Novel strand exchange activity of the human PALB2 DNA Binding Domain and its critical role for DNA repair in cells

Breast cancer associated proteins 1 and 2 (BRCA1, −2) and partner and localizer of BRCA2 (PALB2) protein are tumor suppressors linked to a spectrum of malignancies, including breast cancer and Fanconi anemia. They stimulate RAD51 recombinase during homology-directed repair (HDR). Along with being a hub for a protein interaction network, PALB2 interacts with DNA. The mechanism of PALB2 DNA binding and its function are poorly understood. We identified a major DNA-binding site in PALB2, mutation of which reduces the RAD51 foci formation and the overall HDR efficiency in cells by 50%. PALB2 N-terminal DNA-binding domain (N-DBD) stimulates the RAD51 strand exchange reaction. Surprisingly, it promotes the strand exchange without RAD51. Moreover, N-DBD stimulates the inverse strand exchange and can use both DNA and RNA substrates. Our data reveal a versatile DNA interaction property of PALB2 and demonstrate a critical role of PALB2 DNA binding for chromosome repair in cells.


INTRODUCTION 34
Breast cancer associated proteins 1 and 2 (BRCA1, -2) regulate an efficient non-mutagenic 35 pathway of chromosome break repair and are described as guardians of chromosomal integrity 36 (Venkitaraman, 2014). They initiate RAD51-mediated homologous recombination (HR) ( localizer of BRCA2 (PALB2) protein was discovered as a protein forming a complex with BRCA2 and 40 regulating BRCA2 activity (Xia et al., 2006). Similarly, to BRCA proteins, PALB2 is an essential inhibiting an alternative pathway of non-homologous end joining and initiating homology-directed 46 repair (HDR) through interactions with PALB2/BRCA2/RAD51 (Prakash et al., 2015). 47 BRCA1/BARD1 complex interaction with DNA and RAD51 led to the discovery of the BRCA1/BARD1 68 role in RAD51-mediated strand invasion and D-loop formation (Zhao et al., 2017). Two DBDs were 69 previously identified in PALB2 (Buisson et al., 2010;Dray et al., 2010). The functional role of these 70 domains remains unknown. PALB2 construct lacking 500 amino acids between the BRCA1 and 71 BRCA2 binding motifs does not support BRCA2 and RAD51 foci formation in cells during DNA 72 damage (Sy et al., 2009b). Since both the BRCA1-binding N-terminal and the BRCA2-interacting 73 WD40 C-terminal domains were retained in this mutant, the results points to the potential importance 74 of DBDs in PALB2 function. 75 In the current study, we identified a major DBD of PALB2 (N-DBD) and specific amino acids 76 involved in DNA binding. Mutations of four amino acids significantly reduce RAD51 foci formation and 77 the efficiency of HDR in a model cell system. Surprisingly, we found that N-DBD supports both forward 78 and inverse strand exchange even in the absence of RAD51 and can use RNA as a substrate. 79 Altogether, our data reveal a novel activity of PALB2 and highlight the importance of PALB2 DNA 80 binding in chromosome maintenance in cells. 81

RESULTS 82
The DNA-binding mechanism of PALB2 and its function in DNA repair. 83

The major DNA-binding site of PALB2 is localized in the N-terminal domain (N-DBD). Two truncation 84
fragments of PALB2 were previously reported to interact with DNA, T1 (residues 1-200) and T3 85 (residues 372-561) (Buisson et al., 2010). Both fragments together with the fragment consisting of 86 amino acids 1-573 (PB2-573 in text), which includes both the T1 and T3, were cloned and purified 87 Quantitative measurement of PALB2 interaction with ss-and dsDNA oligonucleotides of different 91 lengths demonstrate that T1 fragment alone interacts with all tested substrates with almost 92 indistinguishable affinity from that of PB2-573 (Fig. 1). The T3 fragment has significantly lower affinity 93 for DNA by itself. The Kd of the T1 and PB2-573 fragments were similar with both ss-and dsDNA 94 substrates. The only difference was observed at an elevated salt concentration of 250 mM NaCl, 95 where the PB2-573 fragment retained partial DNA binding activity (Fig. S2). In both cases, interactions 96 were inhibited by addition of 500 mM NaCl. The T1 fragment will be referred as N-DBD in the text 97 below. Interestingly, N-DBD binds long ssDNA substrates (49 nt) with significantly higher affinity than 98 short ones (20 nt). This suggests an interaction with ssDNA through multiple binding sites, potentially 99 formed by the previously described PALB2 oligomerization (Buisson and Masson, 2012;Sy et al., 100 2009c) or through interaction with multiple binding sites within a monomer (see below). Interaction 101 with dsDNA was length-independent, suggesting that more rigid dsDNA interacts with a single site.  Identification of DNA-binding residues. Since the PALB2 DNA binding is salt dependent, we 116 performed alanine scanning mutagenesis of several clusters of positively charged amino acids to 117 identify the DNA binding site in the N-DBD (Fig. S3). 118 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint Figure S3. Amino  The main DNA-binding cluster is formed by amino-acids R146, R147, K148, and K149. Alanine 127 mutation of these residues reduced binding affinity by two orders of magnitude with a Kd change from  in the full length PALB2 protein and the effect of these mutations was measured in two assays. First, 148 we evaluated RAD51 foci formation in cells after gamma irradiation (Fig. 3A). Endogenous PALB2 149 was depleted by siRNA and cells were transformed with either wild type PALB2 or the DNA-binding 150 mutant (Fig. 3A, bottom panel). PALB2 depletion leads to a severe defect in RAD51 foci formation. 151 WT PALB2 restores RAD51 foci formation, while the DNA-binding PALB2 mutant restores only ~ 50% 152 of RAD51 foci formation. Therefore, mutagenesis of only four positively charged residues in PALB2 153 has a major effect on efficiency of RAD51 recruitment to DNA damage sites. 154 155 156

Figure 3. Effect of a PALB2 DNA-binding mutation on homologous recombination. A) 157
Representative immunofluorescence images of RAD51 foci in PALB2 knockdown HeLa cells 158 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. knockdown and complementation efficiency. ***P<0.01 and ****P<0.001. (Fig S5B). 166 Similarly, we tested the role of PALB2 interaction with DNA for the efficiency of HDR in U2OS 167 cells using a novel LMNA-Clover based assay, where DNA breaks at a specific gene are introduced 168 by the CRISPR system (  Surprisingly, the N-DBD promotes strand exchange at a comparable rate even without RAD51. 206 Reaction products were further analysed by EMSA gel shift to rule out any artefact of protein-specific 207 fluorophore quenching (Fig. 4C). The results were confirmed using DNA with different fluorescent 208 labels (Fig. S7). The strand exchange activity of N-DBD was even more efficient with longer dsDNA 209 substrate (Fig. S8). 210 Figure S7. EMSA of N-DBD-mediated strand exchange products using Cy3-and Cy5-labeled ds35 211

DNA. 212
. CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint E). Furthermore, PALB2 supported both reactions with a ssRNA substrate (Figs. 7C, F). DNA-binding 255 mutant fragment (146AAAA) did not support strand exchange on its own and in the presence of 256 RAD51 (Fig. S10). RAD52 was shown to have different efficiency of forward and inverse reactions 257 with relatively low forward and a more efficient inverse reactions (Mazina et al., 2017). We did not 258 observe this difference with PALB2. The inverse strand exchange was slower than in case of RAD52 259 and comparable to that of RAD51 under optimal conditions. 260  (Fig. S13). Therefore, the simple ability of a protein to interact with ss-and dsDNA 295 is not enough to promote strand exchange and even RMPs, which stimulate the reaction by RecA 296 recombinase, do not support it in the absence of recombinase. 297 In this report, we identify major DNA-binding residues of PALB2 and demonstrate their critical 304 role for the HDR in cells. PALB2 is described as a scaffold protein linking BRCA1 with BRCA2 during 305 HDR and interacting with many other chromatin proteins. However, the mutant with BRCA1 and 306 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint BRCA2 binding motifs without the middle portion of the proteins does not support BRCA2 and RAD51 307 recruitment to DSBs (Sy et al., 2009b). A critical role of PALB2 DNA binding was also suggested by 308 studies of the BRCA2 mechanism , where the "miniBRCA2" construct, which 309 includes only DBDs with two BRC repeats, was 3-4 times less efficient in the absence of PALB2 310 interaction. Moreover, interaction with PALB2 alleviates the requirement of BRCA2 DNA binding, 311 including a deletion of the entire BRCA2 DBD. Here, we demonstrate that mutation of only four DNA-312 binding residues of PALB2 reduces both RAD51 foci and overall HDR efficiency by 50%, even in the 313 presence of endogenous BRCA2. Therefore, PALB2 interaction with DNA is critical for recruitment of 314 the BRCA2 and RAD51 to DSB sites and efficient DNA repair in cells. interaction with dsDNA is less length dependent. We can speculate that binding of dsDNA to more 347 than one monomer in PALB2 oligomer can trigger DNA helix distortion. Indeed, bending of dsDNA 348 was observed upon PALB2 binding to 40 bp dsDNA labelled with a Cy3/Cy5 FRET pair as the 349 increase of FRET signal (Fig. S15). Thus, PALB2 shares several specific structural and DNA not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint 2) Alternatively, human RAD51 gene was cloned into pSMT3 vector using SalI and NdeI cloning sites. 411 pSMT3-Rad51 protein expression was carried out at 16°C overnight by addition of 0.2 mM IPTG. 412 SUMO tagged Rad51 protein was purified according to the steps described for the PALB2 fragments. 413 Purified Rad51 protein was dialysed against storage buffer (25 mM HEPES pH 8.0, 300 M NaCl, 40% 414 glycerol, 1 mM TCEP and 2 mM CHAPS) overnight, aliquoted and stored in -80°C until further use. 415 Proteins from both preparations had comparable properties. Data are shown for experiments 416 performed with the second construct, except the experiment represented in Fig. S10. 417 E. coli RecA was purified exactly as described in (Gupta et al., 2013). E.coli RecO and RecR proteins 418 were purified as described in (Ryzhikov and Korolev, 2012;Ryzhikov et al., 2011). 419 Site-directed mutagenesis: Target amino acids were mutated by site directed mutagenesis of using 420 Stratagene QuikChange TM protocol. Single, double, triple and four residues mutants were generated 421 by single stranded synthesFis (Table S1). PCR samples were subjected to DpnI digestion at 37°C for 422 6 h and annealed gradually by reducing temperature from 95°C to 37°C for an hour with a degree 1C  (Table S2) (Table  443 S3) and a 90mer ssDNA (ss90) with homologous region to plus strand. Alternatively, FAM/Dabsyl 49 444 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint bp DNA was used. For the forward reaction, ss90 (100 nM) was incubated with 2 M (or as mentioned 445 in the figure legends) protein for 10 min in 40 L reaction buffer (40 mM HEPES pH 7.5, 20 mM NaCl, 446 5 mM MgCl2, 1 mM TCEP and 0.02 % Tween 20) at 37°C. Strand exchange was initiated by addition 447 of 100 nM 35bp dsDNA (ds35), plate was immediately placed in plate reader and the intensity of Cy5 448 fluorescence was measured at 30 sec intervals for 1 hour with excitation at 635 nm and emission at 449 680 nm. For reactions with RecA and Rad51, an ATP regeneration system (2 mM ATP, 30 mM 450 phosphoenol pyruvate and 30 U of pyruvate kinase) was used (Sigma-Aldrich, USA). For the inverse 451 reaction, protein was incubated with Cy5/IOWA-dsDNA35 substrate and reaction was initiated by 452 addition of ss90. The strand exchange assay with an ssRNA substrate was performed as described 453 above using a 60 ribonucleotide RNA (table S1) complimentary to that of 35bp DNA. Alternatively, 454 Cy3-and Cy5-labelled DNA oligonucleotides were used to prepare dsDNA substrate and the products 455 were analysed by EMSA PAGE (below). 456 EMSA PAGE. Fluorescent-labelled DNA products of strand exchange reactions were also analysed 457 on EMSA PAGE. After fluorescence measurement on plate reader, the final reaction mix (80 l) 458 products were deproteinated by incubation with proteinase K (0.5 mg/ml) with 0.5 mM EDTA and 1% 459 w/v SDS for 20 min at 37°C and the DNA fragments were separated on 10% PAGE gel in TBE buffer. 460 The gel was imaged using a Typhoon 9400 image scanner (GE) and analysed with ImageJ software. 461 FRET assay: FRET assay was performed in 96 well plate format. 100 nM of dual labelled dT70 (Cy5 462 at 5' end and Cy3 at 3' end) was dispensed into 80 L assay buffer identical to those in strand 463 exchange assay (Table 3). Alternatively, dual labelled 40 bp DNA was prepared by annealing dual 464 labelled 40 nt ssDNA (Cy3 and Cy5 on single strand) with unlabelled complimentary 40 nt ssDNA 465 (Table 3). Excitation was at 540/25nm bandpass. Emission for both Cy3 at 590/35nm bandpass and 466 Cy5 at 680/30 nm were monitored at 30 s intervals for 5 min at 37°C, then for 10 min following the 467 addition of PALB2 N-DBD, then for 10 min following the addition of RAD51 with or without ATP. FRET 468 efficiency was calculated by using the formula = 5  (  5+  3) . For each addition, the plate was 469 removed from the plate reader and returned to the reader within 60 s. Protein concentrations were as 470 described in the figure legends. 471 RAD51 foci assay: HeLa cells were seeded on glass coverslips in 6-well plates at 225 000 cells per 472 well. Knockdown of PALB2 was performed 18 hours later with 50 nM PALB2 siRNA (Table S4) using 473 Lipofectamine RNAiMAX (Invitrogen). After 5 hours, cells were subjected to double thymidine block. 474 Briefly, cells were treated with 2 mM thymidine for 18 hours and released after changing the media. representing SEM, and a classical one-way Anova test was performed. 507 Plasmids and siRNA 508 peYFP-C1-PALB2 was modified to be resistant to PALB2 siRNA by Q5 Site-Directed Mutagenesis Kit 509 (NEB, E0554) using primers JYM3892/3893 (Table S4). The resulting siRNA-resistant construct was 510 then used as a template to generate the mutant construct YFP-PALB2 146AAAA with the primers 511 JYM3909/JYM3910. Flag-tagged PALB2 146AAAA mutant was also obtained via site-directed 512 mutagenesis on pcDNA3-Flag PALB2 (Pauty et al., 2017). 513 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint Table S1. Primers for DNA binding site mutagenesis.

Mutation
Sequence RK146AA F  ACT GCC GAG CGC TCG TGC AAA ACA ACA AAA GC   RK146AA R  GCT TTT GTT GTT TTG CAC GAG CGC TCG GCA GT   RK147AA F  ACT GCC GAG CCG TGC TAA AGC ACA ACA AAA GC   RK147AA R  GCT TTT GTT GTG CTT TAG CAC GGC TCG GCA GT   KKK90AAA F  CGA AAA AAT TGC ACA TAG CAT TGC AGC AAC GGT GGA AG   KKK90AAA R  CTT CCA CCG TTG CTG CAA TGC TAT GTG CAA TTT TTT CG   K90A F  CGA AAA AAT TGC ACA TAG CAT TAA  not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint Table S2. Substrates for DNA binding assay.   not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted January 8, 2019. ; https://doi.org/10.1101/495192 doi: bioRxiv preprint