Tuning the interactions of decavanadate with thaumatin, lysozyme, proteinase K and human serum proteins by its coordination to a pentaaquacobalt( II ) complex cation †

The decavanadate anion, H x V 10 O 28(6 (cid:2) ) (cid:2) ( V 10 ), is one of the most studied vanadium polyoxometalate species. In recent decades several works have pointed to its biological relevance coming mainly from its ability to bind to proteins (such as actin, myosin or ion pumps). On the other hand, non-functional binding was observed in several protein crystal structures, where V 10 was incorporated ‘‘accidentally’’ resulting from the presence of Na 3 VO 4 as a phosphatase inhibitor. In this work we broaden the potential biological applications of V 10 by presenting the synthesis and characterization of two decavanadate species where the anion acts as a ligand: (2- hep H)(NH 4 )[{Cu(H 2 O) 2 (2- hep )} 2 V 10 O 28 ] (cid:3) 4H 2 O ( V 10 Cu ) and (2- hep H) 2 [{Co(H 2 O) 5 } 2 V 10 O 28 ] (cid:3) 4H 2 O ( V 10 Co ) (2- hep = 2-hydroxyethylpyridine). Unlike free decavanadate, the complex anions stay intact in model buﬀer solutions (0.1 M 2-( N -morpholino)ethanesulfonic acid, 0.5 M NaCl, pH = 5.8 and 8.0). It has been shown that V 10 Co is stable also in the presence of proteins and for the first time it was possible to study the interaction of decavanadate with proteins without the interference of lower vanadate oligomers. This allowed comparison of interactions of V 10 and V 10 Co with the model proteins thaumatin, lysozyme, proteinase K, human serum albumin and transferrin under conditions close to biological ones (0.1 M 2-( N -morpholino)ethanesulfonic acid, 0.5 M NaCl, pH = 5.8). The linewidths of the signals at half-height in 51 V NMR spectra reflect the strength of interaction of a vanadium species with a protein, and thus it was shown that V 10 and V 10 Co both bind strongly to thaumatin, V 10 binds to lysozyme and V 10 Co binds to proteinase K. V 10 interacts with both human serum albumin and transferrin, but surprisingly V 10 Co exhibits high aﬃnity to transferrin but does not interact with albumin.


Introduction
Polyoxometalates (POMs) are an important group of metal oxide clusters 1 exhibiting diverse archetypal structures of, particularly, vanadates, molybdates and tungstates. [2][3][4] The tremendous variability of POMs has given rise to their application in distinct areas of materials science, 5,6 catalysis, 7-10 electrochemistry and redox processes, 11,12 photochemistry 13-15 and magnetism. [16][17][18] In recent years, the roles of POMs in biological systems have been intensively investigated. Such studies may be divided into two groups: functional binding and interaction of POMs in biological systems widening the borders of medicinal chemistry; [19][20][21][22][23][24] and non-functional interaction with biomolecules, such as proteins, enhancing the current possibilities in macromolecular crystallography by promoting the crystallization or being useful in obtaining the initial phases while solving the structures of proteins. [25][26][27][28] Decavanadate, H x V 10 O 28 (6Àx)À (V 10 ), is the predominant species formed in vanadate solutions at vanadium(V) concentrations above 1 mM in the pH range of E2-6. 29,30 The structure of V 10 consists of ten face-sharing octahedra (Scheme 1). The symmetrically non-equivalent vanadium atoms V A , V B and V C give rise to three different signals in 51 V NMR spectra depending on the conditions: while the low-field signal of V A atoms stays in a narrow region around À425 AE 3 ppm, the other two peaks representing V B and V C atoms are more sensitive to changes in acidity and exhibit signals at approximately À505 AE 10 ppm and À525 AE 10 ppm, respectively. The oxygen atoms O B and O C are potential sites for protonation, and the atoms O C , O F , O G and O D are the most potential sites for ligation to transition metal ions. 31,32 As such, changes in 51 V NMR parameters reflect the versatility of structural modifications of V 10 . While protonation and coordination of V 10 manifest mostly in peaks' movement, the interaction and binding to proteins result in significant peak broadening defined by linewidths at half-height of the signals, W 1/2 . This is caused by the ligand bulkiness and decreased symmetry of the V 10 species upon interaction. 32 Vanadium is naturally omnipresent in biological matrices 33,34 in a few enzymes such as vanadium dependent haloperoxidases and nitrogenases 35 or acts as a crucial component in the energetic metabolism of Ascidians. 36 Some artificial vanadium compounds, on the other hand, exhibit insulin mimetic properties, antitumor activity, antibacterial activity or anti-HIV activity. [35][36][37][38][39] Specifically decavanadate itself has also been studied with respect to many biological aspects, [40][41][42][43] and it was shown that V 10 binds to several proteins such as actin, 44 myosin, 45 ion pump Ca 2+ -ATPase, 46 bovine serum albumin and gelatine, 47 and microtubule-associated proteins. 48 Protein crystallography revealed the presence of V 10 in the crystal structures of acid phosphatase A (F. tularensis), 49 human activated receptor tyrosine kinase, 50 NTPDase1 (L. pneumophila), 51 NTPDase1 (R. norvegicus), 52 and human TRPM4 channel. 53 In all cases, V 10 was formed from the initially employed Na 3 VO 4 (used as a phosphatase inhibitor) and its role may be explained as stabilization and rigidification of the protein structure. 25 In this work we compare the interaction of both free and ligated decavanadate with commercially available model proteins thaumatin, lysozyme, proteinase K, as well as human serum albumin and transferrin. We utilize 51 V NMR spectroscopy as a powerful tool to investigate the stability of two decavanadates coordinated to metal centres Cu(II) and Co(II) under conditions usually used for protein crystallization.

Materials and methods
All chemicals were of analytical grade and used as received without further purification. All proteins were supplied as lyophilized powders (supplier, reference code): thaumatin from Thaumatococcus danielii (Sigma, T7638; a mixture of thaumatin I and thaumatin II with traces of other sweet proteins), proteinase K from Tritirachium album (Sigma, P6556), lysozyme from chicken eggwhite (Carl Roth GmbH & Co. KG, 8259.3), albumin from human serum (Sigma, A1653) and apo-transferrin from human serum (Sigma, T1147). The determination of C/H/N was carried out by using an 'EA 1108 CHNS-O' elemental analyzer by Carlo Erba Instruments at the Mikroanalytisches Laboratorium, University of Vienna. Metal elements' analyses were performed in aqueous solutions containing 2% HNO 3 using inductively coupled plasma mass spectrometry (PerkinElmer Elan 6000 ICP MS) for Mo and V, and atomic absorption spectroscopy (PerkinElmer 1100 Flame AAS) for Cu and Co. Standards were prepared from single-element standard solutions of concentration 1000 mg L À1 (Merck, Ultra Scientific and Analytika Prague). FT-IR spectroscopy was performed on a Bruker Vertex 70 IR Spectrometer equipped with a single reflection diamond-ATR (attenuated total reflectance) unit in the range of 4000-100 cm À1 .

X-ray diffraction on single crystals
The X-ray diffraction data were collected on Bruker X8 APEXII (V 10 Cu, Mo Ka) and Bruker D8 Venture (V 10 Co, Cu Ka) instruments equipped with multilayer monochromators, Incoatec Microfocus sealed tubes, and Kryoflex and Oxford cooling devices. The structures were solved by direct methods and refined by fullmatrix least-squares. Nonhydrogen atoms were refined with anisotropic displacement parameters. Hydrogen atoms were inserted at calculated positions and refined with riding coordinates except for solvent molecules and H atoms in NH 4 + in V 10 Cu. This was not done because of disorder and partial occupancies of O and N atoms in question. The following software was used for the structure solving procedure: Bruker SAINT software package 54 (frame integration, cell refinement), SADABS 55  material) and Diamond 59 (graphics). Experimental data and CCDC-codes can be found in Table 2.

Syntheses
To the best of our knowledge, the work of Schwendt et al. 61 presenting the synthesis and characterization of (2-hepH) 2 2 O is the only one reporting a coordination compound of decavanadate that stays intact in aqueous solution. Inspired by this work, the synthetic protocol was modified in order to prepare new stable complexes of decavanadate. A reversed addition of the individual reaction components and stepwise acidification of the solution prevented formation of heavy precipitates that were described in the original procedure 61 and such an approach resulted in successful isolation of V 10

Crystal structures
The structure solution and refinement details for compounds V 10 Cu and V 10 Co are summarized in Table 2. The crystal structure refinement of V 10 Cu revealed the presence of the complex anion [{Cu(H 2 O) 2 (2-hep)} 2 V 10 O 28 ] 2À that has been already reported (Fig. 1). 61 The decavanadate anion [V 10   The short distance O8-Cu1 2.889 Å indicates some weak attraction. The crystal structure of V 10 Cu differs from the previously described one 61 in the presence of the NH 4 + cation that was confirmed by elemental analysis and IR spectroscopy. The asymmetric unit of V 10 Co contains one half of the centrosymmetric anion [{Co(H 2 O) 5 } 2 V 10 O 28 ] 2À (Fig. 1), one molecule of (2-hepH) + balancing its charge and two water molecules of crystallization. The decavanadate anion [V 10 O 28 ] 6À is acting as a bridging ligand for two {Co(H 2 O) 5 } 2+ fragments. The slightly irregular octahedral coordination sphere of the Co1 atom is completed by one oxygen atom O1 coming from the terminal VQO group of the decavanadate (atom O F in Scheme 1). Such a coordination fashion is not unknown and was already reported for manganese(II) 72,76 and zinc(II) derivatives. 68  The acidic regime is preferred to ensure that the protein is positively charged and may therefore more profoundly interact with negatively charged POMs. The 51 V NMR spectra of the given solutions are shown in Fig. 2. In the spectrum of V 10 the expected distribution of vanadium into several species can be observed. Due to hydrolysis, the equilibrated species include not only the originally employed decavanadate (À422.5, À498.2 and À513.8 ppm), but also monovanadate H 2 VO 4 À (V 1 , À558.5 ppm), (V 4 , À575.0 ppm) and pentavanadate V 5 O 15 5À (V 5 , À583.0 ppm).
However, due to the high ionic strength decavanadate is still the dominant species and consumes about 80% of V V present in the solution.
The spectra of V 10 Cu and V 10 Co exhibit only peaks that can be assigned to vanadium atoms arising from decavanadate: À422.3, À496.9 and À511.9 ppm for V 10 Cu and À422.3 and À495.9 ppm (broad peak) for V 10 Co. This is the first indication that the complex anions stay intact and do not dissociate off the Cu(II) and Co(II) centres. As a matter of fact, the presence of free decavanadate would evoke vanadate self-condensation reactions and the overall picture of the present species should be similar to that of V 10 . Next, similar to Schwendt et al., 61 we observed significant movement of the high-field signal of the V C atom in V 10 Cu to À511.9 ppm compared to À513.8 ppm in V 10 , and the signal of the V B atom is shifted by 1.3 ppm. For V 10 Co, the shift of the peak is much more obvious and instead of two independent signals for V B and V C atoms we observe only a broad peak with the maximum at À495.9 ppm having a high-field shoulder. Based on peak integration, the integral intensities of the two present signals are in the ratio 2 : 8. In addition, significant peak broadening of the individual signals was observed (Table 3). This can be explained by at least two factors. Firstly, vanadates usually provide narrower lines in comparison to common vanadium(V) complexes because of higher symmetry (i.e. D 2h for ideal [V 10 O 28 ] 6À ). Thus, peak broadening originates in the decrease of symmetry of the coordinated decavanadates. It is also important to note that for V 10 Co the low field signal is broadened by only 10% in comparison to free decavanadate, while the second signal is broader by more than 220% compared to the sum of individual signals of V B and V C in V 10 . This difference is naturally caused by the fact that the V C atoms are the closest ones to the Co(II) atom and the inner V A atoms are less affected by the coordination. The compound V 10 Cu formed a cloudy precipitate in the buffer solution which might have also influenced the peaks' width, but despite this, similar peak broadening as for V 10 Co was observed -14% and 200%, respectively. This leads to the conclusion that the peak coalescence is a consequence of its coordination to the Cu(II) center and not the presence of a precipitate. However, we excluded V 10 Cu from further examination to prevent misinterpretation of protein interaction experiments. On the other hand, peak broadening may also originate in the presence of paramagnetic centres Cu(II) (d 9 ) and Co(II) (d 7 ). The experience has shown, however, that if an extraneous paramagnetic species interferes in a 51 V NMR experiment, this manifests itself also in a lower signal-to-noise ratio and uneven lines (this was not the case).
At pH = 8.0 (0.1 M MES, 0.5 NaCl) profound decomposition of V 10 into lower vanadates was observed, V 10 Cu dissociated off the Cu(II) complex cation, while V 10 Co was still the only species present (see Fig. S1 and Table S1, ESI †).
3.3.2 Interaction of decavanadates with thaumatin, lysozyme and proteinase K. The interaction of V 10 and V 10 Co (1 mM) with model proteins thaumatin (10 mM), lysozyme (10 mM) and proteinase K (3.5 mM) was inspected by 51 V NMR   The linewidth was calculated from the half-height for the low-field and high-field components of the merged broad signal. spectroscopy in 0.1 MES buffer, 0.5 M NaCl at pH = 5.8 one hour after careful addition of the protein solution to the solution of POMs ( Fig. 3 and 4). Table 4 summarizes the data on chemical shifts and linewidths at half-height for the observed peaks. During the experiments, a peak movement larger than AE1 ppm was not observed indicating that no changes in protonation occurred. Table 4 includes data only for decavanadates and [V 4 O 12 ] 4À as peaks corresponding to other vanadates had integral intensities o5%. Importantly, no visible reduction of vanadium(V) was recognized after 1 week of standing solutions.
Both V 10 and V 10 Co bind strongly to thaumatin resulting in significant peak broadening. In the case of lysozyme, very weak binding to V 10 may be deduced from slightly broadened higher-field peaks, but no obvious interaction was observed for V 10 Co. Finally, proteinase K seemed not to interact with free decavanadate, but binds strongly to V 10 Co. In this case we used about 3 times lower protein concentration; therefore, after extrapolation of the data it can be assumed that the binding of V 10 Co to proteinase K is comparatively stronger than to thaumatin.
3.3.3 Interaction of decavanadates with human serum albumin and transferrin. Using the same experimental conditions, the interaction of V 10 and V 10 Co with the main proteins present in human plasma, namely, human serum albumin (1 mM) and transferrin (1 mM) (Fig. 3, 4 and Table 4), was also examined. Interestingly, free decavanadate V 10 binds strongly to albumin, but V 10 Co does not interact. It seems that the Co(II) centers coordinated to VQO groups of decavanadate occupy the binding sites for interaction of decavanadate with this protein and block the interaction. In fact, this was the only case where the interaction was observed only for the free decavanadate but not for the coordinated one. The interaction of V 4 O 12 4À with human serum albumin is much weaker than that of decavanadate based on only about 10% increase in the peak width.   The interaction of V 10 with transferrin is even stronger, and a significant binding to V 4 O 12 4À was also observed. The shift of the peak corresponding to the V B atom of the decavanadate by 1 ppm made it impossible to determine its half-height width. As expected, the presence of Co(II) causes extremely strong binding of V 10 Co to transferrin, resulting in peak broadening comparable to that observed in the case of thaumatin, but at 100Â lower protein concentration. The stability of V 10 Co, V 10 Cu and potentially other coordinated decavanadates in the examined medium at 10 mM total vanadium concentration, in line with the potential of functionalized decavanadates to interact with various proteins, open new possibilities for the investigation of decavanadate's effects in biological systems -for the first time without the side effects of the always present lower oligovanadates (decavanadate may be the only species present in multicomponent solvents). 77 On the other hand, at physiological concentrations (c V = 1 mM and less), vanadate exists only as a monomeric species (VO 2 + , H 2 VO 4 À , HVO 4 À depending on the pH). 32 We therefore checked the stability of V 10 Co at 1 mM concentration (10 mM total vanadium) by 51 V NMR (Fig. S2, ESI †). The chemical shifts corresponding to the decavanadate species (À421.2, À497.9 and À513.1) represent an undisturbed anion indicating that the Co(II) centers are no longer involved in coordination.

Conclusions
In 5 NaCl) at pH = 5.8 making them promising candidates for biological studies. For the first time, an interaction between modified decavanadate and biomolecules without the participation of lower oligovanadates was performed. The pilot interaction studies with several proteins used in model protein crystallization research showed that V 10 and V 10 Co bind to thaumatin, V 10 binds also to lysozyme and V 10 Co binds to proteinase K. As expected, V 10 interacts with human serum albumin and transferrin, but surprisingly V 10 Co exhibits high affinity to transferrin but does not interact with albumin. The isolation and structural characterization of the crystalline products is the ultimate goal necessary for more precise understanding of the interaction between POMs and proteins. It is expected that, in addition to electrostatic interaction of V 10 with proteins, the presence of heterometals may induce a complementary interaction -even a covalent bond -when the coordinated transition metal contains labile ligands such as water molecules (as in V 10 Co and V 10 Cu) or a vacant accessible coordination position (as in V 10 Cu). Furthermore, both V 10 Cu and V 10 Co contain biogenic transition metals that are known to interact with biomolecules to a great extent. In conclusion, the high potential of ligated decavanadate in medicinal chemistry and protein crystallography necessitates the challenging development of synthetic methods leading reliably to stable complexes of decavanadate (i.e. decavanadato complexes).

Conflicts of interest
There are no conflicts to declare.