• Aucun résultat trouvé

A disease-causing single amino-acid deletion in the coiled-coil domain of RAD50 impairs MRE11 complex functions in yeast and humans

N/A
N/A
Protected

Academic year: 2021

Partager "A disease-causing single amino-acid deletion in the coiled-coil domain of RAD50 impairs MRE11 complex functions in yeast and humans"

Copied!
34
0
0

Texte intégral

(1)

HAL Id: hal-03027155

https://hal.archives-ouvertes.fr/hal-03027155

Submitted on 27 Nov 2020

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

A disease-causing single amino-acid deletion in the

coiled-coil domain of RAD50 impairs MRE11 complex

functions in yeast and humans

Marie Chansel-da Cruz, Marcel Hohl, Ilaria Ceppi, Laëtitia Kermasson,

Laurence Maggiorella, Mauro Modesti, Jean-Pierre de Villartay, Talia Ileri,

Petr Cejka, John Petrini, et al.

To cite this version:

Marie Chansel-da Cruz, Marcel Hohl, Ilaria Ceppi, Laëtitia Kermasson, Laurence Maggiorella, et al.. A disease-causing single amino-acid deletion in the coiled-coil domain of RAD50 impairs MRE11 complex functions in yeast and humans. Cell Reports, Elsevier Inc, In press. �hal-03027155�

(2)

1 2 3 4

A disease-causing single amino-acid deletion in the coiled-coil domain of RAD50

5

impairs MRE11 complex functions in yeast and humans

6 7 8 9

Marie Chansel-Da Cruz1, 2, 3, Marcel Hohl4, †, Ilaria Ceppi 5,6, †, Laëtitia Kermasson1,2,

10

Laurence Maggiorella3, Mauro Modesti7, Jean-Pierre de Villartay1, 2, Talia Ileri8,

11

Petr Cejka5, 6, ‡, John. H. J. Petrini4, ‡, Patrick Revy1, 2, *, #

12 13 14 15

1

INSERM UMR 1163, Laboratory of Genome Dynamics in the Immune System, Equipe Labellisée la

16

Ligue contre le cancer, Paris, France

17

2

University of Paris-Sorbonne Paris Cité University, Imagine Institute, Paris, France

18

3

Genomic Vision, R&D innovation department, Bagneux, France

19

4

Molecular Biology Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA

20

5

Institute for Research in Biomedicine, Università della Svizzera italiana (USI), Faculty of Biomedical

21

Sciences, Via Vincenzo Vela 6, 6500 Bellinzona, Switzerland.

22

6

Department of Biology, Institute of Biochemistry, Eidgenössische Technische Hochschule (ETH), 8093

23

Zürich, Switzerland.

24

7

Cancer Research Center of Marseille; CNRS UMR7258; INSERM U1068; Institut Paoli-Calmettes;

25

Aix-Marseille Université, Marseille, France

26

8

Ankara University School of Medicine, Pediatric Hematology and Oncology, Ankara, Turkey

27

† These authors equally contributed to the study

28

‡ These authors equally contributed to the study

29 30 * Lead contact 31 #

Correspondance: Patrick Revy, Imagine Institute, 24 bd du Montparnasse, 75015 Paris, France.

32 Email: [email protected] . 33 Tel. (33) +1 427 542 92 34 35

(3)

36

Summary

37 38

The MRE11-RAD50-NBS1 complex plays a central role in response to DNA double-strand breaks. Here,

39

we identified a patient with bone marrow failure and developmental defects caused by biallelic RAD50

40

mutations. One of the mutations created a null allele while the other (noted RAD50E1035Δled to the loss

41

of a single residue in the heptad repeats within RAD50 coiled-coil domain. This mutation represents a

42

human RAD50 separation-of-function mutation that impairs DNA repair, DNA replication, and DNA end

43

resection without affecting ATM-dependent DNA damage response. Purified recombinant proteins

44

further indicated that RAD50E1035Δ impaired MRE11 nuclease activity. The corresponding mutation in

45

Saccharomyces cerevisiae caused severe thermosensitive defects in both DNA repair and Tel1ATM

-46

dependent signaling. These findings demonstrate that a minor heptad break in the RAD50 coiled-coil

47

suffices to impede MRE11 complex functions in human and yeast. Furthermore, these results emphasize

48

the importance of the RAD50 coiled-coil to regulate MRE11-dependent DNA end resection in humans.

49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71

(4)

Introduction

72 73

DNA double-strand breaks (DSBs) are considered the most toxic form of DNA damage. While unrepaired

74

DSBs can cause cell death, improperly repaired DSBs represent a source of mutation and translocation

75

that challenges genome stability and can ultimately result to cancer development (Scully et al., 2019).

76

MRE11-RAD50-NBS1 (MRN; MRX in yeast for Mre11-Rad50-Xrs2) is a conserved complex that plays

77

a central role in the sensing of and response to DSBs (Lisby et al., 2004; Mirzoeva and Petrini, 2001).

78

MRN is one of the first molecular entities detected at DSBs where it activates the kinase ATM to trigger

79

DNA damage response (DDR) signaling to induce cell cycle checkpoints and apoptosis (Lisby et al.,

80

2004; Stracker and Petrini, 2011). The MRN complex also influences DSB repair via the endo- and exo-

81

nuclease activities of MRE11, initiates DNA end resection to generate, in concert with other nucleases,

82

single stranded DNA ends that drive homology-directed repair (HDR) (Scully et al., 2019; Syed and

83

Tainer, 2018).

84

RAD50 belongs to the highly conserved structural maintenance of chromosome (SMC) family of

85

proteins (de Jager et al., 2004; Hopfner and Tainer, 2003). RAD50 possesses an extended coiled-coil

86

domain that folds back upon itself to form an intramolecular anti-parallel coiled‐ coil structure of about

87

500 angstroms that leads to juxtaposition of the N‐ and C‐ terminal Walker A and B domains and

88

generates a functional ATPase that directly interacts with MRE11. The coiled‐ coil structure, which is

89

conserved across the SMC proteins and all the known Rad50 orthologs (de Jager et al., 2004; Paull, 2018;

90

Stracker and Petrini, 2011; Syed and Tainer, 2018), comprises a heptad repeat pattern wherein the first

91

and the fourth residue are hydrophobic to allow the association of the anti-parallel helices via their

92

hydrophobic faces (de Jager et al., 2001; Truebestein and Leonard, 2016; van Noort et al., 2003). The

93

apex of the RAD50 coiled-coil consists of a zinc hook domain by which two MRE11-RAD50 complexes

94

can dimerize to form intra- and intermolecular complexes that have been proposed to bridge DNA ends

95

(Hohl et al., 2015; Hopfner, 2014; Park et al., 2017; Paull, 2018; Stracker and Petrini, 2011; Syed and

96

Tainer, 2018; Tatebe et al., 2020). It has also been recently suggested that MRE11-RAD50 complexes

(5)

could promote DNA tethering of sister chromatids at stalled forks by facilitating Cohesin loading

98

(Delamarre et al., 2020). Although the flexibility of the RAD50 coiled-coil domain has prevented its

99

whole structure determination at the atomic resolution (Kashammer et al., 2019; Park et al., 2017), several

100

studies suggested that this domain participates in the regulation of MRN functions by propagating

101

information from the Zn-hook domain to the globular domain (Hohl et al., 2015; Hohl et al., 2011; Park et

102

al., 2017). Recent analyses using crystallographic, cryo-electron, and high-throughput single-molecule

103

microscopy suggested that dimers of ATP-bound MRE11-RAD50 intracomplex (M2R2) mostly adopt a

104

ring conformation able to interact and scan DNA homoduplex to recognize DNA end and trigger

105

ATM/Tel1-dependent DDR (Deshpande et al., 2014; Hopfner, 2014; Kashammer et al., 2019; Myler et al.,

106

2017; Tatebe et al., 2020). A recent model suggested that, upon ATP and DNA binding, the

MRE11-107

RAD50 intracomplex undergoes a conformational switch leading to the interaction of intermolecular

108

coiled-coil domains resulting in a rod shaped structure. This structural modification would allow the

109

exposure of the catalytic domain of MRE11 to DNA end and stimulates nuclease activity and DNA end

110

processing (Deshpande et al., 2014; Hopfner, 2014; Kashammer et al., 2019; Lammens et al., 2011; Lim

111

et al., 2011; Liu et al., 2016; Rojowska et al., 2014). Several studies conducted in yeast and bacteria

112

suggested that the RAD50 coiled-coil domain was particularly important during these conformational

113

transitions by propagating spatial information from the Zn hook to the globular domain (Hohl et al., 2015;

114

Hohl et al., 2011; Hopfner et al., 2002; Kashammer et al., 2019; Lee et al., 2013; Park et al., 2017).

115

Nonetheless, despite the recent progress in the structure determination of all or part of the MRN complex,

116

our knowledge of the RAD50 coiled-coil functional role in the regulation of the multiple MRE11

117

complex functions remains limited, especially in mammals (Paull, 2018).

118

Here, we identified an individual, P1, presenting with immunodeficiency and developmental

119

defects and has compound heterozygous mutations in RAD50. One of the mutations creates a null allele

120

while the other appears to be hypomorphic due to the loss of a single amino-acid residue in the coiled-coil

121

domain of RAD50 (noted RAD50E1035Δ). The analysis of the human RAD50E1035Δ mutant as well as its

(6)

yeast counterpart provided unprecedented information, emphasizing the fundamental role of the RAD50

123

coiled-coil domain in modulating MRE11 complex functions. Furthermore, this study reports the first

124

RAD50 separation-of-function mutation that does not affect ATM-dependent DNA damage response but

125

severely impairs DSB repair. These findings demonstrate the crucial role of the RAD50 coiled-coil

126

domain to regulate MRE11-dependent DNA end resection and repair in humans.

(7)

Results

128 129

Clinical features of individual P1

130

This study was initiated by the analysis of an individual P1 who was diagnosed at 7-years of age

131

with bone marrow failure and developmental defects. In the family, the two other children and the parents

132

were healthy (Figure 1A). The P1's cell blood count revealed thrombocytopenia (platelet count of 113

133

G/liter) associated with anemia (hemoglobin level of 10 g/dL), and neutropenia (neutrophil count of 1

134

G/L) (Table S1). Bone marrow smears and biopsies showed reduced cellularity (15%) without

135

myelodysplastic features, consistent with aplastic anemia. Immunophenotyping highlighted a virtual

136

absence of circulating B lymphocytes (CD19+) while T lymphocytes (CD3+) that were normal in number

137

exhibited an increased proportion of central memory T cells (CD3+ CCR7- CD45A-) (Table S1).

138

Conversely, the proportions of both naive CD4+ T cells (evaluated by the CD45RA marker) and recent

139

thymus emigrant (characterized by the CD45RA+CD31+markers) were markedly reduced, suggesting an

140

impaired T lymphopoiesis (Table S1). Moreover, physical examination of P1 revealed failure to thrive

141

(height 108.5cm, <3rd percentile; weight 17.2 kg, <3rd percentile), skin pigmentation, nail dysplasia,

142

leukoplakia, dental loss, microcephaly (cranial perimeter 48cm, <3rd percentile), and dysmorphia. P1 also

143

developed cataract at 10 years of age that necessitated surgical intervention. Collectively, P1's clinical

144

features were reminiscent of a defect in DNA repair and/or telomere maintenance (Glousker et al., 2015;

145

Rivera-Munoz et al., 2007). However, telomere restriction fragment (TRF) analysis indicated that

146

telomere length in P1's peripheral blood cells was not reduced as compared to his parents and his healthy

147

sister (Figure S1).

148 149

Individual P1 carries biallelic RAD50 mutations

150

To determine the molecular etiology of the disease, we performed whole exome sequencing

151

analysis in P1 byconsidering coding sequences carrying rare biallelic variants (frequency less than 0.1% 152

in 1000 genomes, EVS, dbSNP, and our in-house database (8319 individuals)) predicted to be deleterious 153

(Figure 1A). This approach highlighted two mutations in RAD50 that were confirmed by Sanger 154

sequencing and present in a compound heterozygous configuration (Figure 1B). One consisted in a

(8)

nucleotide insertion leading to a frameshift and a premature stop codon (c.2165dup; p.Glu723Glyfs*5)

156

(Figures 1B, 1C) inherited from the asymptomatic mother and present in both healthy sister and brother

157

(Figure S2A). The other mutation was an in-frame deletion of three nucleotides resulting to the loss of a

158

single glutamic acid residue at position 1035 (c.3109_3111del; p.Glu1035del) located in the coiled-coil

159

domain of RAD50 (Figures 1B, 1C and Figure S2A). This deletion corresponded to a de novo mutation

160

since it was absent in siblings and parents (Figure S2A) while microsatellite analysis ascertained for the

161

genetic paternity of the P1's father (Figure S2B). Both RAD50 mutations were absent from gnomAD

162

database (>120,000 individuals tested). Sequencing of RAD50 cDNA from P1's blood cells did not detect

163

the c.2165dup; p.Glu723Glyfs*5 mutation likely owed to nonsense mediated decay (NMD) (not shown).

164

Hence, the only RAD50 transcripts present in P1's cells encoded a RAD50 protein lacking a unique

165

glutamic acid residue in the coiled-coil domain of RAD50 (p.Glu1035del, hereafter dubbed RAD50E1035Δ)

166

(Figure 1C). RAD50 immunoblots with B lymphoblastoid cell line (B-LCL) lysates from P1 and his

167

mother revealed approximately half of RAD50 amount as compared to B-LCL lysate from a healthy

168

donor (Figure 1D). This result was consistent with the presence of a null allele in both P1 and his mother

169

(i.e. c.2165dup; p.Glu723Glyfs*5). Moreover, this finding indicated that RAD50E1035Δ was correctly

170

expressed and as stable as its WT counterpart. Congruent with an interdependent stability of the human

171

MRN components(Stewart et al., 1999; Waltes et al., 2009), we also noticed a reduced amount of NBS1

172

and MRE11 in cell lysates from P1 and his mother (Figure 1D). The RAD50E1035Δ mutation was located

173

in the coiled-coil structure composed of heptad amino acid repeats commonly labeled abcdefg, where a

174

and d represent hydrophobic positions(Truebestein and Leonard, 2016) (Figure 1E). The program

175

Marcoil that calculates coiled-coil probability(Delorenzi and Speed, 2002) indicated that the loss of the

176

residue E1035 led to a break in the heptad repeats (Blue arrow in Figure 1F) that was predicted to disrupt

177

the RAD50 coiled-coil structure nearby the mutation, as inferred by the drop in coiled-coil probability in

178

this region (Blue arrow in Figure 1G). Thus, we decided to investigate the functional consequence of this

179

peculiar mutation in P1's cells.

180 181

(9)

RAD50E1035Δ mutation impairs DNA repair

182

First, we assessed whether P1's cells were able to cope with various DNA lesions produced by

183

distinct genotoxic agents. P1's SV40-transformed fibroblasts, similar to cells from an ATM-deficient

184

patient, exhibited a strong sensitivity to ionizing radiation (IR) suggesting a DNA repair defect (Figure

185

2A). P1's cells also had a pronounced sensitivity to the DSB-inducing agent phleomycin (Figure 2B) and

186

to etoposide (Figure 2C), a drug that generates topoisomerase2-DNA adducts. P1's cells were also

187

sensitive to the DNA interstrand crosslink (ICL)-inducing agent mitomycin C (MMC), although at a

188

lesser extent than cells from a Fanconi anemia patient (Figure 2D), and not associated with an impaired

189

MMC-induced FANCD2 ubiquitination (not shown). Of note, a human HT1080 cell line carrying a

190

nonsense mutation on one allele of the RAD50 gene (RAD50+/-) generated by CRISPR/Cas9 only

191

exhibited a modest sensitivity to genotoxics at higher doses, ruling out haploinsufficiency as the cause of

192

the severe DNA repair defect in P1's cells (Figure S3). Taken together, these results demonstrated that

193

P1's cells have a general DNA repair defect that is likely the cause of the bone marrow failure and

194

developmental anomalies found in this patient.

195

To verify that the RAD50 deficiency in P1’s cells was responsible for the DNA repair defect, we

196

transduced the cells with a vector expressing wtRAD50 (Figure S4). While empty vector had no effect,

197

wtRAD50-expressing vector complemented the phleomycin sensitivity of P1's cells (Figure 2E). We

198

further confirmed this result by a multicolor competition assay after transduction of P1 and control cells

199

with a lentiviral Ires-mCherry vector containing or not the wtRAD50 coding sequence (Smogorzewska et

200

al., 2007). A mix of transduced (mCherry+) and nontransduced (mCherry-) cells was analyzed for

201

selective advantage in culture, as evaluated by the mCherry index of transduced (mCherry+) over

202

nontransduced (mCherry-) cells upon treatment with phleomycin (Figure 2F). Ectopic expression of

203

mCherry/wtRAD50 but not mCherry alone gave in P1's cells a strong selective advantage over

204

nontransduced cells upon treatment with phleomycin, as determined by the 40-fold increase in

205

mCherry/wtRAD50-expressing cells after 18 days in this culture condition (Figure 2F). Taken together,

206

these results established a causal link between the DNA repair defect and the expression of RAD50E1035Δ

(10)

in P1's cells.

208

209

RAD50E1035Δ mutation does not affect DNA damage sensing and signaling 

210

To determine whether the DNA repair defect in P1's cells could be associated with an impaired

211

MRN recruitment at DNA damage sites, we monitored the dynamics of NBS1 at DSBs. We transfected

212

cells with a vector expressing GFP-NBS1 fusion protein and analyzed by live-microscopy the GFP-NBS1

213

behavior following the generation of localized DNA damages induced by a laser microbeam (Zhang et al.,

214

2016). Kinetics and intensity of GFP-NBS1 recruitment at DNA lesions was similar in control and P1's

215

cells (Figures 2G and 2H) indicating that RAD50E1035Δ

216

217

We next assessed whether the DNA repair defect in P1's cells was associated with impaired

ATM-218

dependent DDR functions (Stracker and Petrini, 2011). As expected, KAP1 phosphorylation (P-KAP1),

219

analyzed by immunofluorescence, was undetectable in primary fibroblasts from ATM- and

NBS1-220

mutated patients demonstrating the need for functional ATM and MRN in this process (Figure 2I). To

221

our surprise, IR-induced P-KAP1 was readily detected in primary fibroblasts from P1 suggesting that

222

RAD50E1035Δ did not impair ATM-dependent KAP1 phosphorylation. Furthermore, Western blot analysis

223

demonstrated that P1's SV40-fibroblasts normally induced phosphorylation of NBS1 and CHK2

224

following IR, while ATM-deficient control cells did not (Figure 2J). Consistent with a normal

IR-225

induced CHK2 phosphorylation, P1's SV40-transformed fibroblasts exhibited functional G2/M

226

checkpoint, as determined by the reduction of phospho-histone 3 in G2 phase, a marker of chromosome

227

condensation, following IR (Figure 2K). As expected, ATM-deficient cells used as negative control had a

228

defective G2/M checkpoint (Figure 2K).

229

Collectively, these results demonstrated that P1's cells behaved differently from ATM- and

NBS1-230

mutated cells and provided evidence that the human RAD50E1035Δ mutation, although leading to a DNA

(11)

repair defect, did not impair recruitment of the MRN complex at DNA damage sites or the subsequent

232

activation of ATM.

233 234

RAD50E1035Δ mutation impairs DNA end resection and homology directed repair

235

MRE11 initiates double-stranded DNA end resection via its nuclease activity, a process that

236

culminates in the formation of single stranded DNA (ssDNA) that is used to scan sequence homology and

237

promote homology directed repair (HDR) (Stracker and Petrini, 2011; Syed and Tainer, 2018). During

238

this process the ssDNA molecules are coated by RPA and then by RAD51 (Zhao et al., 2019). To get

239

further insight into the DNA repair pathway that might be affected in this patient, we assessed DNA end

240

resection following IR-treatment in P1's cells. In contrast to P-KAP1, which was readily detected in P1's

241

primary fibroblasts 6h post IR (Figure 3A), RPA (Figure 3B) and RAD51 (Figures 3C and Figure S5;

242

P<0.0001) foci were virtually absent in these cells, suggesting impaired DNA end resection at IR-induced

243

DSB. The impaired RAD51 foci formation upon IR treatment was further confirmed in SV40-hTERT

244

P1's fibroblasts (P<0.0001; Figure 3D). Transduction with wtRAD50 but not with an empty vector

245

complemented the impaired IR-induced RAD51 foci formation in P1's cells, a result demonstrating the

246

causal link between compromised DNA end resection and RAD50 deficiency in these cells (Figure 3D).

247

We next tested HDR pathway by measuring the efficiency of insertion in a chromosomal context of a

248

DNA sequence encoding Clover, a green fluorescent protein variant, mediated by

CRISPR/Cas9-249

mediated HDR within the first exon of Lamin-A gene (Figure S6A) (Pinder et al., 2015). Cells were

250

transfected with a mCherry-expressing vector to gate on transfected cells (mCherry+) in combination with

251

the Clover-donor vector with or without the CRISPR/Cas9 vector targeting the Lamin-A locus (Pinder et

252

al., 2015). This approach induced a significant increase in Clover positive cells in control's cells

253

transfected with both the donor and the CRISPR/Cas9 vectors (P <0.0001), asserting for efficient HDR in

254

these cells (Figure 3E). In contrast, in the same experimental conditions, we did not detect a significant

255

increase in Clover expression in P1's cells, suggesting that HDR was impaired in these cells (Figure 3E).

256

Since the HDR pathway is mainly active during the S and G2 phases, when a DNA template generated by

(12)

DNA replication is available (Jasin and Rothstein, 2013), we analyzed the cell cycle profile of P1's cells.

258

As determined by the combined detection of BrdU incorporation and propidium iodide staining, the cell

259

cycle profiles in control and P1's cells were similar, ruling out that the defective HDR in P1's cells was

260

caused by abnormal cell cycle (Figure S6B).

261

We concluded from these experiments that the DNA repair defect observed in P1's cells was

262

associated with a reduced resection of IR-induced DSBs and impaired HDR efficiency.

263 264

RAD50E1035Δ mutation prevents nascent DNA degradation after replication stress.

265

In the absence of genotoxic stress, spontaneous 53BP1 focus formation is an indicator of

266

replicative stress (Pasero and Vindigni, 2017). Strikingly, we noticed an increase in 53BP1 foci in P1's

267

primary fibroblasts in the absence of any exogenous stress (Figures 4A). Automated detection and

268

quantification of 53BP1 foci reported a mean of 7 events/nucleus in P1's cells (n=9,894) and 2.7 in

269

control (n= 19,632; P <0.0001) (Figure 4B), with 63.9% of P1's cells exhibiting 5 or more 53BP1 foci

270

versus 16.3% in control (P <0.0001; Figure 4C). We then performed replication-combing assay (RCA), a

271

technique that enables the measurement of fork speed, fork asymmetry, and inter-origin distance (Bianco

272

et al., 2012; Schurra and Bensimon, 2009) (Figure 4D). SV40-transformed fibroblasts from P1

273

consistently exhibited defective DNA replication inferred by a significant reduction of fork speed (P

274

<0.0001) (Figure 4E). We also noticed a shortening of inter-origin distance in P1's cells (P <0.01)

275

(Figure 4F), suggesting an increased dormant origin firing to compensate the reduced fork speed, as

276

previously described (Anglana et al., 2003; Courbet et al., 2008; Mokrani-Benhelli et al., 2013). P1's cells

277

also showed an increase in asymmetric replicons (P <0.001) likely attributable to stochastic replication

278

fork stalling and/or collapse (Figure 4G). Transduction of P1's cells with a lentiviral vector allowing the

279

expression of wtRAD50 restored a fork speed comparable to the control's cells, demonstrating a causal

280

link between the replication defects and RAD50 deficiency in P1's cells (Figure 4H).

281

Fork restart depends on fork resection initiated by the nuclease activity of MRE11 that degrades

282

newly synthesized DNA strands (Bryant et al., 2009; Pasero and Vindigni, 2017; Trenz et al., 2006). To

(13)

test whether replicative stress observed in P1's cells could be accompanied by a defect in DNA resection

284

at arrested forks, after successive pulses of CldU and IdU, we treated cells with a high dose of HU for 3

285

hours to provoke replicative stress, and examined DNA resection by molecular combing (Coquel et al.,

286

2019) (Figure 4I). As expected, HU treatment led to nascent DNA degradation in control cells, as

287

revealed by ratio IdU/CldU <1 (Figures 4I-J). Contrastingly, IdU track length did not change after HU

288

treatment in P1's cells (Figure 4I-J) suggesting that RAD50E1035Δ mutation led to an impaired DNA

289

resection at arrested forks. Transduction of wtRAD50 in P1's cells complemented this defect (Figure

290

4K), further supporting the notion that RAD50E1035Δ impaired DNA resection at arrested forks.

291

Collectively, the sharp reduction of DNA end resection at IR-induced DSBs (Figures 3B, 3C,3D,

292

and S5) and at arrested forks (Figures 4J) observed in P1's cells suggested that the RAD50E1035Δ mutation

293

impairs MRE11 nuclease activity.

294 295

Impairment of nuclease activity of recombinant MR(N) complex containing RAD50E1035Δ

296

As complete loss of MRE11 nuclease activity is incompatible with cellular viability (Buis et al.,

297

2008; Hoa et al., 2015; Hoa et al., 2016), we surmised that RAD50E1035Δ could not fully abolish this

298

process. To directly evaluate the impact of RAD50E1035Δ on MRE11 nuclease activity we expressed and

299

purified human MRE11 and RAD50 (MR) as a complex from baculovirus-infected Spodoptera

300

frugiperda 9 (Sf9) cells as previously described (Anand et al., 2016) (Figure S7A). For simplicity we

301

denominated MR WT the wild-type MRE11-RAD50 complex and MR E1035 the MRE11-RAD50

302

complex containing RAD50E1035Δ. As depicted in Figure 5A, MR WT and MR E1035 were similarly

303

purified, reinforcing our previous results showing that the RAD50E1035Δ mutation did not affect protein

304

stability (Figures 1D, 2J, and S4A). Electrophoretic mobility shift assay (EMSA) using a 70bp probe

305

revealed a slight decrease in DNA binding capacity of the MR E1035 complex as compared to its WT

306

counterpart (Figures 5B, 5C). In vitro assay with 50bp unprotected DNA substrates revealed a consistent

307

decreased exonuclease activity of MR complexes with MR E1035 as compared to MR WT (Figures 5D,

308

5E). In particular, the lower products corresponding to the most resected DNA substrates were sharply

(14)

underrepresented with MR E1035 asserting for a reduced exonuclease activity (Figure 5D, 5F). CtIP is

310

a cofactor that potentiates the DNA binding activity of the MR complex especially in its

non-311

phosphorylated form (Anand et al., 2016). We produced recombinant CtIP treated or not with lambda

312

phosphatase to generate either non-phosphorylated CtIP (CtIP) or phosphorylated CtIP (pCtIP),

313

respectively (Figure S7B). As previously described (Anand et al., 2016), the adding of the

non-314

phosphorylated recombinant CtIP led to a strong increase in the DNA binding capacity of the MR

315

complex that, in this condition, was similar with MR E1035 and MR WT (Figures S7C and S7D). This

316

result indicated that in the presence of CtIP, the RAD50 E1035 mutation did not impair DNA binding

317

capacity of the MR complex. However, in the presence of pCtIP, we observed a slight reduction in DNA

318

binding capacity of the mutated MR E1035 complex (Figures S7C and S7D). Next, we added

MBP-319

NBS1 and pCtIP to the complex in order to assess the endonuclease activity of MRE11 as described

320

previously (Anand et al., 2019). This analysis revealed a subtle reduction of endonuclease activity of the

321

MR E1035 complex, evident only at higher concentration (Figures 5G, 5H). Salt titration experiments

322

performed to examine whether the effect of the mutant may be more apparent after more restrictive

323

conditions confirmed the slight decreased endonuclease activity of the MR E1035 complex and further

324

showed that high salt concentration did not worsen this defect (Figures S7E and S7F). The detection of

325

MRE11 endonuclease activity, which depends on ATP hydrolysis (Anand et al., 2016; Cannavo et al.,

326

2019; Deshpande et al., 2017; Hopfner et al., 2000; Liu et al., 2016; Paull and Gellert, 1999), suggested

327

that recombinant RAD50E1035Δ did not abolish ATP hydrolysis, at least in vitro. Accordingly, in vitro

328

ATPase activity was similar with MR E1035 and MR WT (Figure 5I, J).

329

In summary, our experiments using recombinant proteins indicated that the RAD50E1035Δ

-330

containing MRN complex exhibits defects in MRE11 nuclease activity in vitro, a result congruent with

331

the conclusions drawn from the observations obtained in RAD50E1035Δ-expressing P1's cells.

332

333

Modeling the human RAD50E1035Δ mutation in Saccharomyces cerevisiae

334

We next modeled the human RAD50E1035Δmutation in S. cerevisiae by producing a yeast strain deleted

(15)

of the corresponding residue, i.e. the glutamic acid at position 1042 (rad50-E1042) (Figure 6A). In

336

addition, we generated a strain lacking a glutamic acid located nearby, at position 1044 (rad50-E1044;

337

Figure 6A) and also created two other yeast strains carrying missense mutations instead of deletion at

338

position 1042 where the glutamic acid residue was substituted by either an alanine (rad50-E1042A) or a

339

lysine (rad50-E1042K). Western blot and co-immunoprecipitation experiments showed a normal

340

expression and interaction with Mre11 of the deletion and missense mutants at both 30°C and 37°C,

341

indicating that these mutations did not impair Mre11 complex formation and stability (Figure 6B).

342

However, unlike wt strain, rad50-E1042 and rad50-E1044 deletion mutants exhibited high sensitivity

343

to camptothecin (CPT) (Figure 6C), methyl methane sulfonate (MMS) (Figure 6D), and hydroxyurea

344

(HU) (Figure 6E) at 37°C (but not at 30°C), indicating a strong temperature-sensitive effect of these

345

mutations on DNA repair. In striking contrast, both rad50-E1042A and rad50-E1042K missense mutants

346

showed WT levels of survival in the presence of CPT, MMS, and HU at both 30°C and 37°C (Figures

347

6C, 6D, 6E). These observations provided evidence that the glutamic acid residue at position 1042 (or

348

1044) in the coiled-coil domain of Rad50 is not critical for DNA repair. Rather, the presence of any

349

residue at this position is required, presumably to preserve the phasing of the local heptad repetitive

350

disposition of the coiled coil region. Consistent with this result, Marcoil analysis predicted that Rad50

351

deletion mutations caused a heptad break in the coiled-coil domain while this structure was not affected in

352

missense mutants (Figure S8A). Thus, this observation supported the hypothesis that the temperature

353

dependent DNA repair defect observed in rad50-E1042 and rad50-E1044 strains resulted from an

354

abnormal coiled-coil conformation.

355

In S. cerevisiae telomere maintenance relies on a functional Tel1ATM pathway. Both rad50-E1042

356

and rad50-E1044 deletion mutants, but not the rad50-E1042A and rad50-E1042K missense mutants,

357

exhibited shorter telomeres at 37°C (and to a lesser extent at 30°C) than in WT strain, suggesting

358

impaired Tel1 activation in deletion mutants (Figure 6F and Figure S8B). We previously showed that

359

MMS sensitivity of mec1ATRΔ cells is strongly rescued by sae2CtIPΔ, as mec1Δ sae2Δ cells hyperactivate

(16)

Tel1 kinase signaling and compensate for a lack of Mec1 kinase (Park et al., 2017; Usui et al., 2001).

361

Thus, to further examine Tel1 function, we evaluated the phosphorylation of Rad53CHK2 and thesurvival

362

upon MMS treatment of Rad50 mutants deficient for both Mec1 and Sae2. Rad50-E1042Δ mec1Δ sae2Δ

363

and rad50-E1044Δ mec1Δ sae2Δ mutant strains exhibited defective survival (Figure 6G) and Rad53

364

phosphorylation (Figure S8C) upon MMS exposure at 37°C, confirming impaired Tel1 signaling at this

365

temperature. In contrast, rad50-E1042A mec1Δ sae2Δ and rad50-E1042K mec1Δ sae2Δ mutants behaved

366

as WT at both 30°C and 37°C (Figure 6G and Figure S8C).

367

Taken together, the experiments conducted in yeast demonstrated that the loss of a unique residue

368

breaking the heptad repeats in the coiled-coil domain of Rad50 did not impact Rad50 expression or

369

Mre11 complex formation but severely affected DNA repair and Tel1 checkpoint signaling in a

370 temperature-dependent manner. 371 372

Discussion

373 374

In this study, we identified biallelic mutations in the RAD50-encoding gene in an individual (P1)

375

exhibiting bone marrow failure, immunodeficiency, microcephaly, and developmental defects.

376

Consistently with the function of RAD50 as a part of the MRE11-RAD50-NBS1 complex, most of the

377

P1's clinical characteristics were reminiscent, although apparently less severe, than those found in the

378

autosomal-recessive diseases Nijmegen breakage syndrome (NBS) and ataxia-telangiectasia-like disorder

379

(ATLD), respectively caused by biallelic mutations in the genes encoding NBS1 and MRE11 (O'Driscoll,

380

2012; Stracker and Petrini, 2011). The symptoms observed in these disorders as well as in P1 likely

381

originate from the genome instability and DNA damage accumulation that particularly affect proliferating

382

tissues and lead to apoptotic attrition in those contexts (Taylor et al., 2019). Of note, P1 also developed a

383

cataract at 10 years of age that necessitated surgical intervention. To our knowledge, cataracts have not

384

been reported in patients with NBS1 and MRE11 deficiencies. However, a mouse model has implicated

385

Nbs1 in terminal differentiation of the lens fiber cells and cataractogenesis (Yang et al., 2006). Thus, the

386

early onset cataract in P1 suggests that RAD50 might also participate to the development and

(17)

maintenance of the lens in humans.

388

To the best of our knowledge, RAD50 hypomorphism has only been reported twice in two

389

individuals presenting with microcephaly, mental retardation, and short stature but no immunodeficiency

390

(Taylor et al., 2019; Waltes et al., 2009). These patients carried biallelic RAD50 mutations that strongly

391

reduced the expression of RAD50 (Ragamin et al., 2020; Waltes et al., 2009). Waltes et al. and Ragamin

392

et al. demonstrated that the highly pronounced reduction of functional MRN in the cells from these

393

patients caused DNA repair defects, genome instability, and impaired ATM-dependent DDR similar to

394

that observed in cells from patients with ATM, MRE11, and NBS1 mutations (Carney et al., 1998; Gatei

395

et al., 2011; Shiloh and Lederman, 2017; Stewart et al., 1999; Taylor et al., 2019; Varon et al., 1998;

396

Waltes et al., 2009). In sharp contrast, the present study showed that the cells from patient P1 exhibited

397

severe defects in DNA replication, DNA repair, and DNA end resection while the ATM-dependent DDR

398

remained intact (Figure S9A). Since one of the P1's RAD50 mutations generated a null allele, we

399

attributed this unprecedented phenotype to the RAD50E1035Δ mutation producing a normally expressed and

400

stable RAD50 protein that lacks a unique glutamic acid causing a break in the heptad repeats within the

401

coiled-coil domain. Knowing that the complete loss of Rad50 is lethal in mice (Adelman et al., 2009) and

402

having observed that P1's cells were proficient in ATM-dependent DDR, we deduced that

403

RAD50E1035Δmutation is hypomorphic.

404

Remarkably, P1's cells expressing RAD50E1035Δ mirrored the phenotype observed in mouse

405

embryonic fibroblasts (MEFs) expressing the Mre11 nuclease dead mutant (Mre11H129N) (Buis et al.,

406

2008). Indeed, in both Mre11H129N MEFs and RAD50E1035Δ P1's cells, MRN stability and recruitment to

407

DSB as well as ATM-dependent DDR were functional (Buis et al., 2008). Contrastingly, both Mre11H129N

408

MEFs and RAD50E1035Δ P1's cells exhibited increased sensitivity to genotoxics, defective IR-induced

409

RPA and RAD51 focus formation, and reduced HDR efficiency (Buis et al., 2008; Myler et al., 2017).

410

This resemblance is consistent with the interpretation that the RAD50E1035Δ mutation impacted the

411

MRE11 nuclease activity in vivo. This hypothesis was further supported by in vitro assays with purified

412

recombinant proteins that showed impaired MRE11 exonuclease activity when in complex with

(18)

RAD50E1035Δ while endonuclease activity was only slightly affected. However, since animal model

414

expressing the Mre11 nuclease dead mutant at a homozygous status (Mre11H129N/H129N) and human

415

MRE11-/H129N lymphoblast cell lines are unviable (Buis et al., 2008; Hoa et al., 2015; Hoa et al., 2016), we

416

surmise that RAD50E1035Δdoes not totally abolish MRE11 nuclease activity in vivo and/or that an

417

alternative mechanism allows to cope with defective MRE11-dependent DNA end resection in P1.

418

Hence, RAD50E1035Δ mutation represents the first human RAD50 separation-of-function mutation

419

impairing DNA repair, DNA replication, and DNA end resection without affecting MRN recruitment to

420

DSBs, and ATM-dependent DDR (Figure S9A). It is unlikely that RAD50E1035Δ affected RAD50-ATP

421

binding and hydrolysis since (i) this step is required for ATM-dependent DDR (Cannavo and Cejka,

422

2014; Deshpande et al., 2014; Lee et al., 2013), (ii) we observed an in vitro ATP hydrolysis with

423

recombinant RAD50E1035Δ (Figure 5), (iii) in vitro MRE11 endonuclease activity, which depends on ATP

424

hydrolysis (Anand et al., 2016; Cannavo et al., 2019; Deshpande et al., 2017; Hopfner et al., 2000; Liu et

425

al., 2016; Paull and Gellert, 1999), was detected with recombinant RAD50E1035Δ (Figure 5). Our

426

observation that recombinant MRE11-RAD50E1035Δ complex exhibited a reduced DNA binding and an

427

impaired exonuclease activity in vitro (Figure 5) combined with the reduction of IR-induced RPA and

428

RAD51 foci and HDR in P1's cells (Figure 3) rather supports the notion that RAD50E1035Δ impairs the

429

MRN complex nuclease activity required to promote DNA end resection.

430

These results led us to propose a unified speculative model in which the human RAD50E1035Δ

431

mutation creates (or loses) a structural constraint in the coiled-coil domain that propagates to the globular

432

domain to hinder the proper conformational transition of the MRN intracomplex from the ring to the rod

433

shape (Figure S9B). This abnormally structured MRN intracomplex would retain its ATP binding and

434

hydrolysis, as well as most of its capacity to stimulate MRE11 endonuclease activity. However, the

435

mutant would corrupt subsequent conformational change required for MRE11-dependent exonuclease

436

activity and DNA end resection (Shibata et al., 2014). Since it has been suggested that the MRE11

437

intracomplex in its rod shaped conformation clamps DNA (Kashammer et al., 2019), the observation of a

438

reduced capacity of recombinant MRE11-RAD50E1035Δ proteins to interact with DNA in vitro further

(19)

supports this model. One cannot however rule out that other functionally important coiled-coil

440

interactions both within the anti-parallel coiled-coil and between the dimeric coiled-coil assembly could

441

be affected by RAD50E1035Δ. Alternatively, we cannot exclude that RAD50E1035Δ mutation could

442

compromise MRN to interact and/or activate other factors such as EXO1, DNA2, and EXD2 required for

443

efficient DNA end resection in vivo (Broderick et al., 2016; Cejka et al., 2010; Delamarre et al., 2020;

444

Myler et al., 2017; Pasero and Vindigni, 2017; Paull, 2018; Stracker and Petrini, 2011; Syed and Tainer,

445

2018; Zhu et al., 2008).

446

Notably, our analysis of rad50-E1042∆ yeast strain modeling the human RAD50E1035Δ mutation

447

pinpointed a phenotype strikingly different from human cells. Indeed, E1042∆ strain (and

rad50-448

E1044∆) behaved as wt at 30°C while it was almost as severe as rad50∆ strain for both DNA repair and

449

Tel1ATM activation at 37°C. To our knowledge this extremely severe thermosensitive phenotype in

rad50-450

E1042∆ (and rad50E1044∆) is unprecedented in Rad50 mutants. Furthermore, the observation that yeast

451

strains carrying missense mutations at position E1042 (rad50-E1042A and rad50-E1042K) behaved as wt

452

at both 30°C and 37°C proved that this is the deletion of a single amino-acid at this position that is

453

deleterious and not the nature of the residue. From these yeast experiments we propose a model in which

454

the rad50E1042∆ (and rad50E1044∆) mutations, by breaking the heptad repeats, increase the flexibility

455

of the coiled-coil at 37°C but not at 30°C. This temperature dependent change would ultimately

456

destabilize either the coiled-coil interface-dependent dimerization, the conformation of the globular

457

domain, the Zn-hook-mediated dimerization, or a combination of these modifications that would result in

458

the extremely severe phenotype at 37°C. Interestingly, the phenotype of rad50-E1042∆ strain at 37°C was

459

similar to the one of yeast mutant lacking the Zn-hook in which MR intra- and intercomplex formations

460

are abolished (Hopfner et al., 2002; Stracker and Petrini, 2011). Moreover, it has been recently reported

461

that the Zn-hook-proximal coiled-coil participated in the stabilization of intracomplex Mre11 complex

462

assembly (Park et al., 2017). One can therefore hypothesize that the temperature sensitivity of the

rad50-463

E1042∆ mutant results from defective MRN intracomplex assembly at 37°C (Hohl et al., 2011; Wiltzius

464

et al., 2005). Interestingly, other yeast strains carrying mutations affecting the Rad50 Zn-hook and/or

(20)

coiled-coil domains exhibited thermosensitive phenotypes (Hohl et al., 2015; Hohl et al., 2011). However,

466

the molecular cause of the temperature dependent phenotype in these Rad50 mutants remains unclear and

467

future studies are warranted to get further into the structural and functional consequences of these

468

mutants.

469

In summary, our study provided evidence that a single amino-acid deletion that breaks the heptad

470

repeats in the coiled-coil domain of the yeast and human RAD50 compromised MRE11 functions.

471

Furthermore, our demonstration that human RAD50E1035Δ is a separation-of-function mutation that

472

impairs DNA repair, DNA replication, and DNA end resection without affecting ATM-dependent DDR

473

supports the idea that the integrity of the RAD50 coiled-coil domain is essential to enable switch of the

474

MRE11-RAD50 intracomplex toward functional rod shaped conformation promoting DNA clamping and

475

DNA end resection (Hopfner, 2014; Kashammer et al., 2019; Park et al., 2017; Rojowska et al., 2014).

476

Further studies using cryo-EM, atomic force, and high-throughput single-molecule microscopy are

477

warranted to better characterized the structural hindrance caused by RAD50E1035Δ and decipher the rules

478

delineating the communication between the hook, the coiled-coil, and the globular domains of the

479

MRE11-RAD50 complex in mammals. Lastly, the development of a Rad50E1035Δ mouse model should

480

provide crucial information on the functional consequence of compromised Mre11-dependent DNA end

481 resection in vivo. 482 483 484 485 486 487 488 489 490 491

(21)

Acknowledgments

492 493

General: We thank the patient and his family for their generous assistance with samples and information,

494

which made this research possible. We acknowledge M. Garfa Traore and the imaging facility of Imagine

495

Institute for help with microlaser irradiation, the Bioinformatics Department of Imagine Institute for help

496

with Whole exome sequencing, N. Lambert (CEDI, AP-HP, Necker Hospital, Paris, France) for

497

microsatellite analysis, A. Fernandes from the Centre de ressource biologique, Imagine Institute, who

498

generated B-LCL. We thank Dr B. Lopez for RAD50 construct used as matrix and technical advice for

499

RAD51 and RPA immunofluorescence study. We thank Dr G. Dellaire for the HDR assay components

500

(Pinder et al., 2015). Dr P. Pasero is acknowledged for his advice with replication combing assays. P.R.

501

thanks Dr A. Decottignies, Dr P. Kannouche, and Dr E. Brunet for critical reading of the manuscript. P.R.

502

is a scientist from Centre National de la Recherche Scientifique (CNRS).

503

504

Funding: Work performed in the GDIS lab was supported by institutional grants from INSERM, Ligue

505

Nationale contre le Cancer (Equipe Labellisée "LIGUE 2020"), INCa (PLBIO19-027; INCA_13766),

506

GIS-Institut des maladies rares, and state funding from the Agence Nationale de la Recherche under

507

“Investissements d’avenir” program (ANR-10-IAHU-01). This work was supported by GM56888 and

508

MSK Cancer Center Core Grant P30 CA008748 (J.H.J.P.) and Swiss National Science Foundation

509

(31003A_17544) and European Research Council grant (681-630) to P.C. M.M. is supported by the

510

French National League Against Cancer (EL2028.LNCC/MaM), the French National Cancer Institute

511

(PLBIO2017-167). M.C.D-C. beneficiated from scholarships from Association Nationale Recherche

512

Technologie (ANRT) and La Ligue contre le Cancer. P.R. is a scientist from Centre National de la

513 Recherche Scientifique (CNRS). 514 515 516 517 Author contributions 518

M.C.D-C. carried out most of the experimental work on human cells assisted by P.R. and L.K. T.I.

519

identified the affected patient and assisted with related clinical and laboratory studies. I.C. and P.C.

520

produced recombinant MRN complex and performed in vitro experiments. M.H and J.H.J.P conceived

521

and performed experiments on yeast strains. P.R. conceived the project and wrote the manuscript with

522

editing contributions from J-P.V., I.C., M.H., P.C., and J.H.J.P.

(22)

Main figure titles and legends

524

Figure 1. Biallelic RAD50 mutations identified in individual P1. (A) (upper panel) Pedigree of the P1's

525

family. (lower panel) Genetic approach used to identify the disease-causing gene. (B) Sequencing of

526

cloned RAD50 PCR products in patient. (C) Schematic of the domain architecture of human RAD50

527

showing the position of disease-associated mutations. NBD: nucleotide binding domain. (D) (left)

528

Immunoblot of indicated proteins in B-LCL lysates from a healthy donor, P1 and his mother. Ku70 was

529

used as loading control. (right) Relative abundance of indicated proteins normalized to Ku70. (E) (left)

530

Schematic view of a RAD50 protein. (right) Schematic view of MRN complex in its intracomplex

531

conformation13. (F) Marcoil analysis revealed a break in the heptad repeats nearby the E1035Δ mutation

532

(blue arrow). (G) (Up) Representative view of the coiled-coil probability along the full WT and

533

RAD50E1035Δ sequences determined by Marcoil tool. The red area corresponds to the probable coiled-coil

534

regions. (Down). Zoom of the region encompassing the E1035Δ mutation. Blue arrow indicates a break

535

in the coiled-coil probability.

536 537

Figure 2. Defective DNA repair but normal DNA damage sensing and signaling in P1's cells. (A)

538

Survival of SV40-fibroblasts after IR (A), Phleomycin (B), Etoposide (C) and MMC (D) treatment.

539

Results are expressed as the fraction of surviving cells in relation to untreated cells. Each point represents

540

the mean value and standard deviation of three separate determinations. ATM and Cernunnos deficient

541

cells were used as sensitive control. Control sensitive fibroblasts were from ATM, Cernunnos and

542

Fanconi deficient patients. (E) Sensitivity of the P1's cells to phleomycin after wtRAD50 or empty vector

543

transduction. This experiment was performed three times. Control sensitive fibroblasts were from a

544

Cernunnos deficient patient. (F) Functional complementation of phleomycin sensitivity provided by

545

wtRAD50 transduced into the patient’s cells as compared with the empty vector. The selective growth

546

advantage is scored as the increase in the index of mCherry-positive cells/mCherry-negative cells at

547

various times compared with the initiation of the culture (index = 1). This experiment was performed

548

twice. (G) Recruitment of GFP-NBS1 to DNA damage upon laser microirradiation in indicated cells.

549

GFP-NLS (nuclear localization signal) is used as negative control. Irradiated zones are located in the red

550

circles. Representative of three independent experiments. (H) Kinetics of recruitment of GFP-NBS1 and

551

GFP-NLS to DNA damage in control and patient's cells. (I) Immunofluorescence detection of

552

phosphorylated KAP1 (P-KAP1) in primary fibroblasts before and 1 hour after 0.5 Gy irradiation. ATM-

553

and NBS1- deficient cells are used as negative controls. Image representative of three independent

554

experiments. (J) Phosphorylated forms of CHK2 and NBS1 detected by Western blot analysis of whole

555

cell lysates from SV40-fibroblasts from a healthy control, an ATM-deficient patient and patient P1

556

untreated or 1 hour after 5 or 10 Gy irradiation. Vinculin immunoblot is used as loading control. Asterisk

557

indicates a non specific band. (K) (Left) The DNA content and phosphorylation of histone H3 in G2 (pink

558

rectangle) were analyzed by FACS. (Middle) FACS images of P-H3 in G2 in untreated and irradiated

559

cells. (Right) G2/M checkpoint was measured by the % inhibition of entry into mitosis in cells after IR.

560

Results represents the mean and SD from at least three independent experiments. Statistical significances

561

are noted.

562 563

Figure 3. P1's cells exhibit reduced DNA end resection and homology directed repair.

564

Immunofluorescence detection of phosphorylated KAP1 (A), RPA32 (B), and RAD51 (C) in primary

565

fibroblasts from healthy control and P1 before and 6 hours after 10 Gy irradiation. Image representative

566

of three independent experiments. (D) Quantitative analysis of RAD51 foci before and after IR in

(23)

hTERT fibroblasts untransduced or transduced with empty or wtRAD50- expressing vector. The number

568

of nuclei scored and statistical significances are noted. (E) Efficiency of homology directed repair

569

assessed by the Cas9-directed knock-in of Clover in the LMNA coding sequence. Cells were transfected

570

with the mCherry vector alone (used to control transfection efficiency), or in combination with the repair

571

template sequence containing the Clover coding sequence in conjunction or not with the pX330-LMNA1

572

vector encoding Cas9 and the gRNAs targeting the LMNA sequence (noted CRISPR/Cas9). Relative

573

HDR efficiency was measured by the percentage of Clover+cells triple transfected relative to cells

574

transfected without the CRISPR/Cas9-gRNA expressing vector. Results represents the mean and SD of

575

triplicates from three independent experiments.

576 577

Figure 4. DNA replication and resection at stalled forks in P1's cells. (A) Detection of 53BP1 foci in

578

primary fibroblasts from patient and a healthy control at similar passage. (B) Mean of 53BP1 foci in

579

primary fibroblasts from healthy control and patient. The number of nuclei scored and statistical

580

significances are noted. (C) The percentages of fibroblasts from a control and patient P1 with the

581

indicated number of 53BP1 foci are indicated. Statistical significances are noted. (D) (Up) Representative

582

image of combed DNA (yoyo labeling). (Down) Picture representing the detection of newly synthesized

583

DNA by molecular combing (successive pulse labelings of IdU and CldU). DNA is detected by

584

counterstaining (blue). (E) Fork velocity in SV40-fibroblasts from 3 independent experiments in control

585

and P1 are represented. (F) Inter-origin distances in SV40-fibroblasts from control and P1are represented.

586

The statistical significances are noted. (G) Left. Asymmetric replicon with sister forks with more than a

587

25% length difference in SV40-fibroblasts from control and P1. Plots of the distances covered by

right-588

moving and left-moving sister forks during the IdU and CldU pulses from two independent experiments.

589

(Right) Percentage of replicons with more of 25% asymmetry in SV40-transformed fibroblasts from

590

control and P1. The statistical significance is noted. (H) Fork velocity of SV40-transformed fibroblasts

591

from control and P1 before or after transduction with a wtRAD50 expressing vector or empty vector.

592

Results representative of two independent experiments. The statistical significances are noted

(Mann-593

Whitney statistical test). (I) (Up) Protocol used to measure DNA resection after replicative stress

594

induction by HU. (Down) Representative tracks of CldU (red) and IdU (green) incorporation in control

595

and patient's cells with or without HU treatment. DNA is counterstained in blue. (J) DNA resection at

596

stalled forks is determined by the ratio between CldU and IdU track lengths after 3 hours of HU treatment.

597

Results representative of two independent experiments. (K) DNA resection at stalled forks after 3 hours

598

of HU in control and P1 cells before or after transduction with an empty or wtRAD50 expressing vector.

599

Results representative of two independent experiments. The statistical significances are noted

(Mann-600

Whitney statistical test).

601 602 603

Figure 5. Impairment of nuclease activity of recombinant MRE11 and RAD50E1035Δ proteins. (A)

604

Recombinant MRE11-RAD50 (MR) with WT RAD50 (MR WT) or mutated RAD50E1035Δ (MR

605

E1035Δ) used in this study. (B) Representative DNA-binding analysis by electrophoretic mobility shift

606

assay with recombinant MR WT and MR 1035Δ, using 70-bp-long dsDNA as a substrate. (C)

607

Quantitation of experiments shown in (B). Error bars, SEM; n=3. (D) Exonuclease activity of MR WT or

608

MR E1035Δ using a 50-bp-long dsDNA substrate. (E) Plots of signals shown in (D) reveal a reduction of

609

the most resected DNA substrate (higher migration) with MR E1035Δ as compared to MR WT. (F)

(24)

Quantitation of experiments shown in (C). Error bars, SEM; n=3. (G) Representative nuclease assay with

611

MR WT or MR E1035Δ in presence of MBP-NBS1 and pCtIP on 5'-end-labeled 70-bp dsDNA with all

612

ends blocked with streptavidin. (H) Quantitation of experiments shown in (G). Error bars, SEM; n=3. (I)

613

ATP hydrolysis activity of MR WT or MR E1035Δ with or without MBP-NBS1. The MRN complexes

614

containing the RAD50 K40A (noted MR(KA)N) and RAD50 K40R (noted MR(KR)N) mutants, defective

615

in ATP binding and hydrolysis (Cannavo et al., 2019; Chen et al., 2005), were used as negative controls.

616

(J) Quantitation of experiments shown in (I). n=2.

617 618 619

Figure 6. Modeling the human RAD50E1035Δ mutation in Saccharomyces cerevisiae. (A) Schematic

620

representation of the Rad50 structure and multiple sequence alignment of C-terminal coil domain of the

621

Rad50 carrying the human E1035Δ mutation. The human E1035 residue and corresponding amino acid in

622

Saccharomyces cerevisiae and Mus musculus are highlighted in green. (Sc, S. cerevisiae; Mm, M.

623

musculus; Hs, H. sapiens). (B) Coimmunoprecipitation and western blot with Rad50 or Mre11 antisera

624

assessed the Mre11 complex integrity in wild-type (WT) and indicated Rad50 mutants. IP,

625

immunoprecipitation.  (C, D, E) Sensitivities of WT and indicated Rad50 mutant to the indicated

626

concentration of CPT (C), MMS (D), and HU (E). Plates were incubated at 30°C and 37°C as indicated.

627

(F) Telomere lengths of wild-type (WT) and rad50 mutants after 4 days of growth at 30°C and 37°C.

628

Heterozygote diploids were included as 0 generation of growth. (G) Cell survival of rad50 mutants in

629

Mec1- and Mec1- Sae2- deficient backgrounds upon MMS treatment at 30°C and 37°C.

630 631 632 633 634 635 636 637 638 639 640 641 642 643 644

(25)

STAR METHODS 645 646 RESOURCE AVAILABILITY 647 Lead Contact 648

Further information and requests for resources and reagents should be directed to and will be fulfilled by

649

the Lead Contact, Patrick Revy ([email protected]).

650 651

Material availability

652

All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials 653

Transfer Agreement. 654

655

Data and Code Availability

656

All used software is listed in the Key Resources Table. This study did not generate any unique datasets or new code. 657

658

EXPERIMENTAL MODEL AND SUBJECT DETAILS

659

Human fibroblasts used in this study were obtained from skin biopsies from pediatric healthy donors (3

660

years of age) and patients. When indicated transformed (with large T antigen from SV40T) and

661

immortalized (after transduction with a hTERT-expressing vector, Addgene #85140) fibroblasts were used.

662

All cell lines were from male individuals and were checked for mycoplasma contamination. Informed and

663

written consent was obtained from donors and patients. The study and protocols comply with the 1975

664

Declaration of Helsinki as well as with the local legislation and ethical guidelines from the Comité de

665

Protection des Personnes de l’Ile de France II and the French advisory committee on data processing in

666

medical research.

667

Saccharomyces cerevisiae strains used in this paper are listed in Table S2. Yeast strains were grown in

668

liquid cultures and on plates in YPD media supplemented with 100 mg/L adenine. Further specifications

669

are mentioned within the Methods Details section.

670 671 672 METHOD DETAILS 673 674 Study approval 675

Informed and written consent was obtained from donors and patients. The study and protocols comply with

676

the 1975 Declaration of Helsinki as well as with the local legislation and ethical guidelines from the

677

Comité de Protection des Personnes de l’Ile de France II and the French advisory committee on data

678

processing in medical research.

679 680

Cells

681

Control fibroblasts were obtained from skin biopsies from pediatric healthy donors (3 years of age). When

682

indicated transformed (with large T antigen from SV40T as previously described (Buck et al, 2006)) and

683

immortalized (after transduction with a hTERT-expressing vector, Addgene #85140) fibroblasts were used.

684

All cell lines were checked for mycoplasma contamination.

685 686

Yeast

687

Yeast strains were generated by integration of the rad50-coiled coil mutants at their native chromosomal

688

locus in a diploid WT strain. rad50 mutant spores were obtained by tetrad dissection and verified by PCR

689

genotyping and sequencing using genomic DNA. Details of yeast strains and plasmid constructions are

Références

Documents relatifs