Mechanistic insight into cadmium-induced inactivation of the Bloom protein

Cadmium is a toxic metal that inactivates DNA-repair proteins via multiple mechanisms, including zinc substitution. In this study, we investigated the effect of Cd2+ on the Bloom protein (BLM), a DNA-repair helicase carrying a zinc-binding domain (ZBD) and playing a critical role to ensure genomic stability. One characteristics of BLM-deficient cells is the elevated rate of sister chromatid exchanges, a phenomenon that is also induced by Cd2+. Here, we show that Cd2+ strongly inhibits both ATPase and helicase activities of BLM. Cd2+ primarily prevents BLM-DNA interaction via its binding to sulfhydryl groups of solvent-exposed cysteine residues and, concomitantly, promotes the formation of large BLM multimers/aggregates. In contrast to previously described Cd2+ effects on other zinc-containing DNA-repair proteins, the ZBD appears to play a minor role in the Cd2+-mediated inhibition. While the Cd2+-dependent formation of inactive multimers and the defect of DNA-binding were fully reversible upon addition of EDTA, the inhibition of the DNA unwinding activity was not counteracted by EDTA, indicating another mechanism of inhibition by Cd2+ relative to the targeting of a catalytic residue. Altogether, our results provide new clues for understanding the mechanism behind the ZBD-independent inactivation of BLM by Cd2+ leading to accumulation of DNA double-strand breaks.

Bloom's syndrome (BS) is a rare, autosomal and recessive disease resulting from the mutational inactivation of a human RecQ family helicase encoded by the blm gene 1 . BS is characterized by proportional dwarfism, erythema on sun-exposed skin, hyper-or hypo-pigmented skin areas, immunodeficiency and subfertility 2 . Persons with BS have a high predisposition to cancer and increased risk for early-onset type-II diabetes 3 . The blm gene encodes BLM, a 1417-amino acids protein containing several conserved motifs including a zinc-binding domain (ZBD). Previous works have shown that mutation of any of the four conserved Cys residues of the ZBD leads to the BS 4,5 . Moreover, we have previously shown that the ZBD of RecQ helicases plays a key role in protein folding and is involved in DNA-binding 6 . Thus, alteration of the zinc coordination state and potentially metal-catalyzed oxidation could impair BLM-mediated DNA-repair processing events. In addition to numerous cytological characteristics including high rates of loss of heterozygosity [7][8][9] , chromosome abnormalities (telomere fusions, ring chromosomes and quadriradial chromosomes 10 ), the most striking feature of BLM-deficient cells or cells bearing an impaired BLM mutant is characterized by elevated rates of sister chromatid exchanges (SCEs) 11 . Interestingly, it was shown that Cadmium (Cd) also provoked elevated rates of SCEs in human cell cultures 12 . Thus, the effect of Cd 2+ on human cell lines shares cytological characters with BLM-deficient cells, establishing a connection between BLM and Cd 2+ . Cd 2+ is considered as an important health hazard due to its long retention time and bioaccumulation in human body 13 . Epidemiological and animal experiments have revealed multifactorial carcinogenic properties of cadmium 14 . Exposure to Cd 2+ is associated with cancers of lung, prostate, pancreas and kidney 15 . Among the various carcinogenic effects of Cd 2+ , DNA damage accumulation due to inhibition of DNA-repair enzymes is considered as one of the major underlying process 16,17 . Unlike numerous toxic metal compounds, Cd 2+ is considered as weakly mutagenic. Nevertheless, Cd 2+ is known to severely increase the genotoxic effects of various mutagens in mammalian cells, including ionizing radiations and DNA alkylating agents used at low non-cytotoxic concentrations 18,19 . Many studies using yeast or human cells suggest that DNA-repair systems represent highly sensitive targets for Cd 2+ . However, the precise mechanism behind carcinogenicity remains to be determined. Although

Results
The full-length BLM helicase (BLM full-length ) is composed of 1417 amino acid and contains the helicase core (BLM 642-1290 ) harboring two RecA domains and forming ATP-binding sites, the ZBD (pos. 994-1068) and the winged-helix domain (pos. 1069-1189), whereas the N-and C-terminal domains are more related to protein-protein interactions and nuclear localization 27,28 . BLM 642-1290 , used throughout this study together with BLM full-length and E. coli RecQ (RecQ E.coli ), displays full activities comparable to BLM full-length 27 .

Cadmium severely impairs both helicase and ATPase activities of BLM.
To determine the effect of Cd 2+ on BLM activities, we first measured the helicase activities of BLM full-length and BLM 642−1290 in the presence of increasing concentrations of CdCl 2 . DNA unwinding activities of both proteins were strongly inhibited by Cd 2+ as measured by a radioactive assay: 95% and 100% of inhibition at 5 and 10 μ M Cd 2+ , respectively (Fig. 1a). Interestingly, the RecQ E.coli unwinding activity was actually inhibited by Cd 2+ but to a much lesser extent compared to BLM proteins. Indeed, the same inhibition level (>95%) of RecQ E.coli was obtained at 100 μ M CdCl 2 . These results, suggesting that BLM represents a more sensitive target for Cd 2+ , were confirmed by stopped-flow FRET experiments allowing measurements of the unwinding kinetic rate constant of BLM by using partial duplex DNA labeled with fluorescein and hexachlorofluorescein as a donor and acceptor, respectively 29,30 (Supplementary Fig. S1; see Supplementary Table S1 for details about the structure of the DNA substrate): the levels of Cd 2+ -dependent inhibition of BLM full-length and BLM 642−1290 helicase activities were similar and significantly higher than that observed with RecQ E.coli (≈10-fold higher). Thus, Cd 2+ displays a selective profile for BLM and most likely targets its helicase core since no significant difference in the Cd 2+ -dependent inhibitions of BLM full-length and BLM 642-1290 was observed. We next measured the ATPase activity of the three proteins in the presence of increasing CdCl 2 concentrations. Again, Cd 2+ strongly inhibited the ATPase activities of both BLM full-length and BLM 642−1290 , while the ATPase activity of RecQ E.coli was much less inhibited (Fig. 1b). The corresponding IC 50 values (inhibition concentration 50%) determined for BLM full-length and BLM 642−1290 were 6.7 and 7.3 μ M, respectively, whereas the IC 50 was much higher (65 μ M) for RecQ E.coli (Table 1). This result closely parallels, at least qualitatively, the one obtained for the inhibition of DNA unwinding activity as explained above.
Stoichiometry of Cadmium binding to BLM. Before addressing the mechanism of BLM inhibition, we assessed whether BLM is a direct target of Cd 2+ . We then investigated the Cd 2+ :BLM 642−1290 stoichiometry using a fluorescence-based assay (see Methods). The Cd 2+ :RecQ E.coli stoichiometry was studied in parallel for comparison. We found a stoichiometry of 11-12 Cd 2+ per BLM 642−1290 molecule (Fig. 2a, left) and a significantly lower stoichiometry was found for RecQ E.coli : 6-7 Cd 2+ per protein (Fig. 2a, right). Interestingly, each representation displayed two slopes, reflecting two types of sites characterized by distinct accessibilities and affinities.
Previous studies have reported that free sulfhydryl groups of Cys are good candidates for Cd 2+ binding 25 . BLM full-length and BLM 642-1290 contain 30 and 19 Cys, respectively (11 in RecQ E.coli ). To determine whether Cd 2+ actually targets Cys of BLM 642-1290 and RecQ E.coli , stoichiometry experiments were repeated in the presence of N-Ethylmaleimide (NEM), a thiol-alkylating agent that forms stable covalent thioether bonds with sulfhydryls of reduced Cys. NEM treatment significantly altered Cd 2+ :protein stoichiometries with two features: the Cd 2+ :protein stoichiometry was decreased from 11-12 to 5 and from 6-7 to 3-4, for BLM 642-1290 and RecQ E.coli , respectively (Fig. 2b). Second, in contrast to experiments performed without NEM, the stoichiometry curves displayed one slope, corresponding to the low-affinity binding site cluster for both proteins. Altogether, the results show that these helicases display at least two types of Cys clusters. Assuming that (i) Cd 2+ possesses a higher affinity for Cys localized at the surface compared to residues localized in the protein core, (ii) NEM only interacts with surface residues for steric hindrance reason, we conclude that the 1 st Cys cluster (surface) is characterized by high affinity for Cd 2+ without NEM but does not anymore interact with Cd 2+ in the presence of NEM while the 2 nd cluster (protein core) is characterized by weaker affinity for Cd 2+ due to lower accessibility and is not influenced by NEM. We estimated that the 1 st and 2 nd clusters are composed by 6-7 and 5 Cys, respectively, in BLM 642-1290 (3 and 3-4, respectively, in RecQ E.coli ). The remaining Cys that do not interact with Cd 2+ could be related to residues totally buried in the protein structure. The possible implication of the BLM zinc finger motif is addressed in the following section together with the issue of the Cadmium effect on the proper BLM/DNA interaction.
Cd 2+ binding to BLM primarily impairs its DNA-binding activity. To gain further insight into the mechanism of Cd 2+ -induced inhibitions of DNA unwinding and ATPase activities, we examined DNA-binding activities of BLM full-length and BLM 642-1290 in the presence of increasing CdCl 2 concentrations by using a steady-state fluorescence anisotropy assay 31 . We first determined DNA-binding isotherms curves (i) in the absence of Cd 2+ to define the experimental conditions in terms of protein concentrations (i.e. high enough above the K d to ensure DNA saturation) and (ii) in the presence of Cd 2+ and absence of any protein to determine the CdCl 2 concentration range in which the fluorescence anisotropy of the fluorescein-labeled DNA was not influenced by direct  Cd 2+ -DNA interactions (data not shown). A protein concentration of 200 nM and CdCl 2 concentrations up to 100 μ M were found to correspond to optimal conditions and then, were used in subsequent experiments. BLM full-length and BLM 642−1290 displayed similar inhibition profiles of the DNA-binding activities (Fig. 3). The Cd 2+ -dependent inhibitions of the protein-DNA interaction were efficient, with IC 50 values in the 6.6-10.2 μ M range (Table 1). These values were consistent with those derived from the inhibition of helicase and ATPase activities. Moreover, the inhibition profiles were similar, regardless of the nature of the DNA substrate, single-(ss) or double-stranded (ds) DNA (Fig. 3a) or the DNA length, from 18-to 40-mer (Fig. 3b). However, the RecQ E.coli DNA-binding activity was weakly affected by Cd 2+ (IC 50 = 39-44 μ M), as observed for its ATPase or helicase activity, confirming that BLM is more sensitive to Cd 2+ than RecQ E.coli .
The binding of BLM 642-1290 to DNA was next studied in the presence of Cd 2+ and various amino-acids such as Cys, His and Val. Among the different amino-acids, Cys is the residue forming by far the most stable complex 32 . Accordingly, when amino-acids were pre-incubated with both Cd 2+ /DNA before addition of the protein, the BLM 642-1290 DNA-binding activity was fully restored by Cys but only partially restored by His or Val ( Supplementary Fig. S2). The counteracting effect of Cys was also observed on the ATPase activity of BLM 642-1290 (data not shown). By contrast, the addition of amino-acids after pre-incubation of BLM 642-1290 with Cd 2+ /DNA did not restore the DNA-binding activity, regardless of the nature of the amino-acid ( Supplementary Fig. S2). The absence of protective/competition effect by Cys in the latter case can be ascribed to irreversible binding of Cd 2+ to BLM sulfydryl groups and EDTA only was able to reverse the Cd 2+ -mediated inhibition of DNA-binding (see below).
Taking into consideration that BLM contains a zinc finger that is composed of four conserved Cys 5 , we next investigated whether the BLM ZBD could be a target site for Cd 2+ . Mutation of any one of these Cys leads to the BS and inactivates BLM activity both in vitro and in vivo 6 . First, we wondered whether Zn 2+ and Cd 2+ would be able to bind to the same site and whether the replacement of Zn 2+ by Cd 2+ would lead to a conformational change of the BLM ZBD and, consequently, impairs BLM activity. To further investigate the mechanism of Cd 2+ -mediated inhibition and the interplay between Cd 2+ and Zn 2+ , the influences of both metals on the DNA-binding step were compared using the fluorescence anisotropy-based assay. Interestingly, Cd 2+ but also Zn 2+ inhibited the DNA-binding step of BLM full-length , BLM 642-1290 and RecQ E.coli , however to different extents ( Supplementary Fig. S3). Up to 50 μ M, Cd 2+ alone was consistently more potent than Zn 2+ alone for inhibition. Beyond 50 μ M, both metals inhibited BLM full-length and BLM 642-1290 in a similar manner. To note, no inhibition was observed for BLM 642-1290 at low Zn 2+ concentrations (<15 μ M). In this concentration range, the inhibitory effect of Cd 2+ was not reversed by addition of Zn 2+ . The Cd 2+ /Zn 2+ combination was even more efficient for inhibiting DNA-binding activity than Cd 2+ alone, a typical synergistic inhibition phenomenon. Such a behavior Protein-Cd 2+ complexes were then discarded using Q-sepharose beads. The Cd 2+ :protein stoichiometries were deduced from the determination of free Cd 2+ remaining in solution, using the fluorescence-based Measure-iT Cadmium assay as described in Methods.
was also observed with BLM full-length , except that low Zn 2+ concentrations were more efficient for inhibiting the DNA-binding step of BLM full-length compared with BLM 642-1290 . Regarding RecQ E.coli DNA-binding activity which was only partially inhibited by Zn 2+ alone (85% of activity at 100 μ M), the more potent effect of Cd 2+ was not reversed by addition of Zn 2+ . Altogether, these results indicate that, under our experimental conditions, Cd 2+ and Zn 2+ do not bind competitively to the same site. It is important to note that, especially with Cys 4 ZBD (corresponding to BLM or RecQ E.coli ZBDs), Cd 2+ forms much more stable complexes than zinc 33 . Taking into account that a large excess of Zn 2+ over Cd 2+ could be required for efficient competition (a condition which is not compatible with the proper Zn 2+ -dependent inhibition described above), we cannot definitively confirm or ruled out a targeting of the BLM ZBD by Cd 2+ . However, it is unlikely that the ZBD alone accounts for the Cd 2+ -dependent inhibition since BLM and RecQ E.coli are characterized by different susceptibilities to Cd 2+ ; this differential susceptibility appears to be more related to the larger number of targeted Cys in the case of BLM, as suggested by stoichiometry experiments. The relationship between the defect of DNA-binding and the binding of Cd 2+ to surface Cys residues was further investigated below by studying the Cd 2+ effect on the multimeric status of BLM.
Cd 2+ induces BLM oligomerization. During the course of our experiments, we observed that high Cd 2+ concentrations (typically > 30 μ M) induced BLM precipitation in samples. Size-exclusion chromatography analysis showed that while BLM 642-1290 alone was eluted as a monomer as previously reported 34 , its elution profiles when pretreated with Cd 2+ displayed three peaks, corresponding to molecular weights of 75, 150 and >205 kDa, respectively (Fig. 4a), indicating that Cd 2+ induced protein oligomerization. We further studied the oligomeric status of BLM in the absence and presence of Cd 2+ by Dynamic Light Scattering (DLS) (Fig. 4b). DLS analysis confirmed size-exclusion chromatography experiments. Indeed, without Cd 2+ , BLM 642-1290 was monodisperse in solution and characterized by a radius of 3.44 nm, compatible with a monomeric form (peak I). The addition of Cd 2+ dramatically modified the profile of size distribution. The peak I disappeared and was replaced by two peaks (II+ II') corresponding to large BLM multimers, probably non-specific cross-linked aggregates (with MW of 3.2 × 10 5 and 1.3 × 10 9 kDa, respectively). Addition of EDTA on Cd 2+ -treated samples of BLM 642-1290 dissociated higher-order oligomers leading to a recovery of the monomeric form as shown by size-exclusion chromatography (Fig. 4a) or DLS ( Fig. 4b; peak III, radius 3.95 nm). Altogether, these results show that Cd 2+ induces the formation of higher-order BLM oligomers that is reversible upon addition of EDTA.  Dual effect of DTT on the modulation of the BLM DNA-binding activity by Cd 2+ and reversibility with EDTA. DTT has been previously described either as a stimulating or a counteracting agent for Cd 2+ -mediated inhibition 35,36 . All the above-mentioned results were obtained under reducing conditions, i.e. in the presence of DTT. We then compared the Cd 2+ effect on the BLM 642-1290 -DNA interaction under reducing and non-reducing conditions (Fig. 5). In the presence of DTT, Cd 2+ efficiently inhibited the binding of BLM 642-1290 to DNA (Fig. 5a), in accordance with results shown in Fig. 3. To note, NEM that did not lead to significant effect on its own on the BLM 642-1290 DNA-binding activity, fully counteracted the inhibitory effect of Cd 2+ , suggesting that Cys of the 1 st cluster (surface residues as defined above) mainly mediate the Cd 2+ -dependent inhibition.
By contrast, under non-reducing conditions, Cd 2+ did not significantly inhibit the BLM 642-1290 DNA-binding activity (Fig. 5b), suggesting that Cd 2+ -targeted Cys are most likely involved in intra-or inter-molecular disulfide bonds and react with Cd 2+ upon reduction only (in the form of sulfhydryl groups). The effect of Cd 2+ on the BLM 642-1290 DNA-binding activity was further studied under different reducing conditions by varying the DTT concentration. At 0.2 mM DTT (Fig. 5c, right), this effect was comparable to the one previously observed at 2 mM DTT (Fig. 5a), i.e. DTT promoted the Cd 2+ -mediated inhibition of protein-DNA interaction, while no Cd 2+ effect was observed in the absence of DTT (Fig. 5c, left). Interestingly, under limited reducing conditions (0.02 mM DTT; Fig. 5c, middle), the addition of Cd 2+ led to an increase in the steady-state fluorescence anisotropy, instead of a decrease. This dual effect of DTT remains unclear. We can hypothesize that Cd 2+ differentially modulates self-assembly properties of BLM 642-1290 depending on the DTT concentration. As shown above, no free sulfhydryl of surface Cys is available for Cd 2+ binding in the absence of DTT and the presence of intra/inter-molecular disulfide bonds is compatible with DNA-binding. Under moderate DTT conditions (0.02 mM DTT), a part of Cys residues only could be in their reduced forms and Cd 2+ could promote the formation of higher-order BLM 642-1290 multimers of moderate sizes which remain compatible with DNA-binding but leading to higher steady-state anisotropy values due to the size of protein-DNA complexes (larger compared with the size of complexes obtained in the absence of DTT). By contrast, under strong reducing conditions (>0.2 mM DTT), most of surface Cys residues should be in the reduced state and thus, Cd 2+ promotes large multimers/aggregates, not competent for DNA-binding. In other words, the DNA-binding site should be completely hidden in the context of these large BLM multimers/aggregates. Note that both Cd 2+ effects (inhibition or stimulation) were abolished by addition of EDTA (Fig. 5c) in a dose-dependent manner (Supplementary Fig. S4). Consistent with our hypothesis, EDTA was shown to reverse Cd 2+ -induced multimers by size-exclusion chromatography and DLS (Fig. 4).
Cd 2+ -induced inhibition of BLM DNA unwinding activity was not reversed by addition of EDTA. We next addressed the question of whether EDTA could restore the unwinding activity of BLM as measured in the presence of Cd 2+ using the stopped-flow FRET assay. First, Cd 2+ significantly reduced both the unwinding kinetic rate constant and the corresponding reaction amplitude characterizing BLM 642-1290 (Fig. 6a), in accordance with results shown in Fig. 1. Second, we tested the effect of EDTA on the unwinding activity of BLM 642-1290 per se since the Mg 2+ cofactor is required for this activity. As shown in Fig. 6b, although the kinetic rate constant was lower in the presence of EDTA, the reaction amplitude was only slightly affected. This result shows that BLM 642-1290 sustains a significant DNA unwinding activity even in the presence of EDTA. Nevertheless, the Cd 2+ -dependent inhibition of BLM activity was not counteracted by EDTA (Fig. 6b), in contrast to that previously observed for the DNA-binding step. The fact that the helicase activity of BLM 642-1290 cannot be recovered   upon addition of EDTA, suggests that Cd 2+ also inactivates (at least) one additional Cys or another residue, most likely playing a proper catalytic role for DNA unwinding, in an irreversible manner.

Discussion
The BLM helicase plays key roles in numerous cellular processes including DNA double-strand break repair (DSB), Holliday junction resolution and chromosome segregation [37][38][39] . In this context, the study of the toxic effect of Cd on BLM is of particular interest for understanding the pivotal role of BLM in keeping genome stability. Among Cd 2+ , Zn 2+ and Hg 2+ , only Cd 2+ has been reported to induce SCEs, a unique cytological feature of BLM-deficient cells 40 . Furthermore, recent studies highlight that Cd 2+ targets major players of the DNA-repair machinery including proteins involved in the BER, NER or MMR pathways [21][22][23] , although little is known about Cd 2+ effects on proteins involved in the DSB pathway such as BLM. Here, we investigated the interplay between BLM and Cd 2+ at the molecular level. We found two distinct molecular mechanisms accounting for the Cd 2+ -mediated inactivation of BLM. Cd 2+ targets surface Cys in their reduced state and promotes the formation of large BLM multimers and then inhibition of the BLM-DNA interaction. The Cd 2+ -dependent multimerization and DNA-binding inhibition processes were found to be fully reversible upon addition of EDTA. However, the inhibition of the BLM helicase activity was irreversible suggesting another mechanism at the catalytic level, mediated by the targeting of a catalytic residue.
We found that Cd 2+ was able to efficiently inhibit both helicase and ATPase activities of BLM full-length and BLM 642-1290 in the low micromolar concentration range. The corresponding IC 50 values were compatible with values derived from DNA-binding inhibition curves suggesting that the Cd 2+ -dependent inhibition of helicase and ATPase activities could be explained in part by inhibition of the DNA-binding step. The reversibility of the DNA-binding inhibition process by EDTA indicates that the Cd 2+ -dependent inhibition mechanism is not associated with an irreversible structural modification of the protein that could affect DNA-binding properties. It is important to note that, under experimental conditions where Cd 2+ was maintained below 100 μ M, fluorescence anisotropy experiments did not show any significant direct DNA-Cd 2+ interaction, in accordance with previous studies 41 . Furthermore, we found that Cd 2+ inhibits both BLM and RecQ E.coli , however to different extents, with RecQ E.coli much less susceptible to Cd 2+ (≈one order of magnitude) than BLM. This differential susceptibility reinforces the idea that Cd 2+ dose not directly target DNA in our activity or DNA-binding assays. Instead, this differential susceptibility appears to be related to the number of solvent-exposed Cys contained in the protein structure with RecQ E.coli having much less surface Cys (i.e. protected by NEM) than BLM 642-1290 (Fig. 2).
We also tested whether Cd 2+ could replace Zn 2+ in the ZBD based on several statements: (i) Cd 2+ has been previously reported to react with thiol groups, particularly with Cys and glutathione that act as the first line of defense against Cd 2+ in cells 36,42 . (ii) The antagonistic effect of Zn 2+ on Cd 2+ has long been documented and previous studies have shown that Cd 2+ -targeted sites in proteins correspond to zinc finger motifs and that the replacement of Zn 2+ by Cd 2+ may be reversible 43,44 . However, here, we failed to demonstrate that Zn 2+ prevents the inhibitory effect of Cd 2+ on BLM. Taking into account the higher affinity of Cd 2+ over Zn 2+ for Cys 4 ZBD, large excess of Zn 2+ should be required to observe a protective effect against Cd 2+ ; it was not possible to satisfy this condition since, instead to rescue BLM DNA-binding activity, Zn 2+ on its own displayed a significant inhibitory effect although this effect was less important than Cd 2+ . If Cd 2+ target the helicase ZBD, this is probably not the predominant effect for the following reasons: (i) BLM and RecQ E.coli are differentially affected by Cd 2+ , in accordance with their respective number of solvent-exposed Cys (Fig. 2 and Supplementary Table S2) and (ii) we have previously shown that Zn 2+ -coordination to BLM or RecQ E.coli ZBD is strictly required for correct protein folding (during production) but dispensable after for both DNA-binding and helicase activities 6,34 . The mechanism by which Zn 2+ inhibits BLM remains unknown and it appears that Zn 2+ differentially affects helicases of the RecQ family with Zn 2+ having only a slight inhibition effect on the binding of RecQ E.coli to DNA ( Supplementary  Fig. S3). Consistent with our study on human BLM, it was previously shown that Zn 2+ also significantly impairs the helicase activity of BLM yeast homologue, Sgs1 45 . It has also been shown that Zn 2+ enhances the 3′ → 5′ exonuclease activity of another human RecQ helicase, the Werner protein, at the expense of the helicase activity [46][47][48] .
It was previously shown that the binding of BLM and RecQ E.coli to DNA relies on similar mechanisms. Regarding their binding to dsDNA and ssDNA, both proteins involve distinct protein domains with ssDNA-binding sites located along the A1, A2, WH and HRDC domains, whereas the dsDNA-binding site is located near the ZBD 5,34,49 . Although different between BLM and RecQ E.coli , the IC 50 values characterizing protein-DNA interaction inhibitions were similar between ss and dsDNA, reinforcing the idea that the Cd 2+ -dependent DNA-binding inhibition is not strictly related to the ZBD and probably occurs via a more general and common mechanism, regardless of the nature of DNA (ss or ds). Besides the targeting of ZBD, other mechanisms of Cd 2+ -mediated protein inhibitions involved in DNA-repair have been proposed. It was proposed that Cd 2+ inhibits mismatch repair pathway by abrogating the ATPase activity of the MSH2-MSH6 complex, via a mechanism in which Cd 2+ binds in a non-specific manner, leading to a stoichiometry of more than hundred Cd 2+ per protein 50 . In the case of BLM, the number of target sites appears to be limited. In contrast to MSH2-MSH6 complexes that possess a high content of non-specific Cd 2+ binding sites, we found that BLM 642-1290 is characterized by a much lower stoichiometry (11-12 Cd 2+ per protein) and Cys sulfydryl groups are directly involved in this Cd 2+ -binding process. We have characterized two subclasses of Cys targeted by Cd 2+ , based on (i) the presence of two distinct slopes in stoichiometry plots, most likely accounting for differences in accessibility/affinity and (ii) the NEM protective effect. The first class (6-7 residues) is composed by Cys displaying high affinity for Cd 2+ , most likely located at the protein surface while the second class (5 residues) corresponds to Cys with lower affinity for Cd 2+ and probably buried and located in the protein core.
Our data suggest that the first class of Cys is fully responsible for the Cd 2+ -dependent inhibition of BLM DNA-binding since NEM, which prevents the binding of Cd 2+ to Cys belonging to this class only, fully rescues the DNA-binding activity in the presence of Cd 2+ (Fig. 5). The Cd 2+ -mediated inhibition of DNA-binding was strictly Scientific RepoRts | 6:26225 | DOI: 10.1038/srep26225 dependent on the presence of DTT suggesting the involvement of free sulfydryl groups of surface Cys in the binding of Cd 2+ . Most of the class 1 Cys should be engaged in disulfide bridges under non-reducing conditions, explaining the absence of any inhibitory effect of Cd 2+ in the absence of DTT. In parallel, we found that Cd 2+ promotes BLM higher-order multimers or aggregates starting from monodisperse samples of monomers. The mechanism behind at the molecular level is not yet clearly understood and still under investigation. This phenomenon was fully reversed by EDTA, in accordance with the counteracting effect of EDTA on BLM DNA-binding activity. To note, only reducing conditions (to ensure free sulfydryl groups) were compatible with the observation of Cd 2+ -dependent inhibition, suggesting that DNA-binding sites of BLM protomers should be hidden upon the formation of these large non-specific multimers/aggregates. Interestingly, mild reducing conditions, which modulate the number of free -SH at the protein surface, led to an intermediary result between reducing condition (inhibition of DNA-binding by Cd 2+ as measured by a decrease in the anisotropy value) and non-reducing condition (no effect of Cd 2+ on the anisotropy value) (Fig. 5c): in this mild reducing condition, Cd 2+ led to an increase in the anisotropy value, probably accounting for DNA-binding of organized multimers of moderate sizes. Their oligomeric status should be intermediary between monomers and large non-specific multimers/aggregates, with remaining solvent accessible DNA-binding sites. The different number of solvent-exposed Cys between BLM 642-1290 and RecQ E.coli , 6-7 and 3, respectively, highlights the relationship between the number of Cys potentially targeted by Cd 2+ and the extent of the Cd 2+ -dependent DNA-binding inhibition. To note, these numbers of solvent-exposed Cys as determined experimentally agree well with the number of surface Cys based on 3D structures (7 and 3, respectively, without taking into account ZBD Cys; Supplementary Table S2). Nevertheless, the Cd 2+ -dependent inhibition of the BLM unwinding activity was not reversed by EDTA, suggesting an additional catalytic mechanism of inhibition, unrelated to the DNA-binding step. According to the three-dimensional BLM structure 51,52 , we performed single point mutations of Cys residues implicated in DNA binding/unwinding (C895S and C901S). The two BLM mutants displayed modest reduced DNA-unwinding activity and their responses to Cd 2+ were similar to the wild-type BLM (data not shown). This suggests that distinct residues (Cys or other amino-acids) or, alternatively, a combination of the two above-mentioned Cys could be responsible for the irreversible BLM inactivation by Cd 2+ . The underlying mechanism is currently under investigation.
Recombinant proteins. RecQ E.coli and BLM 642-1290 helicases were expressed and purified as previously described 34,53 . BLM full-length was expressed in Saccharomyces cerevisiae JEL-1 strain as previously described 54 with some modifications in the purification protocol. Briefly, cells were thawed at 4 °C and cell disruption was performed using a French press in a Tris-HCl buffer (50 mM, pH 7.5) supplemented with 500 mM KCl, 10% sucrose, 1 mM DTT and protease inhibitors cocktail (Roche). After DNA fragmentation by sonication, the crude extract was subjected to centrifugation at 30,000 g for 45 min using a SS34 rotor (Sorvall). Filtered supernatant (using 0.45 μ m filters) was loaded in 10 mL of Ni 2+ -NTA agarose resin (Qiagen). Beads were washed with 100 mL of K 2 HPO 4 /KH 2 PO 4 buffer (20 mM, pH 7.4), 0.05% Triton X-100, 10% glycerol, 1 mM DTT (= buffer A-20 mM) supplemented with 500 mM KCl and 20 mM imidazole, followed by 100 mL of buffer A-20 mM supplemented with 500 mM KCl and 50 mM imidazole. Proteins were then eluted with 100 mL of buffer A-20 mM supplemented with 500 mM KCl and 300 mM imidazole. Fractions containing BLM were pooled and directly loaded onto a 10-mL Biogel CHT hydroxyapatite column (Bio-Rad). The column was washed with 100 mL of buffer A-20 mM, followed by 100 mL of buffer A-100 mM. Elution was done with 100 mL of buffer A-350 mM supplemented with 50 mM KCl. Fractions containing BLM were pooled and diluted in 25 mL of buffer A-20 mM supplemented with 50 mM KCl and loaded onto 5 mL of Q-sepharose fast Flow (GE Healthcare). Beads were washed with 100 mL of buffer A-20 mM supplemented with 50 mM KCl, followed by 100 mL of buffer A-20 mM supplemented with 230 mM KCl. BLM was eluted using buffer A-20 mM supplemented with 500 mM KCl. Fractions containing BLM were then concentrated before loading onto a S-200 Superdex HR 10/30 gel filtration column (GE Healthcare) equilibrated in buffer A-20 mM supplemented with 500 mM KCl. Fractions containing BLM were concentrated, dialyzed against 50 mM Tris-HCl, pH 7.5, 100 mM KCl, 0.05% Triton X-100, 1 mM DTT, 25% glycerol and stored at − 80 °C. RecQ E.coli , BLM 642-1290 and BLM full-length proteins were > 95% pure as judged by SDS-PAGE analysis and Coomassie staining (Supplementary Fig. S5).
Oligonucleotides. The sequences of DNA substrates used for enzymatic or DNA-binding assays are shown in Supplementary Table S1. PAGE-purified fluorescein-labeled or unlabeled synthetic oligonucleotides were purchased from Eurogentec. Double-stranded DNA substrates were obtained by mixing equimolar amounts of complementary strands in 20 mM Tris-HCl, pH 7.5, 100 mM NaCl. The mixture was heated to 95 °C for 5 min and annealing was allowed by slow cooling to room temperature.
Cadmium binding assay. Cd 2+ binding to helicases was assayed by incubating BLM 642-1290 or RecQ E.coli (0.5 μ M) with increasing concentrations of CdCl 2 (from 0 to 37.5 μ M) in 50 μ l of Tris-HCl buffer (50 mM, pH 8.0) supplemented with 50 mM NaCl and 1 mM DTT, for 5 min at 25 °C. This condition is suitable to measure Cd:protein stoichiometries since both protein and Cd 2+ concentrations were much higher than K d values characterizing Cd·Cys complexes 32 . When required, the concentration of NEM was 0.5 mM. Free Cd 2+ in solution was measured by a fluorescence-based Measure-iT Cadmium assay (Invitrogen) according to the manufacturer's protocol. To avoid any bias in the measurement of free Cd 2+ due to equilibrium displacement (Cd 2+ -protein -> Cd 2+ -sensor), proteins were eliminated from the mixture by using Q-sepharose beads before the measurement. Free Cd 2+ concentrations were deduced from calibration plots using CdCl 2 solutions of known concentrations Scientific RepoRts | 6:26225 | DOI: 10.1038/srep26225 (we checked that interaction between free Cd and beads was negligible: fluorescence intensity values obtained after incubation of CdCl 2 solutions with beads were 95-100% of intensities measured with "input" solutions). The number of protein-bound Cd 2+ was estimated by subtracting the amount of free Cd 2+ to the total amount of Cd 2+ .

Size-exclusion chromatography and dynamic light scattering (DLS) experiments.
The size-exclusion chromatography experiment was performed according to Xu et al. 55 . Briefly, the chromatography was performed at 25 °C, using an FPLC system (GE healthcare), on a Superdex 200 (analytical grade) column equilibrated with buffer S (20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 1 mM DTT and 5% glycerol (v/v)) +/− 2 mM EDTA. 20 μ l of untreated or Cd 2+ -treated BLM (+/− 2 mM EDTA) was loaded on the column (typically in the 2-6 μ M concentration range) and was eluted with buffer S +/− 2 mM EDTA, at a flow rate of 0.4 ml/min; the absorbance was continuously monitored at 280 and 260 nm. The standard molecular markers (Sigma) used for calibration were eluted under identical experimental conditions. DLS measurements were performed using a DynaPro NanoStar instrument (Wyatt Technology, France) equipped with a thermostated cell holder using filtered (0.1 μ m filters) solutions in disposable cuvettes (UVette, Eppendorf). The protein concentration was 1.5 μ M in a Tris-HCl buffer (50 mM, pH 8.0, 200 mM NaCl, 1 mM DTT) (total volume, 50 μ l). The scattered light was collected at an angle of 90°. Recording times were typically between 3-5 min (20-30 cycles in average of 10s each). The analysis was performed with the Dynamics 7.0 software using regularization methods (Wyatt Technology, France). The molecular weight was calculated from the hydrodynamic radius using the following empirical equation 1: where Mw and R H represent the molecular weight (kDa) and the hydrodynamic radius (nm), respectively.
ATPase activity assay. The ATPase activity was assayed by measuring the release of free phosphate during ATP hydrolysis 34  Helicase assay. 1) Radioactive assay: DNA helicase reaction was carried out at 37 °C in a reaction mixture containing 25 mM Tris-HCl, pH 8.0, 50 mM NaCl, 3 mM MgCl 2 , 0.1 μ g/ml BSA, 1 mM DTT, 2 mM ATP. To address the Cd 2+ effect on helicase activity, helicases were preincubated without or with Cd 2+ for 2 min at 37 °C. The unwinding reaction was initiated by addition of 10fmol of the 32 P-labeled partial duplex DNA substrate (3000cpm/fmol) and the reaction mixture was further incubated for 20 min at 37 °C. The reaction was quenched by adding 5 μ l of loading buffer containing 50 mM EDTA, 0.5% SDS, 0.1% xylene cyanol, 0.1% bromophenol blue and 50% glycerol. The reaction products were analyzed by gel electrophoresis using a 12% polyacrylamide gel.
2) Stopped-flow fluorescence measurements: A stopped-flow FRET assay was used for measuring the unwinding kinetic rate constant of BLM, using doubly labeled DNA substrates, with fluorescein and hexachlorofluorescein as a donor and acceptor, respectively 29,30 . The set-up and kinetic data analysis were described in Liu et al. 30 . The standard reaction was performed with 4 nM DNA substrate and 60 nM protein in 25 mM Tris-HCl, pH 7.5, 50 mM NaCl, 2 mM MgCl 2 , 1 mM DTT at 37 °C.
DNA-binding assay. The influence of Cd 2+ on the DNA-binding activity of BLM was assayed by measuring the steady-state fluorescence anisotropy parameter 56-58 , using a Beacon 2000 polarization instrument (PanVera, Madison, Wi), equipped with a temperature-controlled cuvette, according to Xu et al. 31 . Briefly, the 3′-fluorescein-labeled DNA, free in solution and bound to BLM, are characterized by fast (low anisotropy value) and slow (high anisotropy value) rotational diffusion, respectively. The relative change in the anisotropy value allows the calculation of the fractional saturation function. Recombinant RecQ E.coli , BLM full-length or BLM 642-1290 were pre-incubated for 10 min at 25 °C with increasing concentrations of CdCl 2 in a Tris-HCl buffer (50 mM, pH 8.0, 50 mM NaCl, 1 mM DTT) before addition of the 3′ -fluorescein-labeled double-or single-stranded DNA (5 nM in a total volume of 150 μ l). Fluorescence anisotropy was measured under real-time condition (steady-state fluorescence anisotropy values were recorded every 8s). The effect of Cd 2+ on the fractional saturation function (A) (also called relative DNA-binding activity) was calculated using equation 2: x 0 y 0 where A x and A y represent the fluorescence anisotropy values for a given concentration of protein in the presence or absence of CdCl 2 , respectively. A 0 represents the anisotropy value characterizing the fluorescently labeled DNA alone.