PNAS PLUS

Poly(ADP-ribose) polymerase 1 escorts XPC to UV-

induced DNA lesions during nucleotide excision repair
Mihaela Robua,1, Rashmi G. Shaha,1, Nupur K. Purohita, Pengbo Zhoub, Hanspeter Naegelic, and Girish M. Shaha,2
a
Laboratory for Skin Cancer Research, Neuroscience Axis, Centre Hospitalier Universitaire de Québec Research Center-Université Laval, Québec, QC, Canada
G1V 4G2; bDepartment of Pathology and Molecular Medicine, Weill Cornell Medical College, New York, NY 10065; and cInstitute of Pharmacology and
Toxicology, University of Zurich, Zurich, CH-8057, Switzerland

Edited by James E. Cleaver, University of California, San Francisco, CA, and approved July 5, 2017 (received for review April 26, 2017)

Xeroderma pigmentosum C (XPC) protein initiates the global geno- in the genome (1) by association and quick dissociation until it en-
mic subpathway of nucleotide excision repair (GG-NER) for removal of counters the lesion site where it stabilizes due to a stronger associ-
UV-induced direct photolesions from genomic DNA. The XPC has an ation (4). Since the yeast ortholog of XPC forms identical crystal
inherent capacity to identify and stabilize at the DNA lesion sites, and structures with normal and UV-damaged DNA (5), it is proposed
this function is facilitated in the genomic context by UV-damaged that the discrimination between normal and damaged DNA de-
DNA-binding protein 2 (DDB2), which is part of a multiprotein UV–DDB pends on the difference in the kinetics of twisting and opening of
ubiquitin ligase complex. The nuclear enzyme poly(ADP-ribose) poly- the helix by XPC at these two sites (5, 6). Although this physical
merase 1 (PARP1) has been shown to facilitate the lesion recognition verification mechanism could work rapidly in vitro with small and
step of GG-NER via its interaction with DDB2 at the lesion site. Here, naked DNA, it would be too slow to explain the rapid recruitment
we show that PARP1 plays an additional DDB2-independent direct of XPC that is known to occur within minutes at UV-lesion sites in
role in recruitment and stabilization of XPC at the UV-induced DNA a complex eukaryotic chromatin. Therefore, the second proposed
lesions to promote GG-NER. It forms a stable complex with XPC in mechanism is that DDB2 helps XPC in finding the UV lesions in
the nucleoplasm under steady-state conditions before irradiation the genomic context. This is based on the observation that the
and rapidly escorts it to the damaged DNA after UV irradiation in Xeroderma pigmentosum group E (XP-E) cells, which are de-
a DDB2-independent manner. The catalytic activity of PARP1 is not ficient in DDB2 but proficient in XPC, can slowly repair 6–4PP
required for the initial complex formation with XPC in the nucleo- lesions but fail to remove CPD lesions, which constitute the ma-
plasm but it enhances the recruitment of XPC to the DNA lesion site jority of the lesions formed by UV radiation (7, 8). Based on the
after irradiation. Using purified proteins, we also show that the phenotype of XP-E cells and the biochemical studies of XPC with
PARP1–XPC complex facilitates the handover of XPC to the UV- these two types of lesions (2), it is proposed that, although XPC can
lesion site in the presence of the UV–DDB ligase complex. Thus, directly recognize the greater degree of helical distortion in DNA
the lesion search function of XPC in the genomic context is con- induced by 6–4PP lesions, it cannot recognize the smaller degree of
trolled by XPC itself, DDB2, and PARP1. Our results reveal a para- distortion caused by CPD lesions until DDB2 binds to these le-
digm that the known interaction of many proteins with PARP1 sions and increases the degree of helical distortion at the site (3).
under steady-state conditions could have functional significance Finally, the third proposed mechanism is that XPC recognizes the
for these proteins. remodeled chromatin at the lesion site, which is the result of events
initiated by ubiquitination of histones by the UV–DDB ligase
PARP1 | XPC | NER | DDB2 | UV complex (9). This is supported by the observations that decreased
histone ubiquitination impairs the eviction of the UV–DDB ligase

N ucleotide excision repair (NER) is a versatile pathway that

removes a wide variety of DNA lesions including UV radi-
ation (UV)-induced cyclobutane pyrimidine dimers (CPD) and
Significance

6–4 pyrimidine-pyrimidone photoproducts (6-4PP). There are Repair of the majority of UV-induced DNA damage in mamma-
two subpathways of NER: the global genomic NER (GG-NER) lian cells by the nucleotide excision repair pathway starts with
that removes the majority of lesions from the entire genome and rapid recruitment of Xeroderma pigmentosum C (XPC) protein to
the transcription-coupled NER that repairs the minority of total the lesion. However, rapidity of XPC recruitment to the lesion
lesions that occur on the transcribed strand (1). The GG-NER site in a genomic context cannot be fully explained by the
process is dependent on the Xeroderma pigmentosum C (XPC) known properties of XPC or its partner protein DDB2. Here, we
protein, the arrival and stabilization of which at the lesion site, fol- show that the DNA damage-detecting nuclear protein PARP1
lowed by its timely departure, are crucial for permitting the down- forms a stable complex with XPC before DNA damage and
stream NER events (2). XPC accomplishes some of these tasks on its transfers it very rapidly to the DNA lesion site if other repair
own or with the help of processes initiated by two proteins that in- conditions are present. Since PARP1 is known to interact with
dependently reach the lesion site very rapidly, namely UV-damaged many proteins under steady-state conditions, our results reveal a
DNA-binding protein 2 (DDB2) and poly(ADP-ribose) polymerase paradigm that an association with PARP1 could confer a func-
tional advantage to these proteins.
1 (PARP1). It is known that once XPC reaches the vicinity of the
DNA lesion site, the UV–DDB ubiquitin ligase complex containing Author contributions: M.R., R.G.S., N.K.P., and G.M.S. designed research; M.R., R.G.S., and
DDB1, DDB2, Cul4A, and Rbx1 regulates its specific binding and N.K.P. performed research; M.R., R.G.S., N.K.P., P.Z., and H.N. contributed new reagents/
stabilization at the site (2). However, we have the least under- analytic tools; M.R., R.G.S., N.K.P., P.Z., H.N., and G.M.S. analyzed data; and M.R., R.G.S.,
standing of the ability of XPC to rapidly search for and arrive at the N.K.P., P.Z., H.N., and G.M.S. wrote the paper.

very few lesions surrounded by a vast majority of undamaged bases The authors declare no conflict of interest.

in the chromatin structure (3). This article is a PNAS Direct Submission.

CELL BIOLOGY

There are three proposed mechanisms by which XPC could be Freely available online through the PNAS open access option.
rapidly recruited to the lesion site in the genomic context. The first 1
M.R. and R.G.S. contributed equally to this work.
mechanism is based on XPC’s inherent capacity to interrogate the 2
To whom correspondence should be addressed. Email: [email protected].
DNA for verifying the presence or absence of a lesion. It has been This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
suggested that XPC searches for the lesion-induced helical distortion 1073/pnas.1706981114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1706981114 PNAS | Published online July 31, 2017 | E6847–E6856

complex (10) and recruitment of XPC to the lesions site (11). Thus, DDB2 accumulates in the chromatin fraction after UV irradiation
apart from the direct recognition of the DNA damage by XPC, all (Fig. 1A). The XPC-immunoprecipitation (IP) of chromatin ex-
other mechanisms to explain XPC’s ability to rapidly reach most of tracts with equal protein content from control and UV-irradiated
the UV lesions in the genomic context depend on DDB2. How- cells revealed a significant increase in UV-induced association of
ever, XPC must have some DDB2-independent mechanism of PARP1 with XPC on the chromatin (Fig. 1B, Left). We also ob-
recruitment to lesions other than 6–4PP. For example, CPD are served a UV-induced increase in the interaction of DDB2 with
recognized and repaired even after DDB2 is degraded (12) or XPC at the chromatin, which is in agreement with our previous
when the number of lesion sites exceeds the number of DDB2 observation of this interaction identified by DDB2-IP (18). The
molecules in the cell (13). In addition, XPC recognizes and starts reciprocal PARP1-IP using FLAG antibody revealed a strong UV-
NER at other types of DNA damages, such as bulky adducts and induced association of XPC with PARP1 in the chromatin-bound
cross-links that are not likely to be recognized by DDB2, because fraction (Fig. 1B, Middle). Since PARP1 and XPC are present in
these lesions cannot be accommodated in DDB2′s recognition the nucleoplasm before and after irradiation (Fig. 1A), we exam-
pocket (14–16). ined whether they also interact with each other in this subnuclear
Earlier, we showed that PARP1, an abundant nuclear enzyme in fraction. The reciprocal IP for XPC and PARP1 in the nucleo-
higher eukaryotes, is recruited very early to the UV-lesion site and plasmic fraction revealed a strong association of these two proteins
catalytically activated to form polymers of ADP-ribose (PAR) (17, not just after UV irradiation but also under control conditions
18). It has been shown by others and our team that PARP1 and before irradiation (Fig. 1C). Both the IPs failed to pull down
DDB2 reach the lesion site in the same time frame and cooperate DDB2 from control nucleoplasm, indicating a DDB2-independent
with each other to increase the efficiency of GG-NER. More spe- nature of PARP1–XPC nucleoplasmic interaction before irradia-
cifically, DDB2 has been shown to stimulate catalytic activity of tion. Even after UV irradiation, XPC-IP confirmed its lack of in-
PARP1, which in turn PARylates DDB2 and DDB1 (18, 19). The teraction with DDB2 in nucleoplasm, whereas PARP1-IP revealed
inhibition of PARylation has been shown to block the turnover of a feeble interaction of PARP1 with DDB2 (Fig. 1C), reflecting
DDB2 at UV-damaged chromatin (18) and to decrease total cel- some turnover of the PARP1–DDB2 complex from postirradiation
lular levels of DDB2 (19). Additionally, the activated PARP1 chromatin to nucleoplasm.
promotes chromatin decondensation by DDB2 (20) via recruitment The interaction between PARP1 and XPC in UV-irradiated
of the remodeler protein ALC1 at the UV-lesion site (19). To- chromatin was expected, as both are known to bind to UV-
gether, these studies suggest that the interplay of PARP1 and damaged DNA. However, their interaction in nucleoplasm under
DDB2 at UV-damaged DNA could be a mechanism for recruit- steady-state conditions before UV challenge was unexpected; there-
ment and stabilization of XPC at UV-damaged chromatin (18, 20). fore, we examined this interaction in further detail using multiple
Recent studies have shown PARylation of XPC in vitro (21) and in cellular and in vitro models. To exclude the possibility that this could
the cells responding to oxidative DNA damage (22). However, the be an artifact of the expression of exogenous FLAG-tagged PARP1 in
significance of the PARylation of XPC in NER of UV-induced the above model, we examined this interaction in HEK293T cells that
DNA damage is not clear since the higher affinity of XPC for express endogenous XPC, PARP1, and DDB2. The mass spectrom-
larger PAR chains shown in vitro (21) would repel XPC from DNA, etry of the proteins that coimmunoprecipitate (co-IP) with XPC from
and in vivo PARylation of XPC was not observed in UV-irradiated the lysates of unirradiated HEK293T cells confirmed the presence of
cells (22). Thus, studies to date have not shown an unequivocal PARP1 as well as two known partners of XPC, namely HR23B and
direct DDB2-independent role for PARP1 and PARylation in re- Centrin2 (Fig. 1D). The reciprocal IP for PARP1 and XPC validated
cruitment of XPC in GG-NER. the interaction of even the endogenous PARP1 with XPC in this
Here, we show that PARP1 stably interacts with XPC in the fraction (Fig. 1E). Interestingly, the mass spectrometry of XPC
nucleoplasm under unchallenged conditions, i.e., in the absence of eluate did not reveal the presence of DDB2, confirming its lack
any type of exogenously induced DNA damage. The functionally of interaction with XPC under control conditions. The DDB2-
important DNA-binding region of XPC is involved in its interaction independent interaction between endogenous PARP1 and endoge-
with PARP1. Using various cellular models and in vitro assays with nous XPC was further confirmed in the nucleoplasm of DDB2-
purified proteins, we show that, after UV irradiation, PARP1 rap- deficient XP-E cells (Fig. 1F). Thus, using different cellular models
idly escorts XPC to the UV-lesion site and facilitates its handover to expressing endogenous or exogenous PARP1, and the variable status
the damaged DNA in the presence of the UV–DDB ligase complex. of DDB2, we consistently observed a DDB2-independent interaction
We also show that, although the PARP1 catalytic function does not between XPC and PARP1 in the nucleoplasm before and after irra-
influence the initial interaction between these two proteins in the diation and an increased interaction at the chromatin after irradiation.
nucleoplasm, it is required for efficient recruitment of their complex
to the lesion site. Our results reveal that the interaction of XPC with Direct Interaction of XPC and PARP1. PARP1 and XPC are DNA-
PARP1 in nucleoplasm under steady-state conditions facilitates the binding proteins; therefore, we examined the possibility that DNA
search function of XPC for DNA damage in the genomic context. could be mediating their interaction in the nucleoplasm. Unlike
chromatin fraction, the nucleoplasmic fraction of the unirradiated
Results HEK293T cells contained undetectable levels of DNA (Fig. S1).
DDB2-Independent Nucleoplasmic Interaction of PARP1 with XPC Very high doses of micrococcal nuclease (MNase) or benzonase are
Before and After DNA Damage. Within minutes after UV irradia- often used for digesting DNA from chromatin preparations to
tion, XPC, DDB2, and PARP1 are present at the UV-induced DNA demonstrate direct interactions of proteins. In our IP studies, we
lesions in cells. Although independent interactions of DDB2 with prepared a chromatin-bound protein fraction after treatment of
XPC (18, 23) or PARP1 (18, 19) at the lesion site are known, here we chromatin pellets with very low-dose MNase (25 U/mL). Since
examined whether XPC interacts directly with PARP1 on the UV- treatment of HeLa cell extracts with high-dose MNase was shown
damaged chromatin. The FLAG-PARP1–expressing human skin fi- to digest the DNA to mononucleosomes (23), we treated the
broblasts (GMRSiP) were fractionated before and after UV irradi- chromatin fraction of HEK293T cells with different doses of
ation to prepare cytoplasm, nucleoplasm, and chromatin-bound MNase up to 4,000 U/mL or benzonase up to 50 U/mL to digest it
protein fractions (Fig. 1A), as described earlier (18). In this protocol, down to mononucleosomal DNA at 147 bp (Fig. S1). Although
the chromatin proteins such as histone H3 do not leak in the nu- there was no detectable DNA in the nucleoplasmic fraction, we still
cleoplasmic fraction but are extracted in the chromatin-bound treated this fraction from control or UV-treated HEK293T cells
protein fraction. Moreover, we further validated this fraction- with 4,000 U/mL MNase before subjecting it to XPC-IP and ob-
ation protocol by confirming our earlier observation (18) that served that MNase did not break the association of XPC with

E6848 | www.pnas.org/cgi/doi/10.1073/pnas.1706981114 Robu et al.

 PNAS PLUS
Fig. 1. PARP1 interacts with XPC in the nucleoplasmic and chromatin fractions. (A) The GMRSiP cells expressing FLAG-tagged PARP1 were irradiated with
30 J/m2 UVC (or control), and whole-cell extracts (WCE) were fractionated to obtain cytoplasm (Cyt), nucleoplasm (Np), and chromatin-bound (Ch) fractions.
The proteins from each fraction were immunoblotted for XPC, PARP1, and DDB2. Beclin and histone H3 were used as cytoplasmic and chromatin markers,
respectively. The asterisk indicates a nonspecific band. The Ponceau S staining reflects the protein content at the end of each fractionation step. (B and C) The
chromatin (B) and nucleoplasm fractions (C) of GMRSiP cells prepared as described above were subjected to IP for XPC, FLAG (PARP1), and mouse IgG
(negative control), followed by the detection of PARP1, XPC, and DDB2. (D) Table showing the XPC-interacting proteins identified after XPC-IP of
HEK293T cells followed by mass spectrometry analysis. (E) PARP1-IP and XPC-IP were performed in the Np fractions of control and UVC-treated HEK293T cells
as shown in C. The Input and IP eluates were probed for XPC and PARP1. (F) PARP1-IP was performed in the Np fractions of control, and UVC-treated XP-E cells
as described above. The Input and IP eluates were probed for XPC and PARP1. For B, C, E, and F the Ponceau S staining was used as loading control, and results
shown here are representative of results from two to four experiments.

PARP1 (Fig. 2A), confirming DNA-independent interaction of interaction of PARP1 with XPC-C1 fragment, a weak interaction
these two proteins in the nucleoplasm. with XPC-C2 fragment and no interaction with XPC-N fragment,
To identify the domains of XPC involved in interaction with indicating that 496–734 aa is key region of XPC that interacts with
PARP1, we expressed GFP-tagged XPC and its five different par- PARP1 (Fig. 2C, Lower). These results are in agreement with the
tially overlapping fragments in HEK293T cells (Fig. 2B, Top). The above in vivo data (Fig. 2B) showing a comparatively weaker in-
extracts from these cells under control conditions were subjected to teraction with PARP1 for the N-terminal XPC fragment (1–495 aa)
PARP1-IP followed by immunoblotting for GFP. Since the ex- compared with a strong interaction seen with full-length XPC or four
pression levels of full-length XPC and its fragments varied greatly of its fragments that contain the central region of XPC. The results
after transfection (Fig. 2B, Input lanes), the strength of interaction with purified proteins and in vivo data with cells expressing XPC
of each XPC fragment with PARP1 was measured as a fraction of fragments strongly indicate an interaction of the central portion of
total input protein that coimmunoprecipitated with PARP1 and
the XPC protein spanning 496 to 734 aa with PARP1 without in-
expressed relative to the interaction of full-length XPC with
tervention of other proteins or DNA.
PARP1 (Fig. 2B, Lower). While control GFP protein did not in-
To identify the domains of PARP1 implicated in this interac-
teract with PARP1, four of the XPC fragments that span from
tion, its N-terminal (1–232 aa) fragment was expressed in GMSiP
427 to 940 aa showed an interaction with PARP1 similar to that
seen with full-length XPC, whereas a comparatively weaker in- cells in which endogenous PARP1 was depleted by shRNA,
teraction was observed with an N-terminal fragment (1–495 aa). whereas the C-terminal (232–1014 aa) fragment of PARP1 was
The interaction observed above between XPC or its fragments expressed in PARP1−/− A1 cells. For control, we used GMRSiP
and PARP1 after PARP1-IP could not exclude the possibility that cells that express FLAG-tagged full-length PARP1 in the GMSiP
PARP1-interacting proteins may be indirectly mediating their in- cells (Fig. 2D, Top). The XPC-IP of the extracts of these cells
CELL BIOLOGY

teraction. Therefore, we examined the interaction of purified revealed that the C-terminal fragment of PARP1 is implicated in
PARP1 in vitro with equimolar amounts of three purified fragments its interaction with XPC (Fig. 2D, Lower). Collectively, these
of XPC, namely GST-tagged 141–250 aa (XPC-N), His-tagged 496– results demonstrate that the region of XPC that is involved in its
734 aa (XPC-C1), and His-tagged 734–933 aa (XPC-C2) (Fig. 2C, interaction with DNA, DDB2, and HR23B (2) is also involved in
Top). The PARP1-IP of above reaction mixtures revealed a strong its interaction with PARP1.

Robu et al. PNAS | Published online July 31, 2017 | E6849

Fig. 2. Identification of the domains implicated in the interaction between PARP1 and XPC. (A, Left) IP for XPC and rabbit IgG were performed in the
nucleoplasm of HEK293T cells prepared as described in Fig. 1E. (A, Right) Nucleoplasmic extracts were also treated with 4,000 U/mL MNase and subjected to
XPC-IP. The Input and eluates were probed for XPC and PARP1. Ponceau S was used as loading control. (B, Top) Pictogram of GFP-tagged full-length XPC and
its five fragments used in the study. The domains marked as TGD and BHD1-3 represent the transglutaminase homology domain and β hairpin domains 1–3.
(B, Lower) The HEK293T cells were transiently transfected with GFP-tagged full-length XPC or its fragments for 48 h, and cell extracts were subjected to
PARP1-IP. The Input and the IP eluates were analyzed for GFP (XPC) and PARP1. The relative intensity of the IP band was measured as a fraction of the total
input protein. The strength of the interaction between XPC fragments and PARP1 was expressed as relative to the interaction of full-length XPC with PARP1.
(C, Top) The pictogram of the full-length XPC and its fragments used in this study. (C, Lower) The XPC fragments were reacted with pure PARP1 for 30 min at
25 °C, followed by PARP1-IP on magnetic beads. The bead elutes were probed for PARP1, GST (XPC-N), and histidine (XPC-C1 and XPC-C2). (D, Top) Pictogram
showing the domains of full-length PARP1 and its N- or C-terminal fragments used in this study. The DBD, AMD, and CAT are the DNA-binding, automo-
dification, and catalytic domains, respectively. (D, Lower) The PARP1-depleted GMSiPs were transiently transfected with the full-length FLAG-PARP1 or its
N-terminal fragment (GFP-DBD) and the PARP1−/− A1 cells with the C-terminal fragment for 48 h. The cell extracts were subjected to XPC-IP, and the bead
eluates were analyzed for XPC and PARP1 or its fragments (F1-23 and C2-10). The data represent similar results observed in two experiments.

Roles of the PARP1–XPC Complex and Catalytic Activity of PARP1 in Interestingly, the presence of PARPi did not abolish the interac-
Recruitment of XPC to UV-Damaged DNA. PARP1 is known to tion between XPC and PARP1, indicating that catalytic function
rapidly reach damaged DNA in the chromatin context; hence, the of PARP1 is not required for this interaction in the nucleoplasm
PARP1–XPC complex formed in the nucleoplasm before irradi- (Fig. S2B). Using this model, we tracked UV-induced movement
ation could have a role in rapidly escorting XPC to the DNA le- of biotinylated XPC and PARP1 from the nucleoplasmic to
sion site after UV irradiation. To trace the intranuclear movement chromatin fractions of control or UV-treated cells (Fig. 3A). In the
of PARP1-bound XPC, we used the proximity-dependent biotin streptavidin-IP of nucleoplasm, both PARP1 and XPC were de-
identification (Bio-ID) technique (24). The FLAG-PARP1 was tectable before and after irradiation, confirming biotinylation of
cloned in the myc-tagged biotin ligase BirA vector to create a XPC and auto-biotinylation of Bio-ID-PARP1 (Fig. 3A, Left, and
new Bio-ID-PARP1 vector. The expression of Bio-ID-PARP1 in Fig. S2D). The decrease in signal for biotinylated XPC after UV
GMSiP cells depleted of endogenous PARP1 in the nuclear irradiation suggested a movement of XPC away from this fraction.
fraction (Fig. S2A) ensured that any nuclear protein that stably Interestingly, PARPi blocked the reduction in the signal for XPC
associates with or stays within 10 nm of the cloned PARP1 would after UV irradiation, indicating a potential role of catalytic activity
be stably biotinylated. The absence of endogenous PARP1 in of PARP1 in the movement of XPC from the nucleoplasmic
these cells eliminated the possibility that any PARP1-associated fraction to damaged DNA. The absence of biotinylated DDB2 in
protein would not be biotinylated. For optimal biotinylation, we streptavidin-IP (Fig. 3A, Left), once again confirmed that DDB2
incubated these cells for 16–18 h with biotin before irradiation did not interact with Bio-ID-PARP1 or its complex with XPC in
(24) and confirmed biotinylation of XPC and auto-biotinylation the nucleoplasm.
of PARP1 in XPC-IP eluates of the nucleoplasm of control and To examine whether the biotinylated nucleoplasmic XPC rea-
UV-irradiated cells (Fig. S2B). To examine the role of the catalytic ches the UV-damaged DNA, we performed the streptavidin-IP of
activity of PARP1 in this interaction, we also treated the cells with corresponding chromatin fractions from these cells (Fig. 3A,
PARPi PJ-34 under conditions that abolished the signal for PAR- Right). There was some signal for biotinylated XPC under steady-
modified proteins in control or UV-irradiated cells (Fig. S2C). state conditions, which represents a basal level of interrogation of

E6850 | www.pnas.org/cgi/doi/10.1073/pnas.1706981114 Robu et al.

 PNAS PLUS
DNA at all times by the multifunctional XPC protein and an in-
crease in this signal in UV-treated cells, which was suppressed by
PARPi (Fig. 3A, Right). Our results are in agreement with a
previous report that XPC constantly associates–dissociates with
chromatin under steady-state conditions; an additional association
of XPC is seen with chromatin after UV-induced DNA damage
(4). The UV-mediated increase in biotinylated XPC at chromatin
with a corresponding decrease in nucleoplasm suggests that either
PARP1-bound XPC moved from nucleoplasm to UV-damaged
chromatin or a completely independent XPC arrived at the le-
sion site and was biotinylated by PARP1 within 30 min after ir-
radiation when the samples were harvested. However, within the
same time period of 30 min postirradiation, DDB2, which is
known to closely interact with PARP1 at the chromatin immedi-
ately after UV irradiation (18, 19), was not strongly biotinylated by
Bio-ID-PARP1 (Fig. 3A, Right). Thus, the biotinylated XPC that is
deposited on the UV-damaged chromatin must have originated
from nucleoplasmic PARP1–XPC complexes formed before UV
damage, and PARPi suppressed this movement.
To validate the inhibitory effect of PARPi on the movement of
XPC from nucleoplasm to chromatin, we used immunocyto-
logical methods to visualize XPC at the site of local ultraviolet C
(UVC) irradiation up to 3 h in the NER-proficient GMU6 hu-
man skin fibroblasts with or without treatment with PARPi (Fig.
3B). The GMU6 fibroblasts were locally irradiated with UVC
through 5-μm pores in a polycarbonate filter, which produces
distinct subnuclear areas of irradiation that are surrounded by
unirradiated zones in the nucleus (25). Unlike Western blot data
that reveal total XPC molecules that are bound to both UV-
damaged and undamaged portions of chromatin, the immuno-
cytological data with local irradiation reflect the status of XPC
only in the irradiated subnuclear zones, thus excluding the
“noise” of XPC signals from the unirradiated portions of the
nucleus. In the subnuclear irradiated zones identified by staining
with thymine dimer (T-T) CPD-specific antibody, the endogenous
XPC followed a normal kinetics, i.e., initial strong accumulation at
10 min followed by a steady decline to 40% of initial levels in
90 min. The treatment of cells with PARPi, which could abolish
the signal for PAR-modified proteins in control or UV-irradiated
cells (Fig. S2E), also suppressed the initial recruitment of XPC at
10 min by 50%. Moreover, PARPi slowed down the departure of
Fig. 3. Efficient recruitment of PARP1 and XPC to the UV lesion requires
PARP1 catalytic activity. (A) Bio-ID-PARP1 cells expressing myc-Bio-ID-FLAG-
XPC from the lesion site in the first 90 min, compared with the
PARP1 were irradiated (30 J/m2 UVC or control) with or without PARPi (PJ- rapid turnover of XPC in cells not treated with PARPi (Fig. 3B,
34). The nucleoplasm (Left) and chromatin-bound (Right) fractions were Right). Thus, the major impact of PARP inhibition was in partially
subjected to streptavidin-IP. The eluates were analyzed for myc (PARP1), suppressing the initial recruitment of XPC to UV-lesion sites.
XPC, and DDB2. The data represent similar results observed in three exper- Since XPC is in a complex with PARP1 in the nucleoplasm, we
iments, and Ponceau staining provides loading control. The “#” refers to a examined the effect of PARPi on the recruitment of PARP1 itself to
nonspecific band. (B) The XPC kinetics at the UV lesions were monitored in UV-damaged chromatin. Using cells expressing GFP-PARP1 and a
GMU6 cells up to 180 min after local UVC irradiation with 100 J/m2 UVC
through 5-μm pores of a polycarbonate filter with or without PARPi PJ-34.
recently developed in situ extraction technique that can selectively
The background-corrected signal for XPC (green) at T-T spots (red) relative to identify DNA-bound PARP1 (26), we observed that PARPi PJ-
the 10-min signal is represented as mean ± SEM (200–500 spots from three 34 suppressed by 50% the initial recruitment of GFP-PARP1 to
experiments). Note: In all panels of this and subsequent figures, an asterisk (*) local UV-irradiated subnuclear zones, which were identified by
denotes a statistically significant difference with P value < 0.05 with un- immunostaining for T-T (Fig. 3C, Lower). The Z-stack images (Fig.
paired two-tailed t test. (C and D) The GMU6 cells were transiently trans- S3A) and their orthogonal view (Fig. S3B) of the locally irradiated
fected with GFP-PARP1 for 24 h and locally irradiated with 100 J/m2 UVC in
GMU6 cells confirmed the spatial colocalization of the GFP-
the absence or presence of PARPi PJ-34. (C ) The GFP (PARP1) signals at local
T-T spots (red) after background correction were pooled from 200 to
PARP1 with T-T (Fig. 3C). The immunofluorescence image with
300 spots derived from three experiments and expressed relative to the multiple locally irradiated cells presented in 2D and 2.5D format
signal observed in cells not treated with PARPi. (D) A representative 2D- revealed that PARPi treatment significantly reduced the intensity of
merged image for GFP (PARP1) and T-T (red) colocalization and the or- colocalized signal for GFP-PARP1 at the site of DNA damage (Fig.
thogonal view (2.5D image) for the same field is shown to visualize signal 3D). Interestingly, the recruitment of PARP1 itself to a UV-lesion
intensity of T-T and GFP. The x and y axis represent distance in nanometers, site is not dependent on XPC because it occurs to an identical extent
CELL BIOLOGY

and the z axis represents fluorescence intensity in arbitrary units. (E) The
in both XPC-proficient (GMU6) and -deficient (XP-C) cells (Fig.
accumulation of the GFP (PARP1) at T-T lesions was monitored in
GMU6 and XPCU6 cells 10 min after irradiation with UVC at 100 J/m2. The
3E). Collectively, our results indicate that the initial phase of the
background-corrected GFP signal at lesion sites relative to the signal ob- basal level of recruitment of PARP1 and XPC to UV-damaged
served in GMU6 cells is expressed as mean ± SEM derived from ≥150 spots DNA does not require catalytic activity of PARP1, whereas the
from two experiments. second phase occurs in response to PARP1 activation.

Robu et al. PNAS | Published online July 31, 2017 | E6851

PARP1-Mediated Recruitment of XPC to UV-Lesion Sites Is Independent
of DDB2. DDB2 is known to recruit XPC to the UV-lesion site, and
since DDB2 and PARP1 interact with each other to facilitate
NER (27), our results of suppressed XPC recruitment to the
UV-lesion site could also be an indirect effect of PARPi on the
role of DDB2 in recruiting XPC. To exclude this possibility, we
used DDB2-deficient XP-E cells to examine the effect of
PARPi on colocalization of XPC with 6–4PP lesions after local
UV irradiation. At the lesion sites, the signal for XPC declined
rapidly by 50–60% from 10 to 60 min (Fig. 4A). In contrast,
PARPi treatment not only reduced the initial recruitment of
XPC at lesion site by 50%, but also slowed down XPC turn-
over up to 60 min, a trend that was also observed in DDB2-
proficient GM cells (Fig. 3B). An identical profile of suppression
of XPC recruitment and turnover in DDB2-proficient and
-deficient cells indicates that the suppression of XPC re-
cruitment by PARPi is not mediated via DDB2. The biological
end-point of XPC recruitment to the lesion site is the repair of
UV-damaged DNA. Therefore, we measured the kinetics of
removal of 6–4PP lesions up to 8 h following global UV irra-
diation in XP-E cells. As expected, almost all 6–4PP lesions
were removed in XP-E cells by 8 h, but the treatment with
PARPi significantly slowed down this repair process (Fig. 4B).
Since the depletion of PARP1 completely suppressed any PAR
formation in response to UV irradiation without affecting PARP2
expression (18, 28, 29), these results indicate that XPC recruit-
ment is partially controlled by the catalytic activity of PARP1 in
a DDB2-independent manner. Collectively, our findings demon-
strate that the inhibition of PARP1 activity reduces recruitment of
XPC to UV damage with a direct negative consequence on the
repair of UV-induced lesions.
To determine the extent of the contribution of DDB2 and
PARP1 in the recruitment of XPC to DNA lesions, we used four
different cell lines described earlier (29) (Fig. 4C): CHOSiP cells
(deficient in both PARP1 and DDB2); CHOU6 cells (deficient only Fig. 4. XPC recruitment to UV lesions by PARP1 is DDB2-independent. (A) XP-E
in DDB2); GMSiP cells (deficient only in PARP1); and GMU6 cells with or without PARPi (ABT-888) were locally irradiated with 100 J/m2 UVC
(proficient in both DDB2 and PARP1). In each cell type, we ex- and probed for XPC (red) and 6–4PP sites (green) at different times. The
amined the early accumulation of XPC at the local UV-lesion site background-corrected signal for XPC at 6–4PP spots is represented as mean ±
SEM (300 spots derived from three experiments). (B) The DDB2-deficient XP-E
(Fig. 4C, Right). The CHOSiP cells displayed a basal level of ac-
cells, with or without PARPi ABT-888, were globally irradiated with 10 J/m2 UVC
cumulation of XPC at the UV-lesion site, which represents the and immunostained for 6–4PP lesions to determine its repair kinetics up to 8 h.
inherent capacity of XPC to reach the lesions without the help of The data are presented as signal intensity relative to the maximum signal at
DDB2 or PARP1. Relative to this basal level, the presence of only 10 min (mean ± SEM, n ≥ 300 nuclei from three experiments). (C) The four cell
PARP1 (CHOU6) or only DDB2 (GMSiP) increased XPC re- lines with differing status of PARP1 and DDB2, as shown at the Right, were lo-
cruitment by about 40%. However, the presence of both DDB2 and cally irradiated with 100 J/m2 and probed for T-T and XPC at 10 min after irra-
PARP1 in GMU6 cells nearly doubled the XPC accumulation diation. Background-corrected signal for XPC at the T-T under identical exposure
compared with the CHOSiP model, strongly supporting in- conditions was calculated as mean ± SEM derived from 300 to 700 spots from
three experiments. The accumulation of the XPC at the T-T lesion in CHOU6,
dependent roles of DDB2 and PARP1 in this process.
GMSiP, and GMU6 is expressed as fold increase over that observed in CHOSiP. (D)
The total cellular levels of XPC in PARP1-depleted GMSiP and The total cell extracts of the four indicated cell lines were separated on the SDS/
CHOSiP cell lines were similar to PARP1-proficient GMU6 and PAGE and probed for PARP1, XPC, and DDB2. Actin and Ponceau S were used as
CHOU6 cells (Fig. 4D); therefore, the suppression of the recruit- loading controls. (E) The four cell lines from above were globally irradiated with
ment of XPC to the damaged site in PARP1-depleted cells was not 30 J/m2 UVC and fractionated after the time indicated. The chromatin extracts
an artifact of reduced XPC expression in these cells. Since the re- with equal protein content were separated on SDS/PAGE and probed for XPC,
cruitment of the downstream NER proteins depends on XPC XPA, and PARP1. Ponceau S staining was used as loading control.
loading and stabilization at the damage site (1), we measured the
accumulation of Xeroderma pigmentosum A (XPA) protein on the
Characterization of the Handover of XPC from Its Complex with
chromatin-bound fraction of these cells up to 4 h after damage (Fig.
PARP1 to UV-Damaged DNA. Our results show that PARP1 and
4E). In each of these cellular models, the kinetics of XPA re-
cruitment reflected the status of XPC at the UV-damaged chro- XPC form a complex in the nucleoplasm, and the biotin tag on
matin. The XPA accumulation was robust in GMU6 cells and XPC at the UV-lesion site indicates a physical handover of XPC
reduced in the absence of PARP1 (GMSiP cells) or DDB2 (CHO from its complex with PARP1 to UV-damaged DNA. To explore
cells), whereas the weakest accumulation and a rapid turnover of the mechanistic aspect of this transfer in vivo, we carried out GFP
XPA were seen in the CHOSiP cells devoid of both DDB2 and (XPC)-IP of chromatin fractions from HEK293T cells expressing
PARP1. Collectively, our results show that there are three com- GFP-tagged XPC up to 3 h after irradiation and examined the state
ponents of recruitment of XPC to a DNA lesion site: the first is a of association of PARP1 with XPC (Fig. 5A, Left). The IP revealed
basal level of recruitment controlled by XPC itself; the second is a normal kinetics of recruitment and departure of XPC at the UV-
dependent on PARP1 or DDB2 proteins; and the last component damaged chromatin with a strong accumulation at 30 min and a
depends on the catalytic activation of PARP1. significant reduction by 3 h. In contrast, the amount of PARP1 that

E6852 | www.pnas.org/cgi/doi/10.1073/pnas.1706981114 Robu et al.

 PNAS PLUS
is associated with GFP-XPC decreased rapidly from 30 to 90 min, To explore the conditions required for XPC to dissociate from its
and no signal was detected at 180 min. It is noteworthy that, al- complex with PARP1 and bind to UV-damaged DNA, we designed
though a significant amount of PARP1 was still present in the in vitro assays using the factors prevalent at the lesion site in vivo,
chromatin fraction from 30 to 180 min (as seen in the Input namely UV-damaged DNA, DDB2, PARP1, and XPC represented
samples in Fig. 5A, Left), it was not associated with XPC after the by its key fragment 496–734 aa (ΔXPC) that interacts with PARP1
peak period of recruitment of XPC to the lesion. A similar kinetics (Figs. 5 B–D). To examine the endogenous properties of these three
of association and dissociation of XPC and PARP1 was observed in proteins to bind to UV-damaged DNA, we reacted them singly or
chromatin over 3-h period after exposure to 30 J/m2 UVC in GM in combinations with UVC-irradiated plasmid DNA, which was
cells that express endogenous XPC and PARP1 (Fig. 5A, Middle), immobilized on the magnetic beads via T-T antibody (Fig. S4). All
demonstrating the general nature of this observation. Moreover, of the proteins could bind to UV-damaged DNA on their own and
XPC-IP of the chromatin-bound fraction of the same GM cells even in combination with other proteins (Fig. 5B). Although this
10 min after exposure to various UVC doses up to 100 J/m2 assay confirms the inherent capacity of XPC, DDB2, and PARP1 to
revealed a dose-dependent increase in the interaction between bind to UV-damaged DNA, it does not reveal how DDB2 or
XPC and PARP1 at the chromatin (Fig. 5A, Right). On the other PARP1, which are recruited before XPC, could participate in the
hand, despite an abundance of DDB2 in the input chromatin loading of XPC to the lesion site.
fraction at all doses, there was a dose-dependent decrease in as- Since XPC exists as a complex with PARP1 in the nucleoplasm
sociation of DDB2 with XPC (Fig. 5A, Right). The dose-dependent before reaching UV-damaged chromatin, we examined whether
increase in association of XPC and PARP1 at an early time point this PARP1–XPC complex would simulate conditions for in vivo
and their dissociation at a later time support a model that early loading of XPC to UV-DNA. We immobilized the PARP1–ΔXPC
recruitment of XPC occurs as a complex with PARP1 but, having complex on the agarose beads with PARP1 antibody (A beads)
reached the lesion site, XPC gradually dissociates from PARP1 to and reacted it with the above-described UV-damaged DNA
continue with its functions in NER. bound to magnetic beads (M1) in the presence of either PARP1

Fig. 5. Role of PARP1 and the UV–DDB ligase complex in the handover of XPC to the lesion site. (A, Left) HEK293T cells transfected with GFP-XPC plasmid
were irradiated 48 h later with 30 J/m2 UVC (or control) and fractionated to isolate chromatin-bound fraction, which was subjected to IP with GFP-trap beads.
The eluates were probed with PARP1 and GFP antibody. (A, Middle and Right) GMU6 cells were irradiated with either 30 J/m2 UVC (Middle) or different UVC
doses as indicated (Right) and fractionated after the time indicated for the Middle panel or at 10 min for the Right panel. The chromatin-bound protein
fraction was used for immunoprecipitation of XPC. The IP eluates were resolved on the SDS/PAGE and probed for XPC, PARP1, and DDB2. The Ponceau S
staining was used as loading control, and each panel is representative of results from two experiments. (B) UVC-DNA was bound to magnetic beads via T-T
antibody and reacted with the pure proteins PARP1 (P), XPC fragment (ΔX), DDB2 (D), and different protein combinations for 15 min at 25 °C. The beads were
washed, and the bound proteins were eluted, separated on SDS/PAGE, and probed with their respective antibodies. (C and D, Top) The Top panels represent
schematic of the different conditions used in the in vitro assays for examining the separation of XPC from the PARP1–XPC complex and its handover to
UV-DNA. The gray and red circles represent agarose and magnetic beads, respectively. The bead-bound PARP1–ΔXPC complex was prepared either on
magnetic or agarose beads as described in Fig. 2C. The representative results for each model from two to three experiments are shown. (C, Lower) The PARP1–
CELL BIOLOGY

ΔXPC complex was prepared on agarose beads (A), and magnetic bead-bound UV-DNA (M1) was reacted with PARP1 (M2), DDB2 (M3), or both (M4), as
described for B. The beads were mixed as indicated in the Top panel, reacted for 15 min at 25 °C in buffer, separated, washed, and eluted, and the elution was
separated on SDS/PAGE and probed with specified antibodies. (D, Lower) The PARP1–ΔXPC complex was prepared on magnetic beads (M), and the agarose
bead-bound UV–DDB ligase complex without (A1) or with (A2) second PARP1 was prepared by prereacting the agarose bead-bound Cul4A–Rbx1 complex
with free UV-DNA, DDB1, and DDB2 for 15 min at 25 °C. The beads were mixed, reacted for 15 min at 25 °C in buffer, separated, washed, and eluted, and the
elution was separated on SDS/PAGE and probed with specified antibodies.

Robu et al. PNAS | Published online July 31, 2017 | E6853

(M2 beads) or DDB2 (M3 beads) or both PARP1 and DDB2 PARP1 depletion reduces XPC recruitment to the lesion site, and
(M4) (Fig. 5C, Top). After each reaction, the agarose (A beads) PARPi reduces the rapid colocalization of PARP1 and XPC to the
and magnetic beads (M1–M4) were separated and washed, and lesion site in vivo, our results indicate that both PARP1 and its
the proteins present in each of these beads were examined by catalytic function determine the movement of the PARP1–XPC
immunoblotting of bead eluates. The immunoprobing for ΔXPC complex from nucleoplasm to chromatin after irradiation.
in the agarose and magnetic beads after the reaction revealed that In XP-E cells deficient in DDB2 function, the repair of 6–4PP is
none of the above conditions could dissociate ΔXPC from its attributed to the inherent property of XPC to recognize 6–4PP
complex with PARP1 (Fig. 5C). lesions. Nonetheless, some studies demonstrated a reduced level of
At the UV-lesion site on DNA, DDB2 is not recruited alone recruitment of XPC to UV damage in these cells compared with
but as a UV–DDB ligase complex containing DDB2, DDB1, DDB2-proficient cells (7, 8). Additionally, we show that PARPi not
Cul4A, and Rbx1. In addition, DDB2 interacts with PARP1 at the only causes further reduction in initial recruitment of XPC to local
lesion site. Therefore, we examined whether the entire UV–DDB spots of UV-induced DNA lesions but also significantly hampers
complex as well as PARP1 are required for loading of XPC to the repair of 6–4PP lesions in these cells. Thus, in the XP-E model,
UV-DNA (Fig. 5D, Top). We recreated this complex by loading our results clearly reveal a DDB2-independent role of PARP1 in
purified Cul4A–Rbx1 on agarose beads and reacted it with puri- facilitating XPC recruitment to the UV lesions and repair of 6–4PP
fied DDB1, DDB2, and UV-DNA in the absence (A1) or pres- by NER. It has been shown that XP-E cells have very low levels of
ence of PARP1 (A2). On the other hand, the PARP1–ΔXPC PAR and that both ubiquitination and ALC1-mediated chromatin
complex was immobilized via PARP1 antibody on the magnetic remodeling are absent in these cells (19). Hence, the decrease in
beads (M). The immunoblotting confirmed the presence or ab- recruitment of XPC by PARPi could not be related to ubiquiti-
sence of each of the six designated proteins in the input beads M, nation or chromatin remodeling at the damaged site but instead
A1, and A2 (Fig. 5D, Input lanes). The PARP1–ΔXPC (M) beads due mainly to the suppression of movement of the PARP1–XPC
were reacted with A1 or A2 agarose beads, and the beads were complex from nucleoplasm to the lesion site on chromatin in XP-E
separated and washed followed by immunoblotting of each of the cells. Using cells that are DDB2-deficient, PARP1-depleted, or
beads for detection of the six proteins. The immunoprobing for treated with PARPi, we identified that the recruitment of XPC to
His-ΔXPC in the magnetic and agarose beads revealed that the the UV-lesion site in the genomic context is the sum of efforts by
UV–DDB ligase complex with UV-DNA provided favorable multiple factors including XPC itself, DDB2, PARP1, and the cat-
conditions for promoting the dissociation of ΔXPC from its alytic activity of PARP1. In a cell line that is devoid of functional
complex with PARP1 on magnetic beads (Fig. 5D, compare lane DDB2 and PARP1 (CHO-SiP), there is a basal level of recruitment
1 with lane 5) as well as loading of ΔXPC onto the UV-DNA on of XPC to the lesion site, indicating that XPC has some inherent
A1 agarose beads (Fig. 5D, compare lane 2 with lane 6). Addition capacity to reach the DNA lesion site that is not dependent on
of PARP1 to the UV–DDB complex on A2 agarose beads in the DDB2 or PARP1 or its activation. The reduced level of XPC
above reaction did not confer any additional movement of ΔXPC translates to an impaired accumulation of XPA at the lesion site.
to UV-DNA on agarose beads (Fig. 5D, compare lane 6 with lane Nonetheless, adding PARP1 alone in this DDB2-deficient back-
8). Collectively, these in vitro assays with purified proteins reveal ground (CHOU6 cells) or DDB2 alone in a PARP1-deficient
that, although free XPC has an inherent capacity to efficiently background (GMSiP) improves the recruitment of XPC above
bind to UV-DNA, its presence as a complex with PARP1 before the basal level, indicating that each of these two proteins in-
irradiation ensures that XPC is preferably deposited at the UV- dependently participates in XPC recruitment and stabilization.
damaged sites that contain the UV–DDB ligase complex. Finally, in the cells with PARP1 and DDB2, it is the DDB2-
stimulated catalytic activation of PARP1 (18) that provides the
Discussion last boost for recruitment of XPC to the lesion.
For the last 15 years, focused efforts have been made to understand PARP1 has many characteristics that would facilitate the search
how XPC, with or without the help of other proteins, rapidly function of XPC in NER: (i) PARP1 is an abundant protein in the
searches for its target lesions scattered across the entire genome in mammalian nucleus that is rapidly recruited to all types of DNA
higher-order chromatin structure. Many studies indicated a role for damages (34) including UV-induced DNA lesions (26). Hence, an
DDB2 in the proper functioning of XPC with an indirect role for association with PARP1 will allow XPC to be quickly recruited
PARP1 via its ability to participate in chromatin remodeling (20, to different types of DNA lesions anywhere in the genome.
27). The present study reveals a paradigm for the functional role of (ii) PARP1 can detect lesions and become activated to form PAR
physical interaction of PARP1 with XPC before DNA damage in and create a protein-recruiting PAR platform (35), which in turn
the initial recruitment and handover of XPC at UV-induced DNA can bring in more PARP1 with XPC and other PAR-seeking pro-
lesions. Using various cell lines with exogenous or endogenous teins to the site. (iii) Like XPC, the binding of PARP1 to damaged
PARP1 and XPC, we show that PARP1 and XPC interact in the DNA is independent of the sequence or the chemical nature of
nucleoplasmic fraction of the cells even in the absence of DNA DNA damage (1, 36). Moreover, both XPC and PARP1 have
damage and that this interaction is independent of DDB2 and affinity for unusual DNA structures with nonhydrogen-bonded
catalytic activation of PARP1. By using the PARP1 proximity- bases, such as hairpins, stem loops, bubbles, and overhangs
mediated biotinylation model in vivo, we also show that XPC (1, 37). Thus, PARP1 could rapidly recruit XPC to all types of
from the nucleoplasmic PARP1–XPC complex is deposited at the damages that are repaired by NER irrespective of their recogni-
DNA lesion site after UV irradiation. Using PJ-34 as PARPi, we tion by DDB2 (38). (iv) PARP1 is a part of the chromatin struc-
observed that PARP inhibition partially suppresses the initial re- ture with preference for binding to the internucleosomal linker
cruitment of XPC and PARP1 to the UV-lesion site, which is in region (39). The chromatin-bound PARP1 can bind rapidly to
agreement with earlier reports showing decreased recruitment of lesions in this region and help recruit the nucleoplasmic PARP1–
PARP1 to the site of microirradiation-induced DNA damage in XPC complex. Additionally, the presence of the UV–DDB ligase
the presence of PARPi such as PJ-34 (30) and NU-1025 (31). complex in the linker (23) will allow handover and stabilization of
Another study reported an increased signal for PARP1 at dam- XPC at this site. (v) Finally, the role of PARP1 activation in
aged DNA after treatment with the PARPi 4-amino-1,8- chromatin remodeling at the lesion site via recruitment of ALC1
naphthalimide (32). The difference in the end results among (19) and PARylation of histones (34) would subsequently permit
these studies could be attributed to the time of harvesting of the XPC to repair less accessible intranucleosomal lesions.
samples after treatment and the capacity of different PARPi to It has been shown that the handover of the UV lesion from the
immobilize PARP1 on the DNA lesion sites (30, 33). Since UV–DDB complex to XPC requires a transient physical interaction

E6854 | www.pnas.org/cgi/doi/10.1073/pnas.1706981114 Robu et al.

 PNAS PLUS
between DDB2 and the central region of XPC (496–679 aa) con-
taining the domains required for its interaction with DNA (23).
Interestingly, our in vitro studies have identified that the same
central domain of XPC (496–734 aa) mediates its interaction with
PARP1. This indicates a dynamic and cooperative process in which
XPC is released from its complex with PARP1 and transferred to
the lesion site containing the UV–DDB complex. Thus, our in vitro
model faithfully replicates the sequence of events surrounding ef-
ficient stabilization of XPC to the lesion site, which starts with its
dissociation from the complex formed with PARP1, followed by the
formation of a new complex with UV–DDB. The requirement of
the UV–DDB ligase complex at the lesion site ensures that
PARP1-escorted XPC is preferably released at a site that is primed
for GG-NER due to ubiquitination and chromatin-remodeling
events initiated by the UV–DDB ligase complex. Our data do not
exclude the role of the PARylation of proteins as well as changes in
the structure of DNA at the lesion site toward dissociation of the
PARP1–XPC complex and stabilization of XPC. These factors
could play a key role in delivering XPC from its complex with
PARP1 to the damage site in the absence of DDB2. Such events
have been shown to play a role in dissociation of PARP1 from
XRCC1 or APE1 during base excision repair (40, 41).
We propose a model for the roles of PARP1 in the lesion rec-
ognition function of XPC (Fig. 6). Before UV irradiation, PARP1
and XPC coexist as a complex in the nucleoplasm, and DDB2 is not
part of this complex. Since PARP1 is a more abundant protein
compared with XPC (9, 42), there will still be sufficient free PARP1
molecules to separately interact with DDB2 at the lesion site. The
free PARP1 molecules as well as the PARP1–XPC complex will
scan the intact DNA due to the affinity of PARP1 and XPC for
Fig. 6. Model for the role of PARP1 in recruitment and stabilization of XPC
DNA, which explains the presence of basal levels of biotinylated
in NER. See details in Discussion.
XPC and PARP1 in chromatin-bound protein fractions from con-
trol cells. However, the transient binding of the complex to control
DNA will not result in separation of XPC from PARP1 because the PARP1 interactors would be much more informative if these
UV–DDB ligase complex is not recruited to chromatin until DNA analyses were performed before and after DNA damage and in
damage occurs. Upon UV irradiation, free PARP1 as well as XPC- different subnuclear compartments.
bound PARP1 molecules will reach the lesion site and may deposit
XPC from the complex to the lesion site with the help of other Materials and Methods
factors. However, the optimum deposition of XPC from its complex Full details are provided in SI Materials and Methods.
with PARP1 to the UV-lesion site would occur when the UV–DDB
ligase complex is present at the lesion site, a condition that would be Cell Lines, Clones, Plasmids, and Cloning. The SV-40 immortalized GM637 and
observed in normal DDB2-competent cells. The formation of PAR primary XP-E (GM01389) human skin fibroblasts were obtained from Coriell.
by DDB2-stimulated PARP1 (18) provides a platform for addi- HEK293T and CHO cells were from ATCC. The clones PARP1-replete (GMU6,
tional PAR-seeking molecules to accumulate at the damaged site, CHOU6), PARP1-depleted (GMSiP, CHOSiP), and FLAG-tagged PARP1-expressing
including more of the nucleoplasmic PARP1–XPC complex. The (GMRSiP) were described earlier (18, 29). The Bio-ID-PARP1–expressing plasmid
presence of the UV–DDB complex at the lesion site in the linker was generated in pcDNA3.1 mycBioID backbone from Addgene (24) and used
region (23) or in the core region (43) would facilitate prioritization for creating Bio-ID-PARP1 cell lines in the PARP1-depleted GMSiP cells. The cre-
of these sites for initial recruitment of XPC and the repair. Addi- ation of the pGFP-DBD vector expressing the N-terminal fragment of PARP1
(1–232 aa) was described earlier (26). The vector expressing the C-terminal
tionally, the PARylation, DDB2, and UV–DDB ubiquitin ligase
fragment of PARP1 (232–1,014 aa) was cloned from PARP31 vector.
complex-mediated chromatin-remodeling events opens the nucle-
osomal structure to allow the arrival of downstream proteins to UV Irradiation and Immunofluorescence Microscopy Studies. The local UVC
complete the process of NER at all of the remaining lesions in irradiation using a 5-μm polycarbonate filter (Millipore), global UVC irradi-
the genome. ation, the repair kinetics assays for 6–4PP, recruitment of NER proteins and
Much effort has gone into understanding the interaction of PAR GFP-PARP1 to local UVC-induced DNA damage, the image acquisition and
and PARP1 with different proteins in the cells after DNA damage. analyses and software used for analyses of images, and full details of the
However, not much is known about the importance of the in- statistical analyses of images are described in SI Materials and Methods.
teraction of PARP1 with multiple cellular proteins in steady-state
conditions before DNA damage, which has been reported in the Cell Fractionation and co-IP of Proteins in the Cell Fractions. Cell fractionation
proteomics study (44). Here, we clearly show that the interaction of to obtain nucleoplasmic and chromatin-bound protein fractions and the IP
PARP1 with XPC before DNA damage is not a random phe- protocols followed by immunoprobing for proteins in these fractions were
described earlier (18) and are explained in SI Materials and Methods. For
nomenon, but serves a definitive purpose of delivering XPC to the
streptavidin-IP, cells were incubated with 50 μM biotin in the medium for
site of DNA damage within minutes after irradiation for efficient 16–18 h before UV treatment.
NER-mediated repair by XPC. We suggest that similar functional
CELL BIOLOGY

roles are possible for the steady-state interaction of other proteins Identification of XPC-Interacting Proteins by Mass Spectrometry. The prepa-
with PARP1. Our study also highlights the fact that proteins move ration of the cell extracts from HEK293T cells, the XPC-IP, the identification
from one subnuclear compartment to another and thus that they XPC-interacting proteins using LC-MS/MS, along with their quantification
may carry old partners into a new compartment or make new using the appropriate software and the threshold limits, were described
partners in the new location. Hence, the proteomic studies of earlier (45) and also in SI Materials and Methods.

Robu et al. PNAS | Published online July 31, 2017 | E6855

In Vitro Studies to Examine the Handover of XPC from PARP1 to UV-Damaged for interpretation of the proteomics data generated at Université de
DNA. Use of purified bovine PARP1 (Apartosis), XPC fragments (Antibodies Sherbrooke. This work was supported by Discovery Grant RGPIN-2016-
online), GST-DDB1 and GST-DDB2 (Abnova), purification of full-length hu- 05868 and Discovery Accelerator Grant RGPAS-492875-2016 (to G.M.S.) from
man Cul4A, binding of UVC-DNA and proteins to magnetic or agarose beads the Natural Sciences and Engineering Research Council of Canada. N.K.P.
through their respective antibodies, and the handover assays are described in received a foreign student fee-waiver scholarship from the Québec Govern-
SI Materials and Methods. ment and Shastri Indo-Canadian Institute. M.R. and N.K.P. were recipients of
graduate scholarships from the Fonds de Recherche du Québec-Santé and
ACKNOWLEDGMENTS. We thank V. Schreiber for GFP-PARP1 and M. Miwa Neuroscience Axis of Centre Hospitalier Universitaire de Québec Research
for permission to receive 10H hybridoma and F. M. Boisvert and D. Levesque Center-Université Laval, respectively.

1. Schärer OD (2013) Nucleotide excision repair in eukaryotes. Cold Spring Harb Perspect 24. Roux KJ, Kim DI, Raida M, Burke B (2012) A promiscuous biotin ligase fusion protein
Biol 5:a012609. identifies proximal and interacting proteins in mammalian cells. J Cell Biol 196:801–810.
2. Puumalainen MR, Rüthemann P, Min JH, Naegeli H (2016) Xeroderma pigmentosum 25. Moné MJ, et al. (2001) Local UV-induced DNA damage in cell nuclei results in local
group C sensor: Unprecedented recognition strategy and tight spatiotemporal reg- transcription inhibition. EMBO Rep 2:1013–1017.
ulation. Cell Mol Life Sci 73:547–566. 26. Purohit NK, Robu M, Shah RG, Geacintov NE, Shah GM (2016) Characterization of the
3. Sugasawa K (2016) Molecular mechanisms of DNA damage recognition for mam- interactions of PARP-1 with UV-damaged DNA in vivo and in vitro. Sci Rep 6:19020.
malian nucleotide excision repair. DNA Repair (Amst) 44:110–117. 27. Pines A, Mullenders LH, van Attikum H, Luijsterburg MS (2013) Touching base with
4. Hoogstraten D, et al. (2008) Versatile DNA damage detection by the global genome PARPs: Moonlighting in the repair of UV lesions and double-strand breaks. Trends
nucleotide excision repair protein XPC. J Cell Sci 121:2850–2859. Biochem Sci 38:321–330.
5. Chen X, et al. (2015) Kinetic gating mechanism of DNA damage recognition by Rad4/XPC. 28. Ghodgaonkar MM, Zacal N, Kassam S, Rainbow AJ, Shah GM (2008) Depletion of
Nat Commun 6:5849. poly(ADP-ribose) polymerase-1 reduces host cell reactivation of a UV-damaged adenovirus-
6. Velmurugu Y, Chen X, Slogoff Sevilla P, Min JH, Ansari A (2016) Twist-open mecha- encoded reporter gene in human dermal fibroblasts. DNA Repair (Amst) 7:617–632.
nism of DNA damage recognition by the Rad4/XPC nucleotide excision repair com- 29. Shah RG, Ghodgaonkar MM, Affar B, Shah GM (2005) DNA vector-based RNAi ap-
plex. Proc Natl Acad Sci USA 113:E2296–E2305. proach for stable depletion of poly(ADP-ribose) polymerase-1. Biochem Biophys Res
7. Moser J, et al. (2005) The UV-damaged DNA binding protein mediates efficient tar- Commun 331:167–174.
geting of the nucleotide excision repair complex to UV-induced photo lesions. DNA 30. Hanssen-Bauer A, et al. (2011) XRCC1 coordinates disparate responses and multi-
Repair (Amst) 4:571–582. protein repair complexes depending on the nature and context of the DNA damage.
8. Oh KS, et al. (2011) Nucleotide excision repair proteins rapidly accumulate but fail to Environ Mol Mutagen 52:623–635.
persist in human XP-E (DDB2 mutant) cells. Photochem Photobiol 87:729–733. 31. Mortusewicz O, Amé JC, Schreiber V, Leonhardt H (2007) Feedback-regulated
9. Luijsterburg MS, et al. (2007) Dynamic in vivo interaction of DDB2 E3 ubiquitin ligase
poly(ADP-ribosyl)ation by PARP-1 is required for rapid response to DNA damage in
with UV-damaged DNA is independent of damage-recognition protein XPC. J Cell Sci
living cells. Nucleic Acids Res 35:7665–7675.
120:2706–2716.
32. Godon C, et al. (2008) PARP inhibition versus PARP-1 silencing: Different outcomes in
10. Lan L, et al. (2012) Monoubiquitinated histone H2A destabilizes photolesion-
terms of single-strand break repair and radiation susceptibility. Nucleic Acids Res 36:
containing nucleosomes with concomitant release of UV-damaged DNA-binding
4454–4464.
protein E3 ligase. J Biol Chem 287:12036–12049.
33. Gassman NR, Wilson SH (2015) Micro-irradiation tools to visualize base excision repair
11. Wang H, et al. (2006) Histone H3 and H4 ubiquitylation by the CUL4-DDB-
and single-strand break repair. DNA Repair (Amst) 31:52–63.
ROC1 ubiquitin ligase facilitates cellular response to DNA damage. Mol Cell 22:383–394.
34. Luo X, Kraus WL (2012) On PAR with PARP: Cellular stress signaling through poly(ADP-
12. El-Mahdy MA, et al. (2006) Cullin 4A-mediated proteolysis of DDB2 protein at DNA
ribose) and PARP-1. Genes Dev 26:417–432.
damage sites regulates in vivo lesion recognition by XPC. J Biol Chem 281:13404–13411.
35. Pascal JM, Ellenberger T (2015) The rise and fall of poly(ADP-ribose): An enzymatic
13. Nishi R, et al. (2009) UV-DDB-dependent regulation of nucleotide excision repair ki-
perspective. DNA Repair (Amst) 32:10–16.
netics in living cells. DNA Repair (Amst) 8:767–776.
36. Langelier MF, Pascal JM (2013) PARP-1 mechanism for coupling DNA damage de-
14. Shell SM, et al. (2013) Xeroderma pigmentosum complementation group C protein
tection to poly(ADP-ribose) synthesis. Curr Opin Struct Biol 23:134–143.
(XPC) serves as a general sensor of damaged DNA. DNA Repair (Amst) 12:947–953.
37. Lonskaya I, et al. (2005) Regulation of poly(ADP-ribose) polymerase-1 by DNA
15. Payne A, Chu G (1994) Xeroderma pigmentosum group E binding factor recognizes a
broad spectrum of DNA damage. Mutat Res 310:89–102. structure-specific binding. J Biol Chem 280:17076–17083.
16. Scrima A, et al. (2008) Structural basis of UV DNA-damage recognition by the DDB1- 38. Sugasawa K, et al. (2005) UV-induced ubiquitylation of XPC protein mediated by UV-
DDB2 complex. Cell 135:1213–1223. DDB-ubiquitin ligase complex. Cell 121:387–400.
17. Vodenicharov MD, Ghodgaonkar MM, Halappanavar SS, Shah RG, Shah GM (2005) 39. Clark NJ, Kramer M, Muthurajan UM, Luger K (2012) Alternative modes of binding of poly
Mechanism of early biphasic activation of poly(ADP-ribose) polymerase-1 in response (ADP-ribose) polymerase 1 to free DNA and nucleosomes. J Biol Chem 287:32430–32439.
to ultraviolet B radiation. J Cell Sci 118:589–599. 40. Moor NA, Vasil’eva IA, Anarbaev RO, Antson AA, Lavrik OI (2015) Quantitative
18. Robu M, et al. (2013) Role of poly(ADP-ribose) polymerase-1 in the removal of UV- characterization of protein-protein complexes involved in base excision DNA repair.
induced DNA lesions by nucleotide excision repair. Proc Natl Acad Sci USA 110: Nucleic Acids Res 43:6009–6022.
1658–1663. 41. Wei L, et al. (2013) Damage response of XRCC1 at sites of DNA single strand breaks is
19. Pines A, et al. (2012) PARP1 promotes nucleotide excision repair through DDB2 sta- regulated by phosphorylation and ubiquitylation after degradation of poly(ADP-ri-
bilization and recruitment of ALC1. J Cell Biol 199:235–249. bose). J Cell Sci 126:4414–4423.
20. Luijsterburg MS, et al. (2012) DDB2 promotes chromatin decondensation at UV- 42. Kraus WL, Lis JT (2003) PARP goes transcription. Cell 113:677–683.
induced DNA damage. J Cell Biol 197:267–281. 43. Osakabe A, et al. (2015) Structural basis of pyrimidine-pyrimidone (6-4) photoproduct
21. Maltseva EA, Rechkunova NI, Sukhanova MV, Lavrik OI (2015) Poly(ADP- recognition by UV-DDB in the nucleosome. Sci Rep 5:16330.
ribose)polymerase 1 modulates interaction of the nucleotide excision repair factor 44. Isabelle M, et al. (2010) Investigation of PARP-1, PARP-2, and PARG interactomes by
XPC-RAD23B with DNA via poly(ADP-ribosyl)ation. J Biol Chem 290:21811–21820. affinity-purification mass spectrometry. Proteome Sci 8:22.
22. Jungmichel S, et al. (2013) Proteome-wide identification of poly(ADP-ribosyl)ation 45. Dubois ML, Bastin C, Lévesque D, Boisvert FM (2016) Comprehensive characterization
targets in different genotoxic stress responses. Mol Cell 52:272–285. of minichromosome maintenance complex (MCM) protein interactions using affinity
23. Fei J, et al. (2011) Regulation of nucleotide excision repair by UV-DDB: Prioritization and proximity purifications coupled to mass spectrometry. J Proteome Res 15:
of damage recognition to internucleosomal DNA. PLoS Biol 9:e1001183. 2924–2934.

E6856 | www.pnas.org/cgi/doi/10.1073/pnas.1706981114 Robu et al.