1

1

2

Differential and concordant roles for PARP1 and poly(ADP-ribose) in regulating 3

WRN and RECQL5 activities 4

5

6

Prabhat Khadka1¶, Joseph K. Hsu1¶, Sebastian Veith2, Takashi Tadokoro1, 7 Raghavendra Shamanna1, Aswin Mangerich2 , Deborah L. Croteau.1, and 8

Vilhelm A. Bohr1* 9 10

11

12

13

1Laboratory of Molecular Gerontology, National Institute on Aging, NIH, 251 Bayview 14 Blvd, Suite 100m Baltimore, MD 21224, USA 15

2Molecular Toxicology Group, Department of Biology, University of Konstanz, 78457 16 Konstanz, Germany 17

18

19

20

¶These authors contributed equally to this work* To whom correspondence should be 21 addressed.Tel: +1 410 558 8162; Fax: +1 410 558 8157; E-mail: vbohr@nih.gov 22

23

MCB Accepted Manuscript Posted Online 21 September 2015Mol. Cell. Biol. doi:10.1128/MCB.00427-15Copyright © 2015, American Society for Microbiology. All Rights Reserved.

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

2

ABSTRACT 24

Poly(ADP-ribose) polymerase 1 (PARP1) catalyzes the polyADP(ribosyl)ation 25

(PARylation) of proteins, a post-translational modification which forms the nucleic acid-26

like polymer poly(ADP-ribose) (PAR). PARP1 and PAR are integral players in the early 27

DNA damage response, since PARylation orchestrates the recruitment of repair 28

proteins to the damage sites. Human RecQ helicases are DNA unwinding proteins that 29

are critical responders to DNA damage, but how their recruitment and activities are 30

regulated by PARPs and PAR is poorly understood. Here we report that all human 31

RecQ helicases interact with PAR non-covalently. Furthermore, we define the effects 32

that PARP1, PARylated PARP1 and PAR have on RECQL5 and WRN, using both in 33

vitro and in vivo assays. We show that PARylation is involved in the recruitment of 34

RECQL5 and WRN to laser-induced DNA damage and that RECQL5 and WRN have a 35

differential response to PARylated PARP1 and PAR. Furthermore, we show that 36

RECQL5 or WRN loss resulted in increased sensitivity towards PARP inhibition. In 37

conclusion, our results demonstrate that PARP1 and PAR actively, and in some 38

instances differentially, regulate the activities and cellular localization of RECQL5 and 39

WRN, suggesting that PARylation acts as a fine-tuning mechanism to coordinate their 40

functions in time and space during genotoxic stress response. 41

42

43

44

45

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

3

INTRODUCTION 46

Mammalian cells are constantly exposed to various environmental genotoxic stresses 47

that can hamper genomic stability. To maintain genomic integrity, cells have developed 48

various complex DNA repair machineries that effectively identify DNA lesions and 49

activate DNA damage responses (DDRs) and DNA repair (1, 2). One of the early DDR 50

mechanisms is through activation of poly(ADP ribose) (PAR) polymerases (PARPs). 51

PARP1 is the most ubiquitously expressed PARP family member and constitutes the 52

majority of PAR synthesis in human cells (3, 4). In response to DNA damage, such as 53

single or double DNA strand breaks, PARP1 is activated and promotes PAR synthesis. It 54

utilizes nicotinamide adenine dinucleotide (NAD+) to form PAR at the DNA lesions or 55

breaks thereby post-translationally modifying itself and other proteins of interest. 56

Importantly, PAR formation is highly dynamic, because shortly after being synthesized, it 57

is rapidly hydrolyzed by PARPs catabolic counterparts, PAR glycohydrolase (PARG) 58

and ADP-ribosylhydrolase 3 (ARH3) (5-8). Neither enzyme is able to remove the last 59

ADP-ribose moiety from acceptor proteins; macrodomain-containing proteins, such as 60

MacroD1 and MacroD2, fulfill this task, leaving behind an unmodified amino acid that is 61

readily available for the next round of PARylation (9, 10). Formation of PAR can act as a 62

docking site to promote protein recruitment and protein-protein interactions. This 63

docking site serves to integrate diverse cellular signaling pathways, including DNA 64

damage detection and repair, transcription, chromatin remodeling, and cell death (11). 65

PAR formation plays a critical role in the rapid recruitment of the DNA repair proteins 66

such as MRE11(12), NBS1(13), BRCA1(14) and ATM (15) to DSBs. Mice lacking 67

PARP1 are extremely sensitive to genotoxic stresses, show defects in repair of 68

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

4

damaged DNA and display genomic instability with high frequencies of sister chromatid 69

exchange and chromosomal breakage (16). 70

RecQ helicases are a highly conserved family of 3’-5’ DNA unwinding proteins 71

that engage in various roles in DNA metabolic processes and DNA repair pathways to 72

maintain genomic stability. In humans, there are five RecQ helicases: RECQL1, BLM, 73

WRN, RECQL4, and RECQL5 (17, 18). In addition to helicase activity, these RecQ 74

proteins also possess the opposite activity as they can anneal the two DNA strands. 75

Furthermore, unlike the other RecQ helicases, WRN possesses a 3’-5’ exonuclease 76

activity. Activated PARP1 has been previously reported to covalently modify human 77

RecQ helicases and to modulate their biological activities (19-21). For example, 78

activation of PARP1 can strongly repress RECQL1’s ability to restore replication forks 79

(22). The exonuclease and helicase activities of WRN are inhibited by the presence of 80

PARP1 and PAR (23) (24). In addition, PARP1 binds to the C-terminal region of 81

RECQL4 and pharmacological PARP inhibition affects RECQL4 nuclear distribution 82

(20). Interestingly, Parp1-/- / Wrn∆hel/∆hel mice develop spontaneous tumors at early ages 83

and the derived mouse embryonic fibroblasts show increased formation of chromosome 84

fragments and chromosomal rearrangements (25). Currently, it is unknown whether 85

PARP1 plays a role in the regulation of RECQL5. 86

Compared with the other RecQ family members, RECQL5 is the least studied 87

and most poorly understood. Even though no human genetic disorders have been 88

reported to be associated with RECQL5 mutations, human RECQL5 knockdown cells 89

display increased levels of endogenous DNA damage and PAR formation (26). RECQL5 90

deficient mice are more susceptible to cancer and show increased levels of 91

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

5

spontaneous DNA double strand breaks (DSBs) and homologous recombination 92

activities (27). Thus, RECQL5 may act as a tumor suppressor to maintain genomic 93

integrity. 94

RECQL5 interacts with several proteins that participate in DNA replication, 95

transcription, and recombination pathways. Previously, RECQL5 was reported to 96

associate with the Mre11-RAD50-NBS1 complex at DSBs to regulate repair (28). DSBs 97

are critical DNA lesions that require an immediate cellular response because failure to 98

do so can result in chromosomal rearrangements, genomic instability, and even cell 99

death. It has been shown that RECQL5 interacts with RAD51 to inhibit RAD51 induced 100

D-loop formation and sister chromatid exchange (29). Consistent with this finding, 101

RECQL5 prevents RAD51 ssDNA filament formation by promoting synthesis-dependent 102

strand annealing to mediate DSBs repair (30). Our previous studies used confocal laser 103

microscopy to study RECQL5’s recruitment kinetics to DSBs in real time in live cells 104

(31). We demonstrated that its recruitment was independent of the activities of MRE11, 105

RNA polymerase II and ATM. 106

The mechanisms that mediate RECQL5’s and WRN’s recruitment to DSBs still 107

remain elusive and therefore we sought to further investigate this by exploring the role 108

of PARP1 and PARylation in the recruitment of RECQL5 and WRN to DSBs. In this 109

study, we show that RECQL5 is not a substrate for PARP1 but that RECQL5 can bind to 110

PAR in a non-covalent manner. RECQL5’s biochemical activities with or without PARP1 111

activation are systematically compared with that of WRN’s. Both RECQL5 and WRN 112

show helicase and ATPase inhibition by activated PARP1. In contrast, strand annealing 113

by RECQL5 can be stimulated by PAR and to some extent PARylated PARP1 but is 114

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

6

inhibited by PARP1. Additionally, we report that the early recruitment of RECQL5 to 115

DSB sites is impaired when PAR synthesis is inhibited. PARP1 is involved in the same 116

pathway as RECQL5 to regulate HR-mediated DSB repair. 117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

7

MATERIALS AND METHODS 133

Cell lines and transfection. 134

Both HEK 293T and HeLa cells were growing in DMEM media (Gibco) containing 10% 135

FBS and 1% Pen-Strep at 37°C in a 5% CO2 humidified incubator. For live-cell imaging 136

experiments, 5X104 cells were seeded in 2-cm glass bottom plates. When cells were 137

60-70% confluent, they were transfected with the indicated plasmids using 138

Lipofectamine 3000 (Invitrogen, Life Technologies) per manufacturer’s directions. Cells 139

were used for microscopic analysis 24 h after transfection. For co-immunoprecipitation 140

and biotin PAR pull-down assays, HEK 293T cells were plated 60% confluence in 10-cm 141

plates and transfected with10 µg of plasmid per plate as suggested by the jetPRIMETM 142

Polyplus DNA transfection protocol (VWR). Cells were harvest 36 h after transfection. 143

144

Recombinant proteins. 145

As previously described, the recombinant RECQL5 protein was purified from the BL21 146

E. coli system (32) and WRN was purified using a baculovirus/ insect cell expression 147

system (33). High specific activity PARP1 enzyme was purchased from Trevigen. 148

149

Far western with PAR (PAR-overlay). 150

Recombinant RecQ proteins (15 pmol) were separated by SDS-PAGE and then 151

transferred to nitrocellulose membranes. Membranes were incubated with indicated 152

concentrations of unfractionated PAR O/N at 4°C, followed by high stringency washes 153

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

8

with 1 M NaCl in Tris buffer saline plus 0.1% Tween-20 (TBST) (3x 5 min). Blots were 154

blocked in 5% (w/v) milk, followed by incubation with PAR specific antibody 10H 1:300 155

in 5% milk for 1 h (10H antibody is a kind gift from M. Miwa, Nagahama Institute of 156

Bioscience and Technology, Japan, and T. Sugimura, National Cancer Center Research 157

Institute, Tokyo, Japan (34). Blots were washed 3-times for 5 min with TBST, followed by 158

1 h incubation with HRP-tagged anti-mouse secondary antibody (Dako, P0447) 1:2000 159

in 5% milk, followed by 3 washes with TBST. Bound PAR was detected using 160

chemiluminescence detection at a LAS4000 mini using ECL (Lumigen ECL Ultra). 161

162

PAR overlay blot assay. 163

Indicated amounts of recombinant proteins were vacuum-blotted onto a nitrocellulose 164

membrane using a slot-blot manifold. The membrane was dried for 10 min at 50°C, 165

followed by incubation with PAR in TBST O/N at 4°C. Bound PAR was detected as 166

described above, by using 10H PAR antibody. 167

168

Peptide binding study. 169

The PepSPOT Membrane was purchased from JPT Peptide Technologies (Berlin, 170

Germany). Prior to the experiment the membrane was activated in 100% methanol for 5 171

min as recommended by the supplier. Subsequently the membrane was washed of 5 172

min in TNT, followed by one hour incubation in TNT. PAR incubation occured O/N at 173

4°C with 0.2 µM unfractionated PAR in TNT. The membrane was washed 3 times for 5 174

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

9

min in TNT + 1 M NaCl, followed by two 5-min washes in TNT. Next, the membrane was 175

blocked for 1 hour in 5% milk powder in TNT, followed by 1 hour incubation with anti-176

PAR antibody 10-H (1:300 in 5 % milk-TNT). After three 5 min washes in TNT the 177

membrane was incubated for 1 hour in goat anti-mouse-HRP secondary antibody 178

(1:2000 in 5% milk-TNT), again followed by three 5-min washes in TNT. The membrane 179

was then developed using ECL and chemiluminescence signals were detected using 180

aLAS 4000. 181

182

Electrophoretic mobility shift assay with end-biotinylated PAR (PAR-EMSA). 183

The PAR-EMSA was essentially conducted as described previously (45). Briefly, 184

recombinant WRN or RECQL5 were equilibrated in EMSA buffer (40 mM Tris [pH 8.0], 4 185

mM MgCl2, 0.1 mg/ml BSA, 0.1% NP-40, 5 mM diothiothreitol [DTT]) for 10 min at 25°C, 186

followed by addition of 500 fmol end-biotinylated, size-fractionated PAR (30-35-mer). 187

Reaction mixtures were incubated for another 20 min at 25°C to allow complex-188

formation in a total volume of 10 µl, followed by addition of 1 µl 10x loading dye (250 189

mM Tris [pH8.0], 0.2 % (w/v) orange G) and subsequent separation on a 5% native TBE 190

gel. The gel was semi-dry blotted onto a nylon membrane (GE, Hybond-N+), followed 191

by an 1 h incubation at 90°C. The membrane was then blocked in 5% (w/v) milk powder 192

in TBST, followed by 3x washing in TBST for 5-min. Next, the membrane was incubated 193

for 1 h with streptavidin-coupled HRP (GE, RPN1051V) in TBST, followed by 194

chemiluminescence detection. Quantitative analysis of the band shift was performed 195

using IMAGEJ. Relative band shift was calculated by dividing the signal intensity of the 196

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

10

protein-PAR complex by the total signal intensity of unbound PAR and the protein-PAR 197

complex. 198

Co-immunoprecipitation and immunoblotting. 199

HEK 293T cells were transfected with various GFP tagged RECQL5 plasmids. Cell 200

extracts were prepared with NTEN lysis buffer (20 mM Tris-HCl, [pH8.0], 100 mM NaCl, 201

1mM EDTA, and 0.5% NP-40) containing protease inhibitor cocktail (Thermo Scientific 202

Pierce)and 1 µM ADP-HPD (Enzo Lifesciences) to inhibit PARG activity. Lysates were 203

sonicated and then clarified by centrifuged at max speed at 4°C and the cell extracts 204

were obtained. The supernatants were pre-cleared with protein G beads (GE 205

Healthcare) for 30 min at 4°C. Ethidium bromide (5 µg/ml, BioRAD) was added to the 206

lysates to eliminate DNA mediated interactions. The supernatants were incubated with 207

indicated antibodies, as described in the figure legends, and the beads for 6 h at 4°C. 208

The beads were washed five times with NTEN buffer (20 mM Tris-HCl [pH 8.0], 100 209

mM NaCl, 1 mM EDTA, and 0.5% NP-40) and then boiled in 2X Laemmli buffer (Sigma-210

Aldrich). Samples were loaded on SDS-PAGE gels (Bio-Rad) and then transferred to 211

PVDF membranes. Membranes were blocked with 5% nonfat milk in TBST for 20 min at 212

RT followed by incubation with primary antibody overnight at 4°C. The HRP-conjugated 213

secondary antibodies (GE Biosciences) were added to the membranes. Blots were 214

developed using the ECL western blotting substrate (Perkin Elmer) and images were 215

obtained via the Bio-Rad ChemiDoc. 216

217

218

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

11

Biotin PAR pull-down. 219

HEK 293T cells were transfected with various GFP tagged RECQL5 plasmids. Cell 220

lysates were prepared with NTEN lysis buffer (20 mM Tris-HCl [pH8.0], 100 mM NaCl, 221

1mM EDTA, and 0.5% NP-40) with the cocktail protease inhibitors (Thermo Scientific 222

Pierce). Cell lysates were sonicated, centrifuged at max speed at 4°C and supernatants 223

were collected. The cell extracts were incubated with biotinylated-PAR (10 µM, 224

Trevigen) and the pull-down was performed by using Biotinylated Protein Interaction 225

pull-down kit (Thermo Scientific Pierce). Samples were resolved on a SDS-PAGE gel 226

and processed for immunoblotting analysis. 227

228

In vitro poly(ADP-ribosyl)ation assay. 229

Human PARP1 (50 nM, Trevigen) was incubated with 4 ng/ml sonicated linear double 230

stranded DNA (salmon sperm) in the presence or absence of 50 nM RECQL5 in 231

reaction buffer (50 mM Tris HCl [pH8.0], 25 mM MgCl2, 0.5mM dithiothreitol [DTT]). The 232

samples were treated with either 500 µM NAD+ to induce PARylation activity or 5 µM 233

Olaparib to inhibit PARP1 activity followed by 20 min at 37°C incubation. The reactions 234

were terminated by boiling with Laemmli sample buffer. PARylation activity was 235

observed by immunoblotting analysis. 236

237

238

239

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

12

Helicase assay. 240

Recombinant WRN and RECQL5 were purified as described previously (35, 36). 241

PARP1 and PAR were obtained from Trevigen and Enzo. PARylated PARP1 was 242

prepared as described in in vitro poly(ADP-ribosyl)ation assay. Helicase activity of 243

RECQL5 with PARP1, PAR, and PARylated PARP1 proteins (at concentrations shown 244

in the figure) were measured using 0.5 nM radiolabeled fork duplex for 30 min at 37°C 245

in a reaction volume of 10 µl reaction buffer containing 30 mM Tris-HCl [pH 7.4], 50 mM 246

KCl, 5 mM MgCl2, 1 mM dithiothreitol [DTT], 100 µg/ml BSA, 10% glycerol, 5 mM ATP, 247

and forked duplex substrates, which has been described previously (37). The 248

sequences of forked duplex substrates are shown as following, 5′-249

TTTTTTTTTTTTTTTGAGTGTGGTGTACA- TGCACTAC-3′, and 5′-250

GTAGTGCATGTACACCACACTCTTTTTTTTTTTTTTT-3′. Reactions were stopped by 251

the addition of stop buffer (30 mM EDTA, 0.9% SDS, 30% glycerol, 0.05% bromophenol 252

blue, and 0.05% xylene cyanol). The reaction products were separated by 253

electrophoresis on a 10% native polyacrylamide gel in 1× Tris-Borate-EDTA (TBE) 254

running buffer at 200 V for 1 h. The gels were processed as per for the strand annealing 255

assay. 256

257

Strand annealing. 258

DNA strand-annealing activity of RECQL5 was measured using complementary 259

oligonucleotides to create a forked duplex. The top strand was labeled at the 5’-end 260

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

13

using [γ-32P] ATP and T4 polynucleotide kinase. Annealing reactions (10 μl) were 261

carried out for 10 min at 37°C in buffer (30 mM Tris-HCl [pH7.5], 50 mM KCl, 1 mM 262

dithiothreitol [DTT], 5 mM MgCl2, BSA 100 µg/ml) with the indicated amounts of 263

RECQL5, PARP1, PAR, or PARylated PARP1 proteins. The reaction products were 264

separated by electrophoresis on a 10% native polyacrylamide gel in 1×TBE running 265

buffer at 200 V for 1 h. The gels were exposed to a PhosphorImager screen (GE 266

Healthcare, Piscataway, NJ), and imaged with a Typhoon scanner (GE Healthcare, 267

Piscataway, NJ). ImageQuanTL version was employed to analyze the phosphoimages 268

and calculate the percentage of free versus annealed substrate in each reaction and 269

each was normalized against background. Assays were performed at least in triplicate, 270

and representative gels are shown. 271

272

ATPase assay. 273

The recombinant protein concentrations were as per listed in the figure legends. 274

Reactions were incubated with 0.2 µM M13mp18 single stranded DNA (New England 275

Biolabs) and 1 µCi [γ-32P] ATP (PerkinElmer Life Sciences, Waltham, MA) and PAR, 276

PARP1, or PARylated PARP1 (as noted in figure legends) in 10 µl reaction buffer (50 277

mM TrisHCl [pH7.5], 50 mM NaCl, 2 mM MgCl2, 1 mM dithiothreitol [DTT], 50 µg/ml 278

BSA, 50 µM cold ATP) for 1 h at 37°C. Reactions were stopped by adding 166 mM 279

EDTA. Reaction samples (2 µl) were spotted onto the Baker-flex cellulose 280

polyethyleneimine thin-layer chromatography plates (J.T. Baker) and developed using a 281

solution of 0.8 M LiCl and 1 M formic acid. The plates were exposed to a 282

phosphorimager screen then images were captured by the Typhoon Phosphorimager 283

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

14

scanner (GE Healthcare, Piscataway, NJ) and ImageQuant versoin 5.2 (GE Healthcare, 284

Piscataway, NJ) to calculate the percentage of ATP hydrolysis. 285

286

Laser micro-irradiation and confocal microscopy. 287

A Nikon Eclipse 2000E spinning disk confocal microscope with a laser imaging module 288

and CCD camera (Hamamatsu, Tokyo, Japan) was used. The Stanford Research 289

Systems (SRS) NL100 nitrogen laser by Micropoint ablation (Photonics Instruments, St. 290

Charles, IL, USA) was employed for double strand break in vivo targeting as described 291

previously(38, 39). Site specific double stranded DNA damage was induced through 292

SRS NL100 nitrogen laser that passed through a 435 nm dye cell. The laser power was 293

regulated by Velocity software 4.3.1 (Improvision/Perkin Elmer, Coventry, UK). The laser 294

intensity was applied at 12% to initiate double strand breaks. Images were taken and 295

analyzed by Velocity software. 296

297

Cell survival assay. 298

Cells were seeded in 60 cm plates and treated the siRNA (RECQL5 or WRN) and next 299

day equal number of cells were splited from 60 cm plates and seeded in 6 well plates (n 300

=3). Cells were subjected to veliparib treatment as indicated and media exchanged 301

every other day. Cell growth was examined on day 4 using cell counter (Beckman 302

Coulter Z1Coulter Particle Counter). Mean cell numbers from three experiments was 303

expressed as percentage of cell ± SE relative to siRNA-vehicle treated cells. 304

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

15

Immunofluorescence. 305

HeLa cells were plated on Ibidi glass bottom dish Grid-50 at a density of 50,000 cells/ 306

plate for overnight, then the cells were irradiated with 12% laser. After irradiation, the 307

cells were fixed with 3.7% formaldehyde for 10 min at room temperature. Cells were 308

washed three times with PBS for 5 min each followed by 5 min room temperature 309

permeabilization treatment (0.1% Triton X 100 in PBS). Then cells were washed with 310

PBS and blocked with goat serum, and then treated with antibodies. Following are the 311

primary and secondary antibodies used, Rabbit polyclonal anti-53BP1 (1:250; Novus 312

Biologicals), rabbit polyclonal anti-WRN H-300 (1:250; Santa Cruz Biotechnology), 313

rabbit polyclonal anti-RECQL5 (1:250; house made), mouse monoclonal anti-Poly(ADP-314

ribose) 10H (1:250; Enzo Life Sciences), donkey anti-rabbit Alexa Fluor 488 (1:1000; 315

Invitrogen) and donkey anti-mouse Alexa Fluor 647 (1:1000; Invitrogen). The Cells were 316

also co-stained with DAPI. The cells Images were captured with a Nikon Eclipse 2000E 317

confocal microscope and analyzed by Velocity software. 318

319

In vivo DSB repair assays. 320

U2OS/DR-GFP, HCA2 NHEJ-I9A and pEGFP-pem1-Ad2 transfected U2OS cells were 321

used to measure in vivo DSB repair efficiency as described previously (40-42). Cells 322

were knocked down with siRNA by transfecting twice with siControl (Qiagen; cat no. 323

1027310), siRECQL5 (CAGGAGGCUGAUAAAGGGUUA), siWRN 324

(GUAAACAGCUCCUGAAAGA), siPARP1-A (CCGAGAAATCTCTTACCTCAA) and 325

siPARP1-B (ACGGTGATCGGTAGCAACAAA) (GE Healthcare) using jetPRIME. Five µg 326

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

16

I-SceI or empty vector along with or without 0.025 μg DsRed plasmid was transfected 327

using AMAXATM cell line nucleofector kit V (Lonza) and NHDF nucleofector kit (Lonza). 328

Three days after transfection, the GFP positive cells were determined by flow cytometry 329

and FACS analysis were performed by BD AccuriTM C6 flow cytometer. 330

331

In vitro end joining. 332

The assays were performed by incubating 10 ng of 2.6 kb cohesive DNA substrate, 333

generated from SalI (New England Biologicals) linearized pUC18 plasmid (42), with 20 334

μg end-joining extracts for 1 h at 25°C in 20 μl reaction containing 20 mM N-2-335

hydroxyethylpiperazine-N′-2-ethanesulfonic acid [pH 7.5], 10 mM MgCl2, 80 mM KCl, 1 336

mM dithiothreitol (DTT), 1 mM ATP and 50 μM deoxyribonucleotide triphosphates The 337

reaction was stopped with 50 mM EDTA and 80 μg/ml RNaseA at 37°C, and 338

deproteinized by incubating with proteinase K (2 mg/ml) for 60 min at 37°C. Ligation 339

products were separated in a 0.7% agarose gel and stained with SYBR-Gold 340

(Invitrogen). Fluorescence was detected with Molecular Imager Gel Doc XR system 341

(Bio-Rad). 342

343

Data and statistical analysis. 344

All graphs display cumulative data representing three independent experiments and 345

standard error of mean (± SEM) was used to display data variability, unless otherwise 346

stated. 347

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

17

348

RESULTS 349

RECQL5 interacts with PAR and PARP1. 350

It is well known that PARylation can modulate protein-protein interactions in the 351

regulation of cell signaling pathways through recruitment of PAR-binding proteins. PAR 352

regulates protein function through covalent modification, but also by specific high affinity 353

non-covalent interactions (43). WRN has recently been evaluated for its PAR binding 354

via a PAR binding motif (PBM) (24). PBMs are suggested to be a loosely conserved 355

stretch of amino acids, about 22-25 long, that contain positively charged residues 356

followed by basic and hydrophobic rich sequences(44). Previous studies have shown 357

that RECQL1 and WRN can interact with PAR non-covalently and that this interaction 358

regulates their helicase activities (22, 24). Here we performed a bioinformatics search to 359

identify whether all of the RecQ helicases contain putative PBMs. These results 360

revealed that all five human RecQ helicases do in fact contain putative PBM motifs 361

(data not shown). To validate the bioinformatic findings, we examined each of the 362

human RecQ helicase proteins, RECQL1, BLM, WRN, RECQL4, and RECQL5, for their 363

ability to bind PAR using far western analysis. Purified recombinant proteins (15 pmol) 364

were separated on an SDS-PAGE gel and then transferred to a nitrocellulose 365

membrane. Immobilized proteins were then incubated with PAR, and the PAR-protein 366

interactions were visualized though immunoblotting analysis. We found that each 367

human RecQ helicase interacts with PAR in a non-covalent manner (Fig. 1A). Using 368

unfractionated PAR, which consists of a mixture of short, long and branched polymer 369

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

18

with chain lengths up to 200 ADP-ribose moieties, RECQL5 displayed the highest level 370

of PAR binding compared with the other RecQ family members (Fig. 1A). To 371

independently confirm these results for RECQL5, we employed an in vitro PAR-overlay 372

slot blot assay (Fig. 1B). As can be seen low concentrations of RECQL5 were able to 373

capture the PAR probe. EMSA analyses using end-biotinylated PAR of defined chain 374

length (30-35-mer) as a probe (Fig. 1C) confirmed that the RECQL5 and PAR 375

interaction also occurs in solution. From these experiments, we estimated that the Kd 376

value for the RECQ5-PAR interaction in the nanomolar range, similar to the WRN-PAR 377

interaction (see Fig.S1 in the supplemental material) and as observed previously for 378

other PAR-interacting proteins (45). These results suggest that RECQL5 and WRN bind 379

short PAR chains (i.e., 30-35-mer) with similar affinities (Fig. 1C; see also Fig. S1 in the 380

supplemental material), whereas unfractionated PAR, which consists mainly of long (> 381

100mer) and branched polymers, is bound much stronger by RECQL5 compared to 382

WRN (Fig. 1A). Further in vivo co-immunoprecipitation (IP) assays were performed after 383

treatment of cells with hydrogen peroxide and with or without the PARP inhibitor, 384

Veliparib (Fig. 1D). It revealed that GFP-RECQL5 can interact with both PARP1 and 385

PAR. 386

The five putative PAR binding motifs in RECQL5 are shown in Fig. S2 in the 387

supplemental material. We performed more peptide binding studies and found that at 388

least three/four of the PAR binding motifs show strong binding (see Fig. S2 in the 389

supplemental material). As shown in Fig. S2 in the supplemental material, the PBMs at 390

amino acids 5-24, 587-606 and 928-947 each show PAR binding that can be abolished 391

by alanine substitutions within the PBM motif. RECQL5 can be broken down into 392

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

19

several well defined domains called the helicase, KIX and SRI domains (46). The 393

helicase and KIX domains each contain one putative PBM while there are three sites in 394

the SRI domain. To identify which domains of RECQL5 can bind PAR, we tested the 395

fragments of RECQL5 for PAR binding using PAR pull down assays. Each of the 396

fragments of RECQL5 displayed PAR binding and the GFP negative control did not bind 397

(Fig. 1E). 398

Since RECQL5 could interact with PARP1 and PAR, we investigated whether 399

RECQL5 could be post-translationally PARylated by PARP1. In vitro PARylation assays 400

were performed in the presence or absence of NAD+, DNA, PARP1 and RECQL5. 401

PARP1 was activated in the presence of NAD+ and DNA (Fig. 1F, Lane 3) as PARylated 402

PARP1 is clearly evident (Fig. 1F). As expected, PARP1 activation was inhibited when 403

the samples were treated with the PARP1 inhibitor, Olaparib, which works by competing 404

for the NAD+ binding site on PARP1 (47). The same reactions were performed in the 405

presence of RECQL5 to determine if RECQL5 could be a substrate for PARP1. In the 406

presence of NAD+, PARP1 and RECQL5, we still observed the PARP1 activation 407

however, there was no formation of high molecular weight modified RECQL5. These 408

results reveal that RECQL5 is not PARylated by PARP1. 409

410

RECQL5 helicase is inhibited regardless of the modification status of PARP1. 411

Previous studies show that WRN’s catalytic activities are inhibited by PAR and 412

PARP1 (23, 24). Therefore, the effect of PARP1 and PAR on RECQL5 helicase activity 413

was investigated using a forked heteroduplex DNA substrate. RECQL5 exhibits weak 414

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

20

helicase activity with intrinsic strand annealing activity (48), and it is possible that 415

RECQL5’s helicase activity might be masked by its intrinsic strand annealing activity in 416

vitro, as reported for RECQL4 (49, 50). Thus, an excess of unlabeled ssDNA was 417

included to enhance the helicase activity of RECQL5 (see Fig. S3 in the supplemental 418

material). We performed RECQL5 and WRN helicase assays in the presence or 419

absence of PARP1. As shown in Fig. 2A and D, WRN helicase exhibited better 420

unwinding property compared to RECQL5. Both RECQL5 and WRN helicase activities 421

were inhibited with increasing PARP1. 422

Next, the effect of activated PARP1 on RECQL5 (Fig. 2A and B) and WRN (Fig. 423

2D and E) helicase activity were analyzed. PARylated PARP1 has a strong inhibitory 424

effect on RECQL5’s ability to unwind the forked DNA structures. PARylated PARP1 also 425

inhibited WRN helicase, but less than the non-PARylated form (Fig. 2G and H). The 426

results for WRN are consistent with what has been previously reported (47), and 427

suggest that PARP1 differentially regulates the catalytic activities of RECQL5 and WRN. 428

To examine the effects of PAR binding on helicase activities of RECQL5 and 429

WRN, various doses of PAR were incubated with the helicases. We noted that RECQL5 430

helicase activity was inhibited by PAR in a dose dependent manner (Fig. 2C and G), but 431

WRN was only minimal inhibited by PAR treatment (Fig. 2F and H). A previous report 432

showed that WRN helicase activity can be strongly repressed by the presence of PAR 433

(24). The difference between these two results can be explained by the PAR 434

concentrations, levels of PAR were thousand fold lower in our experiments. Therefore, 435

PARP1, PARylated PARP1, or PAR can inhibit the helicase activities of RECQL5 and 436

WRN, but the extent of inhibition varies. Compared with WRN, RECQL5 showed higher 437

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

21

sensitivity because it requires lower amounts of PARP1 or PAR to be regulated. The 438

effects of PARP1, PARylated PARP1 or PAR on RECQL5 or WRN’s helicase activities 439

are summarized in table 1. 440

441

ATPase is inhibited by PARP1 activation. 442

RecQ helicases hydrolyze ATP to drive unwinding of double stranded DNA. 443

Thus, we investigated the effects of PARP1 and its activated form, PARylated PARP1, 444

on RECQL5 and WRN ATPase activities. PARP1 was activated in the presence of 445

NAD+ to form PARylated PARP1. As shown in the western blot analysis Fig. 3A, after 446

incubating PARP1 and NAD+ there was an increase of high molecular weight modified 447

PARP1 and PAR production. In the presence of PARP1, there was no change on WRN 448

or RECQL5’s ATPase hydrolysis, lane 3 versus 4 and lane 7 versus 8. However, when 449

incubated with PARylated PARP1, the ATPase activity of both WRN and RECQL5 450

decreased by 27% and 25%, respectively (Fig. 3B). Additionally, RECQL5’s ATPase 451

activity could be completely inhibited by high doses of PAR (Fig.3C, lane 8, 9, and 10). 452

PAR can partially decrease WRN ATPase activity, but not in a dose-dependent manner 453

(Fig. 3E, land 8, 9, and 10). Increasing concentrations of PARP1 had no effect on 454

ATPase activities of either RECQL5 or WRN. 455

456

457

458

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

22

Strand annealing activities of RECQL5 are stimulated by PAR. 459

Besides helicase and ATPase activities, RECQL5 and WRN also exhibit ATP-460

independent single strand annealing activities. We investigated whether the single 461

strand annealing activities of RECQL5 or WRN are altered in the presence of PARP1, 462

PARylated PARP1 or PAR. As shown in Fig. 4A, the ability of RECQL5 to mediate 463

strand annealing was inhibited by increasing concentrations of PARP1. Interestingly, the 464

effect on RECQL5’s DNA annealing was promoted when there was a low level of 465

PARylated PARP1, but was inhibited at high levels of PARylated PARP1 (Fig. 4B). 466

RECQL5’s strand annealing activity could be slightly enhanced by increased levels of 467

PAR in the reaction, while PAR alone had no effect (Fig. 4C). Thus, PARP1, PARylated 468

PARP1 and PAR show more complex effects on strand annealing, distinct from their 469

uniformly inhibitory effects on the helicase and ATPase activities of RECQL5. 470

On the other hand, WRN showed inhibition of strand annealing activity with 471

unmodified PARP1 (Fig. 4D) and PARylated PARP1 (Fig. 4E). It is very likely that 472

PARylated PARP1 represses WRN strand annealing ability via steric hindrance. Low 473

concentrations of PAR had no effect on WRN strand annealing, whereas high PAR 474

concentrations partially inhibited the strand annealing of WRN (Fig. 4F). Thus, the two 475

RecQ proteins show clear differences with respect to strand annealing activity in the 476

presence of PARP1, PARylated PARP1 and PAR (Fig. 4G and H). The effects of 477

PARP1, PARylated PARP1 or PAR on RECQL5 or WRN’s annealing activities are 478

summarized in table 1. 479

480

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

23

PARP activation mediates the early recruitment of RECQL5-GFP to DSBs. 481

Our previous studies have shown that all RecQ helicase proteins are recruited to 482

laser-induced DSB sites (31, 38). RECQL5’s recruitment to DSBs exhibited the same 483

recruitment kinetic pattern as that of WRN and BLM. PARPs are activated early in 484

response to DNA damage and promote PAR production to allow DNA repair proteins to 485

be recruited to the damage sites. To investigate whether PARylation is important for 486

RECQL5 recruitment to sites of damage, we performed laser micro-irradiation and 487

observed RECQL5 recruitment to DSBs induced by 12% laser strikes from a 435 nm 488

beam. We have previously shown under these conditions we generate a majority of 489

DSBs (39). Our results indicated that PAR co-localized with the DSB marker, 53BP1, at 490

the laser sites 30 second after micro-irradiation (Fig. 5A). Using this methodology, we 491

find that endogenous RECQL5 and WRN as well as PAR can be recruited to the sites of 492

the laser strike. 493

To determine if PAR or PARP1 regulate WRN and RECQL5 recruitment to sites 494

of laser damage, we measured the recruitment kinetics of GFP-tagged RecQs to the 495

DSBs when PAR synthesis was inhibited by either olaparib or veliparib. Both inhibitors 496

compete for PARP1’s NAD+ binding pocket and trap PARP1 at DNA strand breaks (47). 497

In response to laser micro-irradiation, RECQL5-GFP was rapidly recruited to DNA 498

damage sites (Fig. 5B; see also Fig. S4A to C in the supplemental material). However, 499

when cells were pretreated with the PARP inhibitors, RECQL5 did not respond to the 500

laser strikes immediately. RECQL5 eventually recruited to the damage site at the later 501

time points but the extent of recruitment was less. In control cells, 100% of RECQL5 502

recruited to the damage sites 10 min after the laser strikes whereas in PARP inhibitor 503

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

24

treated cells, 30% (olaparib) and 52% (veliparib) of RECQL5 recruited to the sites (Fig. 504

5C). This suggests that PARP activation is mediating the early RECQL5 recruitment 505

and the later recruitment of RECQL5 is independent of PAR synthesis or alternatively 506

that DNA damage processing is delayed when PAR synthesis is inhibited and 507

consequently RECQL5’s recruitment is slowed. 508

In comparison, GFP-WRN was quickly able to relocate to the DNA damage sites 509

right after the laser strikes, and like RECQL5, GFP-WRN’s recruitment to the DNA 510

damage site was also inhibited by PARP inhibitors because the overall extent of 511

recruitment was lower after oliparib or veliparib treatment (Fig. 5D and E; see also Fig. 512

S4B to D in the supplemental material). GFP-WRN recruitment level dropped to 64% 513

(olaparib) and 74% (veliparib). These data demonstrate that PARP activity is required 514

for both RECQL5 and WRN full recruitment to sites of laser-induced DSBs. 515

Pharmacological PARP inhibitors target all DNA damage dependent PARP 516

homologs (i.e., PARP1, PARP2, PARP3). Since both RECQL5 and WRN interact with 517

PARP1, we also wanted to determine whether recruitment of both proteins was 518

dependent on the presence of PARP1. PARP1 knockout HeLa cells generated by 519

TALEN-mediated gene targeting were used to investigate the proteins’ recruitment 520

dynamics. These cells are completely null for PARP1 (see Fig. S5A in the supplemental 521

material). Lack of PARP1 had a moderate effect on the recruitment of RECQL5-GFP 522

and GFP-WRN (see Fig. S5 A to E in the supplemental material). Recruitment of both 523

proteins could be visualized at the earliest time points, however the proteins varied with 524

respect to the extent of recruitment. While GFP-WRN eventually reached its full 525

potential for recruitment between 400-600 sec, RECQL5-GFP did not, even after 10 526

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

25

minutes. In order to further suppress PARylation by other PARPs, veliparib was added 527

to the cells and recruitment of the proteins was monitored. Curiously, veliparib had a 528

more pronounced effect on the recruitment of GFP-WRN than RECQL5-GFP, as GFP-529

WRN showed a delay in its early recruitment dynamics. Decreased levels of RECQL5, 530

however, could be visualized during both the early and later recruitment phases and 531

veliparib treated cells showed no difference compared to the PARP1 KO cells (see Fig. 532

S5D and E in the supplemental material). These results suggest that even though 533

PARP1 is the most ubiquitously expressed PARP enzyme, other PARPs also contribute 534

to PAR synthesis at sites of DNA damage. Finally, we summarized the RECQL5 or 535

WRN’s recruitment at DSB sites in PARP1 KO cell and PARP inhibitors in table 1. 536

Previously, we reported that RecQ helicase proteins are retained for several 537

hours at sites of laser-induced DNA damage (31), therefore, we wanted to determine if 538

the absence or presence of PARP1 could alter retention kinetics. As shown, GFP-539

RECQL5 is normally retained for more than three hours at laser strikes and that is also 540

true in PARP1 KO cells (see Fig. S5F in the supplemental material). In addition, we also 541

examined the effects of DNA repair protein kinases on RECQL5’s recruitment to DSBs. 542

We still observed the recruitment of RECQL5 although the amount of recruitment varied 543

after various treatments. We examined the effects of VE821 (an ATR inhibitor), 544

KU55933 (a PI3K inhibitor), and Torin 2 (an ATR, ATM, DNA-PK, and mTOR inhibitor) 545

(see Fig. S6 in the supplemental material). When co-treating with veliparib, RECQL5’s 546

recruitment, under both high dose KU55933 and Torin 2 conditions, was strongly 547

impaired. Thus, these findings further support the notion that after DSBs the early 548

recruitment of RECQL5 depends on the activation of PARPs, and that the later 549

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

26

recruitment of RECQL5 might be associated with the activation of DNA signaling 550

proteins, like ATM, ATR, and DNA-PK. 551

552

Loss of RECQL5 or WRN is associated with high sensitivity to PARP inhibitor. 553

Since PARP inhibition was associated with significant inhibition of RECQL5 and WRN 554

recruitment to DSB sites, we asked if depletion of RECQL5 and WRN by siRNA would 555

lead to an increased sensitivity to PARP-inhibition. To test this we used two different cell 556

lines (HeLa and HEK293T) that expressed normal levels of RECQL5 and WRN (Fig. 6A 557

and B). In both cell lines, transfection with RECQL5 and WRN siRNA efficiently 558

depleted endogenous proteins (Fig. 6A and B). We next assessed PARP inhibitor 559

sensitivity in the RECQL5 and WRN-depleted cells in the presence of increasing 560

concentration of PARP inhibitor (Veliparib at 0, 1, 10 and 100 nM) using a cell counter 561

(Beckman Coulter Z1Coulter Particle Counter). Upon depletion of RECQL5 and WRN, 562

we consistently observed a decrease in cell viability in the presence of increasing 563

concentration of Veliparib (Fig. 6B) in both cell lines. The HeLa cells were generally 564

more sensitive, so we observed the effect at lower concentrations of PARP inhibition. 565

566

PAR formation regulates the recruitment of RECQL5-GFP domains to DSB sites. 567

Having demonstrated that early recruitment of RECQL5-GFP is impacted by 568

olaparib and veliparib (Fig. 5A and B), we next sought to determine which RECQL5 569

domains are associated with the PAR-dependent DSBs recruitment. The three RECQL5 570

constructs: helicase, KIX, and SRI domains as in Fig. 1, were evaluated. As shown in 571

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

27

Fig. 7, all three GFP-tagged RECQL5 domains displayed fast recruitment to the DNA 572

damage sites. Unlike the full length protein and as we reported previously (31), the 573

three GFP-RECQL5 fragments were not retained but instead completely disappeared 574

from the damage sites after three minutes. Upon treatment of cells with veliparib, there 575

was a dramatic loss of the recruitment signal for each of the GFP-RECQL5 fragments. 576

Interestingly, the GFP-KIX domain showed only minor recruitment to damage sites. 577

These findings are in agreement with Fig.1E that all three of RECQL5 domains can 578

interact with PAR and here we show that recruitment of these fragments is altered by 579

PARP inhibition. Thus, without PAR formation, the helicase, KIX, and SRI domains of 580

RECQL5 were not recruited to DNA damage sites. Taken together, these results 581

strongly demonstrate that more than one domain of RECQL5 is needed for its full 582

recruitment to DSB sites and that PARylation impacts recruitment of each of the GFP-583

RECQL5 fragments. 584

585

RECQL5’s role in homologous recombination DSB repair depends on PARP1. 586

Homologous recombination (HR) and non-homologous end joining (NHEJ) are 587

the two major pathways for DNA DSB repair. RECQL5 has previously been shown to 588

mediate non-crossover HR DSB repair (30). Yet, it is unclear whether PARP1 plays a 589

role in the process of the repair here or, whether RECQL5 has any function in regulating 590

NHEJ mediated repair. Here, we investigated the effect of PARP1 on HR or NHEJ 591

mediated DSB repair plus or minus RECQL5 or WRN knockdown. We adopted 592

established U2OS/DR-GFP system to analyze HR-mediated DNA repair efficiency and 593

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

28

HCA2 NHEJ-I9A and pEGFP-Pem1-Ad2 reporter systems to study NHEJ-mediated in 594

vivo DNA repair capacity (40, 41). After knocking down, the cells were transfected with 595

a plasmid expressing the restriction enzyme I-SceI to induce DSBs in the reporter 596

cassettes or with NHEJ reporter plasmid, pEGFP-Pem1-Ad2 cut with HindIII or I-SceI 597

restriction enzymes, along with a control reporter plasmid DsRed. In vivo DSB repair 598

efficiency was measured based on the expression of the green fluorescent protein 599

(GFP), which is translated from the cassettes repaired either by HR (U2OS/DR-GFP 600

cassette) or NHEJ (HCA2 NHEJ-19A and pEGFP-Pem1-Ad2 cassettes) pathway, by 601

flow cytometry. 602

U2OS lacking RECQL5 or PARP1 showed decreased homology directed repair 603

efficiency, but knockdown of WRN had no effect (Fig. 8B). Interestingly, when both 604

WRN and PARP1 or RECQL5 and PARP1 were depleted, the extent of HR repair was 605

reduced to similar levels as knockdown of PARP1 alone, indicating no additive effects. 606

The importance of RECQL5 in NHEJ was tested in vitro and in vivo in U2OS and HeLa 607

cells knocked down for RECQL5 (Fig. 8C). Examination of end-joined products from in 608

vitro NHEJ assays indicated that knockdown of RECQL5 had no significant effect on the 609

generation of dimer, trimer and higher multimers (HM) (Fig. 8D). Further, analysis of 610

EGFP fluorescence from the siRNA-knockdown cells transfected with HindIII and I-SceI 611

linearized pEGFP-Pem1-Ad2 plasmid suggested that RECQL5 has no significant effect 612

on the end-joining mediated by NHEJ in vivo (Fig. 8E). Similar results were obtained 613

with HCA2 NHEJ-I9A system (data not shown). Together these results suggest that 614

RECQL5 participates in HR-mediated DBS repair pathway as PARP1, but not in the 615

NHEJ. 616

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

29

617

DISCUSSION 618

Recently there has been a lot of interest in defining the role of PARP1 and PAR 619

in the DNA damage response network. Many PAR binding partners have been reported 620

to be DNA repair and checkpoint proteins and PAR formation can act as docking sites 621

for signal recruitment of proteins to initiate DDR (51, 52). Targeting of these proteins to 622

PAR depends on their PAR recognition domains. Thus far, six different protein modules 623

that non-covalently interact with PAR binding motif patterns have been identified. One, 624

called PBM, is approximately 22-26 residues long containing a mix of basic and 625

hydrophobic amino acids (53). A second one is called the macrodomain which is 626

comprised of ~190 amino acids; PARP9, PARP14, PARP15 have been reported to 627

contain macrodomains (53). A third one has been described as a PAR-binding zinc 628

finger motif (51). Furthermore, for proteins with WWE, BRCT or FHA domains it has 629

been demonstrated that their recruitment to DNA damage sites depend on PAR binding 630

(13, 54). Examination of the primary sequence of each of the human RecQ helicases 631

reveals that they have multiple versions of the PBM type motif and RECQL5’s motifs are 632

shown in Fig. S2 in the supplemental material. Consistent with this bioinformatics 633

analysis, all five human RecQ helicases show PAR binding ability (Fig. 1), with Kd 634

values in the nanomolar range and RECQL5 showing the strongest affinity of all RecQ 635

helicases to unfractionated, long and branched PAR. 636

PARP1 and PAR have been reported to regulate the activities of RECQL1 and 637

WRN (22, 24) and PARP1 has been reported to interact with RECQL4 (20), yet there 638

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

30

have been no reports regarding PAR interaction with RECQL5. RECQL5 interacts with 639

PARP1, but RECQL5 is not post-translationally modified by PARP1 (Fig. 1). To further 640

characterize the RECQL5 helicase, KIX, and SRI domains for PAR binding, we found 641

that indeed each fragment was able to bind PAR. Additionally, the three domains all 642

displayed fast recruitment to DSB sites but were poorly retained (Fig. 6). This finding is 643

in contrast to what we observed for the full length protein which is retained for hours 644

(see Fig. S4 in the supplemental material). Previously, we reported that only the 645

helicase and KIX, but not the SRI domains, were recruited to DSBs (31). Perhaps, the 646

two observations can be explained by use of different laser intensities, the previous 647

paper used 7% whereas here 12% was used. We are introducing more damage and 648

more DSBs with the higher intensity. Even at higher laser intensity, recruitment of all 649

three RECQL5 fragments was inhibited in the presence of veliparib indicating that their 650

recruitment depends on PARP activation. This finding supports our PBM computation 651

analysis and biotin-PAR pull-down findings that the three/four RECQL5 domains contain 652

a bona fide PBM site (see Fig. S2 in the supplemental material). 653

RECQL5 and WRN share many biochemical properties: helicase, ATPase and 654

strand annealing. Therefore, we sought to investigate how these activities were 655

modulated by PARP1, PARylated PARP1 and PAR. Both RECQL5 and WRN helicase 656

activities were inhibited by the presence of PARP1, PARylated PARP1, or PAR as 657

shown in Fig. 2. PARylated PARP1, but not unmodified PARP1, partially repressed the 658

ATPase activity of both proteins. Taken together, the results suggest that PARP1, 659

PARylated PARP1, or PAR inhibition of RECQL5 and WRN helicase activity might be 660

simply associated with competition for substrate. Therefore PARP1 might prevent 661

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

31

unwinding of a fork duplex by binding the DNA substrate with higher affinity than 662

RECQL5 or WRN. Thus, the results suggest that the RecQ interaction with PARP1 or 663

PAR structurally interferes with the activities of RECQL5 and WRN. Interestingly, strand 664

annealing analysis revealed differences between the proteins. PAR had no significant 665

effect on WRN’s single strand annealing activity, whereas it promoted single strand 666

annealing by RECQL5. In contrast, PARP1 repressed both RECQL5 and WRN’s 667

abilities to mediate single strand annealing. It has been demonstrated that RECQL5 668

prevents the access of RAD51 to single strand DNA (ssDNA) to form D-loop DNA 669

structures, which can promote homologous recombination events to form genomic 670

instability (27). Our findings indicate that PAR may promote RECQL5’s single strand 671

annealing which may further impede RAD51’s activity. 672

673

Consistent with our previous observation (31), both RECQL5 and WRN exhibit rapid re-674

localization to DNA lesions in response to laser-induced DSBs. PARP1 is well known for 675

its role early in the steps of the DNA repair pathway by acting as a sensor of DNA 676

breaks and becoming enzymatically activated to form PAR. We noted that by using 677

PARP-specific inhibitors, olaparib and veliparib, RECQL5 and WRN recruitment to 678

DSBs is affected. RECQL5 recruitment at the early time points is abrogated to a greater 679

extent than WRN (Fig. 5; see also Fig. S4 in the supplemental material). Additionally, 680

the extent of recruitment for both helicases is greatly reduced when treating cells with 681

olaparib and veliparib. These findings indicate that PARP activation is important for 682

RecQ response to DSBs. In the context of the PARP1 KO cells plus veliparib (see Fig. 683

S6 in the supplemental material), WRN’s early recruitment was more strongly affected 684

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

32

by PARP inhibition, while RECQL5’s later recruitment was nullified. As mutations in 685

checkpoint and DNA repair pathways are associated with cancer, tumor cells defective 686

in HR repair with diminished or deficient BRCA1/2 gene function accumulate DNA repair 687

lesions and are more sensitive to PARP inhibitors (55, 56). We next determined whether 688

lack of RECQL5 or WRN might sensitize RECQL5 and WRN-knockdown human cancer 689

cells (HeLa and HEK293T) to PARP inhibitor. Accordingly, RECQL5 and WRN 690

knockdown cells demonstrated increased sensitivity to PARP inhibitor treatment (Fig. 691

6). Though we detected a similar pattern of sensitivity to HeLa and HEK293T cells at 692

higher concentration of veliparib (100 nM), we found that HeLa cells were more 693

sensitive than HEK293T cell. 694

DNA damage induction elicits cascades of posttranslational modifications, including 695

PARylation, phosphorylation, ubiquitylation, and sumoylation that orchestrate the 696

aforementioned processes. These events are largely initiated by phosphatidylinositol 3-697

kinase–related kinases (PIKKs), such as ATM, ATR and DNA-PKcs which regulate both 698

the activity and recruitment of various DDR proteins to DNA damage sites (57). We 699

therefore sought to assess whether ATM or ATR also regulate RECQL5 and WRN 700

recruitment. Damage induced recruitment of RECQL5 and WRN were clearly impaired 701

in the presence of each of these protein kinase inhibitors (see Fig. S6 in the 702

supplemental material). ATM, ATR and DNA-PKcs inhibitor KU-55933 markedly 703

reduced the early recruitment of RECQL5 and WRN (see Fig.S6 in the supplemental 704

material). Importantly, these inhibitors also impaired the late retention time for RECQL5 705

and WRN compared to PARP inhibitor only (see Fig. S6 in the supplemental material). 706

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

33

These findings support our proposal that both WRN and RECQL5 recruitment is 707

regulated by PARP activation and PIKK proteins. This shows that, other than 708

PARylation, phosphorylation also regulates the recruitment of RECQL5 and WRN at 709

double strand break sites. Previously, it had been reported that RECQL5 enhances non-710

crossover DSB mediated HR repair (30). Indeed, our results support this finding and 711

further show that RECQL5 and PARP1 participate in the same pathway (Fig. 8). The 712

observation of the lack of WRN shown no significant effect on HR directed repair is also 713

consistent with previous findings in WS cells (58). Instead, WRN plays a significant role 714

in regulating NHEJ repair. 715

In general, the cellular activities of RecQ helicases must be tightly controlled until 716

needed, because uncontrolled unwinding of the DNA (or nuclease activity in case of 717

WRN) can cause genome instability, leading to catastrophic consequences for cells. 718

Therefore, it is tempting to speculate that PARPs and PAR control the physico-chemical 719

properties of RECQL5 and WRN in a spatio-temporal manner at the site of the damage. 720

Due to its dynamic nature PARylation provides an excellent means to control RecQ 721

function in time and space. Thus, it is likely that PARylation serves as signal to 722

differentially modulate RecQ activities in order to “choose” the most appropriate RecQ 723

protein at the right time to fulfill the relevant task. Since PARylation is often induced in a 724

dose-dependent manner with increasing levels of genotoxic stress, it is conceivable that 725

this leads to a step-wise control of function and interactions among the different RecQ 726

proteins dependent on the type and load of the cellular DNA damage. Here we provide 727

a comparison of the coordinated control of WRN and RECQL5 by their interaction with 728

PARP1 and PAR on a biochemical and cellular level during laser-induced DNA damage 729

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

34

response. Many open questions remain, in particular on the mechanistic details of these 730

processes. 731

732

ACKNOWLEDGEMENTS 733

HeLa PARP1-/- cells were a gift from Elisa Ferrando-May, A. M., and Alexander Bürkle 734

(University of Konstanz). We thank Dr. Xiao-Fan Wang from Duke University for U2OS 735

DR/GFP cell line and Dr. Gorbunova from University of Rochester for HCA2 NHEJ-I9A 736

cell line. We thank Dr. Thomasz Kulikowicz for the recombinant dual-tagged WRN 737

proteins. We thank Drs Beverly Baptise, and Jaya Sarkar for reading and providing 738

helpful comments on the manuscript.This work was supported in part by the Intramural 739

Research Program of the National Institutes of Health, National Institute on Aging. S.V. 740

was supported by a fellowship of the DFG-funded Research Training Group 1331. 741

742

743

744

745

746

747

748

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

35

REFERENCES 749

1. Sancar A, Lindsey-Boltz LA, Unsal-Kacmaz K, Linn S. 2004. Molecular mechanisms of 750 mammalian DNA repair and the DNA damage checkpoints. Annu Rev Biochem 73:39-85. 751

2. Lord CJ, Ashworth A. 2012. The DNA damage response and cancer therapy. Nature 481:287-752 294. 753

3. Kim MY, Zhang T, Kraus WL. 2005. Poly(ADP-ribosyl)ation by PARP-1: 'PAR-laying' NAD+ into a 754 nuclear signal. Genes Dev 19:1951-1967. 755

4. Gibson BA, Kraus WL. 2012. New insights into the molecular and cellular functions of poly(ADP-756 ribose) and PARPs. Nat Rev Mol Cell Biol 13:411-424.doi: 10.1038/nrm3376. 757

5. Meyer-Ficca ML, Meyer RG, Coyle DL, Jacobson EL, Jacobson MK. 2004. Human poly(ADP-758 ribose) glycohydrolase is expressed in alternative splice variants yielding isoforms that localize to 759 different cell compartments. Exp Cell Res 297:521-532.doi: 10.1016/j.yexcr.2004.03.050. 760

6. Niere M, Kernstock S, Koch-Nolte F, Ziegler M. 2008. Functional localization of two poly(ADP-761 ribose)-degrading enzymes to the mitochondrial matrix. Mol Cell Biol 28:814-824. 762

7. Min W, Cortes U, Herceg Z, Tong WM, Wang ZQ. 2010. Deletion of the nuclear isoform of 763 poly(ADP-ribose) glycohydrolase (PARG) reveals its function in DNA repair, genomic stability and 764 tumorigenesis. Carcinogenesis 31:2058-2065.doi: bgq205 [pii]. 765

8. Barkauskaite E, Brassington A, Tan ES, Warwicker J, Dunstan MS, Banos B, Lafite P, Ahel M, 766 Mitchison TJ, Ahel I, Leys D. 2013. Visualization of poly(ADP-ribose) bound to PARG reveals 767 inherent balance between exo- and endo-glycohydrolase activities. Nat Commun 4:2164. doi: 768 10.1038/ncomms3164. 769

9. Rosenthal F, Feijs KL, Frugier E, Bonalli M, Forst AH, Imhof R, Winkler HC, Fischer D, Caflisch A, 770 Hassa PO, Luscher B, Hottiger MO. 2013. Macrodomain-containing proteins are new mono-771 ADP-ribosylhydrolases. Nat Struct Mol Biol 20:502-507.doi: 10.1038/nsmb.2521. 772

10. Jankevicius G, Hassler M, Golia B, Rybin V, Zacharias M, Timinszky G, Ladurner AG. 2013. A 773 family of macrodomain proteins reverses cellular mono-ADP-ribosylation. Nat Struct Mol Biol 774 20:508-514.doi: 10.1038/nsmb.2523. 775

11. Mangerich A, Burkle A. 2012. Pleiotropic cellular functions of PARP1 in longevity and aging: 776 genome maintenance meets inflammation. Oxid Med Cell Longev 2012:321653. doi: 10. 777 1155/2012/321653. 778

12. Haince JF, McDonald D, Rodrigue A, Dery U, Masson JY, Hendzel MJ, Poirier GG. 2008. PARP1-779 dependent kinetics of recruitment of MRE11 and NBS1 proteins to multiple DNA damage sites. J 780 Biol Chem 283:1197-1208. 781

13. Li M, Lu LY, Yang CY, Wang S, Yu X. 2013. The FHA and BRCT domains recognize ADP-782 ribosylation during DNA damage response. Genes Dev 27:1752-1768. 783

14. Li M, Yu X. 2013. Function of BRCA1 in the DNA damage response is mediated by ADP-784 ribosylation. Cancer Cell 23:693-704. 785

15. Haince JF, Kozlov S, Dawson VL, Dawson TM, Hendzel MJ, Lavin MF, Poirier GG. 2007. Ataxia 786 telangiectasia mutated (ATM) signaling network is modulated by a novel poly(ADP-ribose)-787 dependent pathway in the early response to DNA-damaging agents. J Biol Chem 282:16441-788 16453. 789

16. Rouleau M, Patel A, Hendzel MJ, Kaufmann SH, Poirier GG. 2010. PARP inhibition: PARP1 and 790 beyond. Nat Rev Cancer 10:293-301. 791

17. Brosh RM, Jr. 2013. DNA helicases involved in DNA repair and their roles in cancer. Nat Rev 792 Cancer 13:542-558. 793

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

36

18. Croteau DL, Popuri V, Opresko PL, Bohr VA. 2014. Human RecQ helicases in DNA repair, 794 recombination, and replication. Annu Rev Biochem 83:519-552. 795

19. Adelfalk C, Kontou M, Hirsch-Kauffmann M, Schweiger M. 2003. Physical and functional 796 interaction of the Werner syndrome protein with poly-ADP ribosyl transferase. FEBS Lett 797 554:55-58. 798

20. Woo LL, Futami K, Shimamoto A, Furuichi Y, Frank KM. 2006. The Rothmund-Thomson gene 799 product RECQL4 localizes to the nucleolus in response to oxidative stress. Exp Cell Res 312:3443-800 3457. 801

21. Veith S, Mangerich A. RecQ helicases and PARP1 team up in maintaining genome integrity. 802 Ageing Res Rev. 2014 Dec 30. pii: S1568-1637(14)00141-X. doi: 10.1016/j.arr.2014.12.006. 803

22. Berti M, Ray Chaudhuri A, Thangavel S, Gomathinayagam S, Kenig S, Vujanovic M, Odreman F, 804 Glatter T, Graziano S, Mendoza-Maldonado R, Marino F, Lucic B, Biasin V, Gstaiger M, 805 Aebersold R, Sidorova JM, Monnat RJ, Jr., Lopes M, Vindigni A. 2013. Human RECQ1 promotes 806 restart of replication forks reversed by DNA topoisomerase I inhibition. Nat Struct Mol Biol 807 20:347-354. 808

23. von Kobbe C, Harrigan JA, Schreiber V, Stiegler P, Piotrowski J, Dawut L, Bohr VA. 2004. 809 Poly(ADP-ribose) polymerase 1 regulates both the exonuclease and helicase activities of the 810 Werner syndrome protein. Nucleic Acids Res 32:4003-4014. 811

24. Popp O, Veith S, Fahrer J, Bohr VA, Burkle A, Mangerich A. 2013. Site-specific noncovalent 812 interaction of the biopolymer poly(ADP-ribose) with the Werner syndrome protein regulates 813 protein functions. ACS Chem Biol 8:179-188. 814

25. Lebel M, Lavoie J, Gaudreault I, Bronsard M, Drouin R. 2003. Genetic cooperation between the 815 Werner syndrome protein and poly(ADP-ribose) polymerase-1 in preventing chromatid breaks, 816 complex chromosomal rearrangements, and cancer in mice. Am J Pathol 162:1559-1569. 817

26. Tadokoro T, Ramamoorthy M, Popuri V, May A, Tian J, Sykora P, Rybanska I, Wilson DM, 3rd, 818 Croteau DL, Bohr VA. 2012. Human RECQL5 participates in the removal of endogenous DNA 819 damage. Mol Biol Cell 23:4273-4285.doi: 10.1091/mbc.E12-02-0110. 820

27. Hu Y, Raynard S, Sehorn MG, Lu X, Bussen W, Zheng L, Stark JM, Barnes EL, Chi P, Janscak P, 821 Jasin M, Vogel H, Sung P, Luo G. 2007. RECQL5/Recql5 helicase regulates homologous 822 recombination and suppresses tumor formation via disruption of Rad51 presynaptic filaments. 823 Genes Dev 21:3073-3084. 824

28. Zheng L, Kanagaraj R, Mihaljevic B, Schwendener S, Sartori AA, Gerrits B, Shevelev I, Janscak P. 825 2009. MRE11 complex links RECQ5 helicase to sites of DNA damage. Nucleic Acids Res 37:2645-826 2657. 827

29. Islam MN, Paquet N, Fox D, 3rd, Dray E, Zheng XF, Klein H, Sung P, Wang W. 2012. A variant of 828 the breast cancer type 2 susceptibility protein (BRC) repeat is essential for the RECQL5 helicase 829 to interact with RAD51 recombinase for genome stabilization. J Biol Chem 287:23808-23818. 830

30. Paliwal S, Kanagaraj R, Sturzenegger A, Burdova K, Janscak P. 2014. Human RECQ5 helicase 831 promotes repair of DNA double-strand breaks by synthesis-dependent strand annealing. Nucleic 832 Acids Res 42:2380-2390. 833

31. Popuri V, Ramamoorthy M, Tadokoro T, Singh DK, Karmakar P, Croteau DL, Bohr VA. 2012. 834 Recruitment and retention dynamics of RECQL5 at DNA double strand break sites. DNA Rep 835 11:624-635.doi: 10.1016/j.dnarep.2012.05.001. 836

32. Janscak P, Garcia PL, Hamburger F, Makuta Y, Shiraishi K, Imai Y, Ikeda H, Bickle TA. 2003. 837 Characterization and mutational analysis of the RecQ core of the bloom syndrome protein. J Mol 838 Biol 330:29-42. 839

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

37

33. Orren DK, Brosh RM, Jr., Nehlin JO, Machwe A, Gray MD, Bohr VA. 1999. Enzymatic and DNA 840 binding properties of purified WRN protein: high affinity binding to single-stranded DNA but not 841 to DNA damage induced by 4NQO. Nucleic Acids Res 27:3557-3566. 842

34. Kawamitsu H, Hoshino H, Okada H, Miwa M, Momoi H, Sugimura T. 1984. Monoclonal 843 antibodies to poly(adenosine diphosphate ribose) recognize different structures. Biochemistry 844 23:3771-3777. 845

35. Popuri V, Huang J, Ramamoorthy M, Tadokoro T, Croteau DL, Bohr VA. 2013. RECQL5 plays co-846 operative and complementary roles with WRN syndrome helicase. Nucleic Acids Res 41:881-899. 847 doi: 10.1093/nar/gks1134. 848

36. Tadokoro T, Kulikowicz T, Dawut L, Croteau DL, Bohr VA. 2012. DNA binding residues in the 849 RQC domain of Werner protein are critical for its catalytic activities. Aging 4:417-429. 850

37. Tadokoro T, Rybanska-Spaeder I, Kulikowicz T, Dawut L, Oshima J, Croteau DL, Bohr VA. 2013. 851 Functional deficit associated with a missense Werner syndrome mutation. DNA Rep 12:414-421. 852

38. Singh DK, Karmakar P, Aamann M, Schurman SH, May A, Croteau DL, Burks L, Plon SE, Bohr 853 VA. 2010. The involvement of human RECQL4 in DNA double-strand break repair. Aging Cell 854 9:358-371. 855

39. Liberti SE, Andersen SD, Wang J, May A, Miron S, Perderiset M, Keijzers G, Nielsen FC, 856 Charbonnier JB, Bohr VA, Rasmussen LJ. 2011. Bi-directional routing of DNA mismatch repair 857 protein human exonuclease 1 to replication foci and DNA double strand breaks. DNA Rep 10:73-858 86. 859

40. Mao Z, Bozzella M, Seluanov A, Gorbunova V. 2008. DNA repair by nonhomologous end joining 860 and homologous recombination during cell cycle in human cells. Cell Cycle 7:2902-2906. 861

41. Stark JM, Pierce AJ, Oh J, Pastink A, Jasin M. 2004. Genetic steps of mammalian homologous 862 repair with distinct mutagenic consequences. Mol Cell Biol 24:9305-9316. 863

42. Shamanna RA, Hoque M, Lewis-Antes A, Azzam EI, Lagunoff D, Pe'ery T, Mathews MB. 2011. 864 The NF90/NF45 complex participates in DNA break repair via nonhomologous end joining. Mol 865 Cell Biol 31:4832-4843. 866

43. Malanga M, Althaus FR. 2011. Noncovalent protein interaction with poly(ADP-ribose). Methods 867 Mol Biol 780:67-82. 868

44. Barkauskaite E, Jankevicius G, Ladurner AG, Ahel I, Timinszky G. 2013. The recognition and 869 removal of cellular poly(ADP-ribose) signals. FEBS J 280:3491-3507. 870

45. Fahrer J, Kranaster R, Altmeyer M, Marx A, Burkle A. 2007. Quantitative analysis of the binding 871 affinity of poly(ADP-ribose) to specific binding proteins as a function of chain length. Nucleic 872 Acids Res 35:e143.doi: 10.1093/nar/gkm944. 873

46. Islam MN, Fox D, 3rd, Guo R, Enomoto T, Wang W. 2010. RecQL5 promotes genome 874 stabilization through two parallel mechanisms--interacting with RNA polymerase II and acting as 875 a helicase. Mol Cell Biol 30:2460-2472. 876

47. Murai J, Huang SY, Das BB, Renaud A, Zhang Y, Doroshow JH, Ji J, Takeda S, Pommier Y. 2012. 877 Trapping of PARP1 and PARP2 by Clinical PARP Inhibitors. Cancer Res 72:5588-5599. 878

48. Garcia PL, Liu Y, Jiricny J, West SC, Janscak P. 2004. Human RECQ5beta, a protein with DNA 879 helicase and strand-annealing activities in a single polypeptide. EMBO J 23:2882-2891. 880

49. Xu X, Liu Y. 2009. Dual DNA unwinding activities of the Rothmund-Thomson syndrome protein, 881 RECQ4. EMBO J 28:568-577. 882

50. Rossi ML, Ghosh AK, Kulikowicz T, Croteau DL, Bohr VA. 2010. Conserved helicase domain of 883 human RecQ4 is required for strand annealing-independent DNA unwinding. DNA Rep 9:796-884 804. 885

51. Ahel I, Ahel D, Matsusaka T, Clark AJ, Pines J, Boulton SJ, West SC. 2008. Poly(ADP-ribose)-886 binding zinc finger motifs in DNA repair/checkpoint proteins. Nature 451:81-85. 887

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

38

52. Pleschke JM, Kleczkowska HE, Strohm M, Althaus FR. 2000. Poly(ADP-ribose) binds to specific 888 domains in DNA damage checkpoint proteins. J Biol Chem 275:40974-40980. 889

53. Krietsch J, Rouleau M, Pic E, Ethier C, Dawson TM, Dawson VL, Masson JY, Poirier GG, Gagne 890 JP. 2013. Reprogramming cellular events by poly(ADP-ribose)-binding proteins. Mol Aspects 891 Med 34:1066-1087. 892

54. Wang Z, Michaud GA, Cheng Z, Zhang Y, Hinds TR, Fan E, Cong F, Xu W. 2012. Recognition of 893 the iso-ADP-ribose moiety in poly(ADP-ribose) by WWE domains suggests a general mechanism 894 for poly(ADP-ribosyl)ation-dependent ubiquitination. Genes Dev 26:235-240. 895

55. Bouwman P, Jonkers J. 2012. The effects of deregulated DNA damage signalling on cancer 896 chemotherapy response and resistance. Nat Rev Cancer 12:587-598. 897

56. Farmer H, McCabe N, Lord CJ, Tutt ANJ, Johnson DA, Richardson TB, Santarosa M, Dillon KJ, 898 Hickson I, Knights C, Martin NMB, Jackson SP, Smith GCM, A A. 2005. Targeting the DNA repair 899 defect in BRCA mutant cells as a therapeutic strategy. Nature 434:917-920. 900

57. Polo SE, Jackson SP. 2011. Dynamics of DNA damage response proteins at DNA breaks: a focus 901 on protein modifications. Genes Dev 25:409-433. 902

58. Chen L, Huang S, Lee L, Davalos A, Schiestl RH, Campisi J, Oshima J. 2003. WRN, the protein 903 deficient in Werner syndrome, plays a critical structural role in optimizing DNA repair. Aging Cell 904 2:191-199.doi: 10.1101/gad.2021311. 905

906

907

FIGURE LEGENDS 908

FIG 1 RECQL5 directly binds PAR and PARP1, but is not PARylated by PARP1. (A) 909

Far western analysis showing PAR interaction with all of the human RecQ helicases: 910

RECQL1, BLM, WRN, RECQL4, and RECQL5. Proteins were separated by SDS-PAGE 911

and transferred to nitrocellulose membranes and then incubated with PAR. 912

Immunoblotting was performed to detect the PAR. (B) PAR-overlay slot blot assay to 913

show interaction between PAR and RECQL5. (C) Electrophorectic mobility shift assay 914

with end-biotinylated PAR to examine RECQL5-PAR interaction in solution. Size-915

fractionated PAR (500 fmol) was incubated with increasing concentrations of RECQL5 916

as indicated, separated via native gel electrophoresis then PAR was detected via 917

streptavidin-coupled HRP. Right, quantitative evaluation of the RECQL5 gel shifts. Shift 918

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

39

[%] indicates [signal intensity complexed PAR]/[complexed + free PAR]. Data are 919

expressed as mean +/- SEM of triplicates. (D) RECQL5-GFP overexpressed in HEK 920

293T cells were treated 0.5 mM H2O2 for 10 min with or without 5uM veliparib. Co-921

immunoprecipitation shows that RECQL5 binds PAR and PARP1 in vivo. Input and 922

immunoprecipitated samples were analyzed by western blotting with the indicated 923

antibodies. (E) Biotin-PAR pull-down assays with RECQL5 fragments. Cell lysates 924

overexpressing control GFP, RECQL5-GFP or the various GFP-tagged RECQL5 925

domains (helicase, KIX, and SRI domains) were incubated with biotin labeled PAR then 926

streptavidin beads were used to pull down the biotin-PAR. (F) In vitro PARylation 927

reactions carried out with the indicated purified recombinant proteins. PARP1 (50 nM), 928

DNA (4 ng/ml), 500 µM NAD+, 5 µM olaparib and 50 nM RECQL5 were used as 929

indicated in the figure. 930

FIG 2 Both RECQL5 and WRN’s helicase activities are regulated by PARP1, 931

PARylated PARP1 and PAR. RECQL5’s helicase activity in the presence of increasing 932

concentrations (none, 0.5 nM, 1 nM, 2 nM, 5 nM, 10 nM) of PARP1 (A), PARylated 933

PARP1 (B) or PAR (C). WRN helicase activity in the presence of none, none, 0.5 nM, 1 934

nM, 2 nM, 5 nM, 10 nM of PARP1 (D), PARylated PARP1 (E) or PAR (F). (G) 935

Quantitative analysis of A, B, and C. (H) Quantitative analysis of D, E, and F. PARP1, 936

PARylated PARP1, or PAR show no helicase properties and the symbol (∆) is the 937

denatured substrate control. All experiments were repeated at least three times and the 938

error bars represent standard error of the mean (s.e.m.). 939

FIG 3 Activation of PARP1 alters the ATPase activities of both RECQL5 and WRN. 940

(A) WRN (5 nM) and RECQL5 (5 nM) ATPase activity in the presence or absence of 941

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

40

activated PARP1 (10 nM), as shown. Lower panel, western blotting showing PAR to 942

demonstrate the extent of PARP1 activation. (B) Quantification of fig (A) upper panel. 943

(C) RECQL5 ATPase activity measured in the presence of increasing concentrations of 944

PARP1 and PAR. None (lane 1), 100 nM PARP1 (lane 3), and 500 nM (PAR) show no 945

ATP hydrolysis. RECQL5 alone (10 nM, lane 2) or with the 25 nM, 50 nM and 100 nM of 946

PARP1 (lane 4, 5, 6) or with 100 nM, 250 nM, and 500 nM of PAR (lane 8, 9, 10). (D) 947

Quantification of fig (C) . (E) Similar to figure (C) but in the presence of WRN (5 nM). (F) 948

Quantification of fig (E). The values represent an average of three independent 949

experiments, error bars represent standard error of the mean (s.e.m.). 950

FIG 4 The effects of PARP1, PARylated PARP1 and PAR on single strand 951

annealing activities of RECQL5 and WRN. Strand annealing activity of RECQL5 (10 952

nM) examined in the presence of the increasing concentrations (none, 5, 10, 20, 40, 953

and 80 nM) of PARP1 (A), PARylated PARP1 (B), and PAR (C) with the radiolabeled 954

ssDNA fork DNA and its complimentary single strand DNA. (G) is the quantitative 955

results of (A), (B), and (C). WRN’s (10nM) strand annealing activity measured in the 956

presence of increase doses of PARP1 (D), PARylated PARP1 (E), and PAR (F) same 957

concentrations as for RECQL5. (H) is the quantification. The values represent an 958

average of three independent experiments, error bars represent standard error of the 959

mean (s.e.m.). 960

FIG 5 PARylation regulates the early recruitment of RECQL5 to laser induced DNA 961

damage. (A) RECQL5 or WRN displays co-localization with PAR at the DNA lesions. 962

HeLa cells were struck with 12% laser to induce DNA damage. Cells were 963

immunostained with PAR and 53PB1, RECQL5, or WRN antibodies. GFP tagged (B) 964

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

41

RECQL5 or (D) WRN were transiently transfected into HeLa cells for 24 h. The 965

overexpressed cells were pre-treated for 3h with vehicle control, 5 µM olaparib or 5 µM 966

veliparib, and then targeted with the 12% laser to induce DSBs. The cells were imaged 967

at the indicated time points to observe recruitment of the proteins to the DNA damage. 968

Recruitment kinetics of at least twenty cells from three independent experiments were 969

quantified for RECQL5-GFP (C) and GFP-WRN (E) and the relative signal intensities at 970

the laser line were calculated using velocity software and plotted versus time. The white 971

box with the dotted outline indicates the laser striking area. Absolute values are 972

reported in Suppl. Fig. 4. The error bars represent standard error of the mean (s.e.m.). 973

FIG 6 Cancer cell lines with loss of RECQL5 or WRN show increased sensitivity 974

to PARP inhibitor. (A) Western blot demonstrating the expression status of RECQL5 or 975

WRN in HeLa cell line treated with respective siRNA (Right Panel), (Left Panel) 976

demonstrating the sensitivity to the PARP inhibitor (Veliparib 0 nM, 1 nM, 10 nM, 100 977

nM) in RECQL5 or WRN knockdown in HeLa cell line. (B) Western blot demonstrating 978

the expression status of RECQL5 or WRN in HEK293T cell line, treated with respective 979

siRNA (Right Panel), Left Panel demonstrating the sensitivity to the PARP inhibitor 980

(Veliparib) in RECQL5 or WRN knockdown in HEK293T cell line Tubulin was used as 981

loading control for western blot. All cell lines were treated in triplicates for 4 days. 982

Survival is represented in percentage of the control (DMSO treated). The cell numbers 983

were quantified using a colorimetric method (SRB). The error bars represent standard 984

error of the mean (s.e.m.). 985

986

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

42

FIG 7 RECQL5-GFP fragment’s recruitment to laser induced DSBs depends on 987

PARP1’s PARylation activity. (A) The various C-terminal GFP-tagged RECQL5 988

domain constructs, helicase (1-500), KIX (501-650), and SRI domains (853-991), were 989

expressed in HeLa cells. Before inducing DSB damage with the laser (12% intensity), 990

the cells were treated for 3 h with vehicle control or 5 µM PARP1 inhibitor, veliparib. (B) 991

Graphical quantitation of the recruitment dynamics of the RECQL5-GFP fragments to 992

the laser strike with or without veliparib treatment. The white box with the dotted outline 993

indicates the laser striking area. Three independent experiments were performed and at 994

least twenty cells were quantified for each sample and error bars represent the standard 995

error of the mean (s.e.m.). 996

FIG 8 The effects of PARP1 on WRN or RECQL5 mediated HR or NHEJ DSB repair 997

in vivo. (A) Western blot analysis of the U2OS/DR-GFP cells that were knockdown of 998

WRN, RECQL5, PARP1, WRN & PARP1, and RECQL5 & PARP1. Cells were 999

transfected with the indicated siRNA and cell extracts were analyzed by western blot 1000

and probed with the indicated antibodies. (B) The effect of WRN, RECQL5, and PARP1 1001

on the efficiency of repairing I-SCE mediated DSBs through HR. The scheme of the 1002

reporter cassette. Knockdown U2OS/DR-GFP cells were transfected with a plasmid 1003

expressing I-SCE. The cells which repair the GFP cassette will express GFP and are 1004

measured by flow cytometry. The repair efficiency of HR repair is relative to the siSCR 1005

condition, which is set arbitrarily to 100%. (Fifty thousand events were collected in for 1006

each point and five independent experiments were quantified and error bars indicate the 1007

standard error of the mean (s.e.m.). (C) Cells were transfected with RECQL5 siRNA 1008

and cell extracts were analyzed by western blot and probed with the indicated 1009

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

43

antibodies and the same extracts were used for NHEJ assay. (D) In vitro NHEJ assay 1010

on cohesive-end containing DNA substrate with cellular extracts prepared from 1011

siControl and siRECQL5 HeLa and U2OS cells. The end-joining assays were performed 1012

with SalI-linearized pUC18 DNA substrate. The products were resolved by agarose gel 1013

electrophoresis , the gel shows the total end joining from three independent 1014

experiments. DNA substrate, with end sequence, used in the NHEJ is shown on the top 1015

of gel. EDTA, reaction in the presence of EDTA. (E) In vivo NHEJ assays with cohesive 1016

and non-cohesive end containing DNA substrates. U2OS siControl and siRECQL5 cells 1017

were transfected with either HindIII digested (cohesive ends) or I-SceI digested (non-1018

cohesive ends) linearized pEGFP-Pem1-Ad2 reporter plasmid. Twenty-four-hour post 1019

transfection, the cells were harvested and analyzed by flow cytometry for EGFP and 1020

DsRed positive cells. Bar graph represents the quantification of EGFP-positive cells 1021

normalized to DsRed-positive cells from three independent experiments. The error bars 1022

represent the standard error of the mean (s.e.m.). 1023

. 1024

1025

1026 1027

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from

Interacting Partner

Strand Annealing Helicase Activity

RECQL5 WRN RECQL5 WRN

PARP1

PARylated PARP1

PAR

NA NA

Cell line or Treatment

Recruitment at DSB Site

RECQL5 WRN

PARP1 KO # 1 cell line

Delayed, relative to WT cells

Delayed, relative to WT cells

+ PARP1 Inhibitor

(Veliparib)

Delayed Delayed

+ PARP1 Inhibitor (Oliparib)

Delayed Delayed

Table 1. Summary of interacting partner’s effect on human RecQ protein activities. RECQL5 and WRN’s DNA strand annealing and helicase activities were performed in the presence of PARP1, PARylated PARP1 or PAR. ( ) represents minimal stimulation, ( ) represents greater stimulation, ( ) represents minimal inhibition, ( ) represents greater inhibition, ( ) represents greatest inhibition and NA represents not affected. Lower panel, effect of PARP1 protein or activity loss on recruitment of RECQL5 and WRN to double strand break sites.

on February 10, 2018 by guest

http://mcb.asm

.org/D

ownloaded from