Metal Exchange in the Interprotein ZnII‐Binding Site of the Rad50 Hook Domain: Structural Insights into CdII‐Induced DNA‐Repair Inhibition

Abstract CdII is a major genotoxic agent that readily displaces ZnII in a multitude of zinc proteins, abrogates redox homeostasis, and deregulates cellular metalloproteome. To date, this displacement has been described mostly for cysteine(Cys)‐rich intraprotein binding sites in certain zinc finger domains and metallothioneins. To visualize how a ZnII‐to‐CdII swap can affect the target protein's status and thus understand the molecular basis of CdII‐induced genotoxicity an intermolecular ZnII‐binding site from the crucial DNA repair protein Rad50 and its zinc hook domain were examined. By using a length‐varied peptide base, ZnII‐to‐CdII displacement in Rad50’s hook domain is demonstrated to alter it in a bimodal fashion: 1) CdII induces around a two‐orders‐of‐magnitude stabilization effect (log K12ZnII =20.8 vs. log K12CdII =22.7), which defines an extremely high affinity of a peptide towards a metal ion, and 2) the displacement disrupts the overall assembly of the domain, as shown by NMR spectroscopic and anisotropy decay data. Based on the results, a new model explaining the molecular mechanism of CdII genotoxicity that underlines CdII’s impact on Rad50’s dimer stability and quaternary structure that could potentially result in abrogation of the major DNA damage response pathway is proposed.


Introduction
Cadmiumi saw ell-defined nephrotoxic, pneumotoxic, osteotoxic, and cardiotoxic agenta nd, on top of that, as trong carcinogen, af eature probably stemming from its extensive indirect genotoxicity. [1][2][3][4][5] Its detrimental effects on human health are known,b ut ac learm echanism connecting cadmium's intake and its indirect genotoxicity is still elusive, which seems to be dictatedb yamultivalent effect occurring as ar esult of binding of the Cd II ionb yi ts cellular targets. [6] Cd II as as ofter Lewis acid than Zn II readily displaces it in cysteine(Cys)-rich zinc-binding proteins and biomolecules, which solitarily introduces an enormoust oxic effect inside ac ell. [7] Zn II fluctuations, which impact the exchangeable zinc quota, [8] interfere with redox homeostasis, and deregulate cellular metalloproteome. [9] Furthermore, because Cd II binds preferentially to thiol-containing and other nucleophilic ligands, it interacts with aw ide spectrum of redox signaling and reactive oxygen scavenger proteins, inten-sifyingZ n II -displacement effects. [10] Besides generating ap ool of displaced metal ions, Cd II interferes with its target's biophysical properties, namely stability, flexibility,a nd overall structure. [11,12] Structure-interfering properties of Cd II bindingt oZ n IIcontaining proteins remain controversial because ford ecades Cd II was used as aZ n II -isostructural spectroscopic probe for Zn II -binding proteins in UV spectrophotometric studies and 113 Cd NMR spectroscopy. [13][14][15][16][17] Although Cd II and Zn II ionicr adii are relatively similar and displacementi su sually one-to-one, in terms of stoichiometry and coordination sphere, their binding to at arget protein could potentially generate ad ifferent structural outcome. The presented scientific problem of Zn II -to-Cd II substitution and its impact on zinc-binding proteins have been extensively studied;h owever,a ll published data are focused on intramolecular sites, mostly of zinc finger domains and metallothioneins. Effects of this substitution are not uniform and involven o, or very slight, changes in structure and function, like in the case of Sp1, [18] Tramtrack, [19] SUP37, [20] or Ros87, [21] as well as significant alterations of the domain fold, in the case of XPA, [5,22,23] p53, [24] or MTF-1. [25] Herein, we aim to illustrate for the first time how this phenomenon affects the intermolecular zinc-bindings ite by investigating Cd II binding to the central fragment of Pyrococcus furiosus (P. furiosus)R ad50 protein. Rad50 is ac onstituent of the MR(N/X) (Nbs1/Xrs2 is distinctive for eukaryotes) complex responsible for double-stranded DNA damage signalinga nd repair.A ctive as ad imer,i tc onsistso ft wo protomers with a typical structure for the SMC protein family:averyl ong antiparallel coiled-coil( cc) segment that connects two apexes of the protein (i.e.,t he globularD NA-binding ATPase domain), and the zinc hook domain responsible for dimerizationb yf ormation of at etrathiolate Zn(CXXC) 2 coordination sphere ( Figure 1). [26,27] The overall dimeric assemblyo ft he MR(N/X) complex is not well-defined, and data from electron and atomic force microscopy present multiple conformations of the complex rangingf rom open-circular-like to closed-rodshape-like. [26,[28][29][30] Such alternating behavior seems to be a commonf eature of the SMC protein family and, for Rad50 in particular,could potentially allow mechanically driven functional diversification;f or example, switchingb etween pathways of DNA damage repair.T his process, however,c ould only be accommodated through specific flexibility of both globulara nd zinc hook apexes, the former of which has already been documented [31] but the latter remains hypothetical. We hereby present an insight into molecular bases of Cd II toxicitya nd indirect genotoxicity,t aking into account the impact of Cd II on Rad50's dimer stability and quaternary structure that could potentially result in abrogation of the major DNA damage response (DDR) pathway (and subsequent DNA lesions, chromosomal aberrations, and carcinogenesis).
Previous studies on the central fragment of the zinc hook domain of the Rad50 protein from P. furiosus demonstrated that this fold forms an extremelys table complex with Zn II . [32,33] The molecular basis of this high stabilityh as been analyzed by using as et of hook model peptides (Hk) containing a-CPVCbindingm otif with increased peptide flanks from both ends (Hk4-Hk45). [32] The results indicated that the formation of a hook homodimer (Zn(Hk) 2 )o ccurs in as tepwise manner starting from the formation of the ZnHk complex, followed by associationo fa nother Hk molecule. The increased difference in stability of the two complexes with rising peptidel ength favors the formation of the Zn(Hk) 2 species. The existence of an equimolar complex being in equilibrium with ad imer could be of high importance and suggests ap ossibility of readily exchangeable Rad50 molecules during cellular processes,w hich is believed to be the requirement for MRN native functions. [34,35] Keeping in mind the toxic effect of Cd II on zinc finger domains of transcription factors ando ther zinc-sulfur proteins associated with DNA processing, we aimed to examine the impacto fC d II binding on the Rad50 protein using a well-described model from P. furiosus, [32,33] which would be the first intermolecular zinc-binding site analyzed in terms of Cd II attack. The aim of this study was threefold:1 )tod emonstrate how strongly Cd II ions bind to the Rad50 protein, 2) to show whether or not Cd II ions are able to replaceZ n II in the cellular timescale, and finally,3 )toa nalyze how Cd II bindinga ffects the Rad50 hook structure in aw ay that could translate into functional alterations of Rad50and the entire MRN complex.

Results and Discussion
In the first stage of the study,w ee xamined whether Cd II is capable of binding in the same stoichiometricm odel as Zn II and whetheri tf orms more stable complexes allowing Cd II to exchange Zn II from the hook domain. To assess if the structure and stabilityo fC d II complexes rely on as imilar structural basis to that of Zn II ,w eu sed as eries of zinc hook (Hk) peptides rangingi nl ength from 4t o1 30 amino acid residues (Hk4-Hk130) ( Figure 1). Peptide Hk45 has been recognized as af ragment which, as aZ n II complex, possesses all residues thatf orm intermolecular contacts and contains both a-helical and b-hairpin regionso ft he domain. [33] Peptides Hk4, Hk6, Hk10, and Hk14 wereu sed to investigate metal-coupled formation of the b-hairpin. The remaining peptides( Hk27, Hk31, and Hk37) were appliedf or investigation of any important changes occurring in the vicinity of the region between the b-hairpin and the elongatedd omain (about45a mino acid residues). The lon- Figure 1. Architecture of the MR(N/X)complex with Rad50 fragments (Hk4-Hk130)i nvestigated in this study.Green, blue, and yellow represent two Rad50,t wo MR,and one DNA molecule, respectively.Structuralrepresentation is based on the crystal structure of ac entral 112-amino-acid-long Rad50 (P. furiosus)f ragment with Hg II (PDB code:1L8D) and the globular apex of a MR complex from Methanocaldococcus jannaschii boundt oD NA (PDB code: 5DNY). gest Rad50 protein fragment, containing 130 amino acid residues, wasu sed as aR ad50 protein modelc ontaining al ong coiled-coil fragment to examinew hether effects occurring in minimal fragments are transferred to other protein regions and vice versa.

Spectroscopic analysis of Cd II binding to the hook motif
Binding of Cd II to hook peptides was monitored using several spectroscopic methods including spectrophotometry and spectropolarimetry for acetylated and amidated peptides, as well as fluorimetry and fluorescencea nisotropy in the case of fluorescently labeled peptides. Far-UV circulard ichroism (CD) titrations, performed by the addition of Cd II to metal-free peptides at pH 7.4 in the presence of tris(2-carboxyethyl)phosphine hydrochloride (TCEP)( used as an on-metal-binding reducing agent), [36] demonstrated extensive conformational changesf or most investigated peptides upon Cd II coordination (Figure2a and Figure S1, Supporting Information). Spectrophotometric titration in the same region showed formation of ab and with the maximum between 230 and 250 nm. This band corresponds to ligand-to-metal charge transfer (LMCT) events, and it is typicalf or CdS 4 metal centers found in tetrathiol-containing ligands, such as Zn(Cys) 4 (FigureS2, SupportingI nforma-tion). [37][38][39] Both CD and UV/Vis spectrad emonstrate as harp inflectioni nt he titration curves at a1 :2 Cd II -to-peptide molar ratio, which confirms the formationo fp redominantly Cd IImediated homodimers:M L 2 complexes (Figure 2b). However, in the case of Hk27, Hk31, and Hk45, the formation of CdHk (ML) complexes also occurs. The largest difference in stability between ML and ML 2 complexes is observed for Hk27, whichi s reflected by two visible inflection points, whereas the Hk6, Hk10, and Hk14 peptides demonstrate the strongestt endency to form only dimeric species,s imilarly to the studyo fZ n II complexes [33] (Figure 2a nd Figure S2, Supporting Information). In the case of Hk130, the CD spectra changess lightly (Figure 2a), which complicates stoichiometric analysis. However,s pectrophotometric titration in the UV range clearly shows the formation of the Cd(Hk130) 2 homodimer,w hichi st he predominant species (Figure 2b and Figure S3, Supporting Information). Depending on the examined peptide, the Cd(Hk) 2 complex adopts different conformations, as indicated by thed istinct CD spectra.T he structural changes obtainedd uring Cd II coordination can be compared qualitativelyb ased on the differential spectra obtained through the spectral subtraction of free Hk peptidef rom their complexes with the metal ion ( Figure 2c). Both types of spectra, together with their molar ellipticity,i ndicate the formation of a b-hairpin-like structure in the case of Figure 2. Cd II binding to selected hookp eptides recorded by CD and UV/Vis spectroscopy in 20 mm Tris-HCl buffer,pH7.4, I = 0.1 m.a )CDs pectra of the apo form (blackline) and the complex at aC d II -to-Hk peptide molarratio of 0.5 (red line). b) Dependence of the ellipticity (Hk4, Hk14, Hk45)a nd absorbance (Hk130)atl isted wavelengthso nt he Cd II /Hk molar ratio. c) Differential CD spectrao fh ook peptides obtained by the subtraction of the spectraoffree peptides from the spectra of Cd(Hk) 2 complexes.[V]and e refert om olar ellipticity (in deg cm 2 dmol À1 )and molar absorptionc oefficient (in m À1 cm À1 ), respectively. Cd(Hk14) 2 and mixed b-hairpin and helical structuresi nt he case of Cd(Hk45) 2 .T he case of the longest Hk130 peptidei s certainly different due to al ong helical structure (more likely being ap art of coiled-coil structure) that is already present in the metal-free form. The Cd II binding to this protein fragment causes only small changes in the CD spectra,i ndicatinga dditional structurization in the central metal bindingp art and parallel stabilization of the coiled-coil structure. Althought he peptide backbonec ontribution is evident,o ne has to account for the Cd-S LMCT contribution in the overlapping region. Nonetheless, the backbone-related transitions presented herein are around1 0-20 times stronger;h ence, the Cd II -binding contributions are significantly overshadowed and do not influence the qualitative assessment. [37] Metal-coupled b-hairpin formation in the middle of the zinc hook upon Cd II complexation has been demonstrated by using am inimal N-terminally dansylated Hk14 peptide with at ryptophan residue placed at its C-terminus. Figure 3(inset) shows the peptide's emissions pectra when the tryptophan (Trp) residue is excited at 290 nm. Emission spectra demonstrate af luorescence resonancee nergy transfer (FRET) effect betweenT rp and dansyl( Dns) residues,b ut the efficiency shows that both donor and acceptor are far away in the metal-free peptide, indicatingt he disorganized nature of the Hk14 peptide. [32] When Cd II is bound, and the homodimer is formed, the intensity of Trps ignificantly decreasesw hile the intensity of the acceptor increases, demonstrating as ubstantial efficiency increase. [40,41] Ta king into account the Fçrster radius of the Trp-DnsF RET pair,b eing between 21 and 24 ,i ti sl ikely that this FRET efficiency change may be attributed to the proximity of both fluorophores aroundt his range or rather below it. [42] Indeed,t he molecular dynamics simulation of Hk14 in the Cd II complex (see Supporting Information for more details) shows that the estimated average distance between donora nd acceptori n the Cd II complex is (12.7 AE 2.5) .S imilarF RET efficiency change upon metal binding hasb een observed for the Hk14 peptide and several of its mutants upon Zn II complexation. [32] However, the change occurring in the case of Cd II is much more pronounced than that of itsZ n II counterpart, suggesting either more compacts tructure of ap rotomer or smaller distance between the N-and C-termini of as ingle protomer or between opposite termini of both protomersi nC d(Hk) 2 .T he hook structure, even in its minimal fold, is highly sensitivet oa ny closeoccurring changes. Our previous study showed that alanine scanningi nt he Hk14 sequence may affect the FRET effect significantly,i ndicating that even one non-metal-binding residue substitution may impact the hook's globals tructure. [32] Moreover,c hanges occurring in the further region of the full domain also affect the structure and stability of formed complexes;t his has been shown by using al onger peptidem odel as well as in vivo tests on the full MRX complex. [43] Mutations of the hook domain's Cys-neighboring amino acids alter DNA damage repair and signaling functions of the entire complex, provingt hat alteration of the hook motif is transferred to the globularp arts of the MRN complex, presumably throught he coiled-coil region,a nd render it abrogated. [44,45] To thoroughly investigate the Cd II ion's impact on the Rad50's zinc hook domain'sa ssembly,w et urned to anisotropy decay analysis. Measuring time-resolved anisotropy change provides vast analytical opportunities andg ives insights into protein dynamics, dimensions, as well as protomer arrangement in oligomeric species. N-terminally FAM-labeled (FAM = 5(6)-carboxyfluorescein) P. furiosus Rad50H k14 andH k45 were subjected to an anisotropy decay experiment either as apo forms, Zn II /Cd II -loaded, or analyzed after overnight incubation in as et of metal buffers. We focused our attentiono no ne major parameter derived froma nisotropy decay analysis:r otational correlation time [t r ,E q. (1)],w hich delivers information about molecular dimensions in terms of their diffusion capability in asolvent, for which I is intensity, r inf is residual anisotropy, and t r 1 and t r 2 and B 1 and B 2 are rotational correlation time and amplitude, respectively,f or both exponentials. Because the t r parameter describes how fast the fluorophores rotate,t hus decreasing their anisotropy,i tisd irectly correlated with the hydrodynamic radius of the emitting molecules. [42] The rotational correlationt ime parameter differs significantly for Zn II andC d II complexes,w hich is clearly visible from plots of t r against ar ange of free metal ion concentrations ( Figure 4). These results demonstrate that t r forb oth peptides  (3), according to ap rotocolbyP omorski et al. [52] Dashed linesrepresentaplot confidence of 95 %. increases with free metal ion concentration, as ar esult of metal-ion-induced dimerization, but the increase from metalfree to metal-bound form is around 30 %h igher for aC d II -binding event than that of Zn II .T he experiment suggests that Hk14 and Hk45 molecules in Zn II and Cd II complexes presentd ifferent dimer arrangements with different hydrodynamic diameters, the former being smaller than the latter. [42,46] Ta king into account that the Cd II ionic radius (109 pm)i sslightly larger than the Zn II radius (88 pm), [10] this trait feels natural. However, it seems that swellingo ft he Rad50 dimer interface caused by incorporating ab igger metal ion is transferred further down through both protomers, changingt he overall quaternary structureo ft he analyzed dimers.
Our approachp ossesses one major intrinsic obstacle that prevents precise volumetric measurement of analyzed dimers; as ar esult of the homo-labeling setup, homo-FRETevents are very likely to occur between two FAM-labeled N-termini of Rad50 dimers, which results in an additional route of anisotropy decay.W es peculate that homo-FRET transfer is represented as an additional decay time;h owever,i ti st oo fast to allow quantitative assessment of differences between alternatively composed hook dimers. To resolve whether ah omo-transfer's impact on anisotropy decay overshadows differences in molecular volume, we analyzed initial anisotropy valuesf or both complexes.I nitial anisotropy (not fundamental anisotropy, which is constantf or 5(6)-carboxyfluorescein and equals 0.38) would be significantly affected by the homo-FRET phenomen-on, making it an ideal control parameter to ascertain the validity of this approach. [42] Interestingly,i nitial anisotropy values for complexes with Zn II were al ittle higher than those of complexesw ith Cd II ,s uggesting homo-FRET of highere fficiencyf or the latter ( Figure S4, Supporting Information).T his meanst hat differences in rotational correlation time are probablym ore pronounced, and the t r value for Cd(Hk) 2 may be in fact higher than what we have observed.
Determination of stability constants of Cd II zinc hook complexes As demonstrated in the spectroscopic studies, Cd II titration curveso fH kp eptides demonstrate isotherms of complexf ormationw ith as harp inflection point, which clearly indicate that affinity of the studied hook peptides towards Cd II is high, and direct spectroscopic stability constant determination would not be feasible withoutr isk of their underestimation. [47,48] Therefore, to determine actual stability constant valueso ft he formed Cd II complexes,s everal approaches were used. The first one and the most informativeregardingt he stoichiometric model and cumulative stabilityc onstants (formation constant b ijk of the M i H j L k complex, which includes the protonation state of the ligandsc oordinated to the metal ion) was potentiometry.H owever,a saresult of the limitedn umber of residues with acid-base properties, which is ar equirement for this method, it was applied here only for Hk4-Hk14 peptides. [49] The obtained results confirmed that ML and ML 2 stoichiometries of the complexesa re formed during complexation and are variously protonated depending on the peptide chain length ( Figure 5). Table 1p resents cumulative constants of all peptides investigated potentiometrically as well as the pK a valueso fd eprotonating groups. Species distribution presented in Figure 5s hows that the ML 2 (M = Cd II )c omplex is formed at low pH (from about 4a nd 3.5 forH k4 and Hk14, respectively) and is stable up to basic pH. For comparison, Zn II complexes of the same peptides are formed ata round 0.5 unit higheri np H, indicating as ignificant differenceb etween the two metal ions. This pH differenceindimericcomplex formation is demonstrated by approximately two-orders-of-magnitudel ower stability constantsf or Zn II complexesw hen comparedt oC d II ones, althought his difference varies depending on the peptides' length (Table 1). [33] Given that cumulative stability constants are pH-independent, the direct comparison of affinitiesb etween particular ligands or metal ions at certainpHisi mpossible withoutconsiderationo fK a values of metal-free ligands, which differ between each other.T op resent potentiometric data in ac omparable way (with conditional constantsd etermined spectroscopically), we calculated the formation constant K 12 of the Cd(Hk) 2 complexes valid for pH 7.4 as shown below [Eq. (2)]: in which [Hk] is the sum of all metal-free Hk species with variously protonated states as well as fully deprotonated species   (Table 2). However,t he largest increase in stability is observed between Hk6 and Hk10, similarly to Zn II complexes ( Figure 6), suggest-ing that Cd II -mediated folding of the b-hairpin is the energy force that elevates stability of the hookd omain, analogously to Zn II complexes. [32,33] Although potentiometry is ag ood methodf or determination of stabilityc onstantso fh ighly stable complexes, its application is limited to the number of acid-base groups, and it is difficult or impossible to use for proteins and their larger frag-   . Molar species distribution of Cd II and Zn II complexeswith a) Hk4 and b) Hk14 peptides calculated based on protonation ands tability constants determined potentiometrically.H kp eptides and metali on concentrations were set as 400 and 200 mm,r espectively,a si nt he potentiometry experiments. [32] ments.T herefore, to determine K 12 of Cd II complexes with Hk27, Hk31, Hk37, Hk45, and Hk130 peptides, we applied CDmonitored competitionw ith common metal complexones with well-established stability constants, such as HEDTA, EDTA, and TPEN. [48,50] Because ellipticity changes of the Hk130 peptide are the smallest, CD data were supported by natural tyrosine fluorescencec hange, the intensity of which decreases upon Cd II binding ( Figure S5, SupportingI nformation). Figure 7s hows normalized ellipticity or fluorescence intensity as af unction of free Cd II ions, calculated using Hyperquad softwareb ased on protonation and stabilityc onstants of Cd II -complexone com-plexes. [51] To determine K 12 values, experimental data were first fitted to Hill's equation [32] to obtain half-saturation points (Àlog [Cd II ] free 0.5 ). In the next step, K 12 values were calculated accordingt oE quation (2), assuming that the half-saturation point corresponds to the half of maximal concentrationo f Cd(Hk) 2 . [32,33] Obtained K 12 values with standard deviatione rror are presented in Table 2a nd also illustrated in Figure 6( open red circles), in which they overlap very well with K 12 values obtained from potentiometry fors horterp eptides (solidr ed circles). These results indicate acceptable data coverage between the two methods (potentiometry and spectroscopy).  Because Hk14 was shownt ob et he minimal fold of extreme stability, [32] its complex with Cd II wash erein investigated primarily and resulted in an additional set of competitiond ata gatheredf or the fluorescently labeled Hk14 with Dns andT rp moieties. Thist ype of modification enables assessment of wide dynamic range changes that guarantee the highest-quality data determination. Moreover,t he presence of two bands corresponding to the donor and acceptor of the FRET pair makes this peptideasensitive ratiometric sensor of free Cd II concentration and conformational changes.F igures 3a,b demonstrate Trp/Dns( R Trp )a nd Dns/Trp (R Dns )r atios of intensity changes, respectively,a safunctiono ff ree Cd II .T od etermine the actual Àlog [Cd II ] free 0.5 value from the ratiometric study,wea pplied our normalizing equations that are based on fixed maximal and minimal intensities of Trpa nd Dns bands, independently,a s describede lsewhere. [52] Application of these formulae to fluorescencei ntensity ratios resulted in actual Àlog [Cd II ] free 0.5 values of 15.8 and 15.7 for R Trp and R Dns ,r espectively,a llowing furtherl og K 12 calculation to give 21.17. Formation constants obtainedw ith this methoda re convergent with the values calculated from potentiometric data ( Table 2).

Stabilityc omparison of Cd II -substitutedh ook domain with other Cd II complexes
Our previous resultsh ave shown that the zinc hook domain is among the most stable Zn II complexesf ound in proteins and their natural domains. For example, the Zn(Hk45) 2 complex is more stable than the most stable naturalz inc fingers described so far,s uch as Sp1-3, MTF-1, or Zif268-2, when competing in the micromolar range with Hk45. [33,47,53] When concentrations of both protein fragments decrease, zinc fingersd emonstrate a higher tendency to bind Zn II .S uch behaviorc omes from the fact that the hook domain forms an ML 2 -type complex,w hereas zinc fingersa nd most other proteins form an ML-type complex. Our recent resultso btained on various zinc domains with differentZ n II -to-protein stoichiometry,y et similar affinity, showedc oncentration dependence for metal-mediated complexes. [54] Such concentration-dependent metal association might be an important feature for transient saturation and transientp rotein function. In light of these results, there are important questionso fh ow Cd II ions maya ffect zinc hook structure and function, and how Cd II binding may affect possible transient saturation of the hook domain. To get closer to answering these questions and to properly order the cadmium hook in the stability hierarchy of Cd II complexes with proteins, it is worth analyzing the literature in terms of stabilityd ata of Cd II complexes of other proteins or peptide models. Ta ble 2 presentst he apparent formation constantso fC d(Hk) 2 at pH 7.4, calculated either from potentiometric data or determined by competition with complexones. Because of the fact that the hookm otif forms an ML 2 -type complex, direct comparisono fi ts formation constant( K 12 )w ith constants of complexes with ML stoichiometry of other ligands is impossible. To avoid this problem, we converted all formation constants present in Table 2t ot he competitivityi ndex (CI), whichh as been shown previously to be useful for the comparison of affinities of metal complexes with various stoichiometries. [55,56] In principle, it simplifies any stoichiometry to M x L y under certain reactant concentrations and is valid only when comparable ligands and metal ion are present in the same concentrations. [56] Thus, aq uick comparison of CI values of Cd II hook complexes and other Cd II complexes found in the literature (Table 3) shows that the investigated complexes, Hk14-Hk130, are the most stable found to date. The CI values of these Cd II hook complexes are two-to-four orders of magnitude more stable (CI = 17-19) than the strongest ones with XPAzf( CI = 12.8), [23] CadC (CI = 12.6), [57] CmtR (CI = 12.2), [58] or MT2 (CI = 14.4 for bdomain and 15.8 for a-domain). [59] Moreover,the complexes investigated here are much more stable thant he Cd II complex with EDTA( CI = 13.6), [60] consensus CP1 zf (CCCC)z inc finger (CI = 13.4), [61] or short poly-Cys peptides known to form highly stable complexes with Cd II ,f or instance Ac-YCSSCY or Ac-CC-NH 2 ,f or which CI is 14.8 and 12.6, respectively. [62,63] This brief overview clearly demonstrates that the zinc hook domain with substituted Cd II muste xhibit unique stabilizatione ffects that elevatet hermodynamic stability. Molecular reasonsf or such stabilitya re discussed below.
Zn II -to-Cd II swap in zinc hook domain Data presented above showt hat Cd II complexes of the minimal and elongated hook domain are significantly more stable than Zn II counterparts and other protein, peptide, and low-molecular-weight organic complexes.T his should be reflected by an efficient Zn II swap when Cd II ions are added to the zinc hook domain. To examinee xperimentally how fast ande fficient the substitution of metal ions occurs, we titrated the Zn II complex of Hk14 and Hk45 with Cd II and observedc hanges occurring in UV/Vis and CD spectra.T his was possible due to the formation of energetically lower LMCT bandsa t2 30-250 nm for Cd II in comparison with those of Zn II complexes. [37][38][39] This effect has also been observed in CD spectra by bathochromic shift of ellipticity negative maximaa nd their intensity increase when Cd II swap occurred. To measurethe equilibrated states of the reactiond uring titrationp roperly,t he kineticso fZn II -to-Cd II swap was examined first. Figure S6 (Supporting Information) indicates that in the case of both Zn(Hk14) 2 and Zn(Hk45) 2 complexes the time necessary for metal-ion-exchange completion is lower than 2minutes, indicating rapid kinetics of the metal swap.E xchange for Hk14 was shown to occur under first-order kinetics with ar ate constant calculated to be about 0.015 s À1 ,a nd the rate for Hk45 is even faster and was impossible to determine under standard settings.W hen the Zn(Hk14) 2 complex wast itrated with Cd II ,t he metal exchange observed by an increase in the LMCT band intensity (Figure 8a)r eveals almostd irecte xchange, indicating that the reaction is not only rapid but also highlye fficient, as expected, due to the difference in stabilityc onstants of Cd II and Zn II complexes. Fivefold molar excesso fZ n II over Hk14 slows down the exchange in such aw ay that ah igher concentration of Cd II is necessary to swap the hook-bound Zn II .T his enablesc alculation of the formationc onstant K 12 based on the fixed constant of Zn(Hk14) 2 and total Cd II concentration (Figure 8b). [54] The K 12 values calcu-  [47,53] and Table 3. Affinities of selected low-molecular-weight ligands, peptides, and proteins for Cd II collecteda cross the literature that form highly stable complexes.S tability constants wered eterminedu nder various conditions, which are listed. If not specified, valuesr efer to pH 7.4. RT,p HT,C C, and n.c. refer to reverset itration, pH titration,c ompetitionw ith metalc helator,a nd not calculated, respectively.C Iv aluesd erived from formation constantsp resented as 'not calculated' (n.c.) were determinedf rom published log b ijk data from potentiometric analyses. [38,39,62,63,83] Ligand [a] ReferenceM ethod of determination   were not included in further discussion. Nonetheless, the results prove that Zn II -to-Cd II swap is af ast and efficient process that can occuri nside ac ell, causing Cd II -induced toxic effects.
To illustrate the Zn II -to-Cd II exchange event located inside Rad50's hook domain, we performed isothermal titration calorimetry (ITC) analysisi namanner similar to the previous approach.1 00 mm Hk14 with 2.5-fold molar excess of Zn II was titrated stepwise with 500 mm Cd II .A ss hown in Figure 8c,C d II readily displaces the hook-bound Zn II ion, which is represented as aslowly decreasing exothermic process (inset) of atendency much like the one from the UV spectroscopy experiment (Figure 8b). Thermodynamics associated with the metal exchange reactionisd iscussed separately below.
Living organismsp ossess multiple factors that guard cells from Cd II toxicity,w ith the two most important examples being glutathione (GSH) andm etallothioneins. Regardless of the fact that GSH is presenti nt he cytoplasm in millimolar levels, Cd II maintains the capacity to infiltrate protein targets. It is ak nown fact that Cd II accumulation in mammals occurs in soft organs,e specially in the livera nd kidneys. [64][65][66] Metallothionein, found originally as aC d II -binding protein in the horse kidney cortex, binds this metal ion very tightly in two distinct clusters:f our Cd II ions in the a-cluster and three Cd II ions in the b-cluster.H owever,i solated hepatic metallothionein from cadmium-toxicated mammals demonstrates ah eterogeneous nature. [67] Metallothionein binds at the same time Cd II as well as Zn II ions at variousm olar ratios;h owever,o nly aC d 5 Zn 2 MT speciesw as characterized structurally,a nd the crystal structure of this species is the only X-ray structure of mammalian MTs. [68,69] In this study,b yu sing Cd II -reconstituted metallothionein-2,w ea imed to examine whether tightly bound Cd II ions by this protein can be transferred to the zinc hook domain of the Rad50 protein, which as ac onsequence could shed al ight on Cd II ion distribution in cells.
It is worth mentioning that sevenC d II ions are boundi n MT2 with two distinct affinities:t hree ions corresponding to the b-cluster with averaged log K = 14.4 and four bound in the a-cluster with log K = 15.8, which is in contrastt ot hat of zinc MT2. [58,70,71] Stability data obtained in this study suggest that Cd II could be easily transferredf rom human Cd 7 MT2 to the hook domain. To examine Cd II transfer from MT2, which demonstrates stronga bsorption in the UV range duet oL MCT bands occurring between the Cys residue and Cd II ,w eu sed the fluorescently labeled Hk45 peptidea nd monitored anisotropy changes associated with this process. Our study showed that anisotropy decay of Zn II and Cd II hook complexes differs in terms of rotational correlation time values, which enables its use for type-detection of metal ions bound to peptides.
In the first step of metal swap, similarly to free Cd II ion substitution, we examined the kinetics of the transfer.F luorescein emission changes presented in Figure S7 (Supporting Information) indicate that the kinetics is very fast and that the reaction occurs completely in the time window below 1minute. Kinetics of the swap seem to be equally fast for Cd II alone( i.e.,n ot bound to MT2) and fully Cd II -metalated MT2. To show that both act in as imilar fashion, we examinedZ n(Hk45) 2 and Cd(Hk45) 2 with either addition of Cd II or Cd 7 MT2a. The control sample's rotationalc orrelationt ime, being in this particular case of Cd(Hk45) 2 ,w as similara fter separate addition of Cd II and Cd 7 MT2a in 0.5 and one molar equivalents, respectively ( Figure S8, Supporting Information). However,f or the Zn(Hk) 2 sample the parameterc hanged towards Cd II -specific values, both for Cd II and Cd 7 MT2a additions (Figure 8d). Overall,t he results show that Zn II -to-Cd II swap occurs even with the additional Cd II -chelating component, and we suggestt hat it may be ar elevant cellular process where MT is one of the major barriers against Cd II toxicity.

Structure determination of Zn II -a nd Cd II -substituted hook domain
All spectroscopica nd thermodynamic studies obtained in this report demonstrate that the minimal fragment of theh ook domain (Hk14) forms aw ell-defined and highlys table structure as ac omplex with Cd II ions. Stability constantso btained for longerm odelss how that sequence elongation does not affect the stabilityo ft he complex significantly,s uggesting that folding of the hook domain starts from the short 14-amino-acidlong fragment responsible for Rad50d imerization. Although a P. furiosus Rad50 zinc hookd omain'sc rystal structure was solved moret han two decades ago, we are still missing a structure of aZ n II -loaded complex because the solved one was actually am ercury hook (Hg(Hk112) 2 ,P DB code:1 L8D). [26] In light of the resultsp resented above,w es peculate that the zinc hook may be in fact structurally different in solution, given that the Cd II -loaded hook seemst op ossess ad ifferent dimerica rrangement. Thus, we aimed to determine structures of this motif to provideanew look at this alluring interprotein site. For this reason,w eo btained 5mm Zn II and Cd II complexes of the Hk14 hook model and performed an umber of one-and two-dimensional NMR spectroscopice xperiments, including 1 H-13 CH SQC, TOCSY, and NOESY,o btaining 130 and 190 NOE peaks forZ n II andC d II ,r espectively.F rom these, 20 and 15 peaks for Zn II and Cd II complexes,r espectively,w ere rejected at the initial assignment.
The ten lowest-energy structures of Zn(Hk14) 2 and Cd(Hk14) 2 obtained with PROT and CRYST procedures (with hydrogen-bond restraints obtained from protection factors and crystal structures, respectively,a se xplained in the Supporting Information) are shown in Figure 9. The statisticsofv iolated restraints are collected in Ta ble S5. The energy of violated restraints is relatively low for both Zn(Hk14) 2 and the first conformation of Cd(Hk14) 2 .F or the second conformationo f Cd(Hk14) 2 computed with CRYST restraints, significant (> 0.5 ) violations of two hydrogen bonds( Val7 N-Cys5 Sa nd Arg10 N-Cys8 S) wereo bservedi n3out of 10 structures, leading to as ubstantial rise in the averagee nergy of restraints. Therefore, we ran additional calculations without restraints put on these two weak hydrogen bonds and obtained structures with reasonable mean energy of NOE and hydrogen-bond restraints (see Ta ble S5). Calculations with PROT restraints led to structures with very small violations of H-bond restraints, but violations of NOE restraints increased in comparison to the structures obtained with CRYST restraints;a lthough only 2o ut of 10 computed structures have severely violated NOE restraints (NOE energy close to 200 kcal mol À1 ). Nevertheless, structures obtainedf or the second set of chemical shifts are similar to those obtained for the first set, as can be seen in Figure 9.
The structures of peptides with Cd II andZ n II have clearly different arrangements of chain termini (Figure 9a nd Figure S9, Supporting Information). The N-and C-termini of both Hk protomers bound to Cd II are located closet oe ach other (red structures in middle and lower panelsofF igure 9), whereas the terminio ft he same peptides bound to Zn II are located on the opposite sides of the complex (blue structuresi nu pper panel of Figure 9a nd Figure S9, Supporting Information).D escribed structurald ifferences may also be present in complexes of the investigated ions with longer Hk peptides, leading to different arrangements of helices forming coiled coils and ag lobal structuralc hange of the entire MR(N/X)c omplex.N onetheless, furthers tructurals tudies on longerp rotein fragments need to be undertaken to confirmthis hypothesis.

Molecular bases of elevated stability
Data reported herein clearly show that the Rad50 hook complex with Cd II is significantly more stable, by aroundt wo orders of magnitude in terms of K 12 , compared with its physiological counterpart that is Zn II .M oreover,h aving established the Cd II -binding affinity of the central fragment of P. furiosus Rad50, we therebyd ocumented the most stable Cd II -peptide complex analyzed so far (Table 3). Such extreme stabilityi sl ikely to be generated by multiple molecular factors, including enthalpic and entropic contributions from CdÀSb ond formation and metal-induced nucleation,t on ame af ew.F igure 6a and Ta ble S6 (SupportingI nformation) show that formation constant K 12 values of Cd(Hk) 2 complexes are almostp roportionally shifted towards higher stabilities than those of Zn II hook model complexes.T his indicates that these two types of complexesb ehave similarly when the hook motif is elongated from both ends. In Figure 6a,b lue and red arrows indicate two events of Gibbs free energy contribution for Zn II andC d II complexation, respectively,d emonstrated as an energy difference (DDG8 Cd/Zn )a safunction of peptidel ength. The first event occurs between 4a nd 14 aminoa cid residues in the hook motif, and the second event (less pronounced) occurs between 23 and 45, which correspond to the metal-binding-induced formation of the b-hairpin and nucleation of the coiled-coil region,r espectively. [33] The differenceb etween formation constants (log K 12 Cd II Àlog K 12 Zn II )i ss imilarf or Hk10, Hk14, Hk37, Hk45, and Hk130 and is 1.98 AE 0.04, which corresponds to À2.7 kcal mol À1 of Gibbs free energy and, with ah igh level of confidence,i sc orrelated with increasedc ontribution of bond formation enthalpy for CdÀSc ompared with that for ZnÀS( Ta ble S6, Supporting Information). However,t he log K 12 differences presented in Figure 6b demonstrate the presence of two regions where Cd II complexes are even more stabilized than the rest of the length-differentiated peptides eries. The first extra-elevated regiono ccurs for hook peptidesb etweenH k4 and Hk8 with a maximum at Hk6. This significant stability increase is very likely connected to differences in the ionic radii of both ions and the possibility of generating alternate torsion angleso fp eptide bondso re ven diverse intra-and intermolecular connections. [33] Another,l ess pronounced, region indicating an additional difference between the two metal-ion complexesi sp resent at the amino-acid length of 31. This region is more complicated to explain,b ut we suggest it may be caused by as hift in the coiled-coil nucleation eventt owards shorter hook lengths for Cd(Hk) 2 ,w hich is somewhat supported by the CD datas howing higher capacity for the hook's helical nucleation of Cd II compared with that of Zn II ,j udged by qualitative evaluation ( Figure S10, Supporting Information).
To assessw hat thermodynamic processes are involved and how they affect the stability of Cd(Hk) 2 ,weperformed ITC anal- Figure 9. Ensembles of ten dimeric structures of the Hk14 peptide with Zn II (blue chains) and Cd II (red chains). Structures in the left column wered etermined with the CRYST procedure,a nd structures in the right column were determined with the PROT procedure (see the Supporting Information).C ysteine side chains are showninyellow stick representation,a nd Zn II and Cd II are shown as orange and green spheres, respectively.The relative arrangement of loopso ft wo analogous complexeswith different ions differs significantly(see Results and Discussion). Twos tructures of complexes with Cd II (middle and lower rows)c orrespond to two sets of chemical shifts obtained from the NMRs pectroscopic experiments.I nt he computationp roceduref or the bottom left structure, two weak NÀSh ydrogen-bond restraints were removed. ysis of apo Hk4-45t itrated with Cd II and Zn II (Figures S11, S12, Supporting Information) to illustrate differences between complexation of both metal ions as well as the Zn II complex titrated with Cd II to visualize the energetic outcome of the Zn II -to-Cd II swapping phenomenon ( Figure 8c). Figure 8c shows that Cd II efficiently replaces Zn II inside the hook motif,e ven in excessive Zn II conditions. Moreover,i tp rovest he existence of a favorable enthalpic contribution of the swapping event, which seems to be approximately À2.6 kcal mol À1 ,avalue that corresponds well with the À2.7kcal mol À1 value obtainedf rom potentiometric and spectroscopic studies (Figure 6b and Ta ble S6,S upportingI nformation). On the other hand, independente xperimentsw ith Zn II and Cd II titrationso fa po peptides suggest that complex formation enthalpies of Zn(Hk14) 2 and Cd(Hk14) 2 are À18.1 and À21.6 kcal mol À1 ,r espectively, which gives À3.5 kcal mol À1 for DH Cd II ITC ÀDH Zn II ITC ( Figure 10). The 0.9 kcal mol À1 gap between these two values (the Zn II -to-Cd II swap enthalpys ubtracted from the apo-to-holo metalation enthalpy difference) can be explainedb yt he contributiono ft he folding enthalpy (DH fold ), which our results prove to be different for Zn II and Cd II complexes with Rad50. [72][73][74] During the swap event,t he hooki sa lready present as af olded dimer,a nd the energetic contribution of metal-ion-exchange-induced structural changes is too small to be observed directly through this type of ITC experiment;h ence,t he observable difference is mainly dictated by the enthalpy of the CdÀSb ond formation. On the other hand, direct titration of the apo peptide with Zn II or Cd II Assuming that Cys deprotonation and HEPESp rotonation enthalpies are equal for both Zn II and Cd II titrations, the energetic outcomes of metal-sulfur bond formation and protomer foldingh ave to be different. [75] Taking the above into account, one may conclude that the energetic cost of Rad50 protomer foldinga nd dimerization is in fact different for Zn II andC d II and can be roughly estimated for Cd(Hk14) 2 by the followings ubtraction: DDH ITC Cd II ÀZn II À DH ITC swap ,w hich equals À0.9 kcal mol À1 . Moreover,t he entropic contribution, derived from the Gibbs equationa sTDS8,i ss lightly lower for Cd(Hk14) 2 than for Zn(Hk14) 2 ,w ith 23.13 and 23.94 kcal mol À1 ,r espectively,s uggesting that Cd II may induce alternative hook folding that manifestsa sa ne ntropy loss ( Figure S13, Supporting Information). These calculations demonstrate that the molecular basis of the Cd II complexes' elevated stabilityo riginates from the highere nthalpic contribution of both CdÀSb ond formation and Cd II -induced protomer folding, compared with Zn II complexes, therefore indicating that Zn(Hk) 2 and Cd(Hk) 2 present different structural arrangements.
To evaluate values of thermodynamic parameters obtained in this study,w ea imed to compare them with published ITC data for 4Cys metal-binding sites. Although the literature lacks consistentt hermodynamic data for this particular metal-binding site, as well as the Cys residue alone binding to Zn II and Cd II ions, we managed to find some values. Krizek et al. obtained DG Cd II = À1.9 kcal mol À1 for consensus zinc finger CP1 with the 4Cys metal-binding motif. [61] Another zinc finger with a4 Cys binding motif, XPA, showed lower Gibbs free energy of around À4kcal mol À1 for Cd II complexation. [23] On the other hand, Zn 4 MT3 from Musa acuminata titrated with Cd II gave observable heats of around À3kcal mol À1 . [76] Although the pre- sented values seem to differ quite significantly,t hey are still comparable to our data, especially taking into account that none of the experimentsw ere performed in exactly the same way,n or did they maintain high analytical standards. Nonetheless, the difference in Gibbs free energy values of Cd II and Zn II complexation by the Rad50h ook domain or other 4Cys binding motifs seems to vary at around À3kcal mol À1 .

Biological significance
The Rad50 protein is an integral constituent of the MR(N/X) complexp resent in every living organism analyzed so far.M R as ah eterotetrameric (or heterohexamericM R(N/X) in higher organisms)c omplex, comprised of two units of Mre11n uclease and two units of Rad50, guards genomic integrity by sensing and repairing double-stranded DNA breaks that otherwise left unrepaired would cause deleterious genotoxic effects inevitably leading to cell death. [27] Given that Mre11i sr esponsible for processing of the broken DNA ends, Rad50 plays am ore structural role, acting as al engthy scaffold for the complex and a major dimerization factor through action of the zinc hook domain located at the apex of the MR(N) . complex. It is believed that the MR(N/X) complex is af lexible one with highly dynamic properties that enablei tt oa dopt multiple conformational assemblies. [26,28,29,31,77] Recent findings indicate that conformation-altering signals can be generated both at Rad50's ATPase globular and zinc hook domain apexes and transferred through an immensely long coiled-coil segment that spans 500 . [34,35,43,45,[77][78][79] At this point, two conformational states of the MR(N/X) complex seem to play am ajor functional role: closed conformationt hat binds tightly to DNA substrates and open conformation that enables DNA processing and repair. [34,80] Both assemblies interact differently with DDR proteins and seem to be preferentially activated during different phases of the cell cycle. [81,82] These datai ndicate that the zinc hook domain is much more than as imple dimerization motif and could possibly regulate the functional status of the entire complex. We have previously shown that any kind of aminoacid-substitution-based alterations render the zinc hook abrogated, destabilizingZ n II binding in vitro and DNA damage repair in vivo. [33,43] In this sense,a ny factor that inflicts structural changes upon the zinc hook domain has to be treated as a potentialgenotoxic agent.
Ta king the above into account, Cd II represents an evident threat to Rad50's functionals tatus and therefore at hreat to genomic integrity.C onsidering that Zn II -binding proteins represent approximately 10 %o ft he cell'sp roteome, Cd II -swapping outcomes could be deleterious for multiple biochemical reasons. Intuitively, the most detrimental toxic effect pertainst o DNA-binding proteins and transcriptionf actorst hat ubiquitously harbor Zn II -rich domains. Therefore, Rad50, with its zinc hook domain, poses as aperfect target for Cd II toxicity.
Experimental results presentedh erein prove that Cd II bound to Rad50 alters its structural status. The Cd(Rad50) 2 complex presents an augmented molecular volume that coincides with formation of extended helical regions, compared with Zn(Rad50) 2 .F urthermore, the Cd II ion is bound so tightly that it renders the zinc hook domain effectively blockedi no ne rigid conformation and incapable of any Zn II -o rD DR-protein-driven structuralr earrangements that seem to be crucial for the MR(N/X) complex's multifunctionality.

Conclusion
We hereby documented the mosts table Cd II -peptide complex described so far,w ith sub-zeptomolar affinity as aC d(Rad50) 2 dimer,a nd determined molecular bases of Cd II -induced stability elevation compared with its counterpart that naturallya llocates in the Rad50 hook domain, that is, Zn II .C d II ,a lthough similar to Zn II in terms of physical properties, inducess ignificant structural changes in the Rad50 hook domain. The reported increase in rotational correlation timea nd alteredN MR spectroscopic structure suggest the possibility of Cd II -bindinginduced globalM R(N/X) complex rearrangement, implying that this metal ion is capable of abrogating DNA-damage sensing and repair functions of the MR(N/X) complexi nacellular context. Additionally,t he approximately two-orders-of-magnitude increasei nC d(Rad50) 2 stability compared with that of Zn(Rad50) 2 supports the notion that Cd II is bound almost irreversibly to its target and renders it incapable of Zn II -a nd DDRfactor-governed flexibility that seems to be crucial for proper double-stranded DNA damage repair.O ur resultss how that this particulari nteraction of Cd II may be one of the major reasons for its still-unresolved genotoxicity and show how Cd II ions can impact intermolecular zinc-binding sites in oligomeric proteins.

Experimental Section
Materials lexed buffers and solutions. All buffers were prepared with Milli-Q water obtained with adeionizing water system (Merck KGaA).

Peptides ynthesis
Zinc hook peptides (Hk peptides) were synthesized by solid-phase peptide synthesis (SPPS) using an Fmoc-strategy on aT entaGel R RAM Amide Rink (Rapp Polymere GmbH, Tübingen, Germany) resin (substitution 0.2 mmol g À1 )a nd aL iberty 1m icrowave-assisted synthesizer (CEM) as described previously. [33,85] Peptides were N-terminally acetylated with acetic anhydride or fluorescently modified with ad ansyl (Dns) moiety or 5(6)-carboxyfluorescein (FAM )derivatives. Cleaved peptides were precipitated and washed with cold diethyl ether and purified on aC 18 column (Phenomenex) with a gradient of acetonitrile and 0.1 %T FA using aD ionex Ultimate 3000 HPLC system. The identity of peptides was confirmed with an API 2000 Applied Biosystems ESI-MS instrument. Concentration of peptide stocks in 10 mm HCl was determined using as ulfhydrylgroup reactant, DTNB (e = 14,150 m À1 cm À1 at 412 nm), prior to each experiment. [86] Expression and purification of metal-freeHk130 and human metallothionein-2 The production of P. furiosus Hk130 and human metallothionein-2 (MT2) relied on ap reviously established protocol using the pTYB21 expression vector (IMPACT Protein Purification System, NEB) and Escherichia coli (E. coli)B L21-CodonPlus (DE3)-RIL strain. [71] The expression vector is deposited in the Addgene plasmid repository (https://www.addgene.org), plasmid ID 105693 (MT2a). Transformed cells were cultivated in 4Lof LB or TB medium, respectively,a nd grown at 37 8Cu ntil OD 600 was 0.4-0.5, and then induced with 0.1 mm IPTG. Cultures were incubated overnight at 20 8Cw ith vigorous shaking and subsequently collected by centrifugation at 4500 g for 20 min at 4 8C. The pellets were resuspended in icecold buffer A( 20 mm HEPES, pH 8.0, 500 mm NaCl, 1mm EDTA, 1mm TCEP) and lysed by sonication on ice for 30 min, followed by centrifugation at 20 000 g for 15 min. Clarified cell extracts were incubated overnight with chitin resin at 4 8Cw ith mild shaking. After the incubation, the resin was washed with 20 bed volumes of buffer Aw ith increased salt concentration (1 m NaCl) to reduce nonspecific binding of other E. coli proteins. To induce the cleavage reaction, 25 mL of buffer B( 20 mm HEPES, pH 8.0, 500 mm NaCl, 1mm EDTA, 100 mm DTT) was added to the resin, and the mixture was incubated for 36-48 ha tr oom temperature with mild shaking. Eluted protein solutions were acidified to pH % 2.5-3.0 with 7% HCl and concentrated to as mall volume using Amicon Ultra-4 Centrifugal Filter Units with NMWL of 3kDa (Merck Millipore, USA). Hk130 protein was purified by reverse-phase HPLC in a 0.1 %T FA/acetonitrile gradient (Dionex) followed by lyophilization, and MT2 was purified on an SEC-70 gel filtration column (Bio-Rad, USA) equilibrated with 5mm HCl. [70] The identity of metal-free MT2 and Hk130 protein was confirmed by ESI-MS, using an API 2000 instrument (Applied Biosystems, USA);t he average molecular weight (MW) calculated was 6042.0/6042.2 Da (calcd/exp) and 15217.6/ 15217.8 Da (calcd/exp), respectively.

Reconstitution of cadmium metallothionein-2
To reconstitute the protein with Cd II ,a liquots of thionein in 5mm HCl were mixed with cadmium sulfate at am olar ratio of 1:9u nder an itrogen blanket, and the pH adjusted to 8.6 with 1 m solution of Tris base and purified on an SEC-70 gel filtration column equilibrated with 20 mm Tris-HCl buffer,pH8.6. The concentration of the pu-rified protein was obtained spectrophotometrically,w ith DTNB and PARa ssays for the thiol and Cd II concentration, respectively. [86,87] Additionally,s amples were analyzed by inductively coupled plasma (ICP) analysis (ICP-AES iCAP 7400, Thermo Scientific) to confirm the spectrophotometric results.

Spectroscopic studies
The binding properties of hook peptides to Cd II were examined by electronic absorption spectroscopy and circular dichroism by using aCary 300 spectrophotometer (Varian) and Jasco J-1500 spectropolarimeter (Jasco) with aPeltier heating/cooling system, respectively. Spectra were recorded at 25 8Ci nt he wavelength range of 200-280 nm to observe LMCT transitions [39,85] as well as secondary structure changes. [33] Three accumulations were averaged using a 5nmb and width, a2 00 nm min À1 scanning speed, and a1 .0 nm data pitch. Depending on Hk length, 25-100 mm peptide in 20 mm Tris-HCl buffer (pH 7.4, I = 0.1 m from NaClO 4 )w as titrated with small aliquots of concentrated CdSO 4 to achieve molar ratios from 0t o2over Hk. After each addition of Cd II ,s amples were equilibrated for 2-5 min and the spectra were recorded. All measurements were performed in the presence of fourfold excess of TCEP over Cys residue to avoid their oxidation. TCEP forms av ery weak Cd II complex (log K 7.4 = 3.3) compared with Hk peptides, and its metalbinding ability can be neglected. [36] Zn II -to-Cd II replacement experiments were conducted on either equimolar or fivefold excess (to slow down the exchange) of Zn II molar equivalent over Hk with Cd II titration as described previously.

Potentiometry
The protonation constants of the Hk4-Hk14 zinc hook peptides and stability constants of their Cd II complexes were determined at 25 8Ca t0 .1 m ionic strength (from KNO 3 )b yp H-metric titration over ar ange of 2.5 to 10.8, using aM olspin automatic titrator under argon with standardized 0.1 m NaOH as at itrant. The data were analyzed using SUPERQUAD software. [49] Fluorimetry and fluorescence anisotropys tudy Fluorimetric studies were conducted with aJ obin Yvon Fluoromax-3s pectrofluorimeter (Horiba) equipped with aP eltier-thermostatted cell holder.C ompetition intra-FRET analysis was carried out as follows:1m m concentration of the appropriate Zn II chelator (TPEN, EDTA, HEDTA) was used with various concentrations of ZnSO 4 (0.05-0.95 mm)i n5 0mm HEPES, 100 mm NaCl, and 50 mm TCEP at pH 7.4 to maintain the free Zn II concentration at ac onstant between subnano-and femtomolar levels. The concentration of zinc hook peptides in all studies was 5 mm.T he sets of peptides in metal buffers (1.4 mL) were equilibrated over 48 h. Complex formation was measured by using FRET between Trpa nd Dns residues located at both ends of the zinc hook peptides. For that purpose, samples were excited at 290 nm, and spectra were collected in the range of 295-600 nm with am aximum emission of 543 nm. Anisotropy decay studies were performed with aD eltaFlex TCSPC Fluorimeter (Horiba) equipped with aP eltier-thermostatted cell holder. All measurements were carried out at 25 8Ci nt he buffer described above with 500 nm N-terminally FAM-labeled hook peptides. FAMlabeled peptides (400 mL) were placed in a1mL all-transparent quartz cuvette and excited with al inearly polarized laser from Del-taDiode DD-485L (wavelength of (485 AE 10) nm) through a2nm slit. Emission data were detected at 521 nm wavelength with as equentially changing polarizer from 0t o9 0 8 until accumulation of 10 000 peak difference at a1 00 ns timescale was reached. Decay data were subsequently fitted with DAS6 software using two-or three-exponential fitting, depending on the Àlog [M II ] free and metalto-peptide equilibrium, as aV V + +VH sum, VVÀVH difference, and anisotropy.G iven the very small targets (Hk14-Hk45 peptides), only difference spectra from reconvolution anisotropy analyses were taken into account because this approach most precisely accounts for instrument-generated artifacts (Figures S14 and S15, Supporting Information). Zn II -to-Cd II exchange experiments carried out for Hk45 with equimolar content of Cd 7 MT2a were performed accordingly,t he only difference being faster equilibration times:3 0 min for swap experiments and 36 hf or competition experiments with metal chelators.

Competitive titrations
The apparent formation constants (K 12 )o fC d II biscomplexes with Hk14 and Hk45 peptides were determined fluorimetrically or spectropolarimetrically at 25 8Ci nt he presence of Cd II chelators HEDTA (log K 7.4 = 12.2), EDTA( log K 7.4 = 13.6), and TPEN (log K 7.4 = 15.2). [56,60] In the case of fluorescently labeled peptides, 5 mm Hk was incubated with 1mm chelator with various total Cd II concentrations in HEPES buffer with 50 mm TCEP.F or CD monitoring, 5 mm Hk peptide was incubated with 25 mm chelator and 0-22.5 mm ZnSO 4 in 20 mm Tris-HCl buffer (pH 7.4, I = 0.1 m from NaClO 4 )w ith 100 mm TCEP.S amples were incubated for 36 hu nder nitrogen. The free Cd II concentration present in each sample after equilibration was calculated from the total chelator and metal concentrations, corrected for the Cd II transferred to the Hk peptide during Cd(Hk) 2 complex formation. Calculations were performed using the Hyperquad Simulation and Speciation Software (HySS2009). [51] To obtain the apparent dissociation constants, we first determined the normalized isotherms corresponding to complex formation by fitting with Hill's equation. [33] The obtained concentrations of free Cd II ,r eferring to the half-point complex saturation for which half of the total peptide is in the form of the Cd(Hk) 2 complex and half is in the metal-free form Hk, were subsequently used to calculate the apparent dissociation constants (K 12 ).

NMR spectroscopy
All spectra were measured with aD DR2 Agilent 600 MHz spectrometer equipped with aT RIAX probe. Assignment of 1 Ha nd 13 C signals (Table S2, Supporting Information) was based on the previously described assignment of ap eptide complexed with Zn II . [33] 2D homonuclear TOCSY [88] (mixing time 65 ms), NOESY [89] (mixing time 150 ms), and heteronuclear 1 H-13 CH SQC [90] spectra recorded at 25 8Cw ere used to confirm the assignment (relevant parameters of the NMR spectra are given in Ta ble S3, Supporting Information). All chemical shifts in the 1 HNMR spectra are reported with respect to external sodium trimethylsilylpropanesulfonate ([D 4 ]DSS). Chemical shifts of the 13 Cs pectra were referenced indirectly by using the 0.251449530 frequency ratios for 13 C/ 1 H. [91] Processed spectra were analyzed with SPARKY software. [92] Isothermal titration calorimetry The binding of Cd II and Zn II to Hk peptides was monitored using a nano-ITC calorimeter (TAW aters, USA) at 25 8Cw ith ac ell volume of 1mL. All experiments were performed in HEPES buffer (I = 0.1 m from NaCl) at pH 7.4 with 0.25 mm TCEP used as an on-metal-binding reducing agent. [33,36] The Hk peptide (titrate) concentration was 0.1 mm,w hereas the metal (titrant) concentration was 0.65 mm. After temperature equilibration, successive injections of the titrant were made into the reaction cell with 6.82 mLi ncrements at 400 s intervals with stirring at 200 rpm, with the exception of the Zn II -to-Cd II exchange experiment in which the titrant (CdSO 4 )c oncentration was 1mm and the injection volume was 4.83 mL. Control experiments to determine the heats of titrant dilution were performed using identical injections in the absence of Zn II and Cd II . The net reaction heat was obtained by subtracting the heat of dilution from the corresponding total heat of reaction. The titration data were analyzed using NanoAnalyze (version 3.3.0) and Origin (version 8.1) software. [75] Structure calculation Determination of the structure of as ymmetric dimer faces the well-known problem of ambiguity of restraints. In principle each NOE cross-peak can be ar esult of either contacts between protons of the same chain (intra) or different chains (inter) or both intraand interchain contacts. Therefore, the structures of Hk14 peptide complexes with Cd II and Zn II ions were computed with Aria2 software, [93] which can resolve ambiguous restraints in an iterative procedure. The parametrization of the coordination site with Zn II ions was taken from the Aria2 program, with the exception of partial charges, which were computed using quantum chemistry methods applied to the model systems. For the complex with Cd II ,b ond lengths between sulfur and cadmium (set to 2.636 )w ere set to average distances obtained from the crystal structure, and partial charges were adjusted to match those obtained from quantum chemistry computations. For both peptides, partial charges were manually corrected so that the total charge of the coordination site (ion and four cysteine residues) is equal to À2 j e j .T he procedure of structure calculation consisted of nine simulated annealing cycles followed by refinement in an explicit solvent of 10 with the lowest energy out of 300 structures obtained from the last simulated annealing cycle. Each simulated annealing cycle consisted of 45 ps of high temperature dynamics (2000 K) followed by two cooling cycles:9 0pso fc ooling to 1000 Ka nd 72 ps of cooling to 50 K. Finally,2 00 steps of structure optimization with the Powell algorithm were applied. The problem of ambiguity of applied restraints (inter-or intramonomer) was solved by the iterative procedure of NOE assignment implemented in the Aria2 software. The C2s ymmetry of each complex was forced by application of special noncrystallographic symmetry restraints. Because of the ambiguity problem of NOE restraints, incorporation of additional hydrogenbond restraints was necessary.F or each complex, two types of structure determination procedures were applied. Both of them utilized NOE restraints and differed in the number of applied hydrogen-bond restraints. The first procedure (PROT) used hydrogenbond information obtained from experimentally determined protection factors, [33] whereas in the second procedure (CRYST) it was assumed that the same hydrogen bonds are formed as in the crystal structure (PDB code:1 L8D). All applied hydrogen-bond restraints are shown in Ta ble S4. Notably,t he CRYST method uses only one more hydrogen bond than the PROT method (backbone Gly3 O-Leu12 N). The average distance between two ends of the Hk14 peptide was estimated from 20 ms-long molecular dynamics simulation. The peptide in extended conformation was built with the xleap program from the AMBER18 suite. [94] Interactions between atoms were described with the ff14SBf orce field. [95] The peptide was immersed in at runcated octahedron box filled with TIP3P model explicit water molecules. The system was minimized, heated, and equilibrated in an NPT ensemble followed by a2 0mslong NVT ensemble production run. The distance between ends was estimated as the average distance between a-carbon atoms of the first and last amino acid residue. For error estimation, the trajectory was divided into four 5 ms-long trajectories, and mean dis-tances were calculated for each of them. The uncertainty was estimated as the standard deviation of mean values obtained from each window.S econdary structures were calculated for the whole trajectory with the DSSP method. [96]