Metabolic pathways of carcinogenic chromium.

The products of hexavalent chromium [Cr(VI)] reduction by glutathione (GSH) alone or in the presence of equimolar quantities of aspartate (Asp) and/or glutamate (Glu) and a chromium-containing material extracted from bovine liver were studied by ultraviolet-visible spectrum (UV-vis) studies, electrospray mass spectrometry (ES-MS), electron paramagnetic resonance (EPR), and nuclear magnetic resonance (NMR). Reduction of chromate by GSH was followed by UV-vis and NMR, revealing the formation of a paramagnetic complex in which GSH acts as a ligand. ES-MS and EPR measurements provided unequivocal evidence of a dimeric Cr(V)(2)GSH(2) species in which the two metal ions are bridged by the Gamma-Glu carboxylate. The analysis of the (1)H and (13)C shifts experienced by GSH protons and the values of paramagnetic contributions to proton spin-lattice relaxation rates provided a set of constraints for structural determination. The same experiments were repeated in the presence of an equimolar concentration of Asp, revealing the formation of a dimeric Cr(V) paramagnetic complex in which the two metals are now bridged by Asp. Nuclear magnetic resonance dispersion profiles show that water is not displaced by Asp and that the correlation time of this complex is slowed by the increased complexity. When Glu is also included in the solution in equimolar concentration to GSH and Asp, data are consistent with the formation of many mono- and dinuclear species, with the three ligands competing with each other. Finally, the spectroscopic investigation of the chromium-containing material extracted from bovine liver revealed the presence of a complicate mixture of Cr(IV) or Cr(V) complexes, among which some Cr(V)-GSH species are present alone or with other ligands in the metal coordination sphere.

Human exposure to hexavalent chromium compounds has been firmly linked with the induction of cancer (1,2). Chromate, or dichromate, in fact, easily enters cells through the sulfate channel (3) and is quickly reduced by glutathione (GSH) (4), ascorbic acid (4), or cysteine (Cys) (4,5). In exposed animals, the ultimate step of the metabolic pathway yields Cr(III) inserted within the cell nucleus, where it cross-links DNA to proteins (mainly actin) or GSH (6,7). The intermediate steps have not been thoroughly characterized so far, although Cr(V) and Cr(IV) compounds have been demonstrated to produce reactive oxygen species, thiyl, and carbon-based radicals (8)(9)(10)(11)(12).
LMWCr has been investigated extensively for its chemical and biological features (28,29,34): • The protein is isolated with the best yield from bovine liver, with few separation procedures (10-30 mg of protein from 1 kg of diced liver). • Ultraviolet-visible spectrum (UV-vis), electron paramagnetic resonance (EPR), and nuclear magnetic resonance (NMR) features suggest the presence of a peptide-bound cluster of four oxo-bridged octahedral Cr(III) ions. • The protein a) increases the ability of rat adipocytes to metabolize glucose (in the presence of insulin); b) stimulates glucose metabolism in a way proportional to its chromium contents; and c) does not change the concentration of insulin required for half-maximal stimulation. • The average molecular weight is approximately 1.5 kDa. • The apo-protein may be generated (29), but no delineation has been provided of its chemical and spectroscopic features. All these, and other properties (35), are therefore flawed by the impossibility of determining the primary sequence of the protein (29).
We also considered that the reported composition of LMWCr roughly corresponds to two GSH, two Asp, and two Glu molecules, and that the intermediate steps of metabolic reduction of Cr(VI) are worth further investigation.
For these two reasons, we present our spectroscopic UV-vis, electrospray mass spectrometry (ES-MS), and EPR and NMR studies of a) the products of Cr(VI) reduction by GSH alone or in the presence of equimolar quantities of Asp and/or Glu, and b) the chromium-containing material extracted from bovine liver.
A first report of some preliminary results has been published elsewhere (36).

Materials and Methods
All chemicals were obtained from Sigma Chemical Co. (Milano, Italy) and used without further purification. The extraction of the chromium-containing material from the bovine liver was performed by exactly following the already published protocol (29). Solutions were obtained in deuterium oxide buffered at pH 7.4 and deoxygenated by freeze-thaw cycles.
NMR spectra were obtained on 4.7, 11.7, 14.1, and 18.8 T Bruker NMR spectrometers (Bruker, Rheinstetten/Karlsruhe, Germany) at temperatures controlled at ± 0.2 K. Chemical shifts were referenced to external tetramethylsilane. COSY (correlation spectroscopy), TOCSY (total correlation spectroscopy), NOESY (nuclear Overhauser effect spectroscopy), and ROESY (rotating frame Overhauser effect spectroscopy) two-dimensional experiments were performed by using standard sequences. While TOCSY and COSY relate through bonds-coupled nuclei, NOESY and ROESY relate nuclei close in space coupled via dipolar interaction. Spin-lattice relaxation rates were measured with inversion recovery pulse sequences and calculated by exponential regression analysis of the decay curves of longitudinal magnetization components.
X-band EPR spectra were acquired on a Bruker 200D SRC spectrometer (Bruker) at temperature = 298 ± 0.5 K. Microwave frequencies were measured with an XL Microwave 3120 counter (Jagmar, Krakow, Poland). The spectrometer was interfaced with a PS/2 Technical Instruments Hardware computer, and the data were acquired using the EPR data system CS-EPR, produced by Stelar Inc. (Mede, Italy). The spectra were preliminarily corrected for baseline drift. EPR spectra at 282 GHz were acquired under the following conditions: temperature = 10 K, microwave frequency = 282 GHz, field modulation frequency = 9.50 kHz, modulation amplitude = 40 mA, time constant = 3 s, sweep rate = 0.3 T/min. The simulated spectrum was calculated by using a program written by Weihe described in Jacobsen et al. (37). For two unpaired electrons considered as point dipoles, the following formula can be applied (38): [1] where D is one of the parameters for zero field splitting (the other, E, is zero). This formula yields D = -0.0811 cm -1 when the two dipoles are 0.414 nm apart. The other parameter for the simulation was the single-crystal EPR line width taken at 100 mT.
ES-MS spectra were recorded with a Perkin Elmer Sciex triple quadrupole liquid chromatography/mass spectrometry/mass spectrometry (LC/MS/MS) with a water/ formic acid (equilibrated at pH 7.4) mobile phase. The ion at m/z 717 was selected and passed through a collision cell into a second mass analyzer. Molecular structures were generated by the HYPERCHEM software package (39) implemented on a Pentium 120-MHz PC by using the Huckel semi-empirical method for charge calculations and the MM+ force field for molecular mechanics and dynamics calculations (39).

Results and Discussion
The reduction of chromate by GSH at pH 7.4 was first investigated by UV-vis spectroscopy ( Figure 1). As previously noted (4), the progressive disappearance of the chromate absorption at approximately 345 nm was accompanied by the appearance of a strong broad absorption centered at 435 nm and attributed to a Cr(VI)-thioester species (40). That this last is a transient species is clearly demonstrated by the kinetic plot shown in Figure 2: within the first 10-15 min, it has almost disappeared and only a strong absorption in the UV region at approximately 206 nm is left, suggesting exclusive formation of Cr(V) or Cr(IV) species (4). NMR spectra, recorded at 14.1 T on a solution containing GSH and GSSG (the oxidized form of GSH) at a ratio of 5:1 (pH 7.4) upon progressive addition of sodium chromate (Figure 3), allowed us to make relevant inferences on the formed species. The relevant resonances are labeled in Figure 3 and are related to the chemical formulas in Figure 4. The spectra clearly demonstrate formation of GSSG at the expense of GSH. Lowering and raising, respectively, of the intensities of well-separated lines is in fact observed. However, the extensive broadening and the upfield shift experienced by resonances belonging to GSH indicate the formation of a paramagnetic complex in which GSH acts as a ligand. In fact, both 1 H and 13 C-NMR chemical shifts were selectively affected, and proton spin-lattice relaxation rates were selectively enhanced ( Figures 5, 6).
Assignment of binding donors of GSH was based on analysis of NMR parameters yielding the following evidence: • Cys H α , γ-Glu H α , γ-Glu H β , and γ-Glu H γ were the most upfield-shifted GSH 1 H signals upon addition of chromate. • All proton spin-lattice relaxation rates were consistently enhanced, with protons within the Cys and Glu moieties being more affected than those of Gly. • Cys C α , γ-Glu C α , and γ-Glu C β were the most affected GSH 13 C signals upon addition of chromate. Both the change in chemical shifts and the enhancement of relaxation rates are larger at closer distances from the metal center that provides dipolar and hyperfine local fields at the nearby nuclei (41,42). The dipolar contribution to the chemical shift relies on a large value of the anisotropy of magnetic susceptibility (42), which does not manifest in EPR spectra. Only the contact contribution therefore needs to be considered, given by: [2] where A is the contact coupling constant, h the reduced Planck's constant (≡h/2π), µ B is the electron Bohr magneton, S is the total spin number, γ N is the magnetogyric ratio, k is the Boltzmann constant, and T is temperature. This equation states that an electron spin magnetic moment determines the shift of coupled nuclear resonances. The size of the shift experienced by any nuclear spin in a given metal complex is therefore determined by A, which in turn is proportional to the unpaired electron spin density at the nucleus.
On the other hand, the relatively long electron relaxation times expected for S = 1/2 metal ions (41,42) make the contact contribution to nuclear spin-lattice relaxation negligible. The longitudinal relaxation rate of a nucleus sensing the electron magnetic moment can therefore be approximated by the dipole-dipole interaction, as described by Solomon (43). Solomon's equation considers the dipole-dipole interaction of nuclear and electron magnetic moments at a certain distance r from each other. The obtained relaxation rate is determined by the squared dipole-dipole interaction energy that is modulated by molecular tumbling in solution. The mathematical expression is (43) [3] where µ o is vacuum permeability, ω H and ω e are the proton and electron Larmor frequencies, τ c is the motional correlation time, and r is the proton-metal distance. The isotropic hyperfine coupling of the unpaired electron with the metal nuclear spin modifies the Solomon equation (44); moreover, the Curie mechanism (42) is likely to contribute the nuclear relaxation pathway at the high magnetic field used. However, the dependence upon the inverse of r to the sixth power is maintained in any case. As a consequence, a large relaxation rate enhancement indicates spatial proximity of the paramagnetic center to the corresponding proton.
NMR data therefore support the exclusion of Cys sulfhydryl and Gly carboxylate from the possible donor set to the paramagnetic chromium ion.
The results obtained by ES-MS and EPR (both at 9.5 and 282 GHz) provided unequivocal evidence of the predominance of a dimeric Cr(V) 2 GSH 2 species (36) in which the two metal ions are bridged by the γ-Glu carboxylate, as confirmed by the structure of several polynuclear chromium complexes (34). Structural details of this cluster are shown in Figure 7; it is apparent that the γ-Glu amino and the γ-Glu amide groups complete the coordination donor set to chromium (36).
Repetition of the same experiments in the presence of Asp at a concentration equimolar to GSH yielded the following evidence:  relaxation enhancements of GSH are further enhanced (Figure 8). • EPR spectra indicate occurrence of a similar species with the same electronic g factor value and slightly broader signals (Figure 9). • NOESY spectra give evidence of cross-peaks between Asp protons and GSH protons, indicating that both participate in metal binding ( Figure 10A). • 1 H-NMR dispersion curves of water protons ( Figure 11) demonstrate that water bound to paramagnetic chromium in the Cr(V) 2 GSH 2 species is not displaced by Asp, with consequent decrease in the relaxivity. On the contrary, relaxation rates of water protons are further enhanced, thus indicating that, most probably the correlation time has slowed in the presence of Asp, which enhances the molecular complexity. We therefore conclude that both Asp and GSH bind chromium in a dimeric Cr(V) 2 GSH 2 Asp 2 complex, the structure of which was indeed solved by molecular mechanics and dynamics restrained by constraints obtained from NMR data ( Figure 12). It turns out that the two Cr(V) ions are now bridged by Asp, which lets them reach a Cr-Cr distance of 0.343 nm to be compared with the Cr distance found for the Cr(V) 2 GSH 2 complex (0.414 nm). This shortened distance is consistent with the broader EPR lines and, taken together with the increase in molecular complexity, also with the Asp-induced enhancement of proton relaxation rates of either GSH or bound water.
When Glu was also included in the solution mixture exposed to chromate additions in equimolar concentration with both GSH and Asp, the following results were obtained: • 1 H-NMR signals of GSH, Asp and Glu were all affected by reduced chromate, but the relaxation enhancements of GSH and Asp are somehow reduced (Figure 8).
• No GSH-Glu or Asp-Glu cross-peaks are detectable in NOESY spectra, in which the GSH-Asp connectivities are washed out also ( Figure 10B). • 1 H-NMR dispersion curves of water protons ( Figure 11)     Water R1 (s -1 ) ν 0 (MHz)   of GSH, Asp, and Glu in the same metal complex can be reasonably excluded. We therefore conclude that Glu, Asp, and GSH compete with each other for chromium, with the possible simultaneous formation of many mono-and dinuclear species. Although GSH alone or together with Asp succeeds in clustering chromium, the three ligands together do not cluster chromium and give rise to complex equilibria in solution.
The material extracted from bovine liver was finally investigated. Although the already published protocol (29) was exactly followed and repeated 3 times, what we found was quite different from the published features of LMWCr (28,29,34): • UV-vis spectra ( Figure 13) are very similar to those obtained in the GSH/GSSG mixture upon addition of chromate at pH 7.4 and exclude whatever possibility of having Cr(III) in the solution. • ES-MS spectra ( Figure 14) exclude the occurrence of detectable species at m/z > 786.4. • The first recorded NMR spectrum ( Figure  15A) is consistent with a very broad paramagnetic signal that superimposes to a wellresolved diamagnetic region. Moreover, a down-shifted broad signal at +65 ppm appears that somewhat resembles the one observed with LMWCr (29). However, the spectrum obtained at higher frequency ( Figure 15B) with an NMR "old-fashioned" probe (without gradients) reveals that this feature is an artifact; on the contrary, some features do appear in the upfield-shifted region (-20 ↔ -40 ppm). These findings confirm the absence of Cr(III) and indicate the occurrence of paramagnetic Cr(IV) or Cr(V) compounds. • The EPR spectrum ( Figure 16) is again very similar to that obtained in the GSH/GSSG mixture upon addition of chromate. All the obtained results indicate that what we have extracted from bovine liver is a complicate mixture of Cr(V) complexes, among which some Cr(V)-GSH species are present alone or with other ligands in the metal coordination sphere.
We conclude that what we have investigated so far suggests that GSH not only is a primary target for oxidation by chromate but also acts as an efficient ligand stabilizing Cr(V) in a dimeric bridged cluster. When Asp is also present in the medium, it may act as an additional ligand to Cr(V) with formation of a new type of dinuclear cluster. If Asp and Glu are both present in the medium, together with GSH, the three ligands compete for the metal and a complicate mixture of species in equilibrium is observed. It is worth emphasizing that what we obtain by reducing chromate in vitro with GSH, GSH + Asp, or GSH + Asp + Glu represents a good model for the mixture of species that is detected when dichromate, added to homogenized liver in the first step of the extraction of LMWCr (29), is reduced and complexed by endogenous ligands occurring in the cell cytoplasm.