Vanadium Activates or Inhibits Receptor and Non-receptor Protein Tyrosine Kinases in Cell-free Experiments, Depending on Its Oxidation State POSSIBLE ROLE OF ENDOGENOUS VANADIUM IN CONTROLLING CELLULAR PROTEIN TYROSINE KINASE ACTIVITY”

We have shown that vanadium mimics several insulin effects in rat adipocytes, via a staurosporine sensitive cytosolic protein tyrosine kinase (CytPTK Shisheva, A, and Shechter, Y. (1993) J. Biol. Chem. 268,6463). Here we demonstrate that vanadium effects on protein tyrosine kinases are preserved after cell disintegration. Vanadium inhibits or activates protein tyrosine kinases depending on its oxidation state and the tyrosine kinase studied. Vanadyl (4+) but not vanadate (5+) inhibits receptor tyrosine kinases such as the insulin receptor (IC6,, value = 23 = 4 p ~ ) and the insulin-like growth factor-I receptor (IC5o = 19 2 3 PM). Inhibition is non-competitive with respect to ATP, Md+, or substrate concentrations.

(Received for publication, November 8, 1993, and in revised form, January 10,1994) Gerard Elberg$, Jinping Li, and Yoram Shechters We have shown that vanadium mimics several insulin effects in rat adipocytes, via a staurosporine sensitive cytosolic protein tyrosine kinase (CytPTK Shisheva, A, and Shechter, Y. (1993) J. Biol. Chem. 268,6463). Here we demonstrate that vanadium effects on protein tyrosine kinases are preserved after cell disintegration. Vanadium inhibits or activates protein tyrosine kinases depending on its oxidation state and the tyrosine kinase studied. Vanadyl (4+) but not vanadate (5+) inhibits receptor tyrosine kinases such as the insulin receptor (IC6,, value = 23 = 4 p~) and the insulin-like growth factor-I receptor (IC5o = 19 2 3 PM). Inhibition is non-competitive with respect to ATP, M d + , or substrate concentrations. Preincubation of adipocytes with vanadyl(0.4 mM), and staurosporine (which arrests the cytosolic enzyme) substantially inhibited insulin-stimulated lipogenesis. Vanadyl is readily oxidized to vanadate by hydrogen peroxide.
In contrast, CytPTKs were poorly inhibited by vanadyl, and vanadate stimulated several CytPTKs 2-6-fold. CytPTK derived from rat adipocytes, liver and brain were activated, and CytPTK from Nb, lymphoma cells was not affected. CytPTK extracted from insulin-responsive tissues are more sensitive to vanadate activation (ED,, = 3 2 0.7 PM), whereas the brain enzyme is less sensitive (ED,, = 27 -c 3 PM). Tungstate, molybdate, and phenylarsine oxide also stimulate CytPTK, suggesting that the vanadate effect is secondary to inhibiting protein phosphotyrosine phosphatases.
This study supports a working hypothesis implicating the intracellular vanadyl pool in modulating CytPTK activity. Any physiological conditions converting vanadyl to vanadate (i.e. H,O, production) will activate CytPTK and consequently CytPTK-dependent bioeffects.
There is a growing interest in vanadium in the field of insulin action and diabetes, since vanadium mimics most or all of the actions of insulin in various insulin-responsive cells and tissues * This study was supported in part by grants from the Minerva Foundation (Germany), the Roland Shafer Contribution to Diabetes Research, the Israel Ministry of Health, and the Israel Academy of Sciences and Humanities. The costs of publication of this article were therefore be hereby marked "aduertisement" in accordance with 18 defrayed in part by the payment of page charges. This article must U.S.C. Section 1734 solely to indicate this fact.
istry of Science and Technology in Israel. (reviewed in Ref. 2). Vanadium therapy normalizes blood glucose levels in streptozotocin-treated diabetic rats and alleviates many of the aberrations associated with hyperglycemia (3)(4)(5). More importantly, oral administration of vanadium lowers blood glucose levels in the experimental animals representing Type I1 diabetes and severe insulin resistance (6-9). In several instances, this treatment dramatically increases tissue responsiveness to insulin in insulin-resistant rodents ( i e . Ref. 9). We have recently found that, in addition to the insulin receptor tyrosine kinase, rat adipocytes contain another protein tyrosine kinase (10) which is a water-soluble cytosolic protein (CytPTK)' of an estimated mass of 53 kDa. The activity of CytPTK is supported by Co2+ rather than by Mn". Further studies revealed a 3-5-fold increase in the specific activity of CytPTK in intact adipocytes preincubated with vanadate.
Of the several tyrosine kinase blockers tested, staurosporine was the most potent inhibitor of the soluble kinase (IC6,, = 3 2 0.2 nM), but a poor inhibitor of the InsRTK (IC5o = 8 p~) .
In intact cells, staurosporine inhibited vanadate-induced stimulation of lipogenesis and glucose oxidation, but only had a marginal effect on the insulin-stimulated bioeffects. Overall, it was concluded that vanadate mediates its effects on glucose metabolism via this cytosolic tyrosine kinase by an alternative noninsulin-dependent pathway (1,10). This conclusion was in agreement with several studies indicating that vanadate per se (as opposed to pervanadate) does not activate the insulin receptor in intact cellular systems (9,(11)(12)(13).
There are many unanswered questions with regard to how vanadate mimics the actions of insulin. It is not clear whether vanadate (5+ oxidation state) or vanadyl(4+ oxidation state), is the active species that mimics the hormone. Another issue is whether or not the insulin-like effects of vanadium are secondary to the effects of vanadate in inhibiting cellular protein phosphotyrosine phosphatases (PTPases). At least vanadate (5+ oxidation state) has been shown in several studies, to exert inhibitory actions on some members of this family of enzymes (14)(15)(16). However, it is now clear that vanadate will only slightly inhibit or have no effect on many of the mammalian PTPases tested (16,17). Other vanadate-sensitive phosphohydrolyzes are alkaline phosphatase, Na+,K ATPase, Ca2+/Mg2+ ATPase, and glyceraldehyde-3-phosphate dehydrogenase (reviewed in Refs. 18 and 19). These effects, however, may not occur at all in intact cellular systems, since exogenously added vanadate, after permeating into the cell interior, is efficiently nase; InsRTK, insulin receptor tyrosine kinase; poly(Glu,Tyr), random ' The abbreviations used are: CytPTK, cytosolic protein tyrosine kicopolymer containing L-glutamic acid and L-tyrosine at 4:l molar ratio; IGF-I, insulin-like growth factor I; WGA, wheat germ agglutinin; PTPase, protein phosphotyrosine phosphatase.

Procedures
Rat adipocytes were prepared from fat pads of male Wistar rats (100-200 g) by collagenase digestion, according to the method of Rodbell (23).

Assay of Lipogenesis-Assay of lipogenesis was measured by the incorporation of ~-[U-'~C]glucose into lipids as described in detail in
Ref. 1.
Partially Purified Insulin and IGF-I Receptors-Partially purified insulin and IGF-I receptors were prepared from rat liver or rat ovary membranes following homogenization, solubilization with Triton X-100, and affinity purification on a WGA-agarose column (24). The ovaries for the IGF-I receptor preparations were excised from 25-day-old female rats, 48 h after receiving 10 units of pregnant mare's serum gonadotropin.
Preparation of High Speed Supernatant Fractions from Various Tissues-Rat adipocytes, rat liver, rat brain or cultured Nb, rat lymphoma cells were homogenized (in the absence of detergents) with a Polytron homogenizer in 25 mM Hepes, pH 7.4, containing 0.25 M sucrose. The cell homogenates were then centrifuged a t 12,000 x g for 20 min, and the supernatant was further centrifuged at 40,000 x g for 60 min. The supernatant fractions were stored a t -137 "C until used (1).
Qrosine Kinase Activity Measurements-Tyrosine kinase activity measurements for the insulin and IGF-I receptors and for the cytosolic tyrosine kinases, were performed essentially as described in detail in Ref. 1. The final concentrations of Co", Mn2+, Mg"', ATP, and poly(Glu,Tyr) were specified in the text for each protein tyrosine kinase examined. Phosphotyrosine content in poly(Glu,Tyr) was quantitated by radioimmunoassay procedure according to Ref. 25. Autophosphorylation of InsRTK or IGF-I Receptor Qrosine Kinase in Vitro-Wheat germ agglutinin purified InsRTK or IGF-I receptor tyrosine kinase ( -5 pg of protein) were incubated for 20 min a t 22 "C with the indicated concentrations of vanadate or vanadyl in a final volume of 60 pl containing 25 mM HEPES, pH 7.4, 5 mM MnCI,, 5 mM MgCI,, and 100 p~ ATP with or without 0.2 p~ insulin or IGF-I. Samples were then subjected to 7.5% SDS-polyacrylamide gel electrophoresis (25).
Western Blot Analyses-Western blot analyses were carried out by transferring proteins to nitrocellulose membranes (28). These were sequentially treated with polyclonal antibodies to phosphotyrosine and with protein A-horseradish peroxidase conjugate, and finally subjected to enhanced chemiluminescence Western blot detection system a s described in detail in the manufacturers' instructions.
Protein Concentration-Protein concentration was determined by the method of Bradford (29). All the assays were performed either in duplicate or triplicate. Each figure or table is the result of a representative experiment performed three to five times.

RESULTS
Inhibition of InsRTK-dependent Poly(Glu,Tyr) Phosphorylation by Vanadyl-Addition of increasing concentrations of va- nadyl to WGA-purified insulin receptor inhibited InsRTK-catalyzed phosphorylation of poly(Glu,Tyr) in a concentrationdependent manner (Fig. 1). Half-maximal inhibition was obtained a t 23 f 4 VM, and full inhibition was evident at 90 2 7 J~M . Vanadate (5+ oxidation state) did not inhibit InsRTK-catalyzed poly(Glu,Tyr) phosphorylation at any concentration examined (Fig. 1). Thus, this inhibitory effect seems to be specific to vanadyl (4+ oxidation state) and does not result from vanadyl oxidation to vanadate that may occur, to a small extent, at neutral pH values (19). Similarly, pervanadate (peroxides of vanadate), an agent demonstrated to inhibit PTPases, therefore activating protein tyrosine kinases in intact cells (ll), had no effect on the activity of InsRTK in cell-free experiments (Fig.  1). Fig. 2 summarizes the experiments performed in order to " Vanadyl was added to the assay at time zero (prior to receptor autophosphorylation).
" Vanadyl was added to the assay mixture a t time = 20 min (subsequent to receptor autophosphorylation, and prior to the addition of polyiGlu.,Tyr)). gain more insight into the type of inhibition that vanadyl exerts on InsRTK-catalyzed poly(Glu,Tyr) phosphorylation. The potency of vanadyl to inhibit the enzyme was altered neither at 7-fold higher concentration ofATP nor at increasing concentrations of Mn", indicating no competitive inhibition with respect to both ATP and the metal ion. Similarly, increasing the substrate concentration 10-fold had an insignificant effect on the inhibitory potency of InsRTK by vanadyl (Fig. 2). Vanadyl inhibition is thus not competitive with the substrate, the metal ion, or ATP.
Inhibition of ZnsRTKAutophosphorylation by Vanadyl- Fig.  3 demonstrates that vanadyl inhibits InsRTK autophosphorylation in a concentration-dependent manner. Fifty percent inhibition was obtained at 60-80 PM and maximal inhibition at 100 p , namely at somewhat higher concentrations in which vanadyl inhibited receptor-catalyzed poly(Glu,,Tyr) phosphorylation (Fig. 1). Inhibition of autophosphorylation per se is sufficient to explain the inhibitory action of V02+ on r.  poly(Glu,Tyr) phosphorylation a s autophosphorylation precedes and activates the kinase (30,31). However, we have further observed that vanadyl is almost equipotent in inhibiting poly(Glu,Tyr) phosphorylation, when added subsequent to receptor autophosphorylation and its activation (Table I). This fact again emphasizes the uniqueness of the vanadyl-dependent inhibition. It is also noteworthy that V02+ inhibits autophosphorylation nearly as effectively as it inhibits substrate phosphorylation. This is not expected kinetically if the inhibitor is competitive with the substrate, again stressing the noncompetitive nature of vanadyl in inhibiting InsRTK (see above).
Inhibition of ZGF-Z Receptor Tyrosine Kinase by Vanadyl- Vanadyl blocked IGF-I receptor catalyzed poly(Glu,Tyr) phosphorylation (Fig. 41, as well as the ligand-dependent receptor autophosphorylation (Fig. 5). The concentration dependent inhibitions in both cases were strikingly similar t o that observed with the InsRTK. IC,, values amounted to 19 2 3 p~ for the substrate-catalyzed and 60-80 J~M for the autophosphorylating event. Thus, the sites of vanadyl-binding to both enzymes seem to have a considerable degree of homology within the threedimensional structure of those related receptors (32).
Inhibition of Insulin-stimulated Lipogenesis in Rat Adipocytes by Vanadyl-We wished to determine whether the inhibitory effect of vanadyl upon InsRTK activity can also be produced in the intact cell system. Vanadyl, however, stimulates lipogenesis in an insulin-independent manner, via the staurosporine-sensitive CytPTK (1). To avoid this, the effect of vanadyl was carried out in the presence of staurosporine. . For liver, the mixture contained 40,000 x g supernatant from rat liver (9 pg of protein), 1 mM Co", 100 p~ ATP, and 7 pg of poly(Glu,Tyr). For brain, the mixture contained 40,000 x g supernatant from rat brain (2 pg of protein), 20 mM MgSO,, 1 m~ Mn2+, 1 m~ Co", 100 p~ ATP, and 10 pg of poly(Glu,Tyr). For lymphoma, the mixture contained 40,000 x g supernatant from Nb, cells ( alone had no effect on InsRTK activity (Fig. 7). This indicates that vanadyl is readily oxidized by H,O, to vanadate at neutral pH values, in a stoichiometric fashion and at low concentrations of the reactants. In experiments not shown here, vanadyl oxidation to vanadate were also confirmed by H,O, consumption following its mixing with vanadyl and by the disappearance of the typical vanadyl's electron spin resonance (33) spectrum upon adding H,O,. Competition experiments (not shown here) seem to indicate that vanadyl is several times more sensitive to H,O, oxidation, as compared to several endogenous reducing agents found in mammalian cytosols, including reduced glutathione.' Effects of Vanadyl a n d Vanadate on Non-receptor Cytosolic Tyrosine Kinases-In contrast to the receptor protein tyrosine kinases, vanadyl inhibited CytPTK activity at high concentrations only (50-fold higher concentration, Fig. 8). Moreover, at least one, namely the brain cytosolic tyrosine kinase, was significantly stimulated by vanadyl (Fig. 8 ) . Vanadate (which had no effect at all on the receptor protein tyrosine kinases; Figs. 1 and 4) stimulated three of the four CytPTKs studied, namely liver, fat, and brain CytPTKs. Liver and fat CytPTK were stimulated about 2-4-fold, whereas brain CytPTK was stimulated 4-7-fold in the various experiments.
ED,, values amounted to 3 t 0.7 for the adipocytic and liver CytPTK and 27 2 3 PM for the brain CytPTK, respectively. CytPTK derived from Nb,-lymphoma cells was not activated by vanadate (Fig. 8 ) . It is of interest to note that CytPTKs derived from fat and liver, classical insulin-responsive tissues, were activated by lower concentrations of vanadate (i.e. ED,, values of 3 f 0.4 PM). Brain CytPTK was activated by significantly higher concentrations of vanadate (ED,, = 27 2 3 PM), whereas lymphoma Cyt-PTK was not activated by vanadate. The possibility that fat, liver, and muscle typically possess a vanadate-sensitive system is currently being studied.
It was our further intention to determine whether vanadate activated the CytPTK due to direct VO, binding to the enzymes. The alternative was tyrosyl phosphorylation and activation of the kinases resulting from inhibition of PTPases. To approach this issue, the activating effects of fluoride, tungstate molybdate, and phenylarsine oxide were analyzed. Several types of PTPases were reported to be inhibited by those agents G. Elberg, J. Li, and Y. Shechter, manuscript in preparation. (17,34,35). The results have been summarized in Fig. 9. Fluoride did not activate brain-and adipocyte-derived CytPTK, however, molybdate, tungstate, and phenylarsine oxide stimulated the enzyme to the same extent as did vanadate, at concentrations of 10-100 PM (Fig. 9). Thus, vanadate activation of CytPTKs seems to be secondary, resulting from inhibition of PTPases, whereas vanadyl inhibition of receptor tyrosine kinases is likely to occur by direct vanadyl effect on the enzymes. In control experiments, vanadate, up to a concentration of 10 mM did not inhibit at all dephosphorylation of phosphorylated poly(Glu,Tyr) induced by the 40,000 x g supernatants of the various tissues (not shown). This seems to indicate that the activations observed here are not related to vanadate's effects in blocking poly(Glu,Tyr) dephosphorylation. It also seems to indicate that vanadate's inhibition of PTPases is more effective and specific when the substrate for the PTPase is an autophosphorylated-protein tyrosine kinase.
Since tungstate and molybdate resembled vanadate in stimulating rat adipocytic CytPTK, in the cell-free experiment they were expected (if permeable to cell interiors) to stimulate glucose metabolism as well. Indeed, molybdate and tungstate (but not fluoride) significantly activated lipogenesis in intact rat adipocytes. Their stimulating effects could be blocked by staurosporine.2 DISCUSSION Vanadium mimics most of the biological actions of insulin in various target tissues, and endogenous cellular tyrosine kinase activity is believed to be an essential early component in activating insulin-like bioeffects such as glucose and fat metabolism (36). With our recent discovery of the vanadium-dependent (insulin-independent) CytPTK, the steps linking vanadium pretreatment to activated CytPTK and enhanced rates of glucose metabolism remain to be determined. Due to the overall complexity of the intact mammalian cell, such a task is sometimes possible only if the effect is preserved following cell disintegration. This can be achieved if the stimulus and the intracellular target are proximal to each other.
Here, we have described direct attenuating effects of vanadium on protein tyrosine kinases in the cell-free experiments. Moreover. discriminating effects were seen of either vanadate or vanadyl toward receptor or non-receptor protein tyrosine kinases. Vanadyl (but not vanadate) inhibited receptor tyrosine kinases non-competitively, whereas the non-receptor protein tyrosine kinases were not inhibited by vanadyl and stimulated by vanadate. These results together with those of the several previous studies, support the contention that the activating effects of added vanadium on glucose and fat metabolism are indeed mediated via protein tyrosine kinases. It should be noted, at this point, that the coincidence of finding a substance (such as vanadyl), that is present endogenously in cells and inhibits InsRTK at low concentrations in an ATP and substrate independent manner is unique and may have a physiological significance. This is particularly validated by the fact that vanadyl could also block insulin-stimulating effects in cells, pretreated with staurosporine (Fig. 6). We are especially intrigued by the effectiveness of vanadyl in inhibiting autophosphorylation. This observation is an exception among many InsRTK blockers studies, including those competing with ATP (37) or substrate binding (38,39). Both types are more effective (about 10-fold or more) in blocking phosphorylation of exogenous substrate, rather than inhibiting autophosphorylation (37)(38)(39). The latter is an intramolecular transphosphorylation within the insulin receptor heterotetramer (40). Therefore, inhibitors must compete against high local substrate or ATP concentrations.
Perhaps of more significant interest are our cell-free observations demonstrating that ( a ) vanadate (5+) is the species that activates CytPTK, ( b ) it does so at relatively low concentrations in insulin-responsive tissues, and (c) vanadyl is readily converted to vanadate by hydrogen peroxide. A feasible working hypothesis, currently being studied in our laboratory, is illustrated schematically in Fig. 10. According to this hypothesis, the intracellular vanadium pool may have an essential role in modulating those CytPTKs not controlled by external stimuli. Under resting conditions, the bulk of the intracellular vanadium is in the form of vanadyl (4+) (Refs. 18 and 19), which, as was shown here, exerts no effect on CytPTK activity, keeping the latter in basal resting state. Any physiological conditions, however, that activate NADPH oxidase and lead to the formation of H,O, are expected to oxidize a fraction of the endogenous vanadyl pool to vanadate. This, in turn, will inhibit those vanadate-sensitive PTPases and correspondingly increase the steady states of phosphorylation and activation of CytPTK (Fig. 10). Most protein tyrosine kinases studied, including the rat adipocytic CytPTK (1, lo), seem to be activated as a result of autophosphorylation on tyrosine moieties (41, 42). Although it should be emphasized that a gap exists between the vanadate concentrations activating adipocytic CytPTK in cellfree experiments (i.e. ED,, = 3.0 VM) and the documented intracellular vanadium levels (i.e. 0.1-1.0 p~; Refs. 18 and 19). We assume, however, that activation of a minute fraction of the total CytPTK activity in the cell may suffice to give the desired level of phosphorylation and amplification required for further stimulating glucose and fat metabolism. This may especially be valid, as rat adipocytic cytosol contains an extremely high poly(Glu,Tyr) phosphorylating capacity, which exceeds that of the plasma-membrane fraction by more than 1 order of magnitude (1, 10).
This working hypothesis explains the activating effects of added vanadate on CytPTK activity in intact cells, as well as in cell-free experiments. That vanadyl can be converted to vanadate via an NADPH-oxidative pathway has been previously demonstrated (43), and a link between vanadate, NADPH, and activation of tyrosine phosphorylation in cells was frequently observed (44)(45)(46)(47).