI flow cytometry) of U2OS cellsdepleted of RFWD3 (RFWD3 si-1 and RFWD3 si-2). (B)Quantification of S-phase population from threeindependent experiments. (C, Upper) Representativefluorescence image showing fibers from control(control si) and RFWD3-depleted (RFWD3 si-1 andsi-2) U2OS cells that were labeled with 5-chloro-2′-deoxyuridine (CldU) and 5-iodo-2′-deoxyuridine (IdU).(C, Lower) Fork speed was calculated and plotted as abox plot. The top and bottom of each box representthe 75th and 25th percentiles, respectively, and thebar in each box represents the median. Dots rep-resent outliers. (D) Schematic of the experimentalprotocol for RFWD3 depletion in U2OS cells thatare synchronized using dT block and release. Immu-nolocalization of PCNAwas performed at 7, 8, and 9 hafter release. Percent PCNA-positive cells is calculated.Note that RFWD3-depleted cells continue to show>50% PCNA-positive cells at 9 h. Error bars representSD. (E) Length of S phase quantitated by live-cellimaging of YFP-PCNA in control and RFWD3-siRNA–treated cells. (F) Cell cycle profile by flow cytometryof various U2OS cell lines (HA-RFWD3-WT, HA-RFWD3-PIPm, HA-RFWD3-C315A, and HA-RFWD3-I639K mu-tants) depleted of endogenous RFWD3. Note that onlyWT RFWD3 can rescue the cell cycle defect. (G) DNAfiber assay in HA-RFWD3-WT and HA-RFWD3-PIPm celllines that are depleted of endogenous RFWD3. *P <0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 byunpaired two-tailed Student’s t test.

13284 | www.pnas.org/cgi/doi/10.1073/pnas.1814521115 Lin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

levels of the single-stranded binding protein RPA, DNA poly-merases, and TLS polymerases (Fig. 3A and SI Appendix, Fig.S3A). Similarly, the preinitiation complex component Cdc45showed increased chromatin association (Fig. 3A). All these resultspoint to stalling of the replication forks in the absence of RFWD3.Replication-fork stalling upon RFWD3 depletion was corroboratedby cell biological studies using DNA fiber assay (SI Appendix, Fig.S3B). Also, the increased number of RPA and RAD51 foci inRFWD3-depleted cells supports our model that the loss ofRFWD3 causes replication-fork stalling (Fig. 3 B and C).Genomic instability is associated with loss of FA pathway

genes. We observed an increase in the 4C DNA population in allcancer cells depleted of RFWD3. Metaphase spreads were preparedfrom control and RFWD3-siRNA–treated cells using standardprocedures. Strikingly, the cells depleted of RFWD3 using twoindependent siRNAs showed defects in sister chromatid cohesion(SI Appendix, Fig. S3 C and D). The replisome progression com-plex is known to establish sister chromatid cohesion (22). Our resultsposit that defective DNA replication may cause sister chromatidcohesion defects. However, we cannot rule out the direct involve-ment of RFWD3 in mediating sister chromatid cohesion byubiquitinating relevant substrates.

RFWD3 Mediates Ubiquitination of Fork Components to Enable DNAReplication. We have observed that RFWD3 is required for ef-ficient DNA replication. We demonstrated that RFWD3 inter-acts with PCNA and localizes to the replication fork during Sphase. To address the functional significance of PCNA in-teraction with RFWD3, we depleted PCNA using a previouslyvalidated siRNA oligonucleotide (23) and monitored the levelsand chromatin binding of RFWD3. Cells lacking PCNA showedreduction of both total and chromatin-bound levels of RFWD3,suggesting that the binding of PCNA to RFWD3 stabilizesRFWD3 (Fig. 4A). Moreover, chromatin-associated RPA levelsremained unaltered, suggesting that RPA is not sufficient torecruit and stabilize RFWD3 on chromatin in unperturbed cells(Fig. 4A). Previous work has suggested that RFWD3 is recruitedto stalled forks during interstrand cross-link repair via RPA. Wefound that the depletion of RPA in undamaged cells did notaffect the total or the chromatin-associated pool of RFWD3(Fig. 4B and SI Appendix, Fig. S4). Furthermore, the associationof RFWD3 to PCNA remained unaltered in cells lacking RPA(Fig. 4C), suggesting that RPA-independent mechanisms could

act in recruiting and/or stabilizing RFWD3 on chromatin inunperturbed cells.Next, we depleted RFWD3 in U2OS cells stably expressing

either HA-RFWD3-WT or HA-RFWD3-PIPm. The increasedlevels of chromatin-associated RPA observed in RFWD3-depletedcells were rescued in the WT RFWD3-expressing cells, but notin the PIP mutant lines (Fig. 4D). These results support ourconclusion that the binding of RFWD3 to PCNA is essentialfor DNA replication and that upon abrogating this interaction,replication is stalled.To address the mechanism that causes fork stalling in the

absence of RFWD3, we evaluated the ubiquitination of selectreplisome components, considering that RFWD3 is an E3 ligase.We performed an in vivo ubiquitination assay in control as wellas in RFWD3-depleted cells treated with the proteasome in-hibitor carbobenzoxy-Leu-Leu-leucinal or the p97 inhibitor N2,N4-dibenzylquinazoline-2,4-diamine (to inhibit degradation ofubiquitinated proteins). We observed a significant reduction inRPA ubiquitination upon RFWD3 depletion (Fig. 4E). Therewas a marginal reduction in a specific ubiquitinated form of PCNA(labeled with an asterisk in Fig. 4E), but not in other forms.Our results support previous observations that RFWD3 ubiq-uitinates RPA, although previous reports observed RPA ubiq-uitination in cells treated with a DNA-damaging agent such asmitomycin C. Our results support a model in which RFWD3 as-sociates with PCNA, and this interaction is critical for the stabili-zation of RFWD3 to the fork. At the fork, RFWD3 ubiquitinatessubstrates essential for DNA replication progression. RPA isone of RFWD3’s substrates, and its ubiquitination triggers its re-moval and the faithful completion of DNA replication (Fig. 4F).

DiscussionRFWD3, originally identified as a substrate for ATM/ATR in alarge-scale proteomic screen, is an E3 ligase that plays a crucialrole in the DNA-damage response (9). RFWD3 is known toubiquitinate p53 and to stabilize p53 in response to DNA dam-age (11). Several recent studies have pinpointed the role ofRFWD3 in replication checkpoint control (12, 15, 16). RFWD3interacts with RPA at stalled forks and facilitates RPA-mediatedDNA-damage signaling (13, 14). RPA ubiquitination by RFWD3was found to be critical for HR at stalled forks and to be requiredfor fork restart (12). Other recent studies have highlighted theimportance of RFWD3 in ICL repair (15, 16). HR was found tobe disrupted in RFWD3-mutant cells, and biallelic mutations in

α-Tubulin

SRSF1

RFWD3

CDC45

RPA2

Pol-ε

Pol-δ

S2 S3 P3 S2 S3 P3 S2 S3 P3

A

PCNA

PCNA-Ub

RPA

2R

PA/D

AP

IR

FWD

3-si

RPA2 Rad51 Merge Merge/DAPIC

B

37

75

20

100

50

37

37

100

control-si RFWD3 si-1RFWD3 si-2

control-si RFWD3 si-1 RFWD3 si-2

cont

rol-s

i

Fig. 3. Loss of RFWD3 causes fork stalling. (A) Chro-matin fractionation in RFWD3-depleted U2OS cells(siRNA 1 and siRNA 2), followed by immunoblottinganalysis of various replication proteins (PCNA, RPA2,CDC45, Pol-e, Pol-δ). Note increased PCNA ubiquiti-nation and the total levels of RPA and CDC45. SRSF1and α-tubulin are loading controls. P3, chromatin-boundfraction; S2, cytosolic fraction; S3, nuclear solublefraction. (B) RPA and (C) RPA with RAD51 immuno-fluorescence in control and RFWD3-depleted cells.(RPA staining in control, 47.4% replication and 10.3%damagelike foci; RFWD3-siRNA 1, 50.9% replicationand 23.4% damagelike foci; RFWD3-siRNA 2, 36.4%replication and 31.3% damage-like foci). (Scale bar:15 μm.)

Lin et al. PNAS | December 26, 2018 | vol. 115 | no. 52 | 13285

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

RFWD3 have been linked to FA (10). Specifically, RFWD3 hasbeen implicated in ubiquitinating RPA and RAD51 and in theirsubsequent removal to allow HR progression (16). However, themechanism of how RPA and RAD51 are recognized by RFWD3remains to be determined.RFWD3 and several FA genes function at different stages of

HR. Evidence indicates that FANCD2 and RFWD3 as well asRAD51 and BRCA2/FANCD1 accumulate at the same damagesites and show functional convergence (10). The FA pathway isknown to suppress genome instability upon encountering replication-fork stalling (7). Other than their bona fide roles in ICL repair,FA genes are known to play roles in replication (by promotingfork stability) and during mitosis (by controlling chromosomesegregation) (24). The role of the FA pathway in stabilizing stalledforks is beginning to emerge as an important mechanism for themaintenance of genome stability, and this function is clearly in-dependent of its role in ICL repair.

In this study, we demonstrate that RFWD3 plays an importantrole during unperturbed cell cycle, as cells lacking RFWD3 showdefects in cell survival, S-phase progression, and sister chromatidcohesion. Cells without RFWD3 show slower fork progressionand a prolonged S phase, suggesting its role in DNA replication.Consistent with its role in replication, RFWD3 is enriched at thereplication fork, as observed by nascent strand capture assay(17). We support the model that RFWD3 is a component of theFA pathway that has an essential function in ICL repair, and thatPCNA-mediated recruitment of RFWD3 to the fork and itsstabilization are essential for DNA replication. PCNA is the slidingclamp at the replication fork that is required for the processivity ofDNA replication (25). In addition to PCNA’s primary function totether different replication factors to the DNA template, PCNAalso acts as the interacting scaffold at the center of the replicationfork to coordinate various processes such as nucleosome assemblyand epigenetic inheritance, and plays a role during DNA damage

A

RPA2

RFWD3 si - + - + - +WT PIPm

RFWD3

H4

D

B

E F

ubiquitination

Timely removal of RPA

proper fork progression

replication defects

RFWD3

NO

RFWD3

RFWD3 siFlag-Ub +- ++ +- ++

- 21- - 21-

+MG132+MG132+DBeQ

RPA2

RPA2

PCNA

PCNA

IP:Flag

1%Input

75

100150

50

37

75

100150

50

37

37

37

RFWD3

RPA

PCNA

Polymerase Ubiquitin

*

100

Chromatin bound fraction

37

15

RFWD3100150

PCNA

RFWD3(high)

H4

RFWD3(low)

RPA1

RPA2

α-tubulin

S P

control si RPA1 si

S P

RPA2 si

S P

100

37

15

75

50

10037

PCNA

RFWD3

H4

100

37

15

S P

control si PCNA si

S P

37RPA2

C

contr

ol si

RPA1 si

1% input

RPA2 si

HA IgG50% IP

PCNA

(HA)RFWD3100

37

1 0.71 0.16 0.04

1 0.6 0.62 0.14

HA-RFWD3 -

contr

ol si

RPA1 si

RPA2 si

contr

ol si

RPA1 si

RPA2 si

Fig. 4. PCNA-mediated stabilization of the E3 ligaseRFWD3 to the replication fork is essential for ubiquiti-nation of RPA. (A) Chromatin fractionation in PCNA-depleted U2OS cells and immunoblot analysis. Notethat in the PCNA-siRNA–treated cells, there is signifi-cant reduction of total as well as chromatin-associatedRFWD3. P, chromatin-bound fraction; S, soluble frac-tion. (B) Chromatin fractionation in RPA1- and RPA2-depleted U2OS cells and immunoblot analyses usingRPA, PCNA, and RFWD3 antibodies. (C) IP using HAantibody from U2OS cells stably expressing HA-RFWD3that are treated with control, RPA1, or RPA2 siRNAs.RFWD3 and PCNAwere analyzed by immunoblotting.(D) Depletion of endogenous RFWD3 in U2OS cellsstably expressing HA-RFWD3-WT or HA-RFWD3-PIPm,and immunoblot analyses using RPA and RFWD3antibodies. Note the increase in RPA in the absence ofRFWD3 is rescued by HA-RFWD3-WT, but not by thePIP mutant. (E) In vivo ubiquitination assay in controland RFWD3-siRNA–treated cells [+carbobenzoxy-Leu-Leu-leucinal (+MG132); +MG132 and N2,N4-dibenzyl-quinazoline-2,4-diamine (DBeQ)] transfected withFLAG-ubiquitin (Flag-Ub). IP was performed withFLAG and immunoblotting with RPA2 and PCNA. Notethe reduction in RPA ubiquitination and a specific formof PCNA (labeled with an asterisk). (F) Cartoon dem-onstrating the role of RFWD3 in ubiquitination of RPAand DNA replication progression.

13286 | www.pnas.org/cgi/doi/10.1073/pnas.1814521115 Lin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

repair. Interestingly, PCNA is also known to be monoubiquitinatedby both the RAD18 E3 ligase in response to ICLs and by CRL4(Cdt2)to promote TLS associated with endogenous replication stress(26, 27). We observed that in the absence of RFWD3, manycomponents of the fork showed increased binding to thechromatin, including RPA, DNA polymerases, Cdc45, and TLSpolymerases, all of which point to stalling of the replication fork.Increase in monoubiquitination of PCNA is known to occurinstantaneously after fork stalling (28), and we observe robustmonoubiquitination of PCNA upon depletion of RFWD3. Theuncoupling of the helicase from the stalled polymerase is knownto cause an increased accumulation of ssDNA (as evidenced byincreased RPA), resulting in ubiquitination of PCNA (29, 30). Arecent study has reported the presence of p53 bound to PCNA atstalled forks for suppressing the extension from these forks (31).It is interesting to note that RFWD3 is known to positivelyregulate p53 stability (11). It is a possibility that in the absenceof RFWD3, persistence of stalled forks is because of defects inubiquitination of substrates like p53 and RPA. Identifying theentire repertoire of RFWD3 substrates during unperturbed DNAreplication and during DNA damage would tremendously improveour understanding.Depletion of RFWD3 has been shown to induce Chk1 and

Chk2 phosphorylation in the absence of DNA damage, corrob-orating our results that RFWD3 has a fundamental role in pro-moting the stability of unperturbed replication forks (12). We findthat PCNA is required for stabilizing RFWD3 to the chromatin.PCNA is known not only to influence the association of a largenumber of factors to chromatin, but also to affect their activity inmany instances (26). The PIP domain within RFWD3 is close to

the I639 moiety, a site that was found to be mutated in an FApatient and has previously been shown to be crucial for RPAbinding. Furthermore, it has previously been reported that RPA-mediated recruitment of RFWD3 to stalled forks is critical for ICLrepair (15). We find that RFWD3 can be stabilized on chromatin inthe absence of RPA, and our results support a model in whichRFWD3 associates with PCNA and localizes to the replication fork.The fork-associated RFWD3 is also important for the ubiquitina-tion of replication-fork components, including RPA, to ensureproper fork progression. One model would posit that the ubiquiti-nation of RPA enables the removal of RPA, thus allowing forkprogression. Cells lacking RFWD3 show increased accumulationof RPA, increased PCNA monoubiquitination, and, therefore,fork stalling. We propose that RFWD3 plays a key role in re-solving the DNA breaks and stalled forks that are natural oc-currences during regular DNA replication.

Materials and MethodsA detailed description of all the plasmids, antibodies, and experimental pro-cedures can be found in SI Appendix, Material and Methods. Experimentalprocedures include the iPOND assay, DNA fiber analysis, co-IP, and the in vivoubiquitination assay.

ACKNOWLEDGMENTS. We thank members of the S.G.P. and K.V.P. laborato-ries for discussions and suggestions. We thank Drs. J. Chen, A. Gambus, Z. Gong,A. Maréchal, K. Sato, D. Spector, B. Stillman, M. Takata, Y. Wang, and L. Zoufor providing reagents, protocols, and suggestions. We thank Dr. D. Rivier andMs. S. Adusumilli for critical reading of the paper. This work was supported by NSFAward 1723008 (to K.V.P.) and by NSF Faculty Early Career Development Pro-gram Award 1243372, the NSF Award 1818286, and NIH Grants 1R01GM099669and GM125196 (to S.G.P.).

1. Técher H, Koundrioukoff S, Nicolas A, Debatisse M (2017) The impact of replicationstress on replication dynamics and DNA damage in vertebrate cells. Nat Rev Genet 18:535–550.

2. Toledo L, Neelsen KJ, Lukas J (2017) Replication catastrophe: When a checkpoint failsbecause of exhaustion. Mol Cell 66:735–749.

3. Ulrich HD, Walden H (2010) Ubiquitin signalling in DNA replication and repair. NatRev Mol Cell Biol 11:479–489.

4. Hashimoto S, Anai H, Hanada K (2016) Mechanisms of interstrand DNA crosslink re-pair and human disorders. Genes Environ 38:9.

5. Akkari YM, Bateman RL, Reifsteck CA, Olson SB, Grompe M (2000) DNA replication isrequired to elicit cellular responses to psoralen-induced DNA interstrand cross-links.Mol Cell Biol 20:8283–8289.

6. Taniguchi T, et al. (2002) S-phase-specific interaction of the Fanconi anemia protein,FANCD2, with BRCA1 and RAD51. Blood 100:2414–2420.

7. Ceccaldi R, Sarangi P, D’Andrea AD (2016) The Fanconi anaemia pathway: New playersand new functions. Nat Rev Mol Cell Biol 17:337–349.

8. Federico MB, Campodonico P, Paviolo NS, Gottifredi V (2017) Beyond interstrandcrosslinks repair: Contribution of FANCD2 and other Fanconi anemia proteins to thereplication of DNA. Mutat Res 808:83–92.

9. Matsuoka S, et al. (2007) ATM and ATR substrate analysis reveals extensive proteinnetworks responsive to DNA damage. Science 316:1160–1166.

10. Knies K, et al. (2017) Biallelic mutations in the ubiquitin ligase RFWD3 cause Fanconianemia. J Clin Invest 127:3013–3027.

11. Fu X, et al. (2010) RFWD3-Mdm2 ubiquitin ligase complex positively regulates p53stability in response to DNA damage. Proc Natl Acad Sci USA 107:4579–4584.

12. Elia AE, et al. (2015) RFWD3-dependent ubiquitination of RPA regulates repair atstalled replication forks. Mol Cell 60:280–293.

13. Gong Z, Chen J (2011) E3 ligase RFWD3 participates in replication checkpoint control.J Biol Chem 286:22308–22313.

14. Liu S, et al. (2011) RING finger and WD repeat domain 3 (RFWD3) associates withreplication protein A (RPA) and facilitates RPA-mediated DNA damage response.J Biol Chem 286:22314–22322.

15. Feeney L, et al. (2017) RPA-mediated recruitment of the E3 ligase RFWD3 is vital forinterstrand crosslink repair and human health. Mol Cell 66:610–621.e4.

16. Inano S, et al. (2017) RFWD3-mediated ubiquitination promotes timely removal ofboth RPA and RAD51 from DNA damage sites to facilitate homologous recombina-tion. Mol Cell 66:622–634.e8.

17. Alabert C, et al. (2014) Nascent chromatin capture proteomics determines chromatindynamics during DNA replication and identifies unknown fork components. Nat CellBiol 16:281–293.

18. Sirbu BM, Couch FB, Cortez D (2012) Monitoring the spatiotemporal dynamics ofproteins at replication forks and in assembled chromatin using isolation of proteinson nascent DNA. Nat Protoc 7:594–605.

19. Mailand N, Gibbs-Seymour I, Bekker-Jensen S (2013) Regulation of PCNA-protein in-teractions for genome stability. Nat Rev Mol Cell Biol 14:269–282.

20. Warbrick E (1998) PCNA binding through a conserved motif. BioEssays 20:195–199.21. Brown S, Niimi A, Lehmann AR (2009) Ubiquitination and deubiquitination of PCNA

in response to stalling of the replication fork. Cell Cycle 8:689–692.22. Tanaka H, et al. (2009) Replisome progression complex links DNA replication to sister

chromatid cohesion in Xenopus egg extracts. Genes Cells 14:949–963.23. Prasanth SG, Prasanth KV, Siddiqui K, Spector DL, Stillman B (2004) Human Orc2 lo-

calizes to centrosomes, centromeres and heterochromatin during chromosome in-heritance. EMBO J 23:2651–2663.

24. Naim V, Rosselli F (2009) The FANC pathway and mitosis: A replication legacy. CellCycle 8:2907–2911.

25. Prelich G, et al. (1987) Functional identity of proliferating cell nuclear antigen and aDNA polymerase-delta auxiliary protein. Nature 326:517–520.

26. Choe KN, Moldovan GL (2017) Forging ahead through darkness: PCNA, still theprincipal conductor at the replication fork. Mol Cell 65:380–392.

27. Terai K, Abbas T, Jazaeri AA, Dutta A (2010) CRL4(Cdt2) E3 ubiquitin ligase mono-ubiquitinates PCNA to promote translesion DNA synthesis. Mol Cell 37:143–149.

28. Hedglin M, Benkovic SJ (2015) Regulation of Rad6/Rad18 activity during DNA damagetolerance. Annu Rev Biophys 44:207–228.

29. Chang DJ, Lupardus PJ, Cimprich KA (2006) Monoubiquitination of proliferatingcell nuclear antigen induced by stalled replication requires uncoupling of DNA poly-merase and mini-chromosome maintenance helicase activities. J Biol Chem 281:32081–32088.

30. Zhuang Z, et al. (2008) Regulation of polymerase exchange between Poleta andPoldelta by monoubiquitination of PCNA and the movement of DNA polymeraseholoenzyme. Proc Natl Acad Sci USA 105:5361–5366.

31. Hampp S, et al. (2016) DNA damage tolerance pathway involving DNA polymerase ιand the tumor suppressor p53 regulates DNA replication fork progression. Proc NatlAcad Sci USA 113:E4311–E4319.

Lin et al. PNAS | December 26, 2018 | vol. 115 | no. 52 | 13287

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021