l-Glutamic Acid γ-Monohydroxamate

We report that the vanadium ligandl-Glu(γ)HXM potentiates the capacity of free vanadium ions to activate glucose uptake and glucose metabolism in rat adipocytes in vitro (by 4–5-fold) and to lower blood glucose levels in hyperglycemic rats in vivo (by 5–7-fold). A molar ratio of two l-Glu(γ)HXM molecules to one vanadium ion was most effective. Unlike other vanadium ligands that potentiate the insulinomimetic actions of vanadium,l-Glu(γ)HXM partially activated lipogenesis in rat adipocytes in the absence of exogenous vanadium. This effect was not manifested by d-Glu(γ)HXM. At 10–20 μm l-Glu(γ)HXM, lipogenesis was activated 9–21%. This effect was approximately 9-fold higher (140 ± 15% of maximal insulin response) in adipocytes derived from rats that had been treated with vanadium for several days. Titration of vanadium(IV) withl-Glu(γ)HXM led to a rapid decrease in the absorbance of vanadium(IV) at 765 nm, and 51V NMR spectroscopy revealed that the chemical shift of vanadium(IV) at −490 ppm disappeared with the appearance of a signal characteristic to vanadium(V) (−530 ppm) upon adding one equivalent of l-Glu(γ)HXM. In summary,l-Glu(γ)HXM is highly active in potentiating vanadium-activated glucose metabolism in vitro and in vivo and facilitating glucose metabolism in rat adipocytes in the absence of exogenous vanadium probably through conversion of trace intracellular vanadium into an active insulinomimetic compound. We propose that the active species is either a 1:1 or 2:1l-Glu(γ)HXM vanadium complex in which the endogenous vanadium(IV) has been altered to vanadium(V). Finally we demonstrate that l-Glu(γ)HXM- andl-Glu(γ)HXM·vanadium-evoked lipogenesis is arrested by wortmannin and that activation of glucose uptake in rat adipocytes is because of enhanced translocation of GLUT4 from low density microsomes to the plasma membrane.

Intensive studies have been carried out during the last two decades on the insulinomimetic effects of vanadium (1)(2)(3)(4). Va-nadium salts mimic most of the effects of insulin on the main target tissues of the hormone in vitro and also induce normoglycemia and improve glucose homeostasis in insulin-deficient (5-7) and insulin-resistant diabetic rodents in vivo (5)(6)(7)(8). On the basic research frontier, data continue to accumulate showing that vanadium salts manifest their insulin-like metabolic effects through alternative pathways not involving insulin receptor tyrosine kinase activation or phosphorylation of insulin receptor substrate 1 (9 -19). The key events of this backup system appear to involve inhibition of protein-phosphotyrosine phosphatases and activation of nonreceptor protein-tyrosine kinases (20 -23).
Vanadium salts are seriously considered as a possible treatment for diabetes, and several clinical studies have already been performed. In those studies, because of its toxicity, only low doses of vanadium (2 mg/kg/day) were used. Although ϳ20-fold lower than doses used in most animal studies, several beneficial effects were observed and documented (24 -26). Any manipulation to elevate the insulinomimetic efficacy of vanadium without increasing its toxicity is of major clinical interest for the future care of diabetes (reviewed in Ref. 27).
In the wake of these findings, we have continued our search for more effective vanadium binding agents. Of special interest to us were vanadium chelators that synergize with vanadium both in vivo (i.e. in streptozocin rats) and in vitro (i.e. in isolated rat adipocytes) and therefore enable us to gain insight into the basic mechanism(s) by which such compounds potentiate the insulinomimetic activity of vanadium. Specifically, we have studied hydroxamic acid derivatives. These compounds are involved in the microbial transport of iron and are therefore applied therapeutically in conditions of iron deficiency (34). They are also inhibitors of urease activity and have been used in the treatment of hepatic coma. Monoamino acid hydroxamates are simple, nontoxic derivatives of amino acids. D-Aspartic acid ␤-hydroxamate was shown to have antitumoral activity on murine leukemia L5178Y, both in vitro and in vivo, and is active against Friend leukemia cells in vitro as well (35). L-Glu(␥)HXM is cytotoxic against L1210 cells in culture and remarkably antitumoral against L1210 leukemia and B16 melanoma in vivo (35,36). Streptozocin-treated Rats-Diabetes was induced by a single intravenous injection of a freshly prepared solution of streptozocin (55 mg/kg body weight) in 0.1 M citrate buffer, pH 4.5 (9). The effect of the L-Glu(␥)HXM⅐vanadium complex on blood glucose level was determined 8 days after induction of diabetes by streptozocin.

Materials-D-[U-
Cell Preparation and Bioassays-Rat adipocytes were prepared from the fat pads of male Wistar rats (130 -150 g) by collagenase digestion according to the method of Rodbell (37). Cell preparations showed more than 95% viability by Trypan blue exclusion at least 3 h after digestion. All bioassays were performed as described in figure legends. Glucose transport was carried out using 2-deoxy-D-[G-3 H]glucose uptake (38), and lipogenesis (the incorporation of U-14 C-labeled glucose into lipids) was performed according to Moody et al. (39). Briefly, freshly prepared rat adipocytes were suspended in KRBH, 0.7% BSA buffer and divided into about 50 plastic vials. Each vial contained 0.5 ml of adipocyte suspension (about 1.5 ϫ 10 5 cells). These were incubated for 2 h at 37°C under an atmosphere of 95% O 2 , 5% CO 2 with 0.16 mM [U-14 C]glucose. Each assay contained vials with and without 17 nM insulin and the various test compounds. Lipogenesis was terminated by adding toluenebased scintillation fluid, and the extracted lipids were counted (39). Results are expressed as a percent of maximal insulin response. Only assays in which insulin activated lipogenesis 5-6-fold above basal (basal ϳ4000 cpm/1.5 ϫ 10 5 cells/2 h, V insulin ϭ 20,000 -24,000 cpm/1.5 ϫ 10 5 cells/2 h) were taken into consideration. Insulin activated lipogenesis in this assay at an ED 50 value of 33 Ϯ 3 pM. A concentration of 0.3 nM insulin and above already facilitated maximal (100%) response (i.e. Ref. 16). All assays were performed in duplicate or triplicate.
Western Immunoblot Analysis of GLUT4 in Subcellular Membranes Following Stimulation of Rat Adipocytes-Adipocytes prepared from 6-week-old rats were incubated with and without insulin and with L-Glu(␥)HXM alone and complexed with vanadate as specified in the figure. Cells were then homogenized and fractionated to low density microsomal membrane (LDM) and plasma membrane (PM) fractions by differential ultracentrifugation according to Ref. 40. Membrane proteins were then solubilized in sample buffer for 30 min at 25°C, resolved on 10% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose paper, and immunoblotted with anti-GLUT4 antisera (41). Visualization was performed by phosphoimaging. The relative intensity of bands corresponding to GLUT4 was quantitated using MacBas 1000.

L-Glutamic Acid(␥)Monohydroxamate Potentiates Vanadium-evoked Lipogenesis in Rat
Adipocytes-In this set of experiments, rat adipocytes were incubated for 10 -20 min with submaximal concentrations of vanadate (10 -30 M), L-Glu(␥)HXM (10 -30 M), or an equimolar combination of them. The capacity to activate lipogenesis relative to insulin was then determined. As shown in Fig. 1, the combination was highly synergistic. For example, at 10 M vanadate or L-Glu(␥)HXM, lipogenesis was 17 Ϯ 3 and 9 Ϯ 2%, respectively, whereas the combination produced a marked incredible 93 Ϯ 4% activation of maximal insulin response. At 20 M, the extent of lipogenesis was 37 Ϯ 3, 20 Ϯ 3, and 121 Ϯ 7%, and at 30 M, it was 42 Ϯ 4, 23 Ϯ 4, and 143 Ϯ 7% of maximal. Wortmannin (100 nM), an inhibitor of phosphatidylinositol 3-kinase, fully blocked the activating effects of vanadate, L-Glu(␥)HXM, and its combination with vanadate ( Fig. 1, right columns). Thus L-Glu(␥)HXM potentiated vanadate-evoked lipogenesis about 3.5-5-fold; the higher concentrations reached a level that is about 140% of that achieved by saturating concentrations of insulin or vanadate. A finding of significant interest to us was the ability of L-Glu(␥)HXM to partially activate lipogenesis even in the absence of exogenous vanadium (Fig. 1). This finding is examined in great detail in connection with Fig. 6.
In Fig. 2, lipogenesis in rat adipocytes was evaluated at a fixed, low concentration of vanadate (5 M) with increasing concentrations of L-Glu(␥)HXM. Lipogenesis was negligible at 5 M vanadate or L-Glu(␥)HXM alone (4 -6% of maximal insulin effect) but is augmented to 27.0 Ϯ 3% when they were given in combination (at a molar stoichiometry of 1:1). At 2:1 and 3:1 Glu(␥)HXM⅐vanadium molar stoichiometry, lipogenesis expanded to 43 and 57%, respectively, of maximal response. Thus a substantial synergistic effect is obtained at a 1:1 molar ratio and is increased further at a 2:1 molar stoichiometry and even higher, though much less pronounced, at a 3:1 molar ratio (Fig. 2).

L-Glu(␥)HXM Alone and L-Glu(␥)HXM⅐Vanadate Lead to Translocation of GLUT4 from LDM to PM Fractions in Rat
Adipocytes-Incubation of rat adipocytes with L-Glu(␥)HXM and L-Glu(␥)HXM⅐vanadate led to a decrease in the content of GLUT4 in the LDM fraction and an increase in the PM fraction (Fig. 4). The decrease in GLUT4 content in the low density lipoprotein fraction amounted to 32 Ϯ 3, 3 Ϯ 1, and 68 Ϯ 5% of maximal insulin response upon incubating the cells with L-Glu(␥)HXM (40 M), vanadate (20 M, not shown), and the combination, respectively (calculated from Fig. 4). Under similar experimental conditions, L-Glu(␥)HXM, vanadate, and the combination activated 2-deoxyglucose uptake to an extent of 31 Ϯ 4, 7 Ϯ 0.7, and 117 Ϯ 9% of maximal insulin response (Fig.  3), suggesting a contributing effect of the complex to glucose influx in addition to its effect in recruiting GLUT4 transporters from the low density lipoprotein to the PM fraction. 2 L-Glu(␥)HXM⅐Vanadate Normalizes Blood Glucose Levels in Streptozocin-treated Diabetic Rats-In the experiments summarized in Fig. 5, streptozocin-treated rats received intraperitoneally sodium metavanadate (0.05 mmol/kg body weight), L-Glu(␥)HXM (0.1 mmol/kg body weight), or a combination of the two compounds 8 days after the induction of diabetes. As shown in the figure, vanadate and L-Glu(␥)HXM, at these concentrations, had a rather minor effect in reducing the high circulating glucose levels characterizing these hyperglycemic rats. The combination, however, was highly efficient at normalizing blood glucose levels. Normoglycemia was evident 1 day after the first administration and remained so following two more administrations. The glucose levels then remained close to normal for the next 3 days (Fig. 5).
Activation of Lipogenesis in Rat Adipocytes by L-Glu(␥)HXM in the Absence of Exogeneous Vanadium-L-glutamic acid(␥)HXM also activated lipogenesis in the absence of added vanadium, and this effect was studied in detail (Fig. 6). The dose-response curve (Fig. 6A) indicates that activation is already evident at 5 M L-Glu(␥)HXM and that higher concentrations reach a level of 40 Ϯ 7% of maximal insulin response (median effective dose ϭ 35 Ϯ 4 M). Other amino acid hydroxamates such as L-Tyr(␣)HXM, Gly(␣)HXM, and L-Ile(␣)HXM also activated lipogenesis, but they were considerably less potent (ED 50 ϭ 250 Ϯ 30 M, 40 Ϯ 5% of maximal insulin effect). L-Aspartic acid ␤-monohydroxamate showed higher lipogenic activity compared with the ␣-amino acid hydroxamates and was slightly less potent than L-Glu(␥)HXM (ED 50 ϭ 45 Ϯ 7 M, Fig. 6B). N-acetyl-L-Glu(␥)HXM and L-Glu(␥)HXM-␣-methyl ester were virtually ineffective, indicating the need for a free ␣-amino and, to a somewhat lesser extent, a free FIG. 4. L-Glu(␥)HXM alone or complexed with vanadate induces translocation of GLUT4 from LDM to PM fraction in rat adipocytes. Rat adipocytes were incubated for 30 min at 37°C in the presence and the absence of insulin (17 nM) and the indicated concentrations of L-Glu(␥)HXM or L-Glu(␥)HXM⅐vanadate. Cells were then homogenized and fractionated to PM and LDM by differential ultracentrifugation, and GLUT4 protein was identified by Western immunoblot analysis ("Experimental Procedures"). Immunoreactive GLUT4 proteins were visualized by phosphoimaging (top panels) and were quantitated using MacBas 1000 software (histograms, bottom panels).
␣-carboxyl moiety for the activation of lipogenesis by L-Glu(␥)HXM in the rat adipose cell (Fig. 6C). Stereospecificity appears crucial as well, because the D-isomer of Glu(␥)HXM was ineffective. All these findings indicate that activation of lipogenesis by L-Glu(␥)HXM depends on a specific entry of this L-amino acid analog into the adipose cell. Further investigation has led us to suggest that L-Glu(␥)HXM enters the adipose cell primarily through the non-Na ϩ -dependent glutamine transport system. 2 Several organic chelators, which potentiate the insulinomimetic activity of vanadium either in vitro or in vivo, have been documented. These include acetylacetonate (29), maltol (30,31), picolinate (32,33), and RL-252 (28). In Fig. 6D, we have examined whether they are capable of activating lipogenesis in the absence of exogenous vanadium. Unlike L-Glu(␥)HXM, none of these agents were able to activate lipogenesis in the rat adipose cell at concentrations of 100 M (Fig. 6D) or lower (not shown). Figs. 1-4 have taught us that L-Glu(␥)HXM potentiates the insulinomimetic potency of vanadium and that activation of lipogenesis by L-Glu(␥)HXM alone never exceeds 40 Ϯ 7% of maximal insulin effect (Fig. 6). To examine whether L-Glu(␥)HXM-evoked lipogenesis can be affected by the level of intracellular vanadium, a group of male Wistar rats received daily subcutaneous administrations of vanadate (0.1 mmol/kg/day) over a period of 5 days to raise the level of endogenous vanadium. Rats were then sacrificed 7 h after the last administration. Adipocytes were prepared, and the effect of L-Glu(␥)HXM on lipogenesis was compared with that in nontreated freshly prepared adipocytes. As shown in Spectroscopic Studies-Previously we found in cell-free experiments that vanadium(IV), at neutral pH values, undergoes slow spontaneous oxidation to vanadium(V). This occurs similarly in the presence of 10 mM reduced glutathione, an ineffectual reductant of vanadium(V), at neutral pH values with a t1 ⁄2 value of 1 Ϯ 0.1 h at 25°C (29). The results summarized in Fig.  8 show the V 51 NMR spectra of vanadium dichloride(IV) at pH 7.0 prior to and after the addition of L-Glu(␥)HXM. Vanadium dichloride(IV) appeared as a single peak with a chemical shift of Ϫ490 ppm in its 51 V spectrum, indicating one main species present at Ͼ95% purity. Upon the addition of L-Glu(␥)HXM (1 equivalent), the chemical shift of vanadium(IV) at Ϫ490 ppm disappeared within minutes and the principal chemical shift characterizing vanadium(V) at Ϫ530 ppm appeared (Fig. 8).

Extensive Potentiation of L-Glu(␥)HXM-evoked Lipogenesis in Rat Adipocytes in Vitro Following Enrichment with Vanadium in Vivo-The findings presented in
Vanadium(IV) (i.e. vanadyl sulphate or VOCl 2 ) has a characteristic "blue" absorbance with ⑀ 765 nm ϭ 14 Ϯ 0.3, whereas vanadium(V) does not absorb at all at this wavelength (29). The addition of 2-3 equivalents of L-Glu(␥)HXM to VOCl 2 (IV) (50 mM at pH 7.5) led rapidly to a near total decrease in vanadium(IV) absorbance at 765 nm (Fig. 9). Fig. 8B depicts complex formation as a function of the pH in the range of pH 2-9. Decrease is minimal at pH 4.0, quite significant at pH 5.0, FIG. 7. Activation of lipogenesis by L-Glu(␥)HXM. Comparison between normal adipocytes and vanadium-enriched adipocytes. Male Wistar rats received daily subcutaneously injected NaVO 3 (0.1mmol/kg/day) for 5 days (called enriched vanadium rats). The rats were then sacrificed (7 h after the last administration). Lipogenesis was performed comparing the freshly prepared rat adipocytes (3 ϫ 10 5 cells/ml) from nonenriched vanadium rats with the enriched ones suspended in KRB buffer, pH 7.4, containing 0.7% BSA. The cells were preincubated for 10 min with the indicated concentrations of L-Glu(␥)HXM. The cells were then supplemented with [U-14 C]glucose, and lipogenesis was performed for 2 h at 37°C. Radioactivity incorporated into extracted lipids was then determined. Maximal response (100%) is that obtained in the presence of 17 nM insulin. half-maximal at pH 5.7, and reaches a stable plateau at pH range 7-9 (Fig. 9B). DISCUSSION It has been consistently observed that chelated vanadium compounds are more potent than the free metaloxide in facilitating the metabolic actions of insulin. This was demonstrated in vitro with systems like rat adipocytes, as well as in diabetic rodents such as streptozocin-treated hyperglycemic rats (28 -33, 44). Because of the variations in the experimental models used, the oxidation state of vanadium applied, and the different administration modes, the basis for the higher insulinomimetic potencies of complexed vanadium remained rather speculative. Because this topic has immediate therapeutic relevance, we looked for new vanadium chelators characterized by: (a) higher synergistic potencies than previously documented for vanadium chelators with respect to vanadium-evoked glucose uptake and glucose metabolism both in vitro and in diabetic rats in vivo, (b) low indices of toxicity, and (c) reasonable solubility in aqueous, neutral media after complexation with vanadium.
In this study, we have introduced the L-isomer of glutamic acid(␥)monohydroxamate as it satisfactorily fulfilled the above criteria. It potentiated vanadium-activated hexose uptake, glucose metabolism, and recruitment of GLUT4 transporters from LDM to PM fractions (Figs. 1-4). In vivo it potentiated the efficacy of vanadium to lower blood glucose levels in streptozocin rats (Fig. 5). This amino acid analog has negligible toxicity in mammals. 2 Both L-Glu(␥)HXM alone and its complexes with vanadium are fairly soluble in aqueous media at neutral pH values. An important finding was that L-Glu(␥)HXM alone, in the absence of exogenous vanadium, showed a reasonable amount of insulinomimetic activity in that it activated glucose uptake and glucose metabolism in the rat adipose cell (Figs.  1-3). Further investigation revealed that this activating effect is unique to the L-isomer of Glu(␥)HXM but is not facilitated by the D-isomer. Nonmodified ␣-amino and ␣-carboxyl moieties appear essential. This intrinsic activity is exclusive to L-Glu(␥)HXM not being shared by any of the other vanadium chelators that potentiate the actions of vanadium in vivo or in vitro (Fig. 6, A-D, and Refs. 28 -33). Our assumption that L-Glu(␥)HXM permeates into the cell interior and transforms the "dormant" intracellular vanadium pool into an insulinomimetic-activated species gains credence from the dramatic sen-sitization of vanadium-enriched adipocytes to L-Glu(␥)HXMevoked lipogenesis (Fig. 7).
It should be mentioned at this point that because of the extreme complexity of aqueous vanadium chemistry (reviewed in Refs. 46 -49), the intracellular milieu of the mammalian cell is still "a black box" with respect to the state and the form of entered vanadium. With the endogenously present vanadium pool, experiments have shown that it exists mostly as vanadium(IV), though some researchers may wonder even about this experimental finding because vanadium in its IV oxidation state is only stable at acidic pH values (pH Ͻ 3.0) and readily oxidizes to vanadium(V) at neutral pH even in the presence of high glutathione concentrations (28,46). The intracellular vanadium pool, however, can be preserved in its IV oxidation form at neutral pH values if it is chelated by ascorbic acid (not shown) or to endogenous proteins (50,51). At the low physiological level of intracellular vanadium, the cell should have the capacity to chelate all the endogenous vanadium.
Our experimental findings that L-Glu(␥)HXM alone enhances glucose uptake and glucose metabolism (Figs. 1 and 2) together with the apparent rapid conversion of vanadium(IV) to vanadium(V) upon complexation (Figs. 8 and 9) strongly support the contention that vanadium(V) rather than vanadium(IV), and in a chelated form, is the active insulinomimetic species that facilitates the activation of glucose uptake and its metabolism in rat adipocytes. Although most of our previous cell-free experiments support this conclusion, we were not fully convinced prior to the completion of this study. This is because protein phosphotyrosine phosphatases (with p-nitrophenylphosphate as a substrate) are inhibited by both vanadium(IV) and vanadium(V), free or chelated, at nearly the same concentrations (see Ref. 52). On the other hand, adipose nonreceptor protein-tyrosine kinases, whether cytosolic or membranal, are with one exception activated by vanadium(V) but not at all by vanadium(IV) (22,23). We have only observed vanadium(IV)-evoked activation of nonreceptor protein-tyrosine kinases when membranal protein phosphotyrosine phosphatases were extracted with Triton X-100 and added to the cytosolic protein-tyrosine kinase fraction (29). These experimental conditions, however, are not likely to occur in the intact cell system. For example, broken plasma membrane fragments (or deoxycholate-treated membranal fragments) did not sup- port activation of cytosolic protein-tyrosine kinases in the presence of vanadium(IV) (29).
In summary, L-Glu(␥)HXM appears superior to previously documented organic chelators of vanadium in potentiating its activation of glucose uptake and glucose metabolism in vitro and in vivo. Taken together with earlier studies, this may be attributed to one or more of the following: (a) increased efficiency of this specific combination to permeate into cells or tissues; (b) a favorable 5-coordinated, rather than octahedral topography of this complex in an aqueous, neutral environment (Ref. 50); 2 and/or (c) higher intracellular stability of the L-Glu(␥)HXM-vanadium complex. Finally, we have recently observed that vanadate does not inhibit alkaline phosphatase in the presence of L-Glu(␥)HXM. 2 This inhibitory effect of vanadate (53) is undesirable from our point of view as it may contribute to vanadium toxicity in mammals, but not to the efficacy of vanadium to manifest the metabolic actions of insulin (reviewed in Ref. 54). This and other basic and diabetological aspects raised here are being further investigated.