One and Two Dimensional Pulsed Electron Paramagnetic Resonance Studies of in vivo Vanadyl Coordination in Rat Kidney

The biological fate of a chelated vanadium source is investigated by/n vivo spectroscopic methods to elucidate the chemical form in which the metal ion is accumulated. A pulsed electron paramagnetic resonance study of vanadyl ions in kidney tissue, taken from rats previously treated with bis(ethylmaltolato)oxovanadium(IV) (BEOV) in drinking water, is presented. A combined approach using stimulated echo (3-pulse) electron spin echo envelope modulation (ESEEM) and the two dimensional 4-pulse hyperfine sublevel correlation (HYSCORE) spectroscopies has shown that at least some of the VO2+ ions are involved in the coordination with nitrogen-containing ligands. From the experimental spectra, a 4N hyperfine coupling constant of 4.9 MHz and a quadrupole coupling constant of 0.6 + 0.04 MHz were determined, consistent with amine coordination of the vanadyl ions. Study of VO-histidine model complexes allowed for a determination of the percentage of nitrogen-coordinated VO2+ ions in the tissue sample that is found nitrogen-coordinated. By taking into account the bidentate nature of histidine coordination to VO2+ ions, a more accurate determination of this value is reported. The biological fate of chelated versus free (i.e. salts) vanadyl ion sources has been deduced by comparison to earlier reports. In contrast to its superior pharmacological efficacy over VOSO4, BEOV shares a remarkably similar biological fate after uptake into kidney tissue.


INTRODUCTION
The insulin-enhancing effects of vanadium complexes and salts have been well documented/1/. Despite a large volume of work, the mechanism(s) of action remain(s) to be delineated. Difficulties encountered in such studies arise from, among other factors, an incomplete understanding of the metabolism of chelated (i.e. coordination complexes) and "free" sources of vanadium (i.e. the common V(IV) salt vanadyl sulfate,). The uptake and storage of vanadium species within specific tissues is of considerable interest both fbr the development of a complete metabolic model, and for detection of the putative "active species" (as yet undetected) ultimately responsible for insulin enhancement. The biodistribution and organ accumulation of exogenous vanadium species have been widely studied, particularly after the discovery of vanadium's insulin-enhancing effects. Bone, kidney and liver appear to be the primary organs of vanadium accumulation, irrespective of administered oxidation state and administration route/2-6/. Differences between organically chelated and inorganic sources appear to be minimal; the benchmark vanadium complex bis(maltolato)oxovanadium(IV) (BMOV, Figure 1) was preferentially accumulated in the same three tissues, although at concentrations 2-3 times higher than the inorganic salt VOSO4/5/. These similarities are likely due to an almost immediate dissociation of the chelating maltolato ligand in the bloodstream, as demonstrated in a recent report fi'om our laboratories/7/. Earlier studies were limited due to the use of radioactive labeling, insensitive to both the coordination and the oxidation state. Thus, important information regarding the chemical form of accumulated vanadium is lacking, while such knowledge could shed some light on the structure of the active species.
Spectroscopic methods, however, can be used to gain insight into solid, solution and in vivo chemical structures. These methods must possess high selectivity to avoid saturation of the spectrum with unwanted features. For detection of VO 2+ in vivo, the, paramagne,tism of its complexes can be used to considerable  advantage. Due to the strong electron spin magnetic moment, the relative magnetic susceptibility of an electron is 1800 times greater than that of the proton, therefore, detection limits in biological samples can reach the micromolar range. Additionally, paramagnetic resonance methods are naturally selective for paramagnetic species only; any interference from competing diamagnetic substances (e.g. V(V) complexes generated from in vivo oxidation of an administered V(IV) complex) is completely obviated. Thus, it would seem clear that EPR methods should be exceedingly useful in the study of the in vivo coordination structure of V(IV) species.
One of the first applications of pulsed EPR techniques to a true biological sample was reported by Fukui et al. in 1995/8/. Wistar rats were treated by intravenous injection of VOSO4 for 4 days, after which time they were sacrificed. Kidney and liver samples were then recovered and analyzed by continuous wave (cw) EPR and 2-pulse ESEEM spectroscopies. For both kidney and liver tissue, peaks identified with 4N coupling and inner sphere protons from coordinated water molecules were observed. After assigning each peak, and processing using first order perturbation theory, the isotropic hyperfine 4N couplings were compared to model VO(histidine) solutions, leading to the conclusion that the paramagnetic vanadyl centers had become ligated by one or more amines in these tissues. While previous studies of in vivo vanadium distribution have been conducted/2,3,5/, little information was available regarding the chemical state, namely the oxidation and coordination states, of the metal ions.
The application of pulsed EPR techniques permitted the acquisition of in vivo structural data on vanadyl ions within tissue samples, in turn potentially identifying a short-term storage form or a vanadyl-protein complex responsible for insulin enhancement. The 48V-VOSO4 and 4V-BMOV study by our group revealed that, at least for BMOV, orally administered, chelated vanadyl sources were preferentially accumulated in bone, liver and kidney/5/. The Fukui et al. /8/study was limited in that it used a first-generation (i.e. inorganic) antidiabetic agent (VOSO) coupled with an intravenous route of administration.
We recently reported the in vivo coordination of vanadyl ions in bone mineral/9/, as well as a meticulous spectroscopic investigation of the VOZ+:triphosphate system as a model of the in vivo complex/10/. A wide variety of one (2, 3 and 4-pulse ESEEM) and two dimensional (HYSCORE) pulsed EPR techniques was used. These reports provided the first detailed study of the biological fate of chelated vanadyl sources in bone mineral, the primary site of vanadium accumulation. In this report, we present a spectroscopic study of vanadyl ions in BEOV-treated rat kidney samples. While earlier work reported on VOSO4-treated rat kidney samples/8/, the work reported herein resolves several deficiencies in the previous study in that we present full 2 and 3-pulse one dimensional ESEEM spectra in addition to high resolution HYSCORE spectra for the complete delineation of the in vivo coordination of vanadyl ions in tissue. Further, the in vivo and model system spectra are presented herein as the Fourier transformation, eliminating signal distortion arising from application of other linear prediction analysis techniques such as the maximum entropy method (MEM). MEM processing produces very idealized, noiseless spectra from which information regarding the shape, width and intensity of ESEEM harmonics is lost; such spectra do not give any impression about the real quality of the frequency-domain data. Lastly, a more detailed analysis of the experimental spectra is provided, including comparisons to pulsed EPR studies of applicable chemical systems reported in the more recent literature. previously flushed with Ar. Samples were sealed with parafilm, then re-frozen in dry ice and kept on dry ice or at -20 C prior to spectroscopic study. EPR spectra (both cw and pulsed) were obtained with a Bruker Elexsys E580 X-band spectrometer, interfaced with a Bruker ER035M teslameter for field calibration and a built-in Bruker frequency counter for microwave frequency measurement. Liquid He temperature studies were carried out using an Oxford Instruments CF 935 cryostat and an ITC 502 temperature controller. The samples were studied by several different ESEEM methods, the aforementioned two-pulse spin-echo sequence, a three-pulse stimulated echo experiment, and a two-dimensional hyperfine sublevel correlation (HYSCORE) four-pulse sequence. The three-pulse spectrum has the distinct advantage over the two-pulse spin echo experiment in that the modulation depth is much less dependent on the phase memory time. In addition, the spectrum is simplified due to the absence of nuclear sum and difference frequencies. The pulse sequence of rt/2<-rt/2-T-r/2-'-echo leads to formation of a stimulated echo observed time, after the third pulse. In the HYSCORE experiment (r/2<-rt/2-t-rt-tz-rt/2--echo)/11/, the intensity of the stimulated echo after the fourth pulse is measured with t2 and t varied, and constant. This two-dimensional (2D) set of echo envelopes gives, after complex Fourier transformation, a 2D spectrum with equal resolution in each direction. The basic advantage of the HYSCORE technique is the creation in 2D spectra of off-diagonal cross-peaks whose coordinates are nuclear frequencies from opposite electron spin manifolds. The cross-peaks significantly simplify the analysis of congested spectra by correlating and spreading out the nuclear frequencies /12/. In addition, the HYSCORE experiment separates overlapping peaks along a second dimension and enhances the signal-to-noise ratio through the application of a second Fourier transform /11,12/. Detailed consideration of the orientationally-disordered (i.e. powder or frozen solution) HYSCORE spectra for 1/2 and nuclei is given elsewhere/13/.

RESULTS AND DISCUSSION
Early in the study it was observed that no spectroscopic differences existed between the CT and DT tissue samples, and so all further discussion refers to spectra of tissue taken from CT rats treated with BEOV.
The first derivative of the field-sweep electron spin echo (FS-ESE) spectrum of BEOV-treated rat kidney is shown in Figure 2. The FS-ESE experiment collects echo intensity from a two-pulse sequence at various magnetic field values and is analogous to the adsorption mode of the EPR spectrum. The spectrum is typical for axial, magnetically dilute vanadyl ions, with spin Hamiltonian values gll 1.945 + 0.005, All 168 + x      complex in liver and kidney tissue, and as a reference for the quantitative determination of the percentage of the coordinated ions over the total VO 2+ ions in the sample/8/. They did not specify the structure of the complex and most likely assumed that the coordination of the histidine molecule occurred via the amine group only. However, detailed 2D ESEEM study of VO-imidazole and VO-histidine complexes has demonstrated that histidine is in fact a bidentate ligand coordinating VO 2+ ions via the ct-amine and imine nitrogens /16/. ESEEM spectra of VO-histidine complexes are composed of the contributions of the coordinated amine nitrogen (with hyperfine coupling ,4 5.0 MHz, and a quadrupole coupling constant K 0.58 + 0.02 MHz) and the coordinated imine nitrogen of the imidazole (,4 6.3 MHz, and K 1.02 :+/-0.07 MHz). The special peculiarity of this system is that the intensity of the lines from the amine nitrogen in twoand three-pulse spectra is significantly larger than from the imine nitrogen/16/; these differences were not recognized in the MEM spectra reported by Fukui  One and Two Dimensional Pulsed Electron Paramagnetic Resonance Studies .?f in vivo Vanad),l Coordination nuclei) relative to the proton frequencies is significantly smaller for the in vivo sample than for the model complexes, thus indicating that only part of the VO 2+ present in the biological sample is coordinated to nitrogen atoms, or that only one nitrogen atom is coordinating the VO 2+ in vivo instead of the two in the model complexes. This difference is also manifested in the corresponding modulus FT plots shown in Contained within each frequency-domain spectrum in Figure 6 is a triplet of lines at frequencies <8 MHz; each triplet includes two dq transitions and a new intense line corresponding to the combination of the sq transitions, Vsq/ ('2) + Vsq. (2'). The frequency of this peak, to a first-order approximation, is equal to the nitrogen hyperfine coupling constant. Comparison of the spectra of the natural abundance and N-labeled VO(his)2 complexes reveals clear differences between them in the relative intensity of the triplet components as well as in the lineshapes of two lower intensity lines between 10 and 12 MHz. These differences reflect the additional contribution of the 4N imine nitrogens to the ESEEM pattern. Upon SN substitution, this contribution is eliminated, producing only very minor influences on the ESEEM spectra, but at different frequency ranges compared to t4N. 16 As a result, the VO(1-SN-his)2 complex is a more correct model for the quantitative determination of the percentage of nitrogen-coordinated VO 2/ ions. The percentage of the total VO 2/ concentration involved in nitrogen coordination can be estimated from the depth of nitrogen ESEEM in the time-domain spectra. However, the ESEEM spectra in Figure 6 show the presence of peaks from protons with matrix fi-equencies of14 and 28 MHz, which are near exact multiples Sergei A. Dikanov et al.
Bioinorganic Chemistr), and Applications of the dq frequencies of 3.7 and 7.0 MHz. Such concurrences might affect the modulation depth of the nitrogen signals due to coincidences of the modulation minima from the two types of nuclei (InN and H), especially at the initial part of the ESEEM patterns where the proton modulation is deep. Filtration of the high frequencies (> 12 MHz) from the two-pulse ESEEM patterns of Figures 5a, 5c and 5e yields the patterns shown in 5b, 5d and 5f, respectively, eliminating modulation depth distortions caused by the presence of proton signals in the ESEEM patterns. The decay of the signal in each of these patterns has been fitted with an arbitrary exponential decay curve, for the estimation of the initial modulation depth for each sample (vide infra).
If we define Vr(x) as the normalized ESEEM with VN(0) from one amine nitrogen, then the experimental echo intensity is described by Equation 2. Ema--x [VN(/)lmax (4) d2 Emin fill. N (,i;)]mi n }2 If it is assumed that in the in vivo sample only part of the VO 2+ ions are nitrogen-coordinated, then each normalized intensity is described by Equation 6 (x is the percentage of the vanadyl ions in the sample which are nitrogen-coordinated).
The modulation depth in this case is described by Equation 7.

Resonance Studies of in vivo lffmadyl Coordination
If we introduce the related parameter,, or modulation amplitude, we can obtain Equation 8.
By expanding upon the binomial fraction shown above we obtain the modulation amplitude relative to dl Because the exact function Vu(x)max is unknown, an estimate of the value of x using available experimental data must be employed. Equations 1-10 detailed above provide a method by which reasonably accurate estimates can be obtained from experimental spectra. In orientationally-disordered samples, the modulation depth is inversely proportional to the interpulse time; the modulations decrease with increases in the time between the two pulses of the ESEEM experiment. Therefore, the most accurate estimate of x can be obtained from d and d measured at the shortest possible times x of the experimental envelopes where VN(X)max is close to 1.
The measurement of modulation depth was performed on two-pulse ESEEM patterns obtained (after filtration of the proton frequencies, vide supra) for the in vivo sample and two model complexes (VO(his These values would progressively increase as the value of VN(X)max decreases. If, for instance, VN(X)max was found to be 0.8, then the estimate for x gives 0.67 and 0.50. Thus, by using the unlabeled VO(his)2 complex as a basis for a quantitative comparison of the nitrogen-coordinated vanadyl ions in the biological sample, Fukui et al. underestimated this value by 10-17%. Our comparison, however, was made with BEOV-treated tissue, obtained from rats subjected to a profoundly different dosing procedure; we discuss these differences in the next section. This work provides intriguing comparisons to earlier reported in vivo data on the bioaccumulation of vanadyl species. We have shown in the previous section that oral administration of BEOV and subsequent bioaccumulation in kidney tissue results in at least partial degradation of the complex; as much as 67% of the detected ions are found in the nitrogen-coordinated form. This percentage is in fact higher than that previously determined for intravenously-administered VOSO4 /8/, indicating that the oral administration pathway subjects vanadyl complexes (at least those studied to date) to significant biotransformation activities that result in loss of ligand(s) prior to organ accumulation. This conclusion is supported by our earlier study of vanay! ons (from BEOV) in bone mineral which also demonstrated complex degradation prior to accumulation in the bone matrix/9/. Fukui et al. reported 2-pulse ESEEM spectra of kidney samples, taken from rats previously treated (via intraperitoneal injection) with bis(picolinato)oxovanadium(IV) (VO(pic))( Figure 1) and VOSO 4 /2/.
Contained within the ESEEM spectra of VO(pic) samples was a weak signal corresponding to a coordinated imine nitrogen, suggestive therefore of the presence of one of the original picolinato ligands in the first coordination sphere of the paramagnetic ions. We do not detect imine coordination at all in our study, which would seem to corroborate the observations of Fukui et al./25/. Their observation, however, does raise some interesting mechanistic considerations in light of our results. In the current work, BEOV was administered via drinking water to rats, therefore the paramagnetic vanadyl ions must have been absorbed from the. gastrointestinal tract, transported to the kidneys via the bloodstream, and ultimately absorbed again into the renal tissue. Administration by intraperitonea! injection, however, would serve to bypass most if not all but the final process. The VO(pic) found in the kidney tissue would therefore bypass several potential sites of biotransformation and hence potentially arrive in the tissue as the intact complex. Current data in our group and others indicates that the BMOV family of complexes is susceptible to degradation via reaction with serum proteins such as apo-transferrin and albumin/26/. Since the thermodynamic stability of VO(pic) is actually several orders of magnitude lower than that of BEOV (log [3 11.99 for VO(pic)7 versus 16.43 for BEOV) /7/, it is implausible that VO(pic) would survive as the intact complex in the bloodstream. Additionally, greater than 60% of the vanadyl ions present in the kidney sample (from the original BEOV source) is found in the amine-coordinated form; considering the greater thermodynamic stability of BEOV over VO(pic) and the vigorous route of administration (oral versus intraperitoneal injection), it seems likely, all conditions being equal, that an orally-administered VO(pic) sample would undergo a much greater degree of biotransformation than that indicated in the Fukui et al. study. Thus, the possibility that the imine-.amine coordinated vanadyl species (i.e. the original VO(pic) complex) detected in the Fukui et al. study is in some way responsible for the augmented anti-diabetic effects is remote. Further, since the kidneys themselves have little involvement in carbohydrate and lipid metabolism, it is in fact much more likely that the aminecoordinated species detected in both studies is an end-product of in vivo transformation of the administered complex, downstream of the anti-diabetic effect(s). Volume 1. No. 1,2003 One and Two Dinlensional Pldsed Electron Paratnag. netic Resonance Studies qfin vivo Vanadyl Coordination CONCLUSIONS By utilizing pulsed EPR techniques, we have demonstrated that the biological fate of chelated vanadium sources such as BMOV is remarkably similar to that of free sources (e.g. VOSO4) in rat kidney tissue. Administered VO 1+ from either BMOV or VOSO4 becomes nitrogen-ligated, in approximately the same proportion. A detailed consideration of the time-domain spectra of the in vivo sample and two model complexes shows that modulation depths can be used to obtain reasonable estimates of the percentage of the total detected paramignetic species interacting with the nucleus of interest. Such an analysis requires, however, isolation of the undistorted modulation depth arising from the nucleus under study. We have presented an improved method for determining the relative proportion of a particular species, and shown that a previously reported method underestimates this value. With the biological fate in the kidney of both chelated and free vanadium sources remarkably similar, the superior insulin-enhancing activity of chelated sources such as BMOV must be a result of either a key difference in some other metabolic pathway, or merely increased gastrointestinal absorption via oral administration. Due to the predominantly excretory role of the kidneys, it is likely that this difference(s) occurs prior to accumulation of administered vanadium in these organs.