Introduction

One of the most important applications of vanadium compounds is their potential use in the therapy of patients suffering from type II diabetes mellitus [1–6], which affects 150 million people worldwide according to the estimates of the World Health Organization [7]. Since type II diabetes is accompanied by an increasing lack of response to insulin, insulin injections can ultimately become ineffective and alternative methods of treatment are desirable. About 30 years ago, an in vitro study on the insulin-enhancing action of vanadate was published, revealing the vanadate-stimulated uptake and degradation of glucose by adipocytes (fat cells), the stimulation of glycogenesis in the liver, and the inhibition of hepatic gluconeogenesis [8]; these effects were shortly thereafter related to the reduction of vanadate to vanadyl (VO2+), and to the action of VO2+ ion as an inhibitor of a cellular phosphatase [9]. Vanadium species have the advantage over insulin of possible oral administration.

After the initial use of vanadium(IV) and vanadium(V) inorganic salts, not easily applicable as antidiabetic drugs because of their toxicity and very low absorption rate, several peroxovanadium(V) complexes with (N,N), (N,O), and (O,O) ligands [10, 11] and neutral vanadyl species (VOL2, where L is an organic carrier) with bidentate anionic ligands were tested [2–6]. Most of these compounds stimulate glucose intake into cells, inhibit free fatty acids release and gluconeogenesis, stimulate lipogenesis, and lower or restore blood glucose level. Which is the active species, vanadyl (VO2+) or vanadate (H2VO4 −), and the mechanism of action of such compounds are two questions still remaining to be answered: it seems, however, that the effect is based on the phosphate-like activity and/or interference with phosphatases [8–12] and mainly on the inhibition of protein tyrosine phosphatase, which is responsible for the inhibition of the signal transduction pathways and glucose transport into the cells in the absence of insulin [2, 3, 13–16].

VO2+ complexes with VOL2 stoichiometry were found to be very effective and well tolerated in all the animal models of diabetes and resulted in a more reliable glucose lowering than VOSO4 [7]. Among them, [VO(ethylmaltolato)2], or BEOV, is undergoing phase IIa clinical trials [17], and cis-[VO(pic)2(H2O)] and [VO(6-mepic)2] (pic and 6-mepic indicate picolinate and 6-methylpicolinate) are strong inhibitors of the mobilization of the fatty acids and are effective in the treatment of rats affected by diabetes induced by streptozotocin [19–22].

An important factor in assessing the mechanism of action of an insulin-enhancing vanadium compound is its interaction with serum components present under physiological conditions; these include high molecular mass components, such as human serum apotransferrin (hTf) and human serum albumin (HSA), and low molecular mass components of the blood serum; among these latter, some of us demonstrated that only lactate and citrate can interact with the binary complexes of transferrin and albumin to form mixed species generally indicated as VO2+–hTf–lact, VO2+–hTf–citr, VO2+–HSA–lact, and VO2+–HSA–citr, where lact and citr indicate lactate and citrate, respectively [23]. Both high molecular mass and low molecular mass components can partly or fully displace the original carrier L− of the insulin-enhancing agent VOL2 to yield binary or ternary species [24–27], which could be the pharmacologically active species or, at least, the main way that vanadium is delivered to cells. The existence of a mixed complex between [VO(mal)2], where mal denotes the deprotonated form of maltol, and albumin was proposed by Willsky et al. [28], who observed that at least one of the maltol ligand remains coordinated to vanadium without specifying the stoichiometry of the species; subsequently, Liboiron et al. [26] reported the formation of adducts with [VO(mal)2(HSA)] stoichiometry, with one imidazole nitrogen of a histidine residue that occupies the fourth equatorial position of VO2+ ion. These suggestions appear to be in agreement with in vivo blood circulation monitoring–EPR measurements in rats [29] and with the recent demonstration that in conditions close to the physiological ones the bis-chelated complexes with cis-octahedral geometry can form ternary species such as cis-VO(pic)2(HSA), cis-VO(dhp)2(HSA), and cis-VO(dhp)2(hTf), in which a histidine nitrogen replaces the water molecule in the equatorial plane [dhp is the deprotonated form of 1,2-dimethyl-3-hydroxy-4(1H)-pyridinone] [30]. These last insights were based on the interpretation of EPR spectra of the ternary systems VO2+/hTf/carrier and VO2+/HSA/carrier and on their comparison with model systems containing 1-methylimidazole (1-MeIm), which coordinates VO2+ ion like an imidazole nitrogen of a histidine residue [26]; these data also allowed an explanation of the results of Jakusch et al. [31], who proposed the compositions (VO)hTf(dhp), (VO)2hTf(dhp), and (VO)2hTf(dhp)2, with dhp bound to one lobe or both lobes of the protein, for the mixed species formed by transferrin and 1,2-dimethyl-3-hydroxy-4(1H)-pyridinone.

In this work, we studied the biospeciation in the presence of the two most important high molecular mass components (transferrin and albumin) and one low molecular mass component (lactate) of four among the most representative insulin-enhancing vanadium compounds, two with VO(N2O2) coordination and two with VO(O4) coordination: [VO(6-mepic)2] [32, 33], [VO(pic)2(H2O)] [6, 14], [VO(acac)2] [34, 35], and [VO(dhp)2] [36] (acac is the deprotonated form of acetylacetone) (Scheme 1).

Scheme 1
scheme 1

Insulin-enhancing vanadium compounds studied in this work

Full size image

To quantitatively describe the biotransformation of an insulin-enhancing vanadium drug, it is necessary to know the stability constants associated with the several complexes that are formed. The values of log K (or log β) can be used to calculate the percent composition of a system with a high number of components or with a low concentration of VO2+, in conditions for which the characterization of the species present in solution should be difficult with any instrumental technique.

Chasteen et al. [25] found that in vitro transferrin binds VO2+ ion approximately 6 times more strongly than albumin. Recently Jakusch et al. [31] calculated the values of 13.4 for log K 1(hTf) and 11.8 for log K 2(hTf), whereas the binding constant of albumin for (VO)HSA [log K 1(HSA)] was around 10 if it was measured with spectroscopic methods [37, 38] and 9 in ultrafiltration studies [31]. The 3–4 orders of magnitude between K 1(hTf) and K 1(HSA) implies that all of VO2+ ion should be bound to transferrin [31, 37–39]. However, their results were based on the approximation that ternary complex formation between VO2+, hTf, and dhp could be neglected [30, 31]. Therefore, we believe that log K 1(hTf) and log K 2(hTf) must be recalculated in a system where the formation of mixed species is lacking: for example, that containing acetylacetonate is surely a valid choice, since acac does not form ternary complexes with hTf [30]. For the same reason, the value of log K 1(HSA), measured in a system where the formation of cis-VO(pic)2(HSA) is detected [30, 37], must be rediscussed and redetermined.

Moreover, here we report for the first time the stability constants of the (1) dinuclear species formed by albumin, (VO) 2 d HSA [40], (2) ternary species formed by transferrin with lactate or picolinate, where the two ligands replace the bicarbonate ion in the active site of transferrin with a COO− group [23], and (3) ternary species formed by transferrin or albumin with the carrier, VO(6-mepic)(HSA)(OH), cis-VO(pic)2(HSA), cis-VO(dhp)2(hTf), and cis-VO(dhp)2(HSA) [30]. In the examination of the biospeciation of an insulin-enhancing agent in the blood serum, besides the above-mentioned stability constants, those reported in the literature for the insulin-enhancing drugs in their original form [VO(carrier)2] [41–44], for the mixed species formed by carrier and lactate such as [VO(carrier)(lactH−1)]− [41, 44], and for the binary complexes formed by VO2+ and lactate [45] must be taken into account.

Finally, the mixed species formed at physiological pH by 1-MeIm were used as models for the coordination of transferrin or albumin through an imidazole nitrogen belonging to a surface residue of histidine [26, 30]. The stability constants for such species were compared with those of the analogous complexes formed by hTf and HSA [30].

We believe that the results of this study can help to quantitatively estimate the biotransformation of an insulin-enhancing VO2+ compound in blood serum and to identify and evaluate the importance at physiological conditions of the various species present, some of which could be revealed to be the pharmacologically active form.

Materials and methods

Chemicals

Water was deionized prior to use through a Millipore Milli-Q Academic purification system. VO2+ solutions were prepared from VOSO4·3H2O following literature methods [46].

HSA and hTf were obtained from Sigma. Apotransferrin (98%) was obtained as a lyophilized powder with a molecular mass of 76–81 kDa (Sigma T4283). Albumin (97–99%), containing only trace amounts of fatty acids, was crystallized and lyophilized with a molecular mass of 66 kDa (Sigma A9511). The concentrations of the protein solutions were estimated from their UV absorption [ε 280(hTf) = 92,300 M−1 cm−1; ε 278(HSA) = 42,000 M−1 cm-1] [25, 47].

6-Methylpicolinic acid, picolinic acid, acetylacetone, 1,2-dimethyl-3-hydroxy-4(1H)-pyridinone, lactic acid, 1-MeIm, NaHCO3, and 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid were of the highest grade available and were used as received.

Preparation of the solutions

The solutions were prepared by dissolving VOSO4·3H2O and the organic carrier in ultrapure water. The vanadium concentration used was in the range necessary to observe insulin-enhancing activity (approximately 1–400 μM) [7, 48–50]. When desired, lactate was also added. Argon was bubbled through the solutions to ensure the absence of oxygen and avoid the oxidation of VO2+ ion. Appropriate amounts of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid and NaHCO3 were added to the solution to obtain final concentrations of 1.0 × 10−1 and 2.5 × 10−2 M, respectively.

To 1 mL of these solutions, again carefully purged with argon, an amount of transferrin or albumin was added to obtain the correct hTf to VO2+ and HSA to VO2+ ratios; the pH was adjusted to approximately 7.4.

To calculate the stability constants of the several species, the following solutions at pH around 7.4 were considered: (1) 1:1:4 VO2+/hTf/acac (2.5 × 10−4 M VO2+) and 1:1:10 VO2+/hTf/acac (2.5 × 10−4 M VO2+) to calculate log β 1 of (VO)hTf; (2) 2:1:5 VO2+/hTf/acac (5.0 × 10−4 M VO2+) and 2:1:15 VO2+/hTf/acac (5.0 × 10−4 M VO2+) to calculate log β 2 of (VO)2hTf; (3) 4:1:10 VO2+/HSA/pic (1.0 × 10−3 M VO2+) and 4:1:8:20 VO2+/HSA/pic/1-MeIm (1.0 × 10−3 M VO2+) to calculate log β of (VO)HSA; (4) 1:1:2:5 VO2+/HSA/pic/1-MeIm (7.5 × 10−4 M VO2+) to calculate log β of (VO) 2 d HSA; (5) 2:1:5 VO2+/hTf/dhp (5.0 × 10−4 M VO2+) to calculate log β of cis-VO(dhp)2(hTf); (6) 4:1:10 VO2+/HSA/dhp (1.0 × 10−3 M VO2+) to calculate log β of cis-VO(dhp)2(HSA); (7) 1:1:40.8 VO2+/hTf/lact (2.5 × 10−4 M VO2+) to calculate log β of (VO)hTf(lact); (8) 2:1:40.8 VO2+/hTf/lact (VO2+ 5.0 × 10−4 M) to calculate log β of (VO)2hTf(lact)2; (9) 1:1:2.5 VO2+/hTf/pic (2.5 × 10−4 M VO2+) to calculate log β of (VO)hTf(pic), and (10) 2:1:5 VO2+/hTf/pic (5.0 × 10−4 M VO2+) to calculate log β of (VO)2hTf(pic)2.

EPR spectra of all the model systems (VO2+/carrier, VO2+/lact, and VO2+/carrier/lact) were recorded with the same molar ratio used with transferrin or albumin.

EPR spectroscopy

EPR spectra were recorded on frozen solutions at −150 °C with an X-band (9.4-GHz) Bruker EMX spectrometer. The addition of dimethyl sulfoxide was not necessary and no improvement in the resolution of the spectra was obtained. All the solutions were degassed before measuring the anisotropic spectra. After the samples had been transferred into the EPR tubes, the spectra were immediately recorded. Only low-temperature EPR spectra were measured to minimize the oxidation of VO2+ ion to vanadium(V). To increase the signal to noise ratio, signal averaging was used [23, 30, 40]. During the time needed to record the spectra of the frozen-solution samples no appreciable decrease of the signal intensity was observed.

A power saturation study was undertaken and the results showed that with the microwave power used in this study (10 dB) the signals were not saturated. No correction for the phase of the signal for each individual spectrum was needed because it was adjusted by the autotune routine of the acquisition program. This was periodically checked and the phase was always the one giving the highest intensity signals.

The values of A z were calculated through simulations with the computer program Bruker WinEPR SimFonia [51].

The intensity of the EPR signals was measured in the M I = −7/2, 5/2, 7/2 parallel transitions through the double integration of the first-derivative spectra. The examination of the other parallel resonances or of the perpendicular resonances was not possible for the extended superimposition of the signals belonging to the species present in aqueous solution.

The signals of the species (VO) 2 d HSA (S = 1) have lower intensity with respect to (VO)HSA (S = 1/2) [40], and therefore the values obtained with double integration for the two complexes are not directly comparable. To determine the amount of VO2+ bound as (VO) 2 d HSA complex in a system containing both (VO) 2 d HSA and (VO)HSA species, the solution containing VO2+, picolinic acid, and 1-MeIm, with the pH adjusted to approximately 7.4, was divided into two fractions: anisotropic EPR spectra were recorded for the first fraction, where an appropriate amount of HSA was added to form the species (VO) 2 d HSA, cis-[VO(pic)2(1-MeIm)], and [VO(pic)2(OH)]−, and for the second one. In this way the vanadium concentrations in both samples were exactly the same and the fraction present as (VO) 2 d HSA can be obtained from the difference between the intensity signals of the mononuclear species. This procedure was validated on systems in which the relative amounts of mononuclear and dinuclear complexes were known, for example, those containing tartaric or citric acid [52].

The complete EPR spectra in Figs. 2, trace c, 3, trace a, and 5, traces c and d, and in Figs. 3, trace c and 4, traces a and b are reported in the supporting information of [23] and [30], respectively, whereas those in Figs. 2, traces a, b, and d, 3, traces b and d, 4, trace c, and 5, traces a and b are shown in Figs. S1–S7. The microwave frequency used was around 9.4 GHz.

Potentiometric measurements of VO2+/carrier/1-MeIm systems

The stability constants of the proton and VO2+ complexes were determined by pH-potentiometric titrations on 25.00-mL samples. The ligand to metal molar ratio was between 1:1 and 8:1 and the VO2+ concentration was 0.001–0.004 M. Titrations were performed from pH 2.0 until precipitation or very extensive hydrolysis occurred by adding carbonate-free KOH of known concentration (approximately 0.2 M KOH) [53]. The pH was measured with a Metrohm 6.0234.100 combined electrode, calibrated for hydrogen ion concentration by the method of Irving et al. [54]. The water ionization constant, pK w, was 13.76 ± 0.01 under the conditions employed. Measurements were carried out at 25.0 ± 0.1 °C and at a constant ionic strength of 0.20 M KCl with a Radiometer pHM 84 instrument and a Metrohm 715 Dosimat burette. Purified argon was bubbled through the samples to ensure the complete absence of oxygen and carbon dioxide. The number of experimental points used in the calculations was 150–200 for each system and the reproducibility of the points included in the evaluation was within 0.005 pH units in the whole pH range measured.

The stability constants, β pqr  = [(VO) p B q H r ]/[VO]p[B]q[H]r for the binary complexes and β pqrs  = [(VO) p A q B r H s ]/[VO]p[A]q[B]r[H]s for the ternary species, where VO stands for VO2+ ion, A and B are the deprotonated forms of the carrier and 1-MeIm, respectively, and H is the proton, were calculated with the aid of the PSEQUAD program [55]. The values reported for the species formed in the binary systems with 6-methylpicolinate, picolinate, and dhp, determined in identical conditions, were used to estimate log β for the ternary complexes with 1-MeIm [41, 42, 44]. Standard deviations were calculated by assuming random errors. The conventional notation was used: a negative sign for protons indicates either the dissociation of groups which do not deprotonate in the absence of VO2+ coordination or hydroxo ligands. Hydroxo complexes of VO2+ were taken into account and the following species were assumed: [VO(OH)]+ (log β = −5.94), [(VO)2(OH)2]2+ (log β = −6.95), with stability constants calculated from the data of Henry et al. [56] and corrected for the different ionic strengths by use of the Davies equation [57], [VO(OH)3]− (log β = −18.0), and [(VO)2(OH)5]− (log β = −22.0) [58, 59].

Results

The VO2+/1-MeIm and VO2+/carrier/1-MeIm systems

In the literature it has been suggested that 1-MeIm can be considered as a valid model for the binding of an imidazole nitrogen belonging to a histidine residue present in a polypeptide chain [26]. Therefore, in this work the interaction of 1-MeIm with VO2+ ion and the insulin-enhancing compounds [VO(carrier)2] was studied with the combined application of potentiometric (pH titrations) and spectroscopic (EPR) techniques. The stability constants of proton and VO2+ complexes are reported in Table 1.

Table 1 Stability constants (log β) of proton, binary VO2+ complexes of 1-methylimidazole (1-MeIm) and ternary complexes of VO2+/1-MeIm with the deprotonated forms of 6-methylpicolinic acid (6-mepic), picolinic acid (pic), and 1,2-dimethyl-3-hydroxy-4(1H)-pyridinone (dhp) at 25.0 ± 0.1°C and I = 0.20 M (KCl)
Full size table

The pK a value of 1-MeIm (indicated as HB+ in the fully protonated form in aqueous solution) is slightly higher than that of imidazole, 6.95 [60], owing to the presence of an electron-releasing methyl group on nitrogen in position 1 of the aromatic ring.

As expected, 1-MeIm is a weak ligand towards VO2+ and even at a ligand to metal ratio of 8:1 the hydrolysis starts around pH 4.5. Therefore, the points below pH 4.0 were used for the calculations and from these data only log β for [VOB]2+ can be determined (Table 1); in this complex the aromatic nitrogen behaves as a monodentate donor. Its A z value (176 × 10−4 cm−1) agrees well with the data in the literature [61]. Only at very high ratio (100:1) can the EPR signals of [VOB]2+, [VOB2]2+, [VOB3]2+, and [VOB4]2+, in which one, two, three, or four nitrogens occupy the equatorial positions, be revealed [61].

The potentiometric and spectroscopic study of the ternary systems VO2+/6-mepic/1-MeIm, VO2+/pic/1-MeIm, and VO2+/dhp/1-MeIm shows that mixed complexes are formed. In particular, [VO(6-mepic)(1-MeIm)(H2O)]+ and [VO(6-mepic)(1-MeIm)(OH)] are detected in the system with 6-methylpicolinate (A z 168 × 10−4 and 163 × 10−4 cm−1, respectively), with the organic carrier coordinating the metal ion with the (N, COO−) donor set, a monodentate imidazole nitrogen, and a water or a hydroxide ligand completing the equatorial position of vanadium; the deprotonation of the water molecule takes place with a pK of 5.4.

In the ternary systems containing picolinate and dhp, [VO(pic)2(1-MeIm)] and [VO(dhp)2(1-MeIm)] are observed. Their A z value measured in the EPR spectra is lower than that of cis-[VO(pic)2(H2O)] and cis-[VO(dhp)2(H2O)] (159 × 10−4 versus 165 × 10−4 cm−1 for picolinate [42] and 163 × 10−4 versus 166 × 10−4 cm−1 for 1,2-dimethyl-3-hydroxy-4(1H)-pyridinonate [44]); this suggests that in the mixed species the equatorial water molecule has been replaced by the imidazole nitrogen of 1-MeIm. Recent density functional theory results also confirm this attribution [30]. The structures of the ternary complexes formed by 1-MeIm are shown in Scheme 2. As an example, the distribution curves as a function of pH for the species formed with dhp are depicted in Fig. 1.

Scheme 2
scheme 2

Ternary complexes formed by 1-MeIm with 6-mepic, pic, and dhp

Full size image
Fig. 1
figure 1

Species distribution for the VO2+/dhp/1-MeIm system as a function of pH with a molar ratio of 1:2:4 and a VO2+ concentration of 1 mM

Full size image

Determination of the stability constants for (VO)hTf and (VO)2hTf

The binary system VO2+/hTf has been widely studied through spectroscopic techniques [24, 25, 28, 40, 62–67]. EPR results suggest that there are two binding sites, called A and B [62, 63], with B being further split into B1 and B2 [40, 65, 66]. Analogously to Fe3+, VO2+ needs HCO3 − for the binding to the A and B sites of the protein; bicarbonate can be replaced by other anions, called synergistic anions, which must have certain structural features, such as a carboxylate and an electron-withdrawing group [68, 69].

The first values reported in the literature for the binding constants for binding to the two lobes of transferrin were by Kiss et al. [39], who used a linear free-energy relationship between log K 1(hTf) and the stability constant for the binding of hydroxide ion, log K OH [69]: values of 13.2 ± 1.6 and 12.2 ± 1.6 were obtained for log K 1 and log K 2, with a rather high uncertainty degree [39]. Such values were subsequently corrected to 14.3 ± 0.6 and 11.7 ± 0.6 through the examination of the partial displacement reaction between (VO)hTf and [VO(dhp)2]; these measurements were based on the assumption that VO2+, hTf, and dhp do not form ternary species [66]. Recently, log K 1 and log K 2 were fixed at 13.4 and 11.8, respectively; in this last work, however, the authors attributed the stoichiometries (VO)hTf(dhp), (VO)2hTf(dhp), and (VO)2hTf(dhp)2 to the mixed species detected in solution at physiological pH [31]. In contrast, some of us recently proposed that the species formed under such conditions is cis-VO(dhp)2(hTf), with an imidazole nitrogen of a histidine residue of the transferrin chain occupying the fourth equatorial position [30]. Therefore, incorrect knowledge of the composition of the ternary system certainly affects the estimated values of the stability constants relating to the VO2+–hTf interaction.

To get correct values of log β 1 and log β 2 for the formation of (VO)hTf and (VO)2hTf, a system where mixed species are not observed must be chosen. It was recently found that in the ternary VO2+/hTf/acac system only the formation of binary complexes between VO2+ and transferrin, on one hand, and between VO2+ and acetylacetonate, on the other, can be detected [30].

To estimate log β 1 for the formation of (VO)hTf, EPR spectra at different molar ratios were recorded; in particular the system VO2+/hTf/acac with ratios of 1:1:4 and 1:1:10 was considered (see “Materials and methods”). The results are presented in Fig. 2.

Fig. 2
figure 2

High-field region of the X-band anisotropic EPR spectra recorded at physiological pH on aqueous solutions containing a 1:1 VO2+/hTf (2.5 × 10−4 M VO2+), b 1:1:4 VO2+/hTf/acac (2.5 × 10−4 M VO2+), c 2:1 VO2+/hTf (5.0 × 10−4 M VO2+), and d 2:1:15 VO2+/hTf/acac (5.0 × 10−4 M VO2+). e is the superimposition of the spectra reported in traces b (black) and a (red). I, Ia, and II indicate the resonances of (VO)hTf, (VO)2hTf, and [VO(acac)2], respectively

Full size image

As it is possible to see from an examination of Fig. 2, both (VO)hTf and [VO(acac)2] species exist, with the relative amount depending on the molar ratio of transferrin to acetylacetonate.

The displacement reaction is

$$ [ {\text{VO(acac)}}_{ 2} ] {\text{ + hTf }} \to ( {\text{VO)hTf + 2}}\,{\text{acac}} $$
(1)

and log β 1 (equal to log K 1) can be calculated through the equation

$$ \log\,\beta_{1} = { \log }{\frac{{ [ ( {\text{VO)hTf]}}}}{{ [ {\text{VO}}^{{ 2 { + }}} ] {\text{ [hTf]}}}}}={\text{log}}\left( {{\frac{{ [ {\text{VO(acac)}}_{ 2} ]}}{{ [ {\text{VO}}^{{ 2 { + }}} ] {\text{ [acac]}}^{ 2} }}} \times {\frac{{ [ ( {\text{VO)hTf] [acac]}}^{ 2} }}{{ [ {\text{VO(acac)}}_{ 2} ] {\text{ [hTf]}}}}}} \right)={\text{log}}\,\beta_{ 2} [ {\text{VO(acac)}}_{ 2} ] {\text{ + log}}\,K_{\text{D}} $$
(2)

where K D indicates the constant for the displacement reaction. The value of log β 2 for the formation of [VO(acac)2] is 16.27 [43], whereas K D can be determined by measuring the ratio of the concentrations of (VO)hTf and [VO(acac)2], in turn calculated through the integration of the respective EPR signals.

The relative amount of (VO)hTf and [VO(acac)2] is calculated from the first, seventh, and eighth parallel transitions (M I = −7/2, 5/2 and 7/2, respectively). It must be kept in mind, moreover, that the effective concentration of the deprotonated form of acetylacetonate can be determined from the experimental one ([acac]exptl) through the following equation:

$$ \left[ {\text{acac}} \right] = {\frac{{\left[ {\text{acac}} \right]^{\text{exptl}} }}{{\left[ { 1 +{ 10}^{{ - ({\text{pH}} - {\text{p}}K_{\text{a}} )}} } \right]}}} = {\frac{{\left[ {\text{acac}} \right]^{\text{exptl}} }}{{\left[ { 1+ {10}^{{ - ({\text{pH}} - 8.76)}} } \right]}}} $$
(3)

For example, in the system VO2+/hTf/acac with a molar ratio of 1:1:4 and a concentration of the metal ion of 2.5 × 10−4 M, the amounts of (VO)hTf and [VO(acac)2] measured in the M I = 5/2, 7/2 parallel transitions were 73.9 and 26.1% and 73.7 and 26.3%, respectively. From these data, the values of −3.28 and −3.29 for log K D and of 12.99 and 12.98 for log β 1 can be determined. Using the data obtained for two different ratios (1:1:4 and 1:1:10, see “Materials and methods”), the mean value for log β 1 is 13.0 ± 0.5.

To estimate log β 2, the systems with molar ratios of 2:1:5 and 2:1:15 for VO2+, hTf, and acetylacetonate were studied. From the percentages of [VO(acac)2] with respect to (VO)2hTf, calculated from M I = −7/2, 5/2 and 7/2 resonances in the EPR spectra, a mean value of 25.5 ± 0.5 is obtained.

The final values for log β 1 and log β 2 are summarized below:

$$ \log\,\beta_{1} [ ( {\text{VO)hTf] = 13}} . 0\pm 0. 5 $$
(4)
$$ \log\,\beta_{2} [ ( {\text{VO)}}_{ 2} {\text{hTf] = 25}} . 5\pm 0. 5 $$
(5)

The binding constants calculated exhibit reasonably good agreement with the values recently reported by Jakusch et al. [31] (log β 1 = 13.4 and log β 2 = 25.2). Our data suggest, however, that the affinities of the two sites are very similar (log Κ 1 = 13.0 and log Κ 2 = 12.5), in agreement with the observation that the two sites of transferrin differ by approximately one unit in their log K values [69]. For Fe3+, the site associated with the higher value of log K (site A) has been identified as the C-terminal [69].

Determination of the stability constants for (VO)HSA and (VO) 2 d (HSA)

Several papers have been devoted to the examination of the binary system VO2+/HSA [26, 28, 38, 40, 70]. In equimolar solution or with an excess of albumin, the EPR spectra are characterized by the presence of signals attributable to a dinuclear species [denoted as (VO) 2 d HSA] with spin state S = 1, whereas with an excess of VO2+, the EPR resonances are assigned to a multinuclear complex [denoted as (VO) x m HSA, with x = 5–6] characterized by an S = 1/2 spin state [40].

In the multinuclear species (VO) x m HSA the presence of one “strong” and five “weak” binding sites with different EPR parameters was firstly proposed, with the “strong” binding site associated with the primary site for Cu2+ [26, 28, 70]. The detection of only one set of EPR resonances at physiological pH when the VO2+ to HSA ratio is higher than 1 suggests that the five to six different metal ions bound by albumin have the same coordination and we called this site “high pH” [40]. Its experimental A z value (165 × 10−4 cm−1) allows the (NH2, N−, N−, Nimid) coordination to be ruled out, suggesting that the binding sites are relatively nonspecific with respect to that of Cu2+ or Ni2+ [71]: the coordination of three or four imidazole nitrogens, or of two imidazole nitrogens plus a carboxylate group belonging to an aspartate or a glutamate residue, has been proposed [40].

These results indicate that the affinities of these five to six sites towards VO2+ ion are very similar and, therefore, only a mean value for the association constant of (VO) x m HSA can be calculated. The real values of log β x for the five to six equilibrium steps should differ only slightly from each other (in our opinion, not more that 0.1–0.2 log units) and from the mean value (around 0.3–0.4 log units). Using the same notation as Jakusch et al. [31, 37, 38] on grounds of expediency [40], such a mean value will refer to the species (VO)HSA.

To determine the stability constant for (VO)HSA, a ternary system containing albumin and a ligand forming a binary complex with comparable stability must be studied. For this purpose, Kiss et al. [37] used picolinic acid, assuming that the formation of ternary species between albumin and picolinate was negligible. In the system VO2+/HSA/pic in aqueous solution, at pH 7.4, the species (VO) 2 d HSA [23, 40], (VO)HSA [23, 31, 37, 40], and cis-[VO(pic)2(OH)]− [42] exist and we recently demonstrated that also a mixed complex with the composition cis-VO(pic)2(HSA) is formed, having a histidine nitrogen bound in the equatorial position and two picolinate ligands in an equatorial–equatorial and equatorial–axial arrangement [30]. As mentioned above, the presence of (VO) 2 d HSA or (VO)HSA depends on the VO2+ to HSA ratio (a ratio lower than 1 favors the first complex, and a ratio higher than 1 favors the second species).

For this reason, the value of the stability constant obtained by Kiss et al. (log K 1 = 10 ± 1 through EPR measurements [37, 38] and log K 1 approximately 9 through ultrafiltration studies [31]) must be recalculated: in fact, they supposed that no ternary species exist in aqueous solution at physiological conditions.

The first system from which we calculated log β for VO(HSA) is VO2+/HSA/pic with a molar ratio of 4:1:10. In these experimental conditions, only (VO)HSA, cis-[VO(pic)2(OH)]−, and cis-VO(pic)2(HSA) exist in solution (see Fig. 3 in [30]). We do not know the exact value of log β for cis-VO(pic)2(HSA), but in the first approximation it can be considered equal to that of the mixed complex cis-[VO(pic)2(1-MeIm)], whose coordination mode is the same [(N, COO−); (N, COO−ax); Nimid]; the similarity between log β of cis-[VO(dhp)2(1-MeIm)], cis-VO(dhp)2(HSA), and cis-VO(dhp)2(hTf) (see below) supports this assumption, which, of course, results in a rather high value of the uncertainty on the stability constant. By using log β for [VO(pic)2(OH)]− (5.13) and cis-VO(pic)2(HSA) (14.96), one can calculate that for (VO)HSA. The result is log β[(VO)HSA] = 9.1 ± 1.0.

The second method to calculate the value of log β[(VO)HSA] is the analysis of the system VO2+/HSA/pic/1-MeIm. It is possible to convert cis-VO(pic)2(HSA) into cis-[VO(pic)2(1-MeIm)], whose log β is known (14.96, Table 1), using an excess of 1-MeIm so that the nitrogen of 1-MeIm replaces that of the histidine in the equatorial position of the octahedral VO2+ ion. In this way, only cis-[VO(pic)2(OH)]−, cis-[VO(pic)2(1-MeIm)], and (VO)HSA are present in aqueous solution at equilibrium. From the knowledge of the stability constants of the first two complexes [42] (Table 1), a value for log β 1[(VO)HSA] of 9.0 ± 1.0 is found.

It can be observed that the values of log β 1 determined with the two different methods are comparable. The mean of the two values gives

$$ \log\,\beta_{ 1} [ ( {\text{VO)HSA]}} = 9 . 1\pm 1.0 $$
(6)

If the EPR spectra of the quaternary system VO2+/HSA/pic/1-MeIm are recorded with a ratio of VO2+ to HSA of 1:1, the characteristic resonances of (VO) 2 d HSA with S = 1 appear (Fig. 3, trace b).

Fig. 3
figure 3

High-field region of the X-band anisotropic EPR spectra recorded at physiological pH on aqueous solutions containing a 1:1 VO2+/HSA (7.5 × 10−4 M VO2+), b 1:1:2:5 VO2+/HSA/pic/1-MeIm (7.5 × 10−4 M VO2+), c 1:2:4 VO2+/pic/1-MeIm (1.0 × 10−3 M VO2+), and d 1:2.5 VO2+/pic (1.0 × 10−3 M VO2+). I (and dotted lines), II, and III indicate the resonances of (VO) 2 d HSA, cis-[VO(pic)2(1-MeIm)], and cis-[VO(pic)2(OH)]−, respectively

Full size image

In this case, the presence in solution of (VO) 2 d HSA, [VO(pic)2(OH)]−, and cis-[VO(pic)2(1-MeIm)] must be considered. From the stability constants of cis-[VO(pic)2(OH)]− and cis-[VO(pic)2(1-MeIm)] [42] (Table 1) and from their relative amounts, as evaluated by the double integration of the respective EPR signals, the following value for log β[(VO) 2 d HSA] is calculated:

$$ { \log }\,\beta [ ( {\text{VO)}}_{ 2}^{\text{d}} {\text{HSA] = 20}} . 9\pm 1.0 $$
(7)

This is the first time that log β for the formation of (VO) 2 d HSA has been reported in the literature.

Determination of the stability constants for cis-VO(dhp)2(hTf) and cis-VO(dhp)2(HSA)

Jakusch et al. [31] recently proposed for the mixed species formed by transferrin and 1,2-dimethyl-3-hydroxy-4(1H)-pyridinone stoichiometries of (VO)hTf(dhp), (VO)2hTf(dhp), and (VO)2hTf(dhp)2, with dhp bound to one lobe or both the lobes of the protein. Recent EPR evidence suggests, however, that in the ternary system VO2+/hTf/dhp, at physiological conditions, a mixed complex cis-VO(dhp)2(hTf) is formed, in which the equatorial water molecule of the parent cis-[VO(dhp)2(H2O)] is replaced by a histidine nitrogen of hTf according to the reaction [30]

$$ {\text{VO(dhp)}}_{ 2} ( {\text{H}}_{ 2} {\text{O) + hTf }} \to {\text{ VO(dhp)}}_{ 2} {\text{hTf + H}}_{ 2} {\text{O}} $$
(8)

The log β value of VO(dhp)2(hTf) can be calculated through the equation

$$ { \log }\,\beta [ {\text{VO(dhp)}}_{ 2} ( {\text{hTf)] = log}}\,\beta_{ 2} ( {\text{VO(dhp)}}_{ 2} ) +{\text{log}}\,K_{\text{D}} ={22} . 8 3 + {\text{log}}\,K_{\text{D}} $$
(9)

The comparison of the spectra recorded for the binary VO2+/dhp and ternary VO2+/hTf/dhp system is shown in Fig. 4.

Fig. 4
figure 4

High-field region of the X-band anisotropic EPR spectra recorded at physiological pH on aqueous solutions containing a 4:1:10 VO2+/HSA/dhp (1.0 × 10−3 M VO2+), b 2:1:5 VO2+/hTf/dhp (5.0 × 10−4 M VO2+), and c 1:2.5 VO2+/dhp (1.0 × 10−3 M VO2+). I, II, IIIa (and dotted lines), and IIIb (and dotted lines) indicate the resonances of cis-VO(dhp)2(HSA), cis-VO(dhp)2(hTf), cis-[VO(dhp)2(H2O)], and [VO(dhp)2], respectively

Full size image

In the system with a VO2+, hTf, and dhp ratio of 2:1:5 and a VO2+ concentration of 5.0 × 10−4 M, the amounts of the bis-chelated species of dhp and of VO(dhp)2(hTf) are 54.2 and 45.8% in the M I = 5/2 transition, and 55.9 and 44.1% in the M I = 7/2 transition. If, as supposed, transferrin coordinates the metal ion with an imidazole nitrogen, it must be taken into account that the number of histidine residues exposed on the protein surface is 12 [72], so the effective concentration of histidine involved in the complexation is 3.0 × 10−3 M. These data allow the value of log β for the formation of VO(dhp)2(hTf) to be calculated:

$$ \log\, \beta [ {\text{VO(dhp)}}_{ 2} ( {\text{hTf)]}} = 25 . 5 { } \pm { 0} . 6 $$
(10)

It is worth noting that this value is very close to that obtained by pH-potentiometry for the mixed complex [VO(dhp)2(1-MeIm)], log β = 25.40 (Table 1), indicating that it is only the number of histidine residues present on the protein surface which influences the formation and stability of such a kind of species. Moreover, this result proves that 1-MeIm is a good choice to model the interaction between a metal ion and the histidine residues of a protein.

In the ternary system VO2+/HSA/dhp, a mixed species analogous to that described for transferrin with stoichiometry cis-VO(dhp)2(HSA), in which a histidine nitrogen is equatorially bound to vanadium, is observed [30].

In Fig. 4, trace b is the spectrum recorded for the ternary VO2+/HSA/dhp system. The molar ratio used was 4:1:10 with a VO2+ concentration of 1.0 × 10−3 M. The equations used and the discussion for the calculation of log β are similar to those just given for transferrin. The only difference that must be considered is that, in the case of albumin, the number of histidine residues on the protein surface is six [73] for an effective concentration of 1.5 × 10−3 M. The value of log β is

$$ \log\,\beta [ {\text{VO(dhp)}}_{ 2} ( {\text{HSA)]}} = 25. 9\pm 0. 6 $$
(11)

In this case too the stability constant for the formation of VO(dhp)2(HSA) is comparable to that of [VO(dhp)2(1-MeIm)] and is similar to that found for VO(dhp)2(hTf). Of course, this depends on the fact that the structure of the three species is the same, with the two dhp ligands adopting an equatorial–equatorial and an equatorial–axial arrangement and an imidazole nitrogen (belonging to 1-MeIm or to a histidine residue present on the surface of transferrin or albumin) occupying the fourth equatorial position. The small difference between the log β values for VO(dhp)2(hTf) and VO(dhp)2(HSA) can be attributed to a different exposure of the histidine residues on the protein surface and/or to charge and solvation effects.

Determination of the stability constants for (VO)hTf(lact) and (VO)2hTf(lact)2

It has long been known that ligands with a carboxylate group, behaving as synergistic anions, can replace bicarbonate in the iron sites of transferrin [24, 68, 69]. Lactate is one of these ligands and in this work we were able to measure the stability constants of the mixed complexes formed with transferrin. From EPR spectra, it is not possible to determine the exact stoichiometry of such species, indicated generally as VO2+–hTf–lact. However, according to chemical considerations and literature comments [31], the stoichiometries (VO)hTf(lact) and (VO)2hTf(lact)2, with the first being detected when the VO2+ to hTF ratio is 1:1 and the second when the ratio is 2:1, can be proposed. Since lactate replaces, at least partly, bicarbonate in the sites occupied by vanadium, the determination of the stability constant of (VO)2hTf(lact), in which only one lactate anion is bound to VO2+, is not possible.

As previously reported [68], it is very difficult to remove bicarbonate from solutions containing transferrin and a residual fraction always remains, affecting the binding of lactate. The procedure for removing completely HCO3 − ion was recently described [26], but it is not particularly useful in the present experiments. In fact, if bicarbonate was completely removed from a solution containing lactate, only the VO2+/hTf/lact ternary species would be obtained (since a synergistic anion is needed for the VO2+ binding to hTf), and log β for such a mixed complex could not be determined. In this work the evaluation of the stability constants by EPR spectroscopy was based on displacement reactions: thus, we need two species to be present in solution, and log β for one of them must be known. Therefore, the present experiments can be considered as a competition between bicarbonate and lactate and, for this reason, a large excess of the latter is necessary to observe the coordination.

EPR spectra recorded in the VO2+/hTf system with or without lactate are shown in Fig. 5. The coordination of lactate can be revealed by shoulders at lower field with respect to the M I = 5/2 and 7/2 transitions of (VO)2hTf and by an A z value slightly lower than that of site A of transferrin, around 167 × 10−4 cm−1 [23, 30]. From the double integration of the spectral lines, it was possible to estimate the relative amount of (VO)hTf (with HCO3 − bound to vanadium in the active sites of transferrin) and (VO)hTf(lact) (VO2+ to hTF ratio of 1:1) and of (VO)2hTf (HCO3 − bound to VO2+) and (VO)2hTf(lact)2 (VO2+ to hTF ratio of 2:1). In traces e and f in Fig. 5 the enlargement of the spectral lines of (VO)hTf and (VO)2hTf, due to the presence of the mixed complexes of lactate, is easily observable. The experimental data allow one to calculate the following values of log β for the formation of such species:

Fig. 5
figure 5

High-field region of the X-band anisotropic EPR spectra recorded at physiological pH on aqueous solutions containing a 1:1 VO2+/hTf (2.5 × 10−4 M VO2+), b 1:1:40.8 VO2+/hTf/lact (2.5 × 10−4 M VO2+), c 2:1 VO2+/hTf (5.0 × 10−4 M VO2+), and d 2:1:40.8 VO2+/hTf/lact (5.0 × 10−4 M VO2+). e is the superimposition of the spectra reported in traces b (black) and a (red) and f is the superimposition of the spectra reported in traces d (black) and c (red). I, Ia, II, and IIa indicate the resonances of (VO)hTf, (VO)2hTf, (VO)hTf(lact), and (VO)2hTf(lact)2, respectively

Full size image
$$ { \log }\,\beta [ ( {\text{VO)hTf(lact)]}} = 14 . 5\pm 0.8 $$
(12)
$$ { \log }\,\beta [ ( {\text{VO)}}_{ 2} {\text{hTf(lact)}}_{ 2} ] = 28 . 5\pm 0.8 $$
(13)

Determination of the stability constants for (VO)hTf(pic) and (VO)2hTf(pic)2

Analogously to lactate, picolinate possesses the features of a synergistic anion [68, 69] and, therefore, forms ternary complexes with VO2+ and hTf [30]. In this case too, the stoichiometry of such species depends on the VO2+ to hTf ratio and can be considered as (VO)hTf(pic) when the ratio is 1 and (VO)2hTf(pic)2 when it is 2.

The procedure for the determination of the stability constants of such species is the same as described for lactate. The results of the measurements give the following values for the stability constants of (VO)hTf(pic) and (VO)2hTf(pic)2:

$$ { \log }\,\beta [ ( {\text{VO)hTf(pic)]}} = 15 . 6\pm 0.8 $$
(14)
$$ { \log }\,\beta [ ( {\text{VO)}}_{ 2} {\text{hTf(pic)}}_{ 2} ] = 30 . 4\pm 0.8 $$
(15)

A comparison of the log β values measured for the mixed complexes formed by lactate and picolinate allows the following comments to be made: (1) the values of log K for the two systems are comparable (14.5 and 14.0 for lactate and 15.6 and 14.8 for picolinate), with the picolinate ligand showing a slightly greater capability to act as synergistic anion and (2) the difference between the log K values for the two complexation steps is, as expected, less than one log unit (0.5 for lactate and 0.8 for picolinate) and similar to that found for the binary species of transferrin (0.5). This indicates that for a ligand to behave as synergistic anion, it is more important that it fulfills a structural requirement (such as the presence of a carboxylate group and the possibility of being stabilized in the pocket occupied by bicarbonate) rather than being a strong donor [24, 69].

Stability constants for VO(6-mepic)(HSA)(OH) and cis-VO(pic)2(HSA)

The other two species that an insulin-enhancing compound can form in the presence of albumin are VO(6-mepic)(HSA)(OH) and cis-VO(pic)2(HSA), as confirmed by density functional theory calculations [30].

For these two species we were not able to calculate the stability constants from the examination of EPR spectra; thus, in a first approximation, the value of log β for the analogous complexes formed by 1-MeIm [3.82 for VO(6-mepic)(HSA)(OH) and 14.96 for cis-VO(pic)2(HSA), Table 1] will be used.

Discussion

The values of the stability constants reported in this work can be used to calculate the distribution of the species formed in some complicated systems like those recently discussed by some of us [23, 30, 40].

As a first example, we will compare the data predicted for the ternary VO2+/hTf/HSA system with different molar ratios, 2:1:1 and 2:1:17, with those experimentally obtained by EPR spectroscopy [30, 40]. The results are listed in Table 2, together with the percent amount of the species expected on the basis of the log β values calculated in [31].

Table 2 Predicted and measured percent distribution of the VO2+ species in the ternary system VO2+/hTf/HSA
Full size table

It can be noticed that the experimental spectroscopic data are better fitted by the values reported in this work, mainly for the system with the physiological ratio of hTf to HSA (1:17). In particular, it emerges that the presence of the dinuclear species (VO) 2 d HSA should be considered, otherwise the fraction of VO2+ not complexed by transferrin will be incorrectly attributed to (VO)HSA.

Subsequently, the values of the thermodynamic stability constants reported above can be used to evaluate in which form the insulin-enhancing VO2+ compounds are transported in the human body. EPR spectra on the quinary systems formed by an insulin-enhancing agent with [VO(carrier)2] composition (in particular, [VO(6-mepic)2], cis-[VO(pic)2(H2O)], [VO(acac)2], and [VO(dhp)2]), by both the high molecular mass components of blood serum (transferrin and albumin), and one low molecular mass component (lactate or citrate) were presented recently [30]. The spectroscopic results obtained when the molar ratio of the components VO2+/hTf/HSA/carrier/lact was 2:1:3:4:40.8 could be compared with those expected on the basis of the log β values calculated in this work. The data are reported in Table 3.

Table 3 Predicted and measured (in parentheses) percent distribution of the main species formed from the biotransformation of the insulin-enhancing VO2+ agents in the quinary system VO2+/hTf/HSA/dhp/lact
Full size table

In Table 3 it is observed that the predicted and measured percent values of the main species are in good agreement. In particular, the following comments can be made:

  1. 1.

    As suggested by the EPR spectra, in all the systems the presence of (VO)hTf(lact) and (VO)2hTf(lact)2 is expected, and their calculated amounts are in satisfactory agreement with those obtained experimentally [30].

  2. 2.

    As observed in the EPR spectra, in the system with 6-methylpicolinate a small amount of (VO) 2 d HSA is predicted [30].

  3. 3.

    The importance of mixed species formed by carrier and low molecular mass bioligands must not be underestimated, as demonstrated by the case of picolinate: the presence of [VO(pic)(lactH−1)]− species in solution can be clearly detected by EPR spectroscopy [30] and is expected on the basis of the thermodynamic stability constants.

  4. 4.

    The data for dhp agree well with the EPR results obtained in the ternary systems VO2+/dhp/hTf and VO2+/dhp/HSA, which can be explained by considering in solution an equal amount of cis-[VO(dhp)2(hTf)] or cis-[VO(dhp)2(HSA)] and [VO(dhp)2]; only a small percentage of (VO)hTf, (VO)2hTf, (VO)hTf(lact), and (VO)2hTf(lact)2 is expected [30].

  5. 5.

    In the system with dhp, the EPR spectra do not allow cis-VO(dhp)2(HSA) to be distinguished from cis-VO(dhp)2(hTf) owing to their similar A z values [30]; from the stability constants calculated in this work, the mixed species formed by albumin should be the predominant species at physiological conditions (hTF to HSA ratio of 1:17).

  6. 6.

    In the systems with picolinate and dhp, only the sum of (VO)hTf(lact) and (VO)2hTf(lact)2, (VO)hTf(pic), and (VO)2hTf(pic)2, on one hand, and that of (VO)hTf and (VO)2hTf, (VO)hTf(lact), and (VO)2hTf(lact)2, on the other, can be measured from EPR spectra owing to the similarity of their A z value [30]; the calculated sums are, however, in very good agreement with those measured experimentally.

  7. 7.

    With the stability constants of the previous model for the species formed by dhp, in the system VO2+/hTf/HSA/dhp/lact with ratio 2:1:3:4:40.8 only [VO(dhp)2] (53.7%) and the ternary species (VO)hTf(dhp), (VO)2hTf(dhp), and (VO)2hTf(dhp)2 (sum 45.8%) are expected [31]. The EPR spectra reported in Fig. 4, instead, clearly show the presence of (VO)hTf, (VO)2hTf, and the mixed species formed by lactate, (VO)hTf(lact) and (VO)2hTf(lact)2, besides [VO(dhp)2] and cis-VO(dhp)2(hTf) or cis-VO(dhp)2(HSA) [30].

  8. 8.

    The only difference with the experimental data is the possible presence of a small amount of the bis-chelated species formed by lactate, [VO(lactH−1)2]2− [74].

In Table 4 the data expected on the basis of the stability constants reported in this work for the species formed after the biotransformation of [VO(6-mepic)2], cis-[VO(pic)2(H2O)], [VO(acac)2], and [VO(dhp)2], when the concentration of the vanadium drug is in the range necessary to observe insulin-enhancing effects, 1–100 μM [7, 48–50], and that of the bioligands is as in the blood serum, are summarized.

Table 4 Predicted percent distribution of the main species formed from the biotransformation of the insulin-enhancing agents at physiological conditions
Full size table

Until now, it has been affirmed in the literature that when the concentration of the insulin-enhancing agent is between 1 and 10 μM and the carrier is weak (6-methylpicolinate) or of intermediate strength (picolinate and acac), almost all of the VO2+ ion exists in the forms (VO)hTf and (VO)2hTf [31, 37–39]; moreover, the previous results indicate that also with maltol for a concentration lower than 10 μM all of the metal ion should be present in solution as binary species of transferrin [31]. Our data partly confirm these assumptions and suggest that when the concentration of [VO(6-mepic)2], cis-[VO(pic)2(H2O)], [VO(acac)2] is between 1 and 10 μM, most of the VO2+ is bound as binary complex to transferrin, but 4–5% should be present in the form of (VO)hTf(lact) ternary species. When the carrier is strong (dhp), for concentrations around 10 μM the undissociated form of the insulin-enhancing agent and the mixed complex cis-VO(dhp)2(HSA) become important, whereas the concentration of (VO)hTf and (VO)2hTf decreases. For dhp, our results are different from those of Jakusch et al. [31], who found a smaller amount of the binary complexes of hTf and a larger amount of the supposed species (VO)hTf(dhp).

If the concentration of the insulin-enhancing compounds is around 100 μM, both with the weak and the intermediate-strength carriers, the amounts of the species formed by albumin increase, in particular the amount of (VO) 2 d HSA, favored by a lower ratio of HSA to VO2+ [40]; for these ligands, it is possible that mixed species formed by the carrier and low molecular mass bioligands, such as [VO(pic)(lactH−1)]−, or the undissociated form of the insulin-enhancing agent, such as [VO(acac)2], can be important. With strong ligands, such as dhp, [VO(dhp)2] and cis-VO(dhp)2(HSA), in which approximately 85% of the VO2+ is bound, become prevalent, whereas the remaining metal ions exist as the binary complex of transferrin [10.8 and 3.2% of (VO)hTf and (VO)2hTf, respectively].

It could also be noticed that with increasing concentration of insulin-enhancing agent (from 1 to 100 μM) with respect to that of transferrin (37 μM), the percent amount of (VO)hTf decreases (e.g., from 93.3 to 2.2% for the weakest carrier, 6-methylpicolinate) and that of (VO)2hTf increases (from 1.9 to 69.1% for 6-methylpicolinate).

Conclusions and outlook

In this work a quantitative study of the biotransformations in conditions mimicking blood serum of four among the most representative insulin-enhancing vanadium drugs, [VO(6-mepic)2], cis-[VO(pic)2(H2O)], [VO(acac)2], and [VO(dhp)2], was presented. To achieve this goal, the determination of the stability constants of the several species formed after the interaction with high molecular mass and low molecular mass bioligands of the serum was necessary. For the first time in the literature, the value of log β for (VO) 2 dHSA, cis-VO(dhp)2(hTf), cis-VO(dhp)2(HSA), (VO)hTf(lact), (VO)2hTf(lact)2, (VO)hTf(pic), and (VO)2hTf(pic)2 was reported. The correctness of the stability constants measured was confirmed by comparison of the calculated percent amounts in the VO2+/hTf/HSA/carrier/lact quinary system with those measured experimentally from the examination of the EPR spectra.

The data obtained allow a good picture of the real conditions of the organism to be provided when the concentration of the vanadium compound is that necessary to show insulin-enhancing effects. Differently from the conclusions reached previously in the literature, our results suggest that VO2+ ion is mainly bound to transferrin only for weak and intermediate-strength carriers and for concentrations in the range 1–10 μM; in these situations, however, the possibility for the formation of mixed species with a low molecular mass component, such as lactate, must be taken into account. With strong carriers and in all the cases when the concentration approaches 100 μM, the binary complexes of albumin, the mixed species formed by the low molecular mass bioligands, and the ternary complexes formed when the carrier stabilizes the cis-octahedral arrangement in aqueous solution, for example, cis-VO(pic)2(HSA), cis-VO(dhp)2(hTf), and cis-VO(dhp)2(HSA), can become important and must be considered.

Finally, we would like to stress that a future improvement of the values of log β with respect to those reported in this work, the determination of the stability constants not reported in this work, and in vivo studies could help to provide an even better description of the biotransformation and transport of an insulin-enhancing vanadium agent in blood serum.