Electron paramagnetic resonance as a key tool to understand the solution behavior of vanadium compounds: a chemical biology perspective

1 Introduction

Vanadium (symbol V) represents a very special case and occupies a unique position in chemical biology (Crans et al., 2004; Rehder, 2015; Treviño et al., 2019; Ścibior et al., 2020). On one hand, it is a trace element in many organisms, an essential component of V-dependent haloperoxidases and nitrogenases (Eady, 2003; Hu et al., 2012; Leblanc et al., 2015) and a metal ion with a wide range of pharmacological effects (Costa Pessoa et al., 2015a; Crans et al., 2019; Amante et al., 2021; Aureliano et al., 2021; Sharfalddin et al., 2022; Amaral et al., 2023; Rehder, 2023; Ribeiro and Correia, 2024; Miller and Crans, 2025); on the other hand, the chemical form with which it is present in the biological systems is often unknown or not known with certainty (Costa Pessoa and Correia, 2021; Sanna and Garribba, 2021; Dinda et al., 2025). In recent decades, vanadium compounds (VCs) have been investigated for their therapeutic properties, in particular as potential anti-diabetic, anti-cancer, and anti-parasitic agents, but to date, the mechanisms of action remain only partially understood (Costa Pessoa and Correia, 2021; Dinda et al., 2025). A recurring theme in this field is that the biological activity of VCs rarely correlates straightforwardly with the structure of the administered species, precluding the use of rational strategies for the development of V-based drugs (Scior et al., 2016).

The origin of this complexity largely lies in vanadium solution chemistry. In an aqueous solution, V presents three oxidation states, +III, +IV, and +V, that can interconvert into each other. It forms complexes with ligands provided by O- and/or N-donors, ubiquitous in biological systems, including carboxylates, phosphates, amino acids, sugars, and others (Dinda et al., 2025). A plethora of phenomena, such as hydrolysis, ligand exchange, redox reactions, and binding with biomolecules, can occur in biological media, which result in the formation of a mixture of species with different oxidation states, geometries, and chemical features (Dinda et al., 2025). As a consequence, identifying the dominant V species present under physiological conditions and understanding how these species interact with targets in the organism remains a central challenge (Costa Pessoa and Correia, 2021; Sanna and Garribba, 2021).

In this scenario, techniques capable of providing information on the possible transformations in water-containing media are fundamental to understanding the biological role of vanadium and its potential in the medical and therapeutic contexts. Electron paramagnetic resonance (EPR) can partially or completely address this challenge. Similar to other techniques, such as UV-vis and circular dichroism (CD) spectroscopy, nuclear magnetic resonance (NMR), X-ray absorption spectroscopy, X-ray crystallography, and mass spectrometry (MS), EPR presents several pros and cons, which we try to highlight in this Perspective. The technique offers oxidation-state selectivity and sensitivity to the coordination environment of V^IV species, which are central to biological activity because they have a high stability in biologically relevant conditions and can be derived from the oxidation of V^III and the reduction of V^V (Dinda et al., 2025). Moreover, the amount of a sample to be used in an EPR experiment is normally quite small when compared to other techniques, and often no pre-treatment is necessary.

The spectroscopic foundations of EPR, the applications to V^III, V^V, and, mainly, V^IV systems, with the discussion of some case studies, and the advantages and limitations with respect to possible future scenarios will be discussed from a critical point of view. In this Perspective, the role of EPR will emerge, which should be regarded not merely as a supporting technique but rather as a central methodological pillar for elucidating vanadium behavior in chemical biology.

2 Spectroscopic foundations: what EPR reveals about vanadium in solution

EPR spectroscopy is applicable to the study of chemical species containing one or more unpaired electrons, and is, therefore, particularly valuable for the investigation of transition-metal complexes. The theoretical background of the technique is described in several authoritative textbooks and review articles (Mabbs and Collison, 1992; Weil and Bolton, 2007; Drescher and Jeschke, 2012; Rizzi et al., 2016; Goldfarb and Stoll, 2018; Roessler and Salvadori, 2018; Ju et al., 2024; Salikhov et al., 2024). For a mononuclear species with a single unpaired electron, the EPR spectrum is primarily described by the g factor, which determines the resonance positions, and by the hyperfine coupling (HFC) constant A, which reflects the interaction between the unpaired electron and the nuclear spin of the metal center. For rapidly tumbling species, these parameters are expressed as isotropic values (g _iso and A _iso), whereas under rigid-limit or anisotropic conditions, they are resolved into their tensor components (g _*x*, g _*y*, g _*z*, and A _*x*, A _*y*, A _*z*). Because both the *g and A tensors are highly sensitive to the electronic structure of the metal center and to the nature of the coordinated ligands, their analysis provides detailed insight into coordination geometry and bonding features. The tensor A* is characterized by three terms, arising from the isotropic Fermi contact (A ^FC), anisotropic or dipolar hyperfine interaction (*A* ^D), and second-order spin-orbit (SO) coupling (*A* ^SO): *A* = A ^FC 1 + *A* ^D + *A* ^SO, where 1 is the unit tensor (Neese, 2003).

For vanadium(IV) species, with an [Ar]3d ¹ electronic configuration, the HFC interaction is due to the coupling of the electron spin (S = 1/2) with the nuclear spin of the ⁵¹V nucleus (I = 7/2), with the values of A being usually negative. In V^IVO²⁺ complexes with approximate C _4v symmetry and a ground state dominated by the d _*xy* orbital, such as the aquation [V^IVO(H₂O)₅]²⁺, a tetragonal EPR pattern is observed, featuring two distinct g and A values; the expected order of the spin-Hamiltonian parameters is g _*z* < g _*x* ≈ g _*y* < g _e, where g _*e* = 2.0023 is the g factor of a free electron, and |A _*z*| > |A _*x*| ≈ |A _*y*| > 0. The components g _*z* and A _*z* are strongly sensitive to the changes in the equatorial ligand environment: stronger electron-donating ligands generally lead to a reduction in |A _*z*| and a concomitant increase in g _*z*. Experimentally, |A _*z*| values range from approximately 182 × 10⁻⁴ cm^–1 for [V^IVO(H₂O)₅]²⁺ with 4×O coordination to approximately 134 × 10⁻⁴ cm^–1 for [V^IVO(ethane-1,2-dithiolate)₂]²⁻ with 2×(S⁻, S⁻) coordination, passing through intermediate values for [V^IVO(ethylenediamine)₂]²⁺ (155 × 10⁻⁴ cm^–1, 2×(N, N) coordination). The “additivity relationship” assigns individual contributions to equatorial donor atoms and allows estimating A _iso or A _*z* within approximately ±3 × 10⁻⁴ cm^–1 from experimental data (Chasteen, 1981; Smith et al., 2002). In the case of six-coordinate, non-oxido, or “bare” V^IV complexes, the values of |A| are significantly lower, typically falling in the range of (90–155) × 10⁻⁴ cm^–1, in dependence on the coordination geometry and nature of the singly occupied molecular orbital (SOMO) (Sanna et al., 2016).

In dinuclear V^IVO species, the magnetic exchange coupling between the two paramagnetic centers can yield a singlet spin state (S = 0, 2S + 1 = 1) due to an antiferromagnetic coupling of the two unpaired electrons and a triplet state (S = 1, 2S + 1 = 3) due to a ferromagnetic interaction (Rodríguez-Fortea et al., 2004). In this last case, the complex is EPR-active, and the spectral pattern differs markedly from that of mononuclear S = 1/2 systems. The spin-Hamiltonian must include a zero-field splitting (ZFS) term, D, which removes the degeneracy of the M _*S* = ±1 and M _*S* = 0 levels even in the absence of an external magnetic field. Conventional X-band EPR spectra, which use a microwave frequency in the range 9–10 GHz, often show weak or broadened signals. In some cases, a characteristic ΔM _*S* = ±2 transition can be observed around half the magnetic field of the main resonances with ΔM _*S* = ±1.

The anisotropic spectrum of a dinuclear complex with tetragonal symmetry consists of four absorptions: the external lines, corresponding to the orientation of the highest symmetry axis parallel to the magnetic field, are separated by 2D, and the internal lines, corresponding to the perpendicular orientation, are separated by D (Smith and Pilbrow, 1974), which reflects the strength of the magnetic exchange interaction and the spatial arrangement of the two V^IV centers, making EPR a sensitive probe of V···V coupling and the structural features of a dinuclear system. The number of lines for each resonance, when resolved, can be calculated using the formula (2nI + 1), where n is the number of magnetic centers and I is 7/2, with an HFC constant ca. half of that observed for the mononuclear complex with the same equatorial coordination (Garribba et al., 2006). From D, the approximate value of the distance between the two paramagnetic centers (R) can be calculated using the equation D = 1.39 × 10³ (g/R ³) (Eaton et al., 1983).

3 Prediction of spin-Hamiltonian parameters by computational techniques

Spin-Hamiltonian parameters, particularly the HFC *A* tensor, can be accurately predicted by computational techniques, among which is density functional theory (DFT) (Munzarová and Kaupp, 2001; Neese, 2003). The synergistic application of EPR spectroscopy and DFT has proven to be a highly effective approach for determining coordination environments in uncharacterized complexes and has been extensively employed to identify the binding sites of VCs or V fragments within proteins and enzymes.

Benchmark studies on V^IVO and “bare” V^IV complexes showed that different classes of compounds require different levels of theory to reproduce the HFC constants. Meta-GGA functionals tend to underestimate HFC constants, while hybrid functionals with high Hartree–Fock contributions and, above all, half-and-half hybrid functionals, such as BHandHLYP and BHandH, provide the best agreement with the experimental values of A (Saladino and Larsen, 2003; Aznar et al., 2004). The superior performance of these functionals compared to other approaches stems from their improved description of the Fermi contact contribution, which is governed by indirect core-level spin polarization induced by the unpaired spin density in the V-d orbitals and is underestimated by standard DFT treatments. DFT calculations performed at the BHandHLYP/6-311g(d,p) level yield deviations from the experimental |A _z| values below 5% and, in the majority of cases, approximately 3% (Micera and Garribba, 2011b).

For non-oxido V^IV species, the double-hybrid functional B2PLYP was identified as the most accurate for predicting both the experimental A _iso and A _i values (with i = x or z). When combined with the VTZ or 6-311g(d,p) basis sets, this approach reproduces the HFC constants with a percent deviation of approximately 7–8% for |A _iso| and 3–4% for |A _i|; for example, |A _iso| and |A _*z*| of naturally occurring compound amavadin are predicted with deviations of −1.1% and −2.0%, respectively, relative to the experimental data (Sanna et al., 2016).

In contrast, predicting the *g* tensor remains more challenging, as accurate treatment of the SO coupling is required. While DFT offers a practical compromise, multiconfigurational methods provide only limited improvements, highlighting the need for further methodological developments (Singh et al., 2018).

4 V^III complexes

For V^III complexes, which contain two unpaired electrons and have total spin S = 1, ZFS terms must be considered; their magnitude determines whether a conventional X-band EPR spectrum can be detected. In the majority of cases, the large ZFS values result in EPR-silent spectra, or, in some lucky situations, weak or broad signals can be obtained. Therefore, V^III species have received little attention in EPR studies, despite their relevance as reduced forms of V and involvement in the redox cycles. A comprehensive discussion is provided in an authoritative methodological work (Goldfarb and Stoll, 2018).

Two approaches are usually employed for these systems: the use of high-frequency EPR (HFEPR) to overcome the problem related to large ZFS values or an indirect application by monitoring the appearance or disappearance of V^IV signals.

The HFEPR technique enables both qualitative and quantitative analysis, allowing ZFS parameters to be determined and correlated with the underlying electronic structure, ligand field strength, and coordination symmetry (Krzystek et al., 2015). Low temperature is often required to fully characterize these species. For example, the reduction of the potential anti-diabetic agent, [V^VO₂(dipic)]^–, where dipic is the dipicolinate ligand, with one or an excess of the reductant L-ascorbate (Asc), yields the corresponding V^IVO and, finally, mixed-ligand V^III–dipic–Asc species (Horton et al., 2014). A series of V^III complexes, possible models of V-nitrogenase, formed by tetradentate ligands with PS₃ coordination, have been studied with HFEPR; the microwave frequencies used were 106 MHz and 224 MHz (Ye et al., 2010).

When V^III is not directly observable, its presence can be inferred indirectly through changes in V^IV signals, highlighting the diagnostic value of EPR. An indirect application concerned the marine ascidians, where V is intracellularly accumulated in vanadocytes at remarkably high concentrations (Ueki et al., 2015). Monitoring the changes in the EPR signal recorded on vanadocytes after oxidation and reduction processes, Michibata demonstrated that V is present predominantly in the +III oxidation state (Hirata and Michibata, 1991). These results were corroborated by vanadium K-edge X-ray absorption spectroscopy (XAS) (Frank et al., 2008), highlighting the importance of integrating different spectroscopic techniques. In a similar manner, the reduction of V^IV to V^III by the biological reductant cysteine methyl ester has been demonstrated (Islam et al., 2007).

5 V^IV complexes: lessons from model systems

Vanadium(IV) complexes with an S = 1/2 electronic ground state and a magnetically active ⁵¹V nucleus can be easily studied by EPR. V^IVO species can undergo variations in coordination number, typically from five-coordinate geometries (square-pyramidal or trigonal-bipyramidal) to six-coordinate (distorted octahedral) structures. In an aqueous solution, the speciation of V^IVO–L complexes is governed by a combination of factors, including solvent properties, ligand donor strength and denticity, metal concentration, pH, and the presence of competing ligands. These variables determine the distribution of species, the coordination geometries, and the extent of ligand exchange processes. The transformation in solution of V^IVO–L compounds, where L is, for simplicity, a monoanionic bidentate ligand, determines which V-containing species are able to bind to biomolecules. In aqueous media, V^IVO(L)₂ complexes can undergo successive equilibria after the administration, leading to the formation of V^IVO(L)⁺ and, ultimately, V^IVO²⁺, with weak ligands, low pH values, and low V concentration favoring the aqua ion. Moreover, depending on L, transformations or equilibria between the square-pyramidal, cis-octahedral and trans-octahedral species are possible, each of them potentially relevant because they lead to different interactions with the (bio)ligands (Costa Pessoa et al., 2021; Dinda et al., 2025).

5.1 Transformations from square-pyramidal to octahedral V^IVO complexes

EPR spectroscopy unveils changes in axial coordination of V^IVO compounds. Binding of an additional axial ligand, such as a solvent molecule, an organic ligand, or, in a biological system, a protein donor, usually results in a significant decrease in |A _z| and a concomitant increase in g _z, reflecting weakening of the V=O bond and redistribution of the spin density. For example, it was shown that, for the potential anti-diabetic V^IVO complexes formed by certain picolinate derivatives, such as 6-methylpicolinate (mepic), quinaldinate, and kynurenate ligands, an equilibrium is established in solution between the five- and six-coordinate species, according to the reaction [V^IVO(L)₂] + Solv ⇄ trans-[V^IVO(L)₂(Solv)]. EPR spectroscopy provides direct evidence for this transformation as the trans-[V^IVOL₂(Solv)] species exhibits an unusually low |A _z| value. In the case of mepic, |A _z| for [V^IVO(mepic)₂] is 160.8 × 10^–4 cm^–1, in good agreement with the “additivity relationship,” whereas values in the range (148.6–149.0) × 10^–4 cm^–1 are observed for the trans-[V^IVO(mepic)₂(Solv)] with Solv = H₂O, DMF, or DMSO (Gorelsky et al., 2010; Lodyga-Chruscinska et al., 2011). The reduction in |A _z| upon axial coordination of a solvent molecule arises from a decrease in the isotropic hyperfine constant |A _iso|, due to core-shell and valence-shell spin-polarization effects (Munzarová et al., 2000).

5.2 Conversion from square-pyramidal to cis-octahedral V^IVO complexes

Many biologically and pharmacologically relevant V^IVO complexes adopt square-pyramidal geometry in the solid state, such as [V^IVO(acac)₂], [V^IVO(ma)₂], [V^IVO(ema)₂], and [V^IVO(dhp)₂], where acac, ma, ema, and dhp are acetylacetonate, maltolate, ethylmaltolate, and 1,2-dimethyl-3-hydroxy-4(H)-pyridinonate, respectively (Sanna and Garribba, 2021). These VCs can undergo significant coordination changes in solution depending on the solvent and ligand features. The anti-diabetic [V^IVO(ma)₂] adopts a penta-coordinated, square-pyramidal geometry in the solid state and non-coordinating media such as dichloromethane but converts to a cis-octahedral species in coordinating solvents such as water, methanol, or pyridine with two ligands in an equatorial–equatorial and an equatorial–axial arrangement and a solvent in the fourth equatorial position (Hanson et al., 1996; Buglyo et al., 2000). This behavior was later demonstrated to be general for a series of bidentate ligands, including kojate, 8-hydroxyquinolinate, bipyridines, and phenanthrolines (Dinda et al., 2025).

EPR spectroscopy provides direct support for these transformations as the conversion from aqua to hydroxide complexes is accompanied by a decrease in |A _z|, around the deprotonation pK of the equatorial H₂O ligand; for example, |A _z| goes from (170.1–170.8) × 10^–4 cm^–1 for [V^IVO(ma/ema)₂] to (167.1–167.3) × 10^–4 cm^–1 for [V^IVO(ma/ema)₂(OH)]^–. pH-potentiometric measurements confirm ligand-dependent deprotonation of the coordinated water molecule in the pH range 5–11 with the formation of cis-[V^IVO(L)₂(OH)]⁻ (Sanna et al., 2012b). In contrast, no changes in |A _z| are expected for square-pyramidal [V^IVO(L)₂] complexes.

5.3 Temperature and solvent dependence of the equilibria between V^IVO complexes

For some systems, square-pyramidal, cis- and trans-octahedral 1:2 V^IVO complexes have a comparable stability. Variations in temperature or solvent can favor the formation of one or more of these VCs. For example, the bis-complex with 2-hydroxypyridine-N-oxide (hpo), [V^IVO(hpo)₂], is square-pyramidal and pink in the solid state. In solution, at 298 K (RT), it is square-pyramidal in DMF and transforms into cis-[VO(hpo)₂(DMSO)], green, in DMSO. When the temperature is lowered to 77 K (LNT), it forms cis-[VO(hpo)₂(DMF)] in DMF, while it remains penta-coordinated when dissolved in CH₂Cl₂. These transformations can be followed by EPR: the values of |A _z| are 152 × 10^–4 cm^–1 for [V^IVO(hpo)₂] and (167.8–168.8) × 10^–4 cm^–1 for cis-[VO(hpo)₂(Solv)] (Sanna et al., 2012d).

The predominant 1:2 complex of maltolate in MeOH at 297 K is cis-[V^IVO(ma)₂(MeOH)] (I in Figure 1) with |A _iso| = 96.3 × 10^–4 cm^–1, together with a significant amount of trans-[V^IVO(ma)₂(MeOH)] with |A _iso| = 88.0 × 10^–4 cm^–1 (II). The relative amount of trans-octahedral isomer increases markedly with temperature, becoming predominant over the cis isomer at 335 K (trace B of Figure 1). Note that, with increasing temperature, the rapid rotational motion of molecules in solution averages the anisotropic components of the spin-Hamiltonian, providing isotropic signals characterized by progressively narrower lines. In contrast, at 297 K in a weakly coordinating solvent such as chloroform, only the solvent-free complex [V^IVO(ma)₂] is observed (III), characterized by an intermediate isotropic HFC constant (|A _iso| = 90.4 × 10^–4 cm^–1). DFT calculations confirm the experimental observations (Sanna et al., 2021).

5.4 Geometric distortions and steric effects for V^IVO complexes

Spin-Hamiltonian EPR parameters are sensitive to distortions within a given coordination number, which can be important in biological systems. For example, the inhibition of phosphatases by vanadate(V) involves the formation of trigonal-bipyramidal adducts with the covalent binding of a protein residue (McLauchlan et al., 2015). Deviations between g _*x,y* and A _*x,y* are indicative of distortions from square-pyramidal to trigonal-bipyramidal geometry; |g _*x* − g _*y*| and |A _x–A _y| are 0 for a perfect C _4v symmetry and are ≠ 0 for distorted complexes with C _2v symmetry. An example of this transformation has been documented for [V^IVO(L)₂] complexes formed by α-hydroxycarboxylates, salicylaldimine derivatives, and aminophenolates, with distortion increasing with the steric bulk of the ligands and geometric constraints on the V center (Cornman et al., 1997; Garribba et al., 2003; Hagen et al., 2004). When the 1:1 species formed by the biologically relevant glycolate (glyc) and lactate (lact), [V^IVO(glyc/lact) (H₂O)₂], transforms into [V^IVO(glyc/lact)₂]^2– in the presence of a ligand excess, |g _*x* − g _*y*| becomes 0.012, and |A _*x* − A _*y*| to (6.0–7.4) × 10^–4 cm^–1 (Garribba et al., 2003). Parallel changes are associated with the UV-vis spectra (Micera and Garribba, 2011a). The anisotropic spectra of [V^IVO(glyc)(H₂O)₂] and [V^IVO(glyc)₂]^2–, with tetragonal and rhombic symmetry, respectively, are reported in Figure 2.

5.5 Non-oxido V^IV species

Non-oxido V^IV complexes have been isolated from bidentate ligands, such as o-catechols, o-mercaptophenols, and dithiolenes with (O,O), (O,S), and (S,S) donor sets and, also, from tridentate ligands with (O,X,O), with X = N, S, Se, P, and (O,N,S) donors (Dinda et al., 2025). EPR spectra are characterized by A values in the range (90–155) × 10^–4 cm^–1, depending on the metal geometry and composition of the SOMO, allowing for monitoring the transformation of V^IVO into “bare” V^IV species (Sanna et al., 2016).

For example, [V^IVO(cat/dhp)₃]^2–/0, where cat is catecholate, with an HFC constant of (154–157) × 10⁻⁴ cm⁻¹, yields [V^IV(cat/dhp)₃]^2−/+ with (107–120) × 10⁻⁴ cm⁻¹ when CH₃COOH conc. is added to the solution (Buglyo et al., 2002; Rangel et al., 2006). For [V^IV(L)₂] complexes formed by tridentate dianionic (O,N,S) dithiocarbazates, EPR demonstrated that a mixture of mer and fac isomers exists in solution, with mer-[V^IV(L)₂] having a lower |A _*z*| value, (110.2–110.6) × 10⁻⁴ cm⁻¹, than the fac-[V^IV(L)₂] species, (120.6–120.9) × 10⁻⁴ cm⁻¹. These results allowed the determination of the relative stability of the two forms from the spectral signal intensity, consistent with DFT calculations (Banerjee et al., 2023).

Using EPR, it has been demonstrated that in the system V^IVO²⁺/dihydroxyfumarate, a non-oxido V^IV complex with |A _*z*| = 118 × 10⁻⁴ cm⁻¹ is observed, whose formation can be rationalized only by postulating a ligand rearrangement to dihydroxymaleate, which has its two–OH groups in a cis instead of a trans configuration (Sanna et al., 2011b).

Notably, the EPR spectrum of amavadin, a compound isolated from mushrooms of the genus Amanita (da Silva et al., 2013), initially attributed to a V^IVO species due to its large |A _*z*| (≈153.0 × 10⁻⁴ cm⁻¹), was subsequently interpreted in terms of a “bare” V^IV, based on the determination of the stability constant by potentiometry and of the crystallographic structure (Anderegg et al., 1987; De Carrondo et al., 1988; Smith et al., 1997; Berry et al., 1999).

5.6 Ligand exchange in V^IVO species

Ligand exchange is a distinctive aspect of vanadium chemistry, and V^IVO–L complexes, owing to their intrinsic lability, readily form mixed-ligand species in ternary or more complex systems. These transformations occur rapidly in solution and biological media (Dinda et al., 2025). Such behavior is particularly relevant in biological environments, where V-based drug candidates may interact with endogenous low-molecular-mass bioligands (bLs). In blood serum, the most relevant V-binding bLs include lactate, phosphate, citrate, oxalate (ox), and amino acids, while in the cellular environment, reduced glutathione (GSH) and ATP are particularly important (Jakusch and Kiss, 2017).

The conversion of a 1:2 complex, such as [V^IVO(L)₂], to a mixed-ligand species with the formulation [V^IVO(L)(bL)]^*x*–, whose charge depends on bL, has been demonstrated for numerous ligands, such as maltolates, picolinates, pyridinonates, bipyridines, and phenanthrolines (Dörnyei et al., 2006; Jakusch et al., 2011; Sanna et al., 2012a; Sanna and Garribba, 2021). With lactate and citrate (citr), EPR can distinguish between the two possible coordination modes, (COO⁻, OH) or (COO⁻, O⁻), with the hydroxide -OH group protonated or deprotonated. For example, [V^IVO(Me₂phen)(SO₄)], known also as Metvan, one of the first VCs proposed in the treatment of cancer (Narla et al., 2001), when dissolved in a solution containing citr with a metal-to-bL ratio of 1:2, forms [V^IVO(Me₂phen)(citrH)], [V^IVO(Me₂phen)(citr)]⁻, and [V^IVO(Me₂phen)(citrH₋₁)]²⁻. The value of |A _z| of [V^IVO(Me₂phen)(citrH)] and [V^IVO(Me₂phen)(citr)]⁻, formed between pH 2–4 and 3–6, respectively, is the same, 165.1 × 10^–4 cm^–1, the donor set being identical (COO⁻ _eq, OH_eq, COO⁻ _ax). Above pH 6, [V^IVO(Me₂phen)(citrH_–1)]⁻ is revealed with (COO⁻ _eq, O⁻ _eq, COO⁻ _ax) coordination and |A _z| = 158.1 × 10^–4 cm^–1 (Sanna et al., 2017a). The values of |A _z| and the spectral pattern are different from those of the corresponding binary systems.

Similarly, lactate forms with two mixed species with the potential anti-diabetic [V^IVO(pic)₂(H₂O)] complex, [V^IVO(pic)(lact)] with coordination mode (COO⁻, OH) and |A _*z*| = 173 × 10^–4 cm^–1, and [V^IVO(pic)(lactH_–1)]⁻ with (COO⁻, O⁻) coordination and |A _*z*| = 160 × 10^–4 cm⁻¹ (Kiss et al., 2000), with different spin-Hamiltonian parameters than the 1:2 species of pic and lact.

5.7 Di- and oligonuclear V^IVO complexes

The disappearance of the eight-line hyperfine pattern upon dimerization, or the replacement by broadened or weak signals, provides a direct signature of oligomer formation. A feature of V^IVO complexes is the tendency for the hydrolysis of [V^IVO(L)(H₂O)₂]⁺, with two adjacent equatorial sites occupied by water molecules, to yield, at near-neutral pH, di-μ-hydroxide species [(V^IVO)₂(L)₂(μ-OH)₂] or [(V^IVO)₂(L)₂(μ-OH)₂(H₂O)₂] (Felcman and Fraústo da Silva, 1983). Such behavior has been reported for a wide range of biologically relevant ligands, including phosphates, carboxylates, maltolates, picolinates, iminodiacetates, and pyrimidinonates (Gätjens et al., 2003; Kiss et al., 2003; Gonçalves et al., 2013; Sanna and Garribba, 2021; Dinda et al., 2025), and has been substantiated by potentiometry to account for the base consumption observed near neutral pH during the titrations of V^IVO²⁺/L systems. The resulting complexes may be either EPR-active or EPR-silent, depending on the geometry of the bridge and relative orientations of the V=O groups (Plass, 1996; Rodríguez-Fortea et al., 2004). For complexes with bipy, phen, and pyrazinecarboxylates, two broad bands in the ΔM _S = ±1 region centered at g ∼ 2 are observed, plus a 15-line pattern in that ΔM _S = ±2, at ∼1650 G; the bands are separated by HFC constants of (80–84) × 10^–4 cm^–1, approximately half of those observed for the corresponding mononuclear V^IVO species (Garribba et al., 2006; Sanna et al., 2012c). These spectral features are consistent with an anti-coplanar arrangement of the dinuclear core, whereas for ma, koj, pic, and their derivatives, an EPR-silent spectrum is observed, suggesting an anti- or syn-orthogonal bridge structure.

EPR studies of the systems containing D- and L-tartrate (D- and L-tart) are also interesting for their potential anti-diabetic activity and biological relevance (Sakurai et al., 2005) and have been carried out since the 1970s (McIlwain et al., 1977). The isotropic EPR spectra suggest the formation of [(V^IVO)₂(D-tart/L-tartH_–2)₂]^4– in the pH range, indicated by 15 lines spaced by (38–40) × 10^–4 cm^–1, approximately half the |A _iso| value observed for the analogous mononuclear complex. In contrast, meso-tartrate adopts a different coordination mode, leading to a trinuclear species [(V^IVO)₃ (meso-tartH_–1)₂ (meso-tartH_–2)]^4–, with an EPR spectrum with 22 resonances (three magnetic centers) and an HFC constant of 31 × 10⁻⁴ cm⁻¹, one-third of the mononuclear complex with (COO⁻, O⁻) coordination (Kiss et al., 1995).

A distinct dinuclear arrangement involving two ferromagnetically coupled V^IVO²⁺ ions has been identified in fatty acid human serum albumin (HSA^f). The HFC constant is 80 × 10^–4 cm^–1, ca. one-half of the value observed for a V^IVO²⁺ ion bound to 2 N and 2 O donors (Sanna et al., 2009). D is 631 × 10^–4 cm^–1 and is compatible with a V···V distance of ca. 3.2 Å, with the two V^IV ions in the multi-metal binding (MBS) site bridged by the μ_1,1-Asp249 residue (Sciortino et al., 2020).

5.8 Determination of the stability constants of V^IVO complexes

In addition to structural characterization, EPR spectroscopy can be employed to determine the overall stability constants (log β) of V species in solution by measuring the intensity of the signals through the double integration of the first-derivative spectra. The determination of log β for the (V^IVO) (Protein) adduct requires a ternary system containing a V^IVO²⁺ ion, a ligand L (bi- or polydentate), and a protein, with the simultaneous presence of the resonances of (V^IVO) (Protein) and V^IVO(L) or V^IVO(L)₂.

An example is given by the determination of log β for (V^IVO) (HTf), studying the systems with V^IVO²⁺, HTf, and acac or dhp. The EPR absorptions of (V^IVO) (Protein) and [V^IVO(acac/dhp)₂] coexist, with their relative abundance governed by the HTf-to-acac/dhp ratio. The equilibrium between the two species can be described by the displacement reaction [V^IVO(acac/dhp)₂] + HTf ⇄ (V^IVO) (HTf) + 2 acac/dhp. The value of log β for (V^IVO)(HTf) can be obtained with the following relationship: log β = log β₂(V^IVO(acac/dhp)₂) + log K _D, with K _D being the constant of the displacement reaction. In the last equation, the values of log β₂ for the formation of [V^IVO(acac/dhp)₂] are well-known, while K _D is found from the ratio of the concentrations of (V^IVO)(HTf) and [V^IVO(acac/dhp)₂] that, in turn, can be calculated through the double integration of their EPR signals (Kiss et al., 2006; Sanna et al., 2010a).

6 V^V complexes

V^V complexes are diamagnetic and, for this reason, EPR spectroscopy cannot be used directly but can provide information on the oxidation of V^IV or the reduction of V^V species. V^V is reduced to V^IV in model systems by reductants such as GSH, Asc, cysteine (Cys), or reduced nicotinamide adenine dinucleotide (NADH), with Asc more effective than GSH in the case of pharmacologically active V^V–ma/pic complexes (Crans et al., 2004; Wilkins et al., 2006; Jakusch et al., 2014; Sanna et al., 2019).

The interaction of vanadate(V) with erythrocytes has been widely studied by EPR, which shows that V^V, after crossing the red blood cell (RBC) membrane via the anion exchanger (AE1) channel, is reduced inside the cytosol to V^IV by GSH and Asc (Macara et al., 1980; Harris et al., 1984; Garner et al., 1997; Yang et al., 2003; Delgado et al., 2005; Sanna et al., 2014). EPR studies identified several VCs in erythrocytes, including two adducts with Hb, (V^IVO)(Hb^β) and (V^IVO)(Hb^γ), and V^IVO–bL species with bL that can be citrate, a carbohydrate, or a 2-hydroxyacid derivative (Sanna et al., 2014; Levina et al., 2015).

The reduction of decavanadate (V₁₀) to V^IVO²⁺ is observed in the system containing G-actin due to the action of Cys272 and Cys374 residues. V^IVO²⁺ ions interact with the catalytic α-site of G-actin, and the spin-Hamiltonian EPR parameters are compatible with the binding of Asp154 and His161, rationalizing the competition between V₁₀ and ATP for the nucleotide-binding pocket (Sciortino et al., 2021).

In contrast, the disappearance of the EPR signal non-oxido [V^IV(L)₂] complexes, formed by tridentate dinegative ONS dithiocarbazates, which are very active against HT-29 and HeLa cancer cell lines, suggests the oxidation to V^VO₂ species, confirmed by ESI-MS (Banerjee et al., 2023).

EPR results suggested that the system with [V^IVO(acac)₂] and hen egg white lysozyme (HEWL) follows two different pathways, depending on the experimental conditions: at pH 7.5, in 2 M sodium formate, the oxidation to V^V is complete, with the disappearance of the spectral signal and formation of the adducts HEWL–[V^V ₄O₁₂] and HEWL–[V₂₀O₅₄(NO₃)] (Ferraro et al., 2023); at pH 4.0, in 1.1 M NaCl, 0.1 M sodium acetate, the oxidation to V^V is partial and an unresolved resonance, revealed at g ca. 2, suggests a magnetic interaction between neighboring paramagnetic V^IV centers, accounting for the formation of the adducts HEWL–[V^IV,V ₂₀O₅₁(OH₂)], HEWL–[V^IV,V ₁₅O₃₆(OH₂)], and HEWL–[V^IV,V ₁₅O₃₃(OH₂)] (Tito et al., 2024).

7 Vanadium complexes in biological media

7.1 Model systems

Model ligands, such as imidazole and carboxylate derivatives used to mimic histidine and aspartate/glutamate coordination, have provided valuable benchmarks for interpreting EPR spectra of V–Protein adducts. The correspondence between spin-Hamiltonian EPR parameters in the model and protein-containing systems can allow for the confirmation of the characterization of the binding sites.

Histidine coordination occurs above pH 6–7, when deprotonation of the imidazole-N takes place. Except for His binding at the iron sites of human transferrin (HTf) and MBS of HSA, this interaction is generally non-specific and involves the most solvent-exposed His residues on the protein surface. Histidine coordination can be effectively modeled using 1-methylimidazole (MeIm) (Liboiron et al., 2005). Notably, both the stability constants and the |A _*z*| values of Protein–[V^IVO(L)₂] adducts closely match those of cis-[V^IVO(L)₂ (MeIm)]. For example, the adducts apo-HTf/HSA/IgG–[V^IVO(ma)₂]_*n*, where IgG is immunoglobulin G, exhibit log β values of 19.2–19.6 and |A _*z*| of (164.6–165.5) × 10^–4 cm^–1, comparable to the values of 19.12 and 164.8 × 10^–4 cm^–1 for cis-[V^IVO(ma)₂(MeIm)] (Sanna and Garribba, 2021), the number n of coordinated fragments depending on the accessible His protein residues. The values of log β for Protein–[V^IVO(L)₂] can be calculated from the intensity of the EPR signals of the adduct Protein–[V^IVO(L)₂] and bis-chelated complex [V^IVO(L)₂(H₂O)] following the same approach discussed in Section 5.8. For comparison, the stability constants (log K) for the reaction Protein + [V^IVO(ma)₂(H₂O)] ⇄ Protein–[V^IVO(ma)₂] + H₂O are in the range 2.9–3.3 for apo-HTf, HSA, and IgG.

7.2 Protein binding of V^IV complexes

The binding of V^IV species with proteins can be non-covalent and covalent (Sciortino et al., 2019; Costa Pessoa et al., 2021). Non-covalent protein binding occurs when the first coordination sphere of the VC remains intact, and no ligand exchange occurs, the interaction being stabilized by secondary forces such as H-bonds and van der Waals contacts (Sciortino et al., 2019; Costa Pessoa et al., 2021). Non-covalent interaction has been demonstrated for a series of VCs, such as [V^IVO(L)(H₂O)]⁺, [V^IVO(L)(H₂O)₂]⁺, [V^IVO(L)₂], and cis-[V^IVO(L)₂(H₂O)] plus POVs with different nuclearity (Costa Pessoa et al., 2015b; Sciortino et al., 2019; Costa Pessoa et al., 2021; Aureliano et al., 2022; Santos and Costa Pessoa, 2023; Dinda et al., 2025; Ferraro et al., 2025). The EPR spectrum at RT can be of two types: if the interaction is weak, the signal is comparable to that of the free VC; if the interaction is strong, the reduced rotational mobility results in an anisotropic or a slow-tumbling spectrum even at RT.

Covalent binding entails coordination of one or more protein donors, originating from side chains or the backbone, that replace labile ligands (e.g., H₂O or L) in the VC. For V^IVO species, the most common coordinating residues are Asn, Asp, Glu, His, and Ser. The binding can be specific, as in the case of iron-binding sites of apo-HTf or MBS of HSA, or non-specific, as happens when accessible Asn, Asp, Glu, His, and Ser donors interact with a V-containing fragment. The covalent binding can be demonstrated by comparing the EPR spectra of V^IVO–L species and V^IVO–L–Protein adducts (Santos et al., 2014; Costa Pessoa et al., 2015b; Sciortino et al., 2019; Costa Pessoa et al., 2021; Sanna and Garribba, 2021; Dinda et al., 2025).

7.2.1 Covalent binding of V^IVO(pic)₂ to proteins

The [V^IVO(pic)₂(H₂O)] complex is a potential anti-diabetic agent (Sakurai et al., 1998). As with other V^IVO complexes, it shows a cis-octahedral geometry, and, in the presence of a protein, the weak equatorial aqua ligand can be replaced by a protein side-chain donor, generally Asp/Glu or His. When cis-[V^IVO(pic)₂(H₂O)] interacts with HSA or IgG at pH 7.4 (Figure 3, trace A), the values of |A _*z*| for HSA–[V^IVO(pic)₂]_*n* and IgG–[V^IVO(pic)₂]_*n* (159.7 × 10⁻⁴ cm⁻¹ and 160.0 × 10⁻⁴ cm⁻¹ (Sanna et al., 2010b, 2011a); see traces E-F of Figure 3) are similar to those of the mixed-ligand species cis-[V^IVO(pic)₂ (MeIm)] (158.8 × 10⁻⁴ cm⁻¹ (Sanna et al., 2010b); see trace D). This means that one or more His-N replace the aqua ligand to yield adducts with composition HSA–[V^IVO(pic)₂]_*n* and IgG–[V^IVO(pic)₂]_*n*, where n corresponds to the number of His accessible on the protein surface.

With HEWL–[V^IVO(pic)₂], |A _*z*| is 163.3 × 10⁻⁴ cm⁻¹ (Sciortino et al., 2017; trace B of Figure 3), intermediate between that of cis-[V^IVO(pic)₂(H₂O)] (165.0 × 10⁻⁴ cm⁻¹; trace A) and of the model complex cis-[V^IVO(pic)₂(MeIm)] (158.8 × 10⁻⁴ cm⁻¹; trace D), suggesting that the protein binds V^IVO²⁺ center with a donor stronger than water-O and weaker than His-N. Based on the EPR parameters, it could be a COO⁻ group stemming from an Asp or Glu residue, suggesting that, when histidines are poorly exposed, as happens for the only histidine residue of HEWL (His15), their coordination is often inhibited. X-ray crystallographic and docking calculations corroborate the EPR prediction, showing that HEWL binds the V^IVO(pic)₂ moiety with Asp52 (Santos et al., 2014; Sciortino et al., 2017). Notably, both the spectra obtained at RT with a suspension of the crystals in buffer and with the same suspension frozen at LNT exhibited a hyperfine pattern, with comparable g and A parameters and |A _*z*| = 164.5 × 10⁻⁴ cm⁻¹ (Santos et al., 2014).

EPR spectra recorded in the system containing bovine pancreatic ribonuclease (RNase A) (Figure 3, trace C) exhibit the resonances of an adduct with |A _*z*| = 163.0 × 10⁻⁴ cm⁻¹ with the possible coordination of an Asp/Glu-COO^–. Crystallographic and computational studies confirm that RNase A binds to the V^IVO(pic)₂ fragment through the Glu111 side chain, which replaces the equatorial H₂O ligand. Notably, the shoulder at a lower magnetic field (at ca. 4057 G, denoted by the hashtag in trace C) indicates that alternative binding modes are plausible, involving coordination of His105 and His119 to the V^IV center to yield another adduct with |A _*z*| = 159.7 × 10⁻⁴ cm⁻¹ (Ferraro et al., 2021).

In light of these conclusions, the spectral pattern of the system with HSA can be reinterpreted for the first time in this Perspective. The resonance with the asterisk (Figure 3, trace E) with |A _*z*| ≈ (163–164) × 10⁻⁴ cm⁻¹ would suggest the coordination of exposed Asp/Glu residues of HSA to cis-V^IVO(pic)₂.

7.2.2 Non-covalent interaction of V^IVO–flavonoid complexes to HEWL

The rotation motion of a complex strongly influences the appearance of EPR spectra. In frozen solutions, for instance at LNT, small VCs generally display rigid-limit spectra due to the absence of molecular motion, whereas at RT, rapid rotational tumbling leads to isotropic spectral features (Sciortino et al., 2018). Conversely, when a V^IV species forms adducts with large biomolecules such as proteins, the reduced rotational mobility results in spectral signals approaching the rigid-limit or slow-tumbling regime even at RT (Campbell and Freed, 1980).

In the case of the VCs with flavonoid ligands, which display significant anti-cancer activity (Naso et al., 2023), such as [V^IVO(que)₂]^2–, [V^IVO(mor)₂], [V^IVO(7,8-dhf)₂]^2–, [V^IVO(chr)₂], or [V^IVO(5-hf)₂] (where que, mor, 7,8-dhf, chr, and 5-hf denote quercetinate, morinate, 7,8-dihydroxyflavonate, chrysinate, and 5-hydroxyflavonate, respectively), the interaction with proteins is predominantly non-covalent. The strength of this interaction is largely determined by the number of hydroxy or carbonyl groups not directly involved in V coordination, which are engaged in H-bonds with polar residues on the protein surface, such as Arg, Trp, Ser, Asn, Gln, Asp, and Glu, and can be determined by EPR. VCs containing four such free groups, as with que and mor ligands, exhibit particularly strong interactions, sufficient to suppress molecular rotation on the EPR timescale and yield spectra characteristic of the rigid-limit regime. When only one free functional group is present, as in the 7,8-dhf, chrysinate, or 5-hf species, the resulting spectra fall between the rigid-limit and slow-tumbling regimes. For comparison, complexes lacking free OH or CO groups, such as [V^IVO(acac)₂] or [V^IVO(cat)₂]^2–, display fully isotropic EPR spectra (Sciortino et al., 2018).

7.2.3 Use of V^IVO as an EPR spin probe

The use of V^IVO²⁺ ions as a spin probe represents a well-established strategy to investigate metal binding sites in proteins, owing to the marked sensitivity of its EPR parameters to the changes of coordination environment (Chasteen, 1981). This method is based on substituting the native, often diamagnetic or EPR-silent metal ion with the paramagnetic V^IVO²⁺, whose *g and A* tensors provide detailed information on ligand identity, coordination geometry, and electronic structure. In particular, variations in the HFC A _*z* reflect the nature of the equatorial donor atoms, allowing O-, N-, S-, or mixed donor sets to be distinguished, while changes in the spectral pattern can reveal the presence of multiple binding sites or equilibria between different species bound to the sites. As a result, studies using a V^IVO²⁺ spin probe enable the indirect characterization of metal binding environments in proteins under conditions in which direct spectroscopic or X-ray analysis is not possible.

For example, in carboxypeptidase A, substitution with V^IVO²⁺ allowed for determining the variation of the binding site of the Zn²⁺ ion with pH. The values of A _*z* demonstrate the coordination of COO⁻ side chains belonging to Asp or Glu residues at acidic pH, while around neutrality His-N donors and water molecules are involved in the catalytically relevant metal binding site (DeKoch et al., 1974). A similar approach has been applied to imidazole glycerol phosphate dehydratase (IGPD), a Mn²⁺-dependent enzyme, where the binding of V^IVO²⁺ to the apoenzyme revealed the presence of multiple sites that coexist in solution. The corresponding EPR signals, characterized by distinct A _*z* values, indicate the coordination of two or three His-N in the equatorial or axial position, reflecting a large flexibility of the metal site (Petersen et al., 2000). Vanadium-dependent bromoperoxidase (VBrPO) was investigated upon reduction of the active form, which contains a trigonal-bipyramidal V^V bound to His418. EPR investigations of the reduced form with V^IVO suggested a pH dependence of A _*z* (De Boer et al., 1988). This behavior has been rationalized in terms of protonation equilibria of the coordinating residues, such as His or Ser, which influence both the geometry and the electronic structure of the metal center (Sciortino and Garribba, 2020).

Finally, it should be noted that systems containing V^IV as an EPR probe can be studied by pulsed techniques. For instance, approaches such as electron-nuclear double resonance (ENDOR), electron spin echo envelope modulation (ESEEM), and hyperfine sublevel correlation (HYSCORE) allowed the detection of nuclei in the first or second coordination sphere weakly coupled with ⁵¹V, providing a complete description of metal-biomolecule interactions (Schweiger and Jeschke, 2001).

7.3 Speciation in serum and cellular media

A major recent advance has been the application of EPR spectroscopy to increasingly complex biological systems, including blood serum. By combining EPR with thermodynamic modeling and complementary analytical techniques, it has become possible to rationalize the distribution of vanadium among competing bLs as a function of total V concentration. These studies demonstrate clearly that vanadium speciation is concentration-dependent.

Extending EPR studies to cellular environments represents an important frontier. Although technically challenging, such studies can give information on the V speciation in the crowded milieu of the cell, providing insights into uptake, transformation, and possible biological targets.

7.3.1 Rationalization of the pharmacological action of non-oxido V^IV complexes

As discussed in Section 5.5, non-oxido V^IV compounds are relatively rare because their formation entails cleavage of the V=O bond. A systematic investigation of the anti-diabetic properties of six V^IV complexes revealed that only three “bare” species have a significant pharmacological activity, in some cases greater than that of V^IVOSO₄, while the other three are ineffective. The results of the anti-diabetic activity can be rationalized using EPR spectroscopy, which has shown that anti-diabetic action is observed only for compounds undergoing conversion to the corresponding V^IVO species, such as [V^IV(hyph^Ph)₂], [V^IV(pdbh)₂], and [V^IV(cat)₃]²⁻, where hyph^Ph = 3,5-bis(2-hydroxyphenyl)-1H-1,2,4-triazolate(2–) and pdbh = pentan-2,4-dione benzoylhydrazonate(2–). The conversion of V^IV to the V^IVO form can be monitored by EPR, which shows a significant increase in the largest HFC constant, shifting from ca. (95–145) × 10⁻⁴ cm⁻¹ to ca. (155–165) × 10⁻⁴ cm⁻¹. In contrast, compounds that maintain the non-oxido form, such as [V^IV(taci)₂]⁴⁺, [V^IV(inoH₋₃)₂]²⁻, and [V^IV(dhab)₂], where taci = 1,3,5-triamino-1,3,5-trideoxy-cis-inositol, ino = cis-inositol, and dhab = 2,2′-dihydroxyazobenzenate(2–), are inactive, highlighting that the role of the V=O group is crucial for anti-diabetic behavior because it allows the inhibition of protein tyrosine phosphatases (PTPs) (Yoshikawa et al., 2014).

7.3.2 Speciation of biologically active [V^IVO(L)₂] in blood serum

EPR spectroscopy allows for following the speciation of pharmacologically active [V^IVO(L)₂] in biological fluids. It can be effectively employed for the quantitative evaluation of the equilibria between V^IVO complexes and biomacromolecules, allowing the determination of the binding constants and the analysis of competitive processes in complex media, including blood serum. This approach enables discrimination between free and protein-bound VCs and provides valuable information on their thermodynamic stability and distribution (Kiss et al., 2008; Jakusch et al., 2011).

When blood serum is incubated with [V^IVO(ma)₂] or [V^IVO(dhp)₂], the systems can be analyzed by EPR, and the results compared with model calculations. The data can be summarized as follows. Upon incubation of serum with [V^IVO(ma)₂] at a V concentration of 454.5 µM (Part 1 of Figure 4), the adduct Protein–[V^IVO(ma)₂] (species III in Figure 4) predominates, where the protein may correspond to apo-, holo-HTF, HSA, or IgG, while smaller amounts of cis-[V^IVO(ma)₂(H₂O)] (species I) and transferrin-bound forms, (V^IVO)(HTf)/(V^IVO)₂(HTf) (IV) are also detected. The results almost perfectly match the predictions based on the thermodynamic stability constants. When the V concentration is reduced to 90.9 µM, more than half of the V^IVO²⁺ ion is associated with transferrin as adduct IV, with the remaining fraction present as Protein–[V^IVO(ma)₂] (III) and cis-[V^IVO(malt)₂(H₂O)] (I), in good agreement with the expected distribution.

Studies on the [V^IVO(dhp)₂] system (Part 2 of Figure 4) further demonstrate a strong consistency between computational predictions and the experimentally observed speciation. At a V concentration of 454.5 µM, the EPR signals of the predominant species correspond to Protein–[V^IVO(dhp)₂] (species VII in Figure 4). Under these conditions, only minor contributions from bis-chelated species, cis-[V^IVO(dhp)₂(H₂O)] and [V^IVO(dhp)₂] (Va and Vb) are observed, while the fraction of V^IVO associated with transferrin is negligible (IV). At a lower V concentration of 90.9 µM, Protein–[V^IVO(dhp)₂] remains the dominant form (VII), whereas the contributions from bis-chelated complexes and transferrin-bound species become comparable (Sanna et al., 2017b).

Comparable agreement between calculated speciation profiles and experimental behavior was observed for other VCs, including [V^IVO(acac)₂], [V^IVO(pic)₂(H₂O)], and [V^IVO(mepic)₂].

7.3.3 Speciation of the potential anti-cancer agent vanadocene in serum

Another clear example of correspondence between models and behavior in serum that can be demonstrated by EPR is provided by [Cp₂V^IVCl₂] or vanadocene (VDC). Because its direct interaction with serum proteins is negligible, the distribution of VDC among the main low-molecular-mass bLs, ox, lact, HPO₄ ^2–, and HCO₃ ⁻, which have a physiological serum concentration of 9.2 μM, 1.51 mM, 1.10 mM, and 24.9 mM, respectively, (Harris, 1992), was studied by EPR spectroscopy using the physiological ratios between the four components. The spectra recorded at VDC/ox ratios of 1:1 and 10:1, corresponding to serum V concentration of 9.2 µM and 92 μM, respectively, showed that at ≈10 μM, [Cp₂V^IV(ox)] dominates in solution; at ≈100 µM, [Cp₂V^IV(CO₃)] is the major species with smaller amounts of [Cp₂V^IV(ox)] and [Cp₂V^IV(lactH_–1)]; finally, at intermediate concentrations (10–100 µM), VDC distributes between oxalate and carbonate. Notably, the experimental spectra match the model system obtained using the stability constants of the four complexes and physiological concentrations of the bLs (Sanna et al., 2015).

7.3.4 Biotransformation of cis-[V^IVO(pic)₂(H₂O)] in red blood cells

Another example of VC transformation in the cellular environment, demonstrated by EPR, involves [V^VO₂(pic)₂]^– and its derivatives in RBCs. When RBCs are incubated with [V^VO₂(pic)₂]^– (pH 7.4), it dissociates into H₂V^VO₄ ⁻ and picolinate, which enter the cell independently, with H₂V^VO₄ ⁻ transported via the AE1 channel. Inside the cytosol, H₂V^VO₄ ⁻ is reduced by Asc and GSH to V^IVO²⁺, which, in the presence of picolinate, yields cis-[V^IVO(pic)₂(H₂O)], whose formation is indicated by |A _*z*| = 165.0 × 10⁻⁴ cm⁻¹. Subsequently, this latter species reacts with hemoglobin, GSH, and other thiol-containing proteins to produce mixed-ligand V^IVO complexes, each one countersigned by specific spin-Hamiltonian EPR parameters (Sanna et al., 2019).

7.3.5 In vivo blood circulation-monitoring EPR experiments

Yasui et al. employed an in vivo blood circulation-monitoring EPR (BCM-EPR) approach to collect EPR signals in real time in order to investigate the relationship between biological activity and speciation of relevant V-based potential anti-diabetic drugs. Application of this method allows the direct observation of V^IVO species, countersigned by eight resonances in the isotropic spectra recorded on the circulating blood of rats after the intravenous administration of [V^IVO(mepic)₂]. Moreover, analysis of the EPR data using compartmental pharmacokinetic models revealed substantial distribution of V^IVO complexes into peripheral tissues and elimination predominantly via urinary excretion (Yasui et al., 2000).

7.4 Application of EPR to vanadium enzymes

Although the majority of EPR investigations of vanadium in biological environments have focused on low-molecular-mass complexes and their interactions with biomolecules, EPR spectroscopy has also contributed significantly to the study of vanadium-dependent enzymes. Among these, V-nitrogenase represents one of the most fascinating systems, as it catalyzes the biological reduction of dinitrogen to ammonia and contains a FeV-cofactor. Recent structural and spectroscopic studies have significantly improved the knowledge of the electronic structure of this cofactor. In particular, EPR spectroscopy has played an important role in identifying not only the oxidation and spin states but also the magnetic interactions within the FeV cluster, providing new insights into the catalytic site and its role in nitrogen reduction (Hoffman et al., 2014; Yang et al., 2021; Davydov et al., 2025; Wahl et al., 2025).

8 Advantages and limitations of EPR in vanadium chemical biology

The main advantage of EPR is its intrinsic selectivity toward paramagnetic species. In vanadium chemistry, this translates into an exceptional ability to probe V^IV species, even in the presence of non-Kramers V^III or diamagnetic V^V forms. This oxidation-state selectivity is a decisive strength in V chemical biology, where redox interconversion is common, and other techniques that determine the total metal content cannot identify the number and structure of the species existing under physiological conditions. The EPR-active V^IV is central in the V chemistry because it is the product of the oxidation of V^III and the reduction of V^V and is stabilized upon complexation by many low-molecular-mass bioligands and proteins. Moreover, the redox processes can be indirectly studied by EPR through the disappearance or the appearance of the spectral signals in a biologically relevant system.

A second major strength is the richness of structural information obtained by spin-Hamiltonian EPR parameters. The *g and A* tensors provide direct insight into chemical form, coordination geometry, ligand donor set, covalency of metal–ligand bonds, and axial ligation. Moreover, EPR reveals the coexistence of multiple species in solution; the number of species in solution corresponds to the number of hyperfine resonances, often well resolved in an anisotropic spectrum, even when other spectroscopic or analytical techniques could suggest a single dominant species or give unresolved bands due to the overlap of several absorptions. In many cases, EPR provides the only direct experimental evidence for subtle geometric features, such as an axial binding or square pyramid-trigonal bipyramid distortion, that otherwise can remain inaccessible in solution. Few spectroscopic techniques reach this level of detail for metal centers in solution, particularly when crystallographic data are unavailable or biologically irrelevant.

EPR is also sensitive to dynamic processes. Changes in line shape and/or in anisotropy can be related to molecular tumbling, ligand exchange, and interaction strength with biomolecules.

A distinctive advantage of EPR over techniques such as UV-vis, CD, and NMR spectroscopy is that it does not require optically transparent samples and can be applied to solids, frozen solutions, viscous media, and heterogeneous systems. In addition, EPR typically requires relatively small sample amounts and minimal sample preparation. These features make it particularly suitable for studying complex matrices and multiphase systems, where the application of many spectroscopic techniques is often limited. The ability of EPR to distinguish coexisting paramagnetic species makes it a unique tool for speciation studies in complex biological environments.

Finally, EPR spectroscopy is highly complementary to other analytical techniques. When combined with thermodynamic modeling, computational chemistry, UV-vis, circular dichroism, NMR, X-ray crystallography, or mass spectrometry, EPR provides an essential piece of the speciation puzzle, often representing the only direct probe for paramagnetic species. The results collected with EPR can be compared with the predictions obtained using the stability constants of the species existing in solution.

Despite these strengths, EPR spectroscopy also has intrinsic limitations that must be taken into account. The most obvious is that EPR is blind to diamagnetic species. In vanadium chemical biology, this means that V^V, which is an important oxidation state under various physiological conditions, is not directly observable by EPR. Interpretation of EPR data, therefore, requires careful consideration of redox equilibria and, ideally, integration with complementary techniques.

Sensitivity is also a limiting factor. Although modern EPR spectrometers are highly sensitive, detection limits may not be satisfactory in biological samples where V concentrations are very low. In healthy individuals, these concentrations typically fall within the nanomolar range (approximately 0.2–15 nM; see Costa Pessoa et al., 2015a), whereas after oral administration of drug candidates, the serum levels reach the micromolar range (approximately 1–10 µM in humans and 40–60 µM in rats treated with V-based agents (Thompson et al., 2003; Zhang et al., 2008; Thompson et al., 2009)). This issue becomes particularly relevant for in-cell or in vivo studies, where signals are very weak due to the low concentration, compromising the interpretation of the data. To partly bypass this problem, EPR spectra with a satisfactory or good signal-to-noise ratio could be obtained through the repeated acquisition of many spectra with weak signals (Sanna et al., 2009).

Another challenge could be the spectral complexity. When multiple V^IV species coexist, EPR spectra may consist of many overlapping signals, and a careful simulation and deconvolution could be required to interpret the results. While this complexity is not a weakness in itself, it demands expertise to avoid possible misinterpretations.

Experimental conditions, such as temperature, solvent composition, and sample preparation, can influence EPR spectra. Therefore, care must be taken to ensure that the measured spectra reflect biologically relevant conditions and are not artifacts arising from the freezing, concentration effects, or non-physiological environments. In this context, the effect of temperature must be outlined. If, on one hand, low temperatures are requested to observe anisotropic signals that provide more detailed insight into (i) the symmetry and coordination geometry of V^IV species, (ii) the nature of the equatorial ligands and (iii) the presence of minor species in solution that could not be detected from an isotropic spectrum, on the other hand, they can also stabilize species that are not favored or may not even exist at RT (see Section 5.3).

Finally, two other points must be highlighted: the relatively high price of an EPR spectrometer, which has, for its nature, a more limited application than other techniques such as UV–vis, NMR, or MS, and the requirement, particularly for HFEPR and pulsed ENDOR, ESEEM, or HYSCORE, of highly specialized researchers to properly interpret the data obtained.

9 Outlook: future directions for EPR in vanadium chemical biology

This Perspective shows that EPR spectroscopy is a cornerstone in the study of vanadium chemical biology. Over the past decades, EPR has evolved from a descriptive to a predictive technique, and the fields of application have shifted from predominantly chemical systems to biological and medicinal applications. It is a uniquely informative and selective tool to unveil speciation, transformations, and redox processes of V under conditions close to physiological. Allowing for directly determining V^IV, occasionally accessing V^III, and indirectly determining the EPR-silent V^V, EPR plays a central role in elucidating the behavior of VCs in biologically relevant systems. In light of the advantages and limitations illustrated in Section 8, the future of EPR spectroscopy in vanadium chemical biology appears to be particularly bright if the methodological advances are applied to relevant biological and pharmacological issues. The next phase will likely exploit these features to address increasingly complex systems and biologically realistic environments.

A major direction of development concerns the application of high-field and high-frequency EPR (HFEPR). Up to now, EPR spectra were recorded at X-band frequencies (9–10 GHz), while measurements at Q-band (usually 34–35 GHz, but up to 50 GHz), W-band (ca. 90–95 GHz), and higher frequencies significantly enhance spectral resolution, allowing for resolving subtle differences in *g and A* tensors, which are often obscured at conventional X-band frequencies. For vanadium systems, this improvement is particularly valuable for resolving overlapping signals arising from multiple V^IV species in equilibrium in solution. Therefore, HFEPR will play a key role in improving speciation models in complex solutions, including biological fluids.

A second important direction of progress involves the use of pulsed EPR techniques, such as ENDOR (Murphy and Farley, 2006), ESEEM (Deligiannakis et al., 2000), and HYSCORE (Schweiger and Jeschke, 2001). These methods enable direct detection of hyperfine interactions with nearby nuclei, including hydrogens, nitrogen, phosphorus, and other heteroatoms commonly found in biological ligands. In the context of vanadium chemical biology, pulsed EPR offers the possibility to determine second-sphere interactions, H-bond networks, and weak coordination modes that are not detectable with continuous-wave (CW) EPR. Such information could be central for understanding how VCs are stabilized or destabilized upon interaction with a biological matrix.

Equally important will be the integration of EPR spectroscopy with computational chemistry. DFT calculations of *g and A* tensors have reached a level of reliability that allows meaningful comparison with experimental data, provided that appropriate structural models and levels of theory are used. In our view, the most powerful applications arise when EPR and computation are combined iteratively: experimental spectra can guide the construction of valuable models, while the calculations, in turn, can help to interpret the behavior and structure of unknown VCs. This approach could be extended by performing molecular dynamics to consider weak interactions and possible conformational flexibility.

In the near future, advances are expected in the improvement of the sensitivity, integration of EPR with complementary spectroscopic and computational methods, and refining quantitative speciation approaches in biologically relevant media. As mentioned above, the V concentration ranges from a few nM in healthy individuals to some μM in humans or rats treated with V drugs, the latter being the limit with conventional CW EPR spectrometers.

The quantitative application of EPR spectroscopy remains challenging, as the direct proportionality between signal intensity and concentration is frequently compromised by relaxation effects, experimental conditions, and different saturation of the paramagnetic species. Addressing these limitations will be essential to fully exploit EPR as a quantitative tool for V^IV speciation in complex biological environments.

The possibility of using EPR in in situ studies and for in-cell measurements seems more remote.

Finally, we emphasize the importance of methodological integration. No single technique can provide a complete description of vanadium behavior in biological systems. Therefore, the future impact of EPR will depend on its combination with complementary methods, such as NMR for diamagnetic species, X-ray absorption spectroscopy for oxidation-state and coordination analysis, and mass spectrometric techniques, such as electrospray ionization–mass spectrometry (ESI-MS) and inductively coupled plasma–mass spectrometry (ICP-MS) for identifying which species exist in solution near physiological conditions and quantifying total V distributions. Within such an integrated framework, EPR will play a central and irreplaceable role as the most direct tool for identifying and characterizing paramagnetic VCs in solution.

By adopting innovative methodologies and integration with other techniques, future EPR studies will not only deepen our understanding of V chemistry in solution but also help to clarify how this versatile metal exerts its biological and pharmacological effects in living systems.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Debbie C. Crans, Colorado State University, United States

Reviewed by: João Costa Pessoa, University of Lisbon, Portugal

Brian Hoffman, Northwestern University, United States