This is the accepted version of the journal article: Sciortino, Giuseppe; Sanna, Daniele; Lubinu, Giuseppe; [et al.]. «Unveil- ing VIVO2+ Binding Modes to Human Serum Albumins by an Integrated Spectroscopic-Computational Approach». Chemistry, Vol. 26, Issue 49 (Septem- ber 2020), p. 11316-11326. DOI 10.1002/chem.202001492 This version is available at https://ddd.uab.cat/record/281916 under the terms of the license FULL PAPER Unveiling VIVO2+ binding modes to human serum albumins by an integrated spectroscopic-computational approach Giuseppe Sciortino,*[a,b] Daniele Sanna,*[c] Giuseppe Lubinu,[b] Jean-Didier Maréchal,[a] and Eugenio Garribba*[b] [a] Dr. G. Sciortino, Prof. J.-D. Maréchal xxx Ref. 1 xxx Ref. 2 xxx Miei commenti/aggiunte xxx Parti da controllare Department de Química Universitat Autònoma de Barcelona 08193 Cerdanyola del Vallés, Barcelona, Spain E-mail: giuseppe.sciortino@uab.cat [b] Prof. E. Garribba, Dr. G. LubinU Dipartimento di Chimica e Farmacia Università di Sassari Via Vienna 2, I-07100 Sassari, Italy E-mail: garribba@uniss.it [c] Dr. D. Sanna Istituto di Chimica Biomolecolare Consiglio Nazionale delle Ricerche Trav. La Crucca 3, I-07040 Sassari, Italy E-mail: daniele.sanna@cnr.it Supporting information and the ORCID identification number(s) for the author(s) of this article can be found under: https://doi.org/10.1002/chem.xxxx. Abstract: Human serum albumin (HSA) is involved in the transport preferential site for Cu2+ and Ni2+ ions with the coordination of of metal ions and potential metallodrugs. Depending on the metal, the terminal amino-NH2, the deprotonated amide-N– of residues several sites are available, among which N-terminal (NTS) and multi- 2 and 3 and imidazole-N of His3. The MBS is located at the metal binding site (MBS). Despite the large number of X-ray interface of subdomains IA and IIA and is the primary site for determinations for albumins, only one structure with Zn2+ is available. Zn2+ and other bivalent ions,[5] with the binding of His67, His247, In this work, the binding to HSA of VIVO2+ ion was studied by an Asn99 and Asp249. Furthermore, other two sites were identified: integrated approach based on spectroscopic and computational site B, where Cd2+ and also Co2+ are bound by one His and methods, which allowed to completely characterize the systems several carboxylate of Asp or Glu residues, [6], [8], [9], [10] and Cys34 even in the absence of an X-ray analysis. The behaviour depends on – the only free thiol of HSA not engaged in intramolecular the type of albumin, defatted (HSAd) or fatted (HSAf). With HSAd ‘primary’ and ‘secondary’ sites were revealed, NTS with (His3, His9, disulfide bridges – that binds Au and Pt compounds with Asp13, Asp255) and MBS with (His67, His247, Asp249, Asn99 or deprotonated -SH group.[11], [12], [13] H2O); with increasing the ratio VIVO2+/HSAd, ‘tertiary’ sites, with one The relationship between the binding of fatty acids (FAs) His-N and others donors (Asp/Glu-O or carbonyl-O) are populated. and some metal ions was suggested:[5, 14], [15] it is proved With HSAf, fatty acids (FAs) cause a rotation of the domains I and II, that Zn2+ coordination to the MBS can be hindered by the which results in the formation of a dinuclear ferromagnetic adduct binding of fatty acids at site FA2 (one of the sites with (VIVO)D2(HSAf) with a -Asp249 and the binding of His247, Glu102, highest affinity for FAs) inducing a rotation of the Glu252 and His67 or Asn99. FAs hinder also the binding of VIVO2+ to subdomains where the MBS is located and separating the the MBS. Zn2+ coordinating amino acids. Such conformational changes can influence the binding of other metal ions and, therefore, the transport properties of albumin. Introduction Among the most promising metals with pharmacological applications, VIVO2+ has shown very interesting anti-diabetic Human serum albumin (HSA) is the most abundant protein in and anti-tumoral properties.[16] After oral or intravenous the blood plasma. Together with human serum transferrin (hTf), administration, these vanadium-containing potential drugs it transports in the organism most of the essential metal ions and reach the bloodstream where VIVO2+ can be released, their compounds, among which metal based potential inorganic mainly at low drug concentrations found in the organism, drugs.[1] Moreover, HSA can bind toxic metals, fatty acids (FAs) and could interact – other than with hTf – with albumin. and endogenous and exogenous ligands. [2] Despite the large Therefore, the study of the binding of VIVO2+ with HSA number of X-ray determinations (XRD) of human and animal assumes great importance since it could affect the transport albumin (around 100 in the Protein Data Bank, PDB), only one of V complexes, and the amount and chemical form which structure with a metal ion, Zn2+, is available.[3] reach the target organs. The binding sites of certain essential (Zn 2+, Cu2+) or toxic (Ni2+, In the literature several papers were published on the Cd2+) metal ions have been identified, while less is known about interaction of VIVO2+ with HSA in its two forms, with or the binding of other metal ions.[4], [5], [6] HSA contains two main without FAs fatty acids (here designed as HSAf and sites: the N-terminal site (NTS or ATCUN motif),[4] and the multi- HSAd);[17] some proposals were advanced about the sites metal binding site (MBS or site A).[7] The NTS is present in most involved in the metal binding, but the systems are far from of all the albumins with a histidine in the third position such as being fully characterized. Chasteen and Francavilla HSA, bovine (BSA) and rat (RSA) serum albumin and is the measured the relative intensity of the EPR resonances of VIVO2+ aquaion and VIVO2+–albumin adducts as a function of 1 FULL PAPER VIVO2+/albumin ratio; with this procedure, they determined systems formed by a metal ion and a protein, for which an the number of the binding sites and their respective binding XRD is not available. constants, distinguishing one “strong” and at least five “weak” binding sites.[17a] Subsequently, these sites were named VBS1 and VBS2, respectively.[17i], [18] Some authors Results and Discussion suggested the involvement of the NTS in the metal coordination,[17i] while evidence based on competition VIVO2+ binding to defatted human serum albumin (HSA d) studies between Zn2+ and VIVO2+ indicated that VIVO2+ has EPR spectra were recorded on the system V IVO2+/HSAd as two types of binding sites and one of them may correspond a function of the molar ratio and pH. In these experiments to the MBS (VBS1).[19] Moreover, EPR spectra recorded on HSAd from Sigma-Aldrich with codes A1887 and A3782 an equimolar solution VIVO2+/HSAf suggested a triplet state were used. In Fig. 1 the spectra measured on an equimolar (S = 1), due to the ferromagnetic interaction of two metal solution containing A1887 are shown; all the full spectra ions in a dinuclear adduct designed as (VIVO)D2(HSA), together with the simulations with the software WINEPR where D stands for dinuclear;[17f], in contrast, when the ratio SimFonia[23] to extract their spin Hamiltonian parameters VIVO2+/HSAf was  2:1, the EPR resonances were attributed are reported in Figs. S1-S4 of the SI. In contrast with what to various “weak” sites in a multinuclear complex was previously reported by some of us on HSA containing (VIVO)xM(HSA) with S = 1/2, withand M means FAs fatty acids,[17f, 17h, 24] only the signals of mononuclear mononuclear, in which the VIVO2+ ions are not magnetically VIVO2+ complexes with S = 1/2 were detected. It can be interacting and are bound to different regions of protein.[17f], observed the formation of one species at pH 5.35 (indicated [17a], [19] On the basis of EPR intensity, Costa Pessoa and with IV in Fig. 1) and three additional adducts at pH > 6.40 coworkers affirmed that most of the paramagnetic V centers (I, II and III). The values of Az are 171.2 × 10–4 cm–1 for IV, were non-detectable even at VIVO2+/HSA ratio of 4/1 and 162.9 × 10–4 cm–1 for I, 166.4 × 10–4 cm–1 for II and 169.0 × concluded that the revealed VIVO complexes are minor 10–4 cm–1 for III. The spin Hamiltonian parameters are listed species in equilibrium with the dimer (VIVO)D2(HSA) and in Table S1 of the SI. The data can be explained postulating hydrolytic products of VIVO2+.[19] However, these that at acidic pH − when histidines are still protonated and conclusions were drawn considering equivalent the cannot interact with the metal ion − the equatorial behavior of HSAd and HSAf. Here, we will demonstrate that coordination positions are occupied by O-donors, for this latter statement is not correct and fatted and defatted example carboxylate groups of Asp and/or Glu residues, HSA differ for their metal binding properties. amidic groups of the backbone and Asn and/or Gln or H2O In the light of these data, several issues remain to be (IV). An increase of pH causes a spectral change due to the explained: i) the influence of the FAs fatty acids in the binding of stronger donors such as His-N (I, II and III) and formation of the metal adducts; ii) the nature of the “strong” these adducts become predominant at pH = 7.40. The first site (VBS1) and if this corresponds to the NTS or MBS; iii) two sites populated at physiological pH are I and II, with a the structure of the dinuclear adduct (VIVO)D2(HSAf); iv) the slight preference for I (named here 'primary' site) with vanadium coordination environment in the “weak” sites VBS2 in (VIVO)xM(HSA). In this paper, the systems VIVO2+/HSAd and VIVO2+/HSAf were studied to unveil these unclear pointsaspects and examined at a molecular level using an integrated spectroscopic and computational multiscale approach. EPR spectroscopy allows the recognition of the type of amino acid donors involved in the metal coordination (for example, His-N, Asp/Glu-COO, or carbonyl-CO, etc.) but not the specific residues nor the three-dimensional structure.[20] [21] Computational methods represent a valuable complementary tool to predict the binding sites and provide with their three dimensional models. Specifically, Density Functional Theory (DFT), Molecular and Quantum Mechanics (MM and QM) based methodologies as well as Molecular Dynamics (MD) and pattern recognition analysis techniques can be used to address the main steps during the metal–protein Figure 1. High field region of the X-band anisotropic EPR spectra recognition process.[22] recorded on frozen solutions (120 K) containing V IVO2+/HSAd at: a) The results allowed to fully characterize the behavior of the ratio 1/1, pH 5.35; b) ratio 1/1, pH 6.40; c) ratio 1/1, pH 7.40 and d) two systems and to clarify the number and properties of the ratio 4/1, pH 7.40. With I, II, III and IV the MI = 7/2 resonances of vanadium binding sites, even in the absence of an XRD. the sites NTS, MBS, ‘tertiary’ (sites B) and ‘quaternary’ are They could provide for useful information about the indicated. VIVO2+ concentration was 7.5 × 10–4 M for the 1/1 ratio transport of vanadium and vanadium compounds in the and 1.0 × 10–3 M for the 4/1 ratio. HSAd used was a Sigma-Aldrich organism. Finally, this approach may be applied to other product with code A1887. 2 FULL PAPER respect to II ('secondary' site). On the basis of the “additivity resonances for the sites I and II (trace b of Fig. 2), relationship”, the empirical rule verified for a lot of VIVO suggesting a slight competition between the two metals and complexes which allows estimating the 51V hyperfine that VIVO2+ remains bound to the 'primary' site I. Only coupling (HFC) constant along the z axis (Azestmtd) from the increasing the Zn2+/VIVO2+ ratio up to 2/1 and 4/1, sum of the contributions of each equatorial donor within ~3 appreciable changes were observed (traces c and d of Fig. × 10–4 cm–1 from the experimental value,[20] the Az's for I 2). In contrast, the addition of one equivalent of Ni 2+ is and II are compatible with two His-N plus a number of O- enough to remove from the 'primary' binding site VIVO2+ that donors stemming from carboxylates (Asp and/or Glu remains bound to the sites II and III (trace e of Fig. 2), residues), carbonyl or water. In contrast, for III, the weaker confirming the conclusions of Costa Pessoa and site or ‘tertiary’ site, the coordination of only one His is coworkers.[19] To further validate this hypothesis, similar expected. EPR experiments were performed using porcine serum The relative strength of the two stronger binding sites I and albumin (PSA) that presents Tyr3 instead of His3 as the II with respect to III was confirmed by the comparison of the third residue of the polypeptide chain, making the NTS not EPR spectra at physiological pH recorded as a function of available for the metal coordination and forcing the metal VIVO2+/HSAd ratio (Fig S5 of the SI): whereas the ions to bind to the MBS. The results on the system resonances I and II are detected when the spectrum is VIVO2+/PSA allowed us to do a very important observation: acquired at VIVO2+/HSAd 1/1 ratio, increasing the ratio up to the resonances I were lacking and only those indicated with 4/1 a broader band, composed by I and II plus the II were revealed (trace f of Fig. 2). When one equivalent of resonances of III, is observed (see also trace d of Fig. 1), zinc is added to this system (trace g of Fig. 2), Zn2+ is able suggesting that when I and II sites with two histidines bound to remove quantitatively VIVO2+ from binding site II, in to the metal are occupied, the additional VIVO2+ equivalents contrast with the behavior of HSAd for which there is no are distributed between the weaker ‘tertiary’ sites significant competition; in this system only the resonances supposedly characterized by only one His and O-donors. of the weak site III are detected. These data, taken To get more information on the VIVO2+ binding sites together, indicate that the ‘primary’ strong VIVO2+ binding location, EPR competition studies were carried out with site is located at the NTS of HSAd (resonances indicated Zn2+ and Ni2+ ions (Fig. 2). The full spectra are represented with I), while the ‘secondary’ site (II) could be found in the in Figs. S6-S12 of the SI. To interpret the data, it is interdomain IA and IIA region and is identified with the necessary to premise that in the defatted human serum MBS, both of the sites involving the coordination of at least albumin the MBS is preserved and that zinc binds two histidine side chains. Non-specific additional ‘weak’ specifically to this site, while nickel to the NTS motif. The binding sites, involving the coordination of one histidine and addition of one equivalent of Zn 2+ to a solution containing O-donors (resonances III), also exist in HSAd. Finally, the VIVO2+ and HSAd with a 1/1 molar ratio does not result in a resonances denoted with IV at pH < 6 can be attributed to the sites with coordination of only O-donors from Asp, Glu, Asn, Gln or from CO of the backbone. Starting from the spectroscopic information, the interaction with VIVO2+ and HSAd was examined with a multiscale computational approach: i) a complete model for the defatted albumin adding the last five N-terminal missing amino acids to the X-ray structure PDB 1AO6 of the free HSAd structure was generated; ii) the model was minimized and subsequently 100 ns of Molecular Dynamics (MD) were carried out to relax the new portion introduced (the RMSD plots of the trajectories were shown in Figs. S13 and S14 of the SI); iii) the MD trajectory was clustered and the representative structures of each pull of snapshots probed for zones in which at least two histidine residues featured - and -carbons within a range of distances from each queried grid point of 2.6-6.4 and 3.4-7.2 Å, respectively;[25] iv) the models that satisfied these criteria were further Figure 2. High field region of the X-band anisotropic EPR spectra evaluated with docking calculations;[26] v) docking solutions recorded at pH 7.4 on frozen solutions (120 K) containing: a) were refined at full Quantum Mechanics (QM) level of VIVO2+/HSAd 1/1; b) VIVO2+/Zn2+/HSAd 1/1/1; c) VIVO2+/Zn2+/HSAd 1/2/1; theory and this step was followed by Density Functional d) VIVO2+/Zn2+/HSAd 1/4/1; e) VIVO2+/Ni2+/HSAd 1/1/1; f) VIVO2+/PSA 1/1 Theory (DFT) simulations of the EPR HFC constants Az that and g) VIVO2+/Zn2+/PSA 1/1/1. With I, II and III the MI = 7/2 resonances of were compared with the experimental ones. the sites NTS, MBS and ‘tertiary’ (sites B) are indicated. VIVO2+ concentration was 7.5 × 10–4 M in all cases. HSAd used was a Sigma- Aldrich product with code A1887. significant change of the position and intensity of EPR 3 FULL PAPER Table 1 'Primary' (NTS) and 'secondary' (MBS) binding sites of HSAd for VIVO2+ ion refined by QM methods, experimental, estimated and calculated Az values, and electronic energies.. Site Donors V−N [a] V−O [a] V−OH2 [a] Azestmtd [b,c] Azcalcd [b,d] Ebinding [e] -165.0 [f] -163.7 [d] -31.7 NTS- His3, His9, Asp13, Asp255 2.236, 2.199 1.975, 2.080 – [f] [d] NTS- His3, His9, Glu6; H2O 2.126, 2.107 1.970 2.103 -168.8 -165.8 -18.4 NTS- His3, His9, Asp13; H2O 2.147, 2.150 1.995 2.071 -168.8 [f] -170.4 [d] -20.9 -170.5 [f] -175.3 [d] -10.2 NTS- His3, His9, His9-CO; H2O 2.162, 2.111 2.097 2.046 -168.8 [g] -165.2 [e] -26.2 MBS1 His67, His247, Asp249; H2O 2.108, 2.112 2.046 2.064 MBS2 His67, His247, Asp249, Asn99-CO 2.135, 2.154 1.964, 2.034 – -166.7 [g] -169.5 [e] -32.9 –4 –1 [a] Distance in Å. [b] Az measured in 10 cm units. [c] Az estimated with the "additivity relationship". [d] Az calculated with DFT methods. [e] Ebinding in kcal/mol. [f] Experimental value for the NTS site is -162.9 × 10–4 cm–1. [g] Experimental value for the MBS site is -166.4 × 10–4 cm–1. Only two of the ten total clusters were found to have the The affinity of the modes NTS-β and NTS- results lower necessary preorganization to bind the VIVO2+ ion with at more than 10 GoldScore units compared to NTS-α, while least two His side chains. The regions containing two the lowest scoring for the mode NTS- is coherent with the potential N-donors were: the interdomain MBS involving decrease in the donor strength replacing a carboxylate with His67 and His247 and the N-terminal site (NTS) with His6 a carbonyl donor. and His9 side chains, that are sufficiently close to each Concerning the interdomain region, the results showed two other. Docking calculations were performed in these potentials VIVO2+ binding modes at the MBS site (indicated regions considering flexible histidines and neighboring with MBS1 and MBS2 in Table 1) with the donor sets (His67- residues (in a radius of 5 Å). Both  and  histidine N, His247-N, Asp249-COO; H2O) and (His67-N, His247-N, tautomers were taken into account. Taking into account the Asn99-CO, Asp249-COO) (Fig. 4 with the QM refined EPR evidences the possibility to bind to the deprotonated structures). In these adducts, the fourth equatorial position ATCUN motif (NH2, N–, N–, His3-N) has not been taken into could be occupied by a water molecule or the amide account (vide infra). Preliminary dockings were performed oxygen of Asn99. To further validate the reliability of the activating all the four equatorial coordination positions in docking approach and to discriminate between the four order to examine the case as general as possible. ForIn the possible solutions at the NTS and two at the MBS region, solutions with the metal binding fromby three protein the most stable docking solutions were refined by QM donors, the first coordination sphere was completed adding cluster method proposed by Siegbahn and Himo [27] a water molecule and docking calculations were repeated to (structures represented in Figs. 3 and 4). The value of confirm the results using the moiety VIVO(H2O)2+, with three Ebinding for the sites NTS and MBS was calculated for activated equatorial positions instead of the ‘bare’ VIVO2+ as reactions 1 and 2: the ligand fragment for the protein. In the N-terminal region four binding modes – named NTS- [VO(H2O)4]2+ + Site ⇄ VO(Site) + 4 H2O (1) , NTS-, NTS- and NTS- – were identified with (His3-N, [VO(H2O)4]2+ + Site ⇄ VO(Site)(H2O) + 3 H2O (2) His9-N, Asp13-COO, Asp255-COO), (His3-N, His9-N, Asp13-COO; H2O), (His3-N, His9-N, Glu6-COO; H2O) and (His3-N, His9-N, His9-CO; H2O) coordination, respectively (in Fig. 3 the structures obtained after the QM refinement are represented). The docking assay (summarized in Table S2 of the SI) suggests that NTS-α is the most stable binding mode with Fmax and Fmean higher than 60 GoldScore units. Figure 3. QM refined binding modes for VIVO2+ ion at the N- Figure 4. QM refined binding modes for VIVO2+ ion at the MBS terminal region of HSAd: a) mode NTS-; b) mode NTS-; c) mode region of HSAd: a) MBS1 and b) MBS2. NTS- and d) mode NTS-. 4 FULL PAPER Table 2 'Tertiary' binding sites (sites B) of HSAd and HSAf for VIVO2+ ion determined by docking methods. Site Residues Coordination V−N [a] V−O [b] Fmax [c] Fmean [d] [e] 1 His367, Glu331; H2O, H2O N, CO; H2O, H2O 1.979 2.597 36.2 33.9 2 His510, Asp512; H2O, H2O [f] N, COO; H2O, H2O 2.193 2.273 30.5 27.8 3 His338, Asp340; H2O, H2O [e] N, COO; H2O, H2O 2.091 2.606 31.4 28.7 4 His535, Glu531, Asn503; H2O N, COO, CO; H2O 2.282 2.169, 2.249 41.0 40.6 [f] 5 His288, Glu153; H2O, H2O N, COO; H2O, H2O 2.352 2.174 36.4 34.9 6 His128, Glu131; H2O, H2O [e] N, COO; H2O, H2O 2.346 2.192 30.3 27.5 [a] Distance in Å. [b] Distance in Å (distance V–OH2 not reported). [c] GoldScore Fitness value obtained for the more stable pose of each cluster (Fmax). [d] Average value of GoldScore Fitness for each cluster (Fmean). [e] Water molecules trans to each other. [f] Water molecules cis to each other. To exclude any bias of the docking simulations, in As a result, the sites with the simultaneous coordination of molecular dynamics (MD) the metal bond forces and (His367-N, Glu313-CO), (His510-N, Asp512-COO), equilibrium parameters were calculated for the solution at (His338-N, Asp340-COO), (His535-N, Glu531-COO, MBS2, and the system submitted to additional 100 ns of MD Asn503-CO), (His288-N, Glu153-COO), and (His128-N, to relax the structure and exclude that the low scoring could Glu131-COO) – the last two not found for Zn 2+ ion[3] – were be originated from a non-correct starting orientation of the predicted and these could be responsible of the resonances amino acid side chains. indicated with III in Figs. 1 and 2. The values of docking The final values of Ebinding for the NTS are -31.7 kcal/mol parameters for these adducts are reported in Table 2 and (mode NTS-), -18.4 kcal/mol (NTS-), -20.9 kcal/mol their structures in Fig. S15 of the SI. These sites are (NTS-), -10.2 (NTS-), -26.2 kcal/mol (MBS1) and -32.9 reasonable candidates to represent could correspond to kcal/mol (MBS2), see Table 1. The results are in good those named B in the literature, with one His-N plus O- agreement with those based on pure scoring and donors, which are the preferred by Co II and MnII and are the population of GoldScore terms and indicate that in all the second choice for CdII.[5-6, 9-10, 15] On the basis of our results, six cases Ebinding is negative. The order of binding energy it is important to note that B is not unique in the HSA d but is MBS2  NTS- > MBS1 >> NTS-  NTS- > NTS-, comprises a number of similar binding sites that share the which means that the binding modes MBS 2 and NTS- will same chemical features. Remembering that Chasteen and be preferred compared to other ones. QM refinement Francavilla found five “weak” binding sites for VIVO2+ ion involve an improvement of the prediction of the angles and (named subsequently VBS2),[17a] it is reasonable to argue distances, which approach those expected for V– that they coincide with the six sites found by dockings and N(imidazole), V–O(carboxylate) and V–O(water) bonds listed in Table 2. (Table 1).[28] Overall, the possible sites predicted resemble Finally, the adducts with coordination of only O-donors were that of V-substituted carboxypeptidase, which has two His examined. Among them, those with the coordination of (one parallel and one perpendicular to V=O bond), one Glu three or two carboxylates, such as (Glu184, Asp187, and one H2O and is characterized by Az of 165.9 × 10−4 Glu188) and (Glu97-COO, Glu100-COO), the same site cm−1,[29] close to those experimentally measured for the found by Handing et al.[3] for Zn2+, are worth being resonances I (attributed to the NTS) and II (MBS). mentioned (Table S3 and Fig S16 of the SI). However, Furthermore, the Az values for the NTS and MBS sites other sites with carboxylate groups exist on the protein were calculated predicted by DFT methods – that today surface and these could also form adducts with the VIVO2+ allow to predict the HFC constants for paramagnetic metals, ion. To these sites, named ‘quaternary’ and occupied and vanadium in particular, with high accuracy [30], [31] – and before the deprotonation of histidine residues, can be compared with the experimental ones (Table 1). Overall, attributed the EPR signals IV at pH < 6 (trace a of Fig. 1). these data and those for Ebinding suggest that in the N- Overall, the results can be summarized as follows. At acid terminal region the most stable site is NTS-, while the two pH the VIVO2+ ion occupies the ‘quaternary’ sites with the best options for MBS (MBS1 and MBS2) differ for the coordination of two or three Asp-COO and/or Glu-COO. binding of CO group of Asn99 instead of a H2O molecule When pH is around the neutrality, after the deprotonation of and can be considered alternative. The HFC constants histidines, the binding mode changes and the ‘primary’, were also estimated with the “additivity relationship”, [20] ‘secondary’ and ‘tertiary’ sites become available. The using the contribution taken from ref. [32], and the values of ‘primary’ and ‘secondary’ sites are observed at low |Az|estmtd confirm these predictions(Table 1). VIVO2+/HSAd ratio (1/1) and correspond to the NTS with To complete the description of the binding of VIVO2+ to (His3, His9, Asp13, Asp255) coordination and MBS with HSAd, further docking calculations were carried out to find (His67, His247, Asp249) plus the fourth binding of Asn99- non-specific binding sites. Taking advantage by the recent CO or H2O ligand. Notably, these two sites are the same as paper by Handing et al.,[3] regions able to coordinate Zn2+ those determined by XRD for Zn 2+ [3] with the difference with at least one His and one O-donor were probed for that, for VIVO2+ and Zn2+, the stability order is inverted. binding with the VIVO(H2O)2+/VIVO(H2O)22+ moieties. These correspond to the sites named ‘strong’ by 5 FULL PAPER Chasteen,[17a] and VBS1 by Costa-Pessoa.[17i] The NTS is in High levels of FAs have been associated to an alteration of the highly disordered N-terminal region of the protein and the zinc homeostasis and the consequent increase of free was found for Zn2+ with equine serum albumin (ESA), while Zn2+ in the blood serum related to the neurodegeneration with HSA the anomalous signal observed during the X-ray and cellular death, to an increased risk of thrombosis and to diffraction study in proximity of His9 and Asp13 could the devolopment of various diseases, such as diabetes.[37] indicate the possible presence of the metal.[3] This explains The distribution of FAs between the binding sites of HSA why there is no competition between V and Zn at low molar depends on the FAs concentration in the blood and on their ratio: in fact, the two metals occupy both the MBS chain length and unsaturation degree. In normal conditions, (preferred by Zn2+) and NTS (slightly preferred by VIVO2+), 1 mol of human serum albumin transports from 0.1 to 2 mol respectively. When the molar ratio VIVO2+/HSAd is of FAsfatty acids,[34] and only at elevated FAs increased, then the ‘tertiary’ sites – formed by one His-N concentrations more than 2 mol are bound to HSA and all and another donor, such as carboxylate or carbonyl – the sites become occupied.[15, 38] Taking into account that a become populated (sites B). These are sites defined ‘weak’ significant fraction of FAs bind to FA4 and FA5 (i.e., those by Chasteen[17a] and VBS2 by Costa Pessoa.[17i] with the highest affinity together with FA2), part of the MBS It is noteworthy that VIVO2+, in contrast with Cu2+ and Ni2+ could remain unaltered and interact with the metal ions ion, does not interact with ATCUN motif since the donor set even in presence of FAs. Therefore, studying both the (Asp1-NH2, N–, N–, His3-N), where N– denotes the forms of albumin appears to be fundamental to completely deprotonated amide groups of residues 2 and 3, should understand the binding and transport of the metal ions. give spin Hamiltonian parameters (gz = 1.960 and Az = 153 Some years ago, in the system VIVO2+/HSAf we reported the × 10–4 cm–1 measured in the model system VIVO2+– formation of a dinuclear species with spin state S = 1, [33] GlyGlyHis ) significantly different from those (VIVO)D2(HSAf) and several hypotheses on its structure experimentally measured for HSA. The “additivity were proposed.[17f, 19] The signals of this adduct were relationship”,[20, 32] suggests |Az|estmtd of 151.4 × 10–4 cm–1, revealed for VIVO2+/HSAf ratio < 2 and are covered at higher very far from the values observed which are in the range ratios by the greater intensity of the resonances of (163-169) × 10–4 cm–1. Moreover, to induce the mononuclear species with S = 1/2 (Fig. S19 of the SI). The deprotonation of amide nitrogens in the tripeptide GlyGlyHis results obtained with HSAf from Sigma-Aldrich with code and reveal the (NH2, N–, N–, His-N), coordination, a ligand- A9511 were displayed also by HSAf with code A1653, to-metal molar ratio of 15/1 must be used and this binding indicating that the presence of long chain FAs determines mode can be clearly detected only at pH > 7.5.[33] Similar the different complexation scheme compared to HSAd. findings were also reached by other authors.[17e] We decided to give insight on the nature of the dinuclear To confirm the data obtained with HSA d A1887, the species using computational techniques. Several PDB experiments were repeated with albumin with code A3782. structures of HSA crystallized in presence of FA with It shows the same complexation scheme with binding at different chain length were analyzed, paying particular physiological pH to the 'primary' and 'secondary' NTS and attention to the relative position of subdomains IA and IIA MBS plus, increasing the VIVO2+/HSAd ratio, the interaction to find regions with at least two histidine residues able to with the 'tertiary' sites with only one His residue (Figs. S17 coordinate a metal ion. This preliminary analysis show that: and S18 of the SI). i) IA/IIA interdomain region does not present significant differences in the six structures analyzed from PDB (1E7I, 1E7F, 1E7E, 1BJ5, 1GNJ, 1GNI) and ii) the regions VIVO2+ binding to human serum albumin containing containing the His-N donors are the same as HSA d, except fatty acid (HSAf) for the MBS, where His247 and His67 are separated. As reported in the literature, the level of medium- and long- Docking calculations were performed in the interface of chain FAs influence the binding of zinc and it was proposed subdomains IA and IIA of the XRD structure crystallized in that they act as a regulatory mechanism of its presence of stearic acid (PDB 1E7I [35a]), considering concentration. Among the seven sites were identified in flexible the histidine side chains and neighboring residues HSA for medium- and long-chain FAs (from C10 to C18) (with a radius of 5 Å). The dinuclear structure prediction named FA1-FA7 (plus two supplementary FA8 for short- was carried out blindly docking a first VIVO2+/VIVO(H2O)2+ chain FAs and F9, available at a saturating FAs moiety on the IA and IIA interdomain region. This analysis concentration.[15] Among them, the sites FA2, FA4 and FA5 suggested three possible binding modes. The first docking exhibit the highest affinity.[34] The binding of FAs to the site assay was followed by subsequent dockings of an FA2, located at the interface of the subdomains IA and IIA, additional VIVO2+/VIVO(H2O)2+ to the three structures causes the rotation of these subdomains IA with respect to individuated in the first step. As a result, four geometrically the subdomain IIA,[35] and the fact that disruption of the site consistent dinuclear candidates were found (indicated with A MBS moving away the coordinating amino acids.[5] [14] [36] 1-4 in Table 3 and shown in Fig. 5). The dimers 1-4 were Therefore, the site A or separating the coordinating amino refined by full QM cluster method proposed by Siegbahn acid of MBS that become is no longer available for Zn2+ or, and Himo.[27] In all of them, Asp249 acts as the bridge; for in general, metal coordination after the binding of FAs. three structures the binding is 1,1 and is supported by a - OH– ion, while the fourth structure presents a 1,3 behavior 6 FULL PAPER Table 3. Structural and magnetic data for (VO)2D(HSAf) dinuclear observed for the mononuclear VIVO species with the same species. equatorial coordination mode.[40] If D has contribution only Dimer J [a,b] Magnetic coupling V···V [c] from dipolar interaction, the approximate value of the 1 19.88 Ferromagnetic 3.164 distance R between the two V centers can be derived with 2 6.99 Ferromagnetic 3.173 eq. 1:[41], [42] 3 0.13 Ferromagnetic 4.876 3  μ   g 2β 2   g  (1) D = π  0   3  = 1.39 104  3  4 -25.42 Antiferromagnetic 3.255 2  4π   R  R  [a] J value in cm–1. [b] J determined with the method reported in refs. [39]. [c] Distance between V atoms in Å. where D is measured in Gauss and R in Å. For the experimental mean value of g, R is 3.49 Å, in line with the calculated value for of Asp249. This site agrees well with the considerations dimers 1, 2 and 4 (even though this latter can be discarded proposed by Costa-Pessoa and co-workers, who suggested because antiferromagnetic). Summarizing the data, the structure that Glu252 might be involved in V coordination.[19] In all corresponding to the dimers 1 and 2 could be close to that the structures Glu100, His247, and Glu252 take part to the formed under physiological conditions for the adduct coordination, to which Asn99 may add; His67 binds V in the (VO)D2(HSAf). The structure of dimer 1 bound to the MBS site of dimers 1, 3 and 4 but not in 2. HSAf is depicted in Figure 6. The predicted models could rationalize the formation of When the ratio VIVO2+/HSAf is > 2, the resonances (VIVO)2D(HSAf) in the system with FAs fatty acids bound to attributable to sites with an only one VIVO2+ ion bound (S = HSA, in which the mutual rotation of the subdomains IA and 1/2) were revealed and these are superimposed to the low IIA results in a separation of His247 and Asp249 residues intensity absorptions of (VIVO)2D(HSAf). When the spectra from His67 and Asn99 – involved in the formation of the are recorded as a function of pH with molar ratio mononuclear adduct with HSAd – by 4-6 Å,[5, 7, 36a] which VIVO2+/HSAf = 4, the ‘quaternary’ sites with only O-donors favors the accommodation of two VIVO2+ ions and formation are observed at pH 5.00 (IV in Fig. S21 of the SI), while of the bridge represented by Asp249 (Fig. 5). The exchange with increasing pH the ‘primary’ site NTS and the various coupling constant J was calculated to determine the nature ‘ternary’ sites with one His-N-donor are revealed (I and III). of the magnetic interaction between the two unpaired Interestingly, the resonances corresponding to the electrons on VIV, ferro- or antiferromagnetic. The results ‘secondary’ site MBS were not detected (Fig. S21 of the SI) show that for three dinuclear structures (1-3 in Table 3) the due to the formation of the dimeric adduct (VIVO)2D(HSAf). coupling is ferromagnetic, in agreement with EPR data that In these species, the binding to the metal of Cys34 with the suggest a triplet state; in contrast, for the dimer 4 the –SH group can be excluded since – for the "additivity interaction is antiferromagnetic and, for this reason, it can relationship" [20] – the value of Az would be much smaller be excluded as a possible candidate for (VO)2D(HSAf). than those experimentally observed. The EPR spectrum of (VO)2D(HSAf) was simulated in this The effect of fatty acids on the complexation scheme of study with a zero-field splitting D of 605 × 10–4 cm–1 (649.8 VIVO2+ was investigated also with HSAf from Sigma-Aldrich Gauss) and gx,y,z = {1.999, 1.999, 1.985} and Ax,y,z = {45.0, with code A1653. Its behavior is identical as HSAf with code 45.0, 80.0} × 10–4 cm–1 (Fig. S20 of the SI). The HFC A9511, confirming the results just discussed (Fig. S22 of constant Az is one half of the value which would be the SI). Figure 6. DFT refined geometry of the dinuclear 1 VIVO2+ at the MBS site of HSAf. Subdomains IA and IIA are shown in blue and Figure 5. DFT optimized geometry of the dinuclear VIVO2+ adducts orange, respectively. at MBS for HSAf: a) dimer 1; b) dimer 2; c) dimer 3 and d) dimer 4. Hydrogen atoms and non-coordinating residues have been omitted for clarity. 7 FULL PAPER Conclusions from carboxylate of Asp or Glu, or from carbonyl groups of the backbone, and are occupied at acid pH. Following the The study of the metal binding to proteins has a nomenclature used in the literature, these could be named fundamental importance in biology, pharmacy and sites C. medicine. In fact, on one hand metal ions, in a free or The presence of FAs fatty acids in HSAf results in a complexed form, are part of the active site of proteins with different behavior and this fact must be taken into account several physiological functions (transport, storage, in the analysis of the transport of metal ions and metal enzymatic, etc.), on the other metal compounds show compounds. FAs Fatty acids cause a rotation of the pharmacological activity and are used or tested for the subdomains IA and IIA and the formation of (VIVO)2D(HSAf), treatment of various diseases, such as ulcer, arthritis, which was revealed by EPR but never rationalized up to diabetes and cancer. To characterize these systems, now,[17f, 19] in which a -Asp249 bridges the two VIVO2+ different instrumental techniques are available, but in most ions with a ferromagnetic coupling. With HSAf, the FAs fatty of cases – in the absence of X-ray determinations – the acids hinder the formation of mononuclear adducts at the information is only partial and the exact location of the MBS and, for this reason, influence the binding of metal binding sites, the identity of the donors involved in the metal ions that prefer this site (Zn2+ and Cd2+ beside VIVO2+), coordination and the complete three-dimensional which are forced to move to the available ‘primary’, ‘tertiary’ description cannot be obtained. To overcome this limitation and ‘quaternary’ sites. and complement the instrumental data, we developed a Finally, we believe that the integrated approach presented methodology to predict the coordination bond between a here could be applied not only to characterize the metal and a protein through dockings methods which can VIVO2+/HSA systems but also to determine the binding of be subsequently refined with QM/MM or QM calculations. [26] other metal species to other proteins, integrating the best Human serum albumin (HSA) is, with transferrin, the most instrumental technique (for example, EPR for a important protein of the blood serum involved in the paramagnetic and NMR for a diamagnetic ion) with transport of various essential and toxic metal ions and docking/QM calculations. Therefore, with this approach potential metallodrugs. Depending on the chemical complex systems such as (metal ion)–protein could be properties of metals, several sites are available, among completely characterized even in the absence of an X-ray which the N-terminal site (NTS) – with which Cu2+ and Ni2+ diffraction analysis. interact – and the multi-metal binding site (MBS or site A), preferred by Zn2+ and Cd2+, plus the not yet identified site B, the strongest for Co2+ and Mn2+.[15] The binding of VIVO2+ Experimental Section ion, which among the three oxidation states +III, +IV, +V revealed in biological systems, is the most stable form Chemicals. Water was deionized prior to use through the under physiological conditions and is probably released by purification system Millipore MilliQ Academic. Metal solutions were antidiabetic and anticancer vanadium drugs, is still elusive prepared from VOSO4·3H2O, ZnSO4·7H2O and NiSO4·7H2O. Metal and, for this reason, we applied an integrated approach salts and 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid based on EPR spectroscopy and docking/QM methods to (HEPES) were Sigma-Aldrich products of the highest grade study its interaction both with defatted (HSAd) or fatted available and used as received. Defatted and fatted human serum (HSAf) albumin. On the whole, the behavior and the binding albumin were purchased from Sigma-Aldrich with codes A1887 and sites of VIVO2+ are very similar to Zn2+, recently determined A3782 (HSAd), A9511 and A1653 (HSAf). Porcine serum albumin by an X-ray diffraction analysis.[3] (PSA) was a Sigma-Aldrich product with code A1830. For HSAd, four types of sites were predicted: i) NTS. This is the ‘primary’ site in the N-terminal region with (His3, His9, EPR measurements. The solutions for EPR measurements were Asp13, Asp255) coordination. ii) MBS or site A. The multi- prepared by dissolving VOSO4·3H2O in ultra-pure water in order to metal binding site (MBS) is the ‘secondary’ site for VIVO2+ get a VIVO2+ concentration in the range (0.75-1.31) × 10-3 M, ion with (His67, His247, Asp249) coordination plus the depending on the ratio VIVO2+/albumin. Argon was bubbled through weak binding of Asn99-CO or H2O ligand. NTS and MBS the solutions to ensure the absence of oxygen and avoid the have a comparable stability with a slight preference for the oxidation of VIVO2+ ion. To the solutions, HEPES buffer of 1.00 × NTS and coincide with those named 'strong' by 10–1 M concentration was added. The value of pH was kept around Chasteen[17a] and VBS1 by Costa-Pessoa.[17i] iii) Sites B. 4.0-4.5 to avoid hydrolysis and, to 1 mL of the solutions with the The sites B are not specific and weaker than NTS and VIVO2+, PSA, HSAf or HSAd were added to obtain the desired MBS, with one His-N and another O-donor such as VIVO2+/albumin ratio. Subsequently pH was raised around 7.4 and Asp/Glu-O or carbonyl-O, populated with increasing the EPR spectra were immediately recorded at 120 K with an X-band VIVO2+/ratio; for the first time, the exact position of such Bruker EMX spectrometer equipped with a HP 53150A microwave sites and the amino acids involved in the interaction have frequency counter. The microwave frequency was in the range been determined. These were proposed previously but not 9.40-9.42 GHz, microwave power was 20 mW, time constant was located,[5-6, 9-10, 15] and for VIVO2+ were named 'weak' by 81.92 ms, modulation frequency 100 kHz, modulation amplitude Chasteen[17a] and VBS2 by Costa-Pessoa.[17i] iv) Sites C. 0.4-1.0 mT, resolution 4096 points. When the samples were These are ‘quaternary’ sites with only O-donors, stemming transferred into the EPR tubes, the spectra were immediately 8 FULL PAPER measured. Signal averaging was used to increase the signal to associated with each pose and ii) the population of the cluster noise ratio.[17f] containing the best pose. The spectra were simulated (Figures S1-S4 and S20 of the SI) with The refinement of the VIVO2+–HSAd and VIVO2+–HSAf adducts found [23] the software WINEPR SimFonia. by dockings was performed cutting out the region with the V IVO2+ ion, and neighbor interacting amino acid side chains and freezing DFT, MD and docking Calculations. The VIVO2+ ion, modeled as the backbone atoms as reported by Siegbahn and Himo.[27] The IV 2+ [V O(H2O)4] , and all the adducts were optimized with Gaussian resulting clusters contain 90 (site NTS-), 70 (NTS-), 45 (NTS-), 09 (revision D.01)[43] at B3P86/6-311++g(d,p) level of theory using 49 (NTS-), 57 (MBS1) and 54 (MBS2) atoms, and 101, 118, 101 the SMD model[44] for water. Frequency calculations were computed 119 atoms for the dimers 1-4. The geometry relaxation and E to ensure that the structure was a minimum in the potential energy calculations were performed at B3P86/6-311++g(d,p) level of theory surface. within the framework of SMD model[44] for water. [45] The model of the defatted albumin were built with Modeller, adding The details of the docking protocol, the atoms parameter libraries for the last five N-terminal missing amino acids to the X-ray structure 1AO6 VIVO2+ atom type, the vanadium “coordination” energetic terms d of the unbound HSA . Molecular Dynamics (MD) simulations were set implemented in the GoldScore scoring function, examples of input up with Xleap,[46] which was instructed to solvate the protein with a and output GOLD files, the force constants for vanadium and cubic box of pre-equilibrated TIP3P water molecules and balance equilibrium parameters derived with Seminario methods, and – the total charge with Cl− ions (ions94.lib library). The AMBER99SB finally – Cartesian coordinates of the DFT optimized structures force field[47] was used for the standard residues, while V-bonding represented in Figures 3-5 are provided in the SI. force constants and equilibrium parameters were obtained through The prediction of the 51V HFC constants (A) was performed the Seminario method, using the MCPB.py. [48] During MDs solvent Gaussian 09 (revision D.01)[43] on the optimized structures at and the whole system were sequentially submitted to 3000 energy BHandHLYP/6-311+g(d) level of theory, according to the minimization steps, then, thermalization of water molecules and procedures previously published.[56] The theory background can be [31] side chains was achieved by increasing the temperature from 100 K found by the interested readers in refs. . The exchange coupling up to 300 K. MDs under periodic boundary conditions were carried constant J for the dinuclear adduct (VO)2D(HSAf) was calculated out during 100 ns with OpenMM engine [49] through OMMProtocol.[50] with the functional B3LYP and 6-311g basis set with the software Analysis of the trajectories was carried out by means of cpptraj ORCA,[57] according to the method reported in the literature.[58] implemented in ambertools16. [46] The implemented clustering Using S1 = S1 = 1/2 in the Heisenberg Hamiltonian Hˆ = − JSˆ1  Sˆ2 the algorithm was used to generate ten clusters taking the RMSD value of J can be expressed as: J = EBS − EHS , where EBS and EHS between frames as distance metric. The representative structure of are the energies of the broken-symmetry solution and triplet state. the most populated cluster was used in the further structural analysis in which each structure was probed for the zones with at least two potential coordinating residues (Asp, Glu, Asn, Gln and Acknowledgements His) using multisite.py.[25] Docking calculations to the HSAd were carried out through GOLD E.G., G.S., D.S., and G.L. thank Regione Autonoma della 5.8 software[51] on the selected frames, while for the analysis of Sardegna (grant RASSR79857) and Università di Sassari HSAf the XRD structure of albumin crystallized in presence of steric acid (fondo di Ateneo per la ricerca 2019) for financial support. (PDB 1E7I [35a]) was used. The PDB structure was cleaned removing G.S. and J.-D.M. thank also Spanish MINECO (grant all the small molecules and crystallographic waters, and hydrogen CTQ2017-87889-P) and Generalitat de Catalunya atoms were added with the UCSF Chimera program. [52] The (2017SGR1323). proteins were docked with VIVO2+, VIVO(H2O)2+, and VIVO(H2O)22+ moieties, obtained from the optimization of the aquaion [VIVO(H2O)4]2+. The equatorial positions were activated replacing Conflicy of interest the equatorial water(s) with dummy hydrogen atoms according to what was recently established.[26, 53] All dockings were computed The authors declare no conflict of interest. considering both the protonation states at  and  nitrogens of His imidazole ring. The docking simulations were carried out constructing in the region of interest an evaluation sphere of 12 and Keywords: oxidovanadium(IV) ion • human serum albumin • 5 Å for HSAd and HSAf respectively, taking into account side chains fatty acids • docking and QM calculations • Electron flexibility using the GOLD implemented rotamers libraries. [54] Paramagnetic Resonance Genetic algorithm (GA) parameters have been set to 50 GA runs and a minimum of 100,000 operations. The other parameters of GA were set to default. The solutions were analyzed by means of References Controllare abbrev. GaudiView.[55] The scoring (Fitness of GoldScore) was evaluated applying the [1] W. Bal, M. Sokołowska, E. Kurowska, P. Faller, Biochim. Biophys. modified version of GoldScore scoring function, which was Acta, Gen. Subj. 2013, 1830, 5444-5455. validated in previously published papers.[26, 53] The best solutions (binding poses) were evaluated through i) the mean (Fmean) and the [2] M. Fasano, S. Curry, E. Terreno, M. Galliano, G. Fanali, P. highest value (Fmax) of the scoring (Fitness of GoldScore) Narciso, S. Notari, P. Ascenzi, IUBMB Life 2005, 57, 787-796. 9 FULL PAPER [3] K. B. Handing, I. G. Shabalin, O. Kassaar, S. Khazaipoul, C. A. Enyedi, T. Kiss, J. Costa Pessoa, J. Mex. Chem. Soc. 2013, 57, Blindauer, A. J. Stewart, M. Chruszcz, W. Minor, Chem. Sci. 2016, 7, 180-191. 6635-6648. [18] J. Costa Pessoa, E. Garribba, M. F. A. Santos, T. Santos-Silva, [4] C. Harford, B. Sarkar, Acc. Chem. Res. 1997, 30, 123-130. Coord. Chem. Rev. 2015, 301–302, 49-86. [5] A. J. Stewart, C. A. Blindauer, S. Berezenko, D. Sleep, P. J. [19] I. Correia, T. Jakusch, E. Cobbinna, S. Mehtab, I. Tomaz, N. V. Sadler, Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 3701-3706. Nagy, A. Rockenbauer, J. Costa Pessoa, T. Kiss, Dalton Trans. 2012, 41, 6477-6487. [6] P. J. Sadler, J. H. Viles, Inorg. Chem. 1996, 35, 4490-4496. [20] a) D. N. Chasteen in Biological Magnetic Resonance, Vol. 3 [7] W. Bal, J. Christodoulou, P. J. Sadler, A. Tucker, J. Inorg. (Eds.: L. J. J. Berliner, J. Reuben), Plenum Press, New York, 1981, Biochem. 1998, 70, 33-39. pp. 53-119; b) T. S. Smith II, R. LoBrutto, V. L. Pecoraro, Coord. [8] E. O. Martins, T. Drakenberg, Inorg. Chim. Acta 1982, 67, 71-74. Chem. Rev. 2002, 228, 1-18. [9] E. Mothes, P. Faller, Biochemistry 2007, 46, 2267-2274. [21] a) B. D. Liboiron in High Resolution EPR: Applications to Metalloenzymes and Metals in Medicine, Vol. 28 (Eds.: L. Berliner, G. [10] M. Sokołowska, M. Wszelaka-Rylik, J. Poznański, W. Bal, J. Hanson), Springer, New York, 2009, pp. 507-549; b) S. C. Larsen, N. Inorg. Biochem. 2009, 103, 1005-1013. D. Chasteen in Metals in Biology: Applications of High-Resolution [11] J. Christodoulou, P. J. Sadler, A. Tucker, Eur. J. Biochem. 1994, EPR to Metalloenzymes, Vol. 29 (Eds.: G. Hanson, L. Berliner), 225, 363-368. Springer, New York, 2010, pp. 371-409. [12] A. I. Ivanov, J. Christodoulou, J. A. Parkinson, K. J. Barnham, A. [22] a) F. Neese, JBIC Journal of Biological Inorganic Chemistry Tucker, J. Woodrow, P. J. Sadler, J. Biol. Chem. 1998, 273, 14721- 2006, 11, 702-711; b) F. Neese, Coord. Chem. Rev. 2009, 253, 526- 14730. 563; c) M. Medhavi, V. Ambarish Sharan, Shankaracharya, Curr. Bioinf. 2011, 6, 444-449; d) S. F. Sousa, A. J. M. Ribeiro, J. T. S. [13] B. P. Espósito, R. Najjar, Coord. Chem. Rev. 2002, 232, 137- Coimbra, R. P. P. Neves, S. A. Martins, N. S. H. N. Moorthy, P. A. 149. Fernandes, M. J. Ramos, Curr. Med. Chem. 2013, 20, 2296-2314; e) [14] J. P. C. Coverdale, J. P. Barnett, A. H. Adamu, E. J. Griffiths, A. I. A. Guedes, C. S. de Magalhães, L. E. Dardenne, Biophys. Rev. J. Stewart, C. A. Blindauer, Metallomics 2019, 11, 1805-1819. 2014, 6, 75-87; f) P. Chaskar, V. Zoete, U. F. Röhrig, J. Chem. Inf. Model. 2014, 54, 3137-3152; g) P. Vidossich, A. Magistrato, [15] S. Al-Harthi, J. I. Lachowicz, M. E. Nowakowski, M. Jaremko, Ł. Biomolecules 2014, 4, 616-645; h) P. Li, L. F. Song, K. M. Merz, J. Jaremko, J. Inorg. Biochem. 2019, 198, 110716. Phys. Chem. B 2015, 119, 883-895; i) V. Muñoz Robles, E. Ortega- [16] a) J. Costa Pessoa, S. Etcheverry, D. Gambino, Coord. Chem. Carrasco, L. Alonso-Cotchico, J. Rodriguez-Guerra, A. Lledós, J.-D. Rev. 2015, 301-302, 24-48; b) E. Kioseoglou, S. Petanidis, C. Maréchal, ACS Catalysis 2015, 5, 2469-2480; j) Z. Wang, H. Sun, X. Gabriel, A. Salifoglou, Coord. Chem. Rev. 2015, 301-302, 87-105; c) Yao, D. Li, L. Xu, Y. Li, S. Tian, T. Hou, Phys. Chem. Chem. Phys. D. Rehder, Future Med. Chem. 2016, 8, 325-338; d) I. E. Leon, J. F. 2016, 18, 12964-12975; k) M. Akhter, JSM Chem. 2016, 4, 1025; l) Cadavid-Vargas, A. L. Di Virgilio, S. B. Etcheverry, Curr. Med. Chem. S. Agarwal, R. Mehrotra, JSM Chem. 2016, 4, 1024; m) N. S. 2017, 24, 112-148; e) D. Crans, C, L. Yang, A. Haase, X. Yang in Pagadala, K. Syed, J. Tuszynski, Biophys. Rev. 2017, 9, 91-102; n) Metallo-Drugs: Development and Action of Anticancer Agents, Vol. Gunseli B. Akcapinar, Osman U. Sezerman, Biosci. Rep. 2017, 37; 18 (Eds.: A. Sigel, H. Sigel, E. Freisinger, R. K. O. Sigel), De Gruyter o) P. Li, K. M. Merz, Chem. Rev. 2017, 117, 1564-1686; p) L. GmbH, Berlin, 2018, pp. 251-280; f) D. Rehder, ChemTexts 2018, 4, Riccardi, V. Genna, M. De Vivo, Nature Reviews Chemistry 2018, 2, 20; g) D. C. Crans, L. Henry, G. Cardiff, B. I. Posner in Essential 100-112; q) L. Alonso-Cotchico, J. Rodrı́guez-Guerra, A. Lledós, J.- Metals in Medicine: Therapeutic Use and Toxicity of Metal Ions in D. Maréchal, Acc. Chem. Res. 2020, 53, 896-905. the Clinic, Vol. 19 (Eds.: P. L. Carver), De Gruyter GmbH, Berlin, [23] WINEPR SimFonia, version 1.25, Bruker Analytische 2019, pp. 203-230. Messtechnik GmbH, Karlshrue, 1996. [17] a) N. D. Chasteen, J. Francavilla, J. Phys. Chem. 1976, 80, 867- [24] D. Sanna, G. Micera, E. Garribba, Inorg. Chem. 2010, 49, 174- 871; b) K. Fukui, H. Ohya-Nishiguchi, M. Nakai, H. Sakurai, H. 187. Kamada, FEBS Lett. 1995, 368, 31-35; c) G. R. Willsky, A. B. Goldfine, P. J. Kostyniak, J. H. McNeill, L. Q. Yang, H. R. Khan, D. C. [25] G. Sciortino, E. Garribba, J. R. G. Pedregal, J. D. Marechal, Crans, J. Inorg. Biochem. 2001, 85, 33-42; d) M. Purcell, J. F. Neault, ACS Omega 2019, 4, 3726-3731. H. Malonga, H. Arakawa, H. A. Tajmir-Riahi, Can. J. Chem. 2001, 79, [26] a) G. Sciortino, J. Rodríguez-Guerra Pedregal, A. Lledós, E. 1415-1421; e) B. D. Liboiron, K. H. Thompson, G. R. Hanson, E. Garribba, J.-D. Maréchal, J. Comput. Chem. 2018, 39, 42-51; b) G. Lam, N. Aebischer, C. Orvig, J. Am. Chem. Soc. 2005, 127, 5104- Sciortino, E. Garribba, J.-D. Maréchal, Inorg. Chem. 2019, 58, 294- 5115; f) D. Sanna, E. Garribba, G. Micera, J. Inorg. Biochem. 2009, 306. 103, 648-655; g) D. Sanna, G. Micera, E. Garribba, Inorg. Chem. 2009, 48, 5747-5757; h) D. Sanna, P. Buglyo, G. Micera, E. [27] P. E. M. Siegbahn, F. Himo, Wiley Interdiscip. Rev. Comput. Mol. Garribba, J. Biol. Inorg. Chem. 2010, 15, 825-839; i) E. Cobbina, S. Sci. 2011, 1, 323-336. Mehtab, I. Correia, G. Gonçalves, I. Tomaz, I. Cavaco, T. Jakusch, E. 10 FULL PAPER [28] L. F. Vilas Boas, J. Costa Pessoa in Comprehensive [43] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Coordination Chemistry, Vol. 3 (Eds.: G. Wilkinson, R. D. Gillard, J. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. A. McCleverty), Pergamon Press, Oxford, 1985, pp. 453-583. Petersson, H. Nakatsuji, M. L. Caricato, X., H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, [29] R. J. DeKoch, D. J. West, J. C. Cannon, N. D. Chasteen, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Biochemistry 1974, 13, 4347-4354. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. [30] a) M. L. Munzarová, M. Kaupp, J. Phys. Chem. B 2001, 105, Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, 12644-12652; b) F. Neese, J. Chem. Phys. 2003, 118, 3939-3948; c) V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. A. C. Saladino, S. C. Larsen, J. Phys. Chem. A 2003, 107, 1872- Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. 1878; d) A. C. Saladino, S. C. Larsen, Catal. Today 2005, 105, 122- Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. 133; e) C. P. Aznar, Y. Deligiannakis, E. J. Tolis, T. Kabanos, M. Bakken, C. J. Adamo, J., R. Gomperts, R. E. Stratmann, O. Yazyev, Brynda, R. D. Britt, J. Phys. Chem. A 2004, 108, 4310-4321. A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. [31] a) G. Micera, E. Garribba, J. Comput. Chem. 2011, 32, 2822- Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, 2835; b) D. Sanna, G. Sciortino, V. Ugone, G. Micera, E. Garribba, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian 09, revision D.01, Inorg. Chem. 2016, 55, 7373-7387. Gaussian, Inc., Wallingford, CT, 2010. [32] E. Garribba, E. Lodyga-Chruscinska, G. Micera, A. Panzanelli, [44] A. V. Marenich, C. J. Cramer, D. G. Truhlar, J. Phys. Chem. B D. Sanna, Eur. J. Inorg. Chem. 2005, 1369-1382. 2009, 113, 6378-6396. [33] E. Garribba, G. Micera, E. Lodyga-Chruscinska, D. Sanna, G. [45] A. Šali, T. L. Blundell, J. Mol. Biol. 1993, 234, 779-815. Sanna, Eur. J. Inorg. Chem. 2005, 4953-4963. [46] D. A. Case, R. M. Betz, D. S. Cerutti, T. E. CheathamIII, T. A. [34] J. R. Simard, P. A. Zunszain, J. A. Hamilton, S. Curry, J. Mol. Darden, R. E. Duke, T. J. Giese, H. Gohlke, A. W. Goetz, N. Biol. 2006, 361, 336-351. Homeyer, S. Izadi, P. Janowski, J. Kaus, A. Kovalenko, T. S. Lee, S. [35] a) A. A. Bhattacharya, T. Grüne, S. Curry, J. Mol. Biol. 2000, LeGrand, P. Li, C.Lin, T. Luchko, R. Luo, B. Madej, D. Mermelstein, 303, 721-732; b) S. Curry, P. Brick, N. P. Franks, Biochimica et K. M. Merz, G. Monard, H. Nguyen, H. T. Nguyen, I. Omelyan, A. Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1999, Onufriev, D. R. Roe, A. Roitberg, C. Sagui, C. L. Simmerling, W. M. 1441, 131-140. Botello-Smith, J. Swails, R. C. Walker, J. Wang, R. M. Wolf, X. Wu, L. Xiao, P. A. Kollman, AMBER 2016, University of California, San [36] a) J. Lu, A. J. Stewart, P. J. Sadler, T. J. T. Pinheiro, C. A. Francisco, CA, 2016. Blindauer, Biochem. Soc. Trans. 2008, 36, 1317–1321; b) C. A. Blindauer, I. Harvey, K. E. Bunyan, A. J. Stewart, D. Sleep, D. J. [47] V. Hornak, R. Abel, A. Okur, B. Strockbine, A. Roitberg, C. Harrison, S. Berezenko, P. J. Sadler, J. Biol. Chem. 2009, 284, Simmerling, Proteins: Struct., Funct., Bioinf. 2006, 65, 712-725. 23116-23124; c) J. P. Barnett, C. A. Blindauer, O. Kassaar, S. [48] P. Li, K. M. Merz, J. Chem. Inf. Model. 2016, 56, 599-604. Khazaipoul, E. M. Martin, P. J. Sadler, A. J. Stewart, Biochimica et Biophysica Acta (BBA) - General Subjects 2013, 1830, 5456-5464; [49] P. Eastman, J. Swails, J. D. Chodera, R. T. McGibbon, Y. Zhao, d) J. P. C. Coverdale, S. Khazaipoul, S. Arya, A. J. Stewart, C. A. K. A. Beauchamp, L.-P. Wang, A. C. Simmonett, M. P. Harrigan, C. Blindauer, Biochimica et Biophysica Acta (BBA) - Molecular and Cell D. Stern, R. P. Wiewiora, B. R. Brooks, V. S. Pande, PLoS Comput. Biology of Lipids 2019, 1864, 532-542. Biol. 2017, 13, e1005659. [37] a) D. De Mel, C. Suphioglu, Nutrients 2014, 6, 3245-3258; b) D. [50] J. Rodríguez-Guerra Pedregal, L. Alonso-Cotchico, L. Velasco- C. Bode, H. F. Stanyon, T. Hirani, M. D. Baker, J. Nield, J. H. Viles, J. Carneros, J.-D. Maréchal, OMMProtocol: A Command Line Mol. Biol. 2018, 430, 919-934; c) O. Kassaar, U. Schwarz-Linek, C. Application to Launch Molecular Dynamics Simulations with A. Blindauer, A. J. Stewart, Journal of Thrombosis and Haemostasis OpenMM, 2018. 2015, 13, 101-110; d) J. Jansen, W. Karges, L. Rink, The Journal of [51] G. Jones, P. Willett, R. C. Glen, A. R. Leach, R. Taylor, J. Mol. Nutritional Biochemistry 2009, 20, 399-417; e) M. Soinio, J. Biol. 1997, 267, 727-748. Marniemi, M. Laakso, K. Pyörälä, S. Lehto, T. Rönnemaa, Diabetes Care 2007, 30, 523-528. [52] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt, E. C. Meng, T. E. Ferrin, J. Comput. Chem. 2004, 25, [38] G. V. Richieri, A. M. Kleinfeld, J. Lipid Res. 1995, 36, 229-240. 1605-1612. [39] a) A. P. Ginsberg, J. Am. Chem. Soc. 1980, 102, 111-117; b) L. [53] a) G. Sciortino, D. Sanna, V. Ugone, G. Micera, A. Lledós, J.-D. Noodleman, J. Chem. Phys. 1981, 74, 5737-5743. Maréchal, E. Garribba, Inorg. Chem. 2017, 56, 12938-12951; b) G. [40] T. D. Smith, J. R. Pilbrow, Coord. Chem. Rev. 1974, 13, 173- Sciortino, D. Sanna, V. Ugone, A. Lledós, J.-D. Maréchal, E. 278. Garribba, Inorg. Chem. 2018, 57, 4456-4469; c) G. Sciortino, D. Sanna, V. Ugone, J. D. Marechal, E. Garribba, Inorg. Chem. Front. [41] S. S. Eaton, K. M. More, B. M. Sawant, G. R. Eaton, J. Am. 2019, 6, 1561-1578; d) G. Sciortino, D. Sanna, V. Ugone, J.-D. Chem. Soc. 1983, 105, 6560-6567. Maréchal, M. Alemany-Chavarria, E. Garribba, New J. Chem. 2019, [42] M. Yoshizawa, K. Ono, K. Kumazawa, T. Kato, M. Fujita, J. Am. 43, 17647-17660; e) V. Ugone, D. Sanna, G. Sciortino, J. D. Chem. Soc. 2005, 127, 10800-10801. Marechal, E. Garribba, Inorg. Chem. 2019, 58, 8064-8078. 11 FULL PAPER [54] S. C. Lovell, J. M. Word, J. S. Richardson, D. C. Richardson, Proteins: Struct., Funct., Bioinf. 2000, 40, 389-408. [55] a) J. Rodríguez-Guerra, Insilichem/gaudiview: Pre-alpha public releas, Zenodo, 2017; b) J. Rodríguez-Guerra Pedregal, G. Sciortino, J. Guasp, M. Municoy, J.-D. Maréchal, J. Comput. Chem. 2017, 38, 2118-2126. [56] a) S. Gorelsky, G. Micera, E. Garribba, Chem.–Eur. J. 2010, 16, 8167-8180; b) D. Sanna, V. L. Pecoraro, G. Micera, E. Garribba, J. Biol. Inorg. Chem. 2012, 17, 773-790. [57] a) F. Neese, ORCA – An Ab Initio, DFT and Semiempirical Program Package, Version 4.0, Max-Planck-Institute for Chemical Energy Conversion, Mülheim a. d. Ruhr, 2017; b) F. Neese, Wiley Interdiscip. Rev. Comput. Mol. Sci. 2017, 8, e1327. [58] E. Ruiz, J. Cano, S. Alvarez, P. Alemany, J. Comput. Chem. 1999, 20, 1391-1400. 12 FULL PAPER Entry for the Table of Contents An integrated approach based on spectroscopic and computational techniques was applied to unveil VIVO2+ binding modes to fatted and defatted human serum albumin. N-terminal (NTS) and multi-metal binding site (MBS) plus 'tertiary' (sites B) and ‘quaternary’ weak sites (sites C) were revealed. The influence of fatty acids in the formation of the adducts and the nature of the "strong" and "weak" sites were also explained. 13 FULL PAPER 14