Issra RashedMakinen Laboratory

Protein tyrosine phosphatase 1B (PTP-1B) helps regulate several cellular processes, including the deactivation of both the insulin and epidermal growth factor receptors. PTP-1B is therefore the object of many inhibitor design studies; however, attempts to find or design potentand specific PTP-1B inhibitors have been quite difficult because the active sites and structural motifs of the known PTPs are heavily conserved. Recently, bis(acetylacetonato)oxovanadium(IV) [VO(acac)2] has been shown to uncompetitively inhibit PTP-1B in the presence of a physiologically relevant, phosphotyrosine (pTyr) containing 12 amino acid substrate, DADEpYLIPQQG. An uncompetitive inhibitor binds to only the enzyme-

substrate (ES) complex, as opposed to a competitive or non-competitive inhibitor that binds to the enzyme alone. Despite the sequential and structural homology of the 100 known PTPs, these phosphatases exhibit non-redundant roles such that each PTP has unique ES complexes.

The object of my studies is to characterize and explore the interaction of VO(acac)2 withthe PTP-1B- DADEpYLIPQQG enzyme-substrate complex, to evaluate the specificity ofVO(acac)2 to PTP-1B in the presence of other substrates, and also to examine the effect of VO(acac)2 on other PTPs.

572

With the introduction of new instruments and improved sensorchip chemistries, surface plasmon resonance (SPR) is findingnew applications for molecular interaction studies. Easy accessto high-quality kinetic and thermodynamic data formacromolecular binding events is providing insights into thefundamental mechanisms of molecular recognition. Progress isbeing made to allow larger-scale interaction studies. Inaddition, combining SPR with other analytical methods isenabling SPR-based analysis of interaction proteomics.

AddressesThe Laboratory of Molecular Biophysics, Department of Biochemistry,University of Oxford, South Parks Road, Oxford OX1 3QU, UK; e-mail: jim@biop.ox.ac.uk

Current Opinion in Chemical Biology 2001, 5:572–577

1367-5931/01/$ — see front matter© 2001 Elsevier Science Ltd. All rights reserved.

AbbreviationsCPWR couple plasmon-waveguide resonanceEGF epidermal growth factorIgE immunoglobulin EMHC major histocompatibility complexSPR surface plasmon resonance

IntroductionThe biological functions of most macromolecules dependon their ability to interact with other molecules. In thepost-genomic era, one of the greatest challenges facing thechemical biology community is a complete description ofthe interaction proteome. This information will provideinsights into the mechanisms of biological processes andprovide opportunities for controlling these processes byinterrupting key molecular interactions.

Surface plasmon resonance (SPR) is a method for charac-terizing macromolecular interactions. It is an opticaltechnique that uses the evanescent wave phenomenon tomeasure changes in refractive index very close to a sensorsurface. The binding between an analyte in solution andits ligand immobilized on the sensor surface results in achange in the refractive index. The interaction is moni-tored in real time and the amount of bound ligand andrates of association and dissociation can be measured withhigh precision. Although SPR is a relatively new biophysi-cal method, with the first commercial instrument beingintroduced in 1990, its growth has been rapid. In the pastseveral years, SPR has taken its place as a mature biophys-ical technique for the analysis of molecular recognitionevents. SPR biosensors have become standard instrumentsin biochemical and biophysical research centers. Therehave been yearly increases in the number of publicationsin which SPR data is reported in an ever-wider variety of

biological systems, and recently there has been an expan-sion of applications into previously rare areas.

In an age where fully sequenced genomes offer a wealth ofpotential opportunities for insights into the molecular basisof biological processes, our next great challenge is thedevelopment of a complete set of proteomic interactionmaps. SPR can contribute to these efforts with rapid andquantitative analysis of molecular interactions.

SPR instrumentation and biosensor chipsA number of commercial SPR biosensor instruments areavailable [1]. Although BIAcore systems have dominatedthe market since their introduction [2•], there are manymore competing SPR instruments now available, includingnew systems introduced recently by Texas Instrumentsand Aviv [3,4••]. A list of manufacturers and their webaddresses can be found in Table 1.

Since the introduction of their instruments, both BIAcoreand Affinity Sensors have marketed sensor chips with acarboxymethylated dextran matrix and a streptavidin-derivatized surface. Although thousands of publicationsdemonstrate the versatility and robustness of these sur-faces, some systems are simply not compatible with theseimmobilization strategies or sensor surface chemistries.Consequently, several new commercial sensor chips havebeen introduced in the past few years, permitting newSPR-based applications. The new surface chemistriesinclude a carboxymethylated surface with a reducedcharge on the dextran matrix, several dextran-free ‘flat’surfaces with different chemistries for immobilizing ligands, a chelated nickel surface for binding His-taggedligands, and several hydrophobic surfaces designed toallow assembly of lipid monolayers and analysis of membrane systems [5,6].

SPR for studying biomolecular interactions:insights into the mechanisms of molecularrecognitionReal strengths of the SPR biosensor technology are its versatility and ease of use. It permits the analysis of receptor–ligand interactions with a wide range of differentmolecular weights, affinities and binding rates, and is com-patible with a myriad of different chemical environments.

SPR is effective in studying interactions for a large range ofmolecular weights of analytes. Experiments with analytemasses ranging from hundreds of daltons [7] to whole-cellbinding [8] have been performed. Although the effectiveaffinity range of SPR has often been quoted to range fromnanomolar to micromolar, it is possible to extend this range

Surface plasmon resonance: towards an understanding of themechanisms of biological molecular recognitionJames M McDonnell

Surface plasmon resonance McDonnell 573

substantially, from sub-picomolar to greater than millimo-lar affinities [9]. The difficulties with measuring very highaffinity interactions are generally related to very slow dissociation rates; for example, when koff = 10–6 s–1 it takesapproximately 104 seconds for 1% of bound material to dissociate. Measuring these small signal changes is experimentally challenging. High-affinity interactions caninstead be characterized using long sample injection timesand measuring equilibrium binding conditions over a rangeof concentrations [9]. Care must be taken when measuringbinding interactions at very low affinities [10], when spe-cific binding may be small compared with non-specificbinding to the sensor matrix or surface and instrumentnoise and drift. Practical advice on improving data-collec-tion methods, running proper controls and optimizing dataanalysis methods has been reviewed recently [11••]. The‘double referencing’ technique, a method for subtractingreference data to remove small systematic signal devia-tions, is particularly helpful in the analysis of weak bindinginteractions or low molecular weight analytes [11••].

SPR interaction analysis can be performed over a widerange of chemical and environmental conditions (tempera-ture, ionic strength, pH etc.). Analyzing binding kineticsand thermodynamics over a range of different conditionscan give unique insights into binding reactions mecha-nisms. Andersson et al. [12] have described a systematicapproach for exploring buffer space, both as a means forregenerating surface sensors and as a means for derivinginformation about the binding event. Several groups havemeasured binding affinities over a range of ionic strengths;these studies describe the role of electrostatics for theinteraction [13,14]. The Debye–Huckel plot — a plot oflog(KA) versus log(ionic strength) — provides informationon the number of ions displaced during the binding event[15]. One of the discriminating characteristics betweenspecific and non-specific protein–DNA binding is the difference in their ionic strength dependence [16]. Thebinding affinities of non-specific protein–DNA interac-tions are highly dependent on ionic strength, because ofthe dominance of the electrostatic contributions mediatedby the negatively charged phosphates of the DNA back-bone. Binding rates and affinities measured at differenttemperatures also provide information on thermodynamicparameters of the interaction. In addition to the wellknown van’t Hoff plots (providing thermodynamic para-meters ∆H, the change in enthalpy, and ∆S, the change inentropy), the temperature dependence of the on- and off-rates can give a direct value for the activation energies(∆H‡,∆S‡,∆G‡) of these processes [17].

It has long been appreciated that protein folding eventscan be described by complex energy landscapes and manyexamples of intermediates along these folding pathwayshave been characterized [18]. In contrast, the energy land-scapes of molecular binding events are less wellcharacterized, although they are thought to be similar innature [19]. The ease of measuring the thermodynamic

parameters and the energy barriers for on- and off-rates bySPR promises new insights into these processes. Usingkinetic and thermodynamic information, collected over arange of temperatures and ionic strengths, Frisch et al.[20••] have tentatively assigned a transition state for theinitial interaction between barnase and barstar. Severalgroups have recently used SPR methods to characterizethe activation energies for both association and dissociationevents [21,22]. These kinds of studies will be critical in understanding the energy landscapes that control macromolecular interactions.

Analysis of kinetics and thermodynamics by SPR can beused to understand the complex mechanisms of molecularrecognition events. Myszka et al. [23•] combined SPR withisothermal titration calorimetry, analytical ultracentrifuge,and structural information from X-ray crystallography todescribe the interaction between CD4 and gp120. Their datasuggest extensive structural rearrangements upon ligandbinding, which may have implications for HIV immune eva-sion and viral entry mechanisms. De Crescenzo et al. [24]observed complex binding kinetics for the interactionbetween TGF-α and the epidermal growth factor (EGF)receptor. They considered several binding mechanisms andfound the biosensor data fit best to a conformational changemodel. According to this model, ligand binding induces aconformational change in the receptor resulting in receptordimerization. This mechanism is consistent with other bio-physical studies of EGF receptor–ligand interactions and isthought be important for EGF receptor-mediated cellularactivation. Gunnarsson et al. [25] used SPR to study confor-mation variants of human α2-macroglobulin, identifying a

Table 1

Manufacturers of SPR instruments.

SPR manufacturer SystemInternet address

BIAcore AB (Uppsala, Sweden) BIAcorehttp://www.biacore.comAffinity Sensors (Franklin, MA) IASys http://www.affinity-sensors.comNippon Laser Electronics SPR-670(Hokkaido, Japan)http://www.rikei.comArtificial Sensing Instruments OWLS (Zurich, Switzerland) http://www.microvacuum.com /products/biosensorIBIS Technologies BV IBIS(Enschede, The Netherlands)http://www.ibis-spr.nlTexas Instruments (Dallas, TX) TISPR http://www.ti.com/spreetaAviv (Lakewood, NJ) PWR-400 http://www.avivinst.comBioTul AG (Munich, Germany) Kinomics http://www.biotul.comQuantech Ltd (Eagan, MN) FasTraQ http://www.quantechltd.com

574 Analytical techniques

site involved in exposure of a ligand recognition site.Deviations from ideal 1:1 Langmuir models have also beenobserved in a large number of other systems and often thisinformation has been used to propose mechanisms for ligand-mediated activation events [26,27] (see Figure 1).Several examples have been identified in which interactionrates are more descriptive of a given biological process thanthe equilibrium binding affinities. For example, Leferinket al. [28] have described growth factor interactions withErbB-1 in which the dynamic rate constants correlate betterwith mitogenic activity than do the equilibrium constants.McDonnell et al. [29] identified a role for the Cε2 domainfrom immunoglobulin E (IgE) in allergic responses. Removalof this domain from IgE has a small effect on the overallaffinity of IgE for its receptor FcεRI but has a marked effect

on the off-rate. Cε2 thus contributes to the unusually slowoff-rate of IgE, which is an important factor in IgE-mediatedmast cell sensitization as part of the allergic response.

The analysis of complex kinetics has been aided by theintroduction of improved data-fitting software. In additionto manufacturer-supplied curve-fitting software, severalexcellent freeware programs are also available includingCLAMP (http://www.cores.utah.edu/interaction/clamp.htm)[30] and SPRevolution (http://www.bri.nrc.ca/csrg/equip.htm) [24]. Using sensible experimental design and datacollection methods [31], and applying the improved dataanalysis, including modelling mass transport effects [32],one can essentially eliminate experimental artefacts, whichwere common in early SPR studies [33,34], and allow

Figure 1

Association AssociationDissociationDissociation

100

80

60

40

20

0

0 200

0 200 300 400100 0 200 300 400100

400 600

100

80

60

40

20

0

100

90

80

60

70

50

40

100

90

80

60

70

50

40

0 200 400 600

(d) (e)

(b)(a) (c)

Current Opinion in Chemical Biology

Time (s) Time (s)

Time (s) Time (s)

RURU

RURU

Insights into molecular recognition processes using SPR. (a) In theinteraction between IgE and FcεRIα, Trp87 makes important energeticcontributions to binding. BIAcore SPR analysis of the Cε3-4 domainsfrom IgE binding to (b) FcεRIα wild type and (c) a Trp87Asp mutantdemonstrate the effect of this mutation on interaction kinetics. Wild type(red circles) and mutant (blue squares) dissociation curves are fit (blackline) to (d) a single-component exponential decay or (e) a two-component

system. The IgE–FcεRI interaction shows biphasic kinetics [26], but whilethe second component makes only a minor contribution to the wild-typeinteraction, this biphasicity is markedly exaggerated by the Trp87Aspmutation. The SPR data demonstrate the effect of this mutation on thefree energy (∆G) of binding, and also offer insights into the complexenergy landscape of this binding event, describing the pathways used inthe transition from free to bound states. RU, resonance units.

confidence in interpreting kinetic data into insights intobiological mechanisms.

SPR in membrane studiesBefore SPR gained popularity as a technique to analyzebiomolecular interactions, plasmon resonance spec-troscopy was used for many years by material scientists formeasurements of surface and optical properties of molecu-lar films and interfaces [35,36]. It seems only natural, then,that SPR could be a powerful approach for studying biological membrane events. The recent introduction of sensor surfaces specifically designed to allow study ofmembrane interactions has resulted in a large number ofnew studies of membrane-associated systems by SPR. Theplanar nature of the sensor surface and the ability to specif-ically orient immobilized ligands may make SPR a bettermembrane surrogate than traditional solution studies. Erbet al. [37] have characterized one of the new BIAcore sensor chips (L1) designed for liposome absorption. Usingatomic force microscopy and fluorescence microscopy, theyhave demonstrated that lipids form a homogeneous mono-layer on the surface of the chip, suggesting a successfulmembrane mimic. Liposome-covered sensor surfaces havebeen used in an assay for lipid absorption for a panel of 27drugs and showed a strong correlation with passive intesti-nal absorption [38]. Celia et al. [39•] established a lipidmonolayer using a chelated-nickel lipid and immobilizedand oriented a major histocompatibility complex (MHC)molecule through a His-tag tail, and then measured bind-ing events between the T cell receptor and the orientedMHC molecule. X-ray diffraction studies of the two-dimensional crystals of the monolayer-bound MHCmolecules established that the protein had maintained thedesired orientation. Several recent reports describe thecharacterization of membrane-integrated G-protein-cou-pled receptors [40–42]. Using a variation of SPR on ahome-built instrument (the predecessor of the commercialsystem from Aviv) Salamon et al. [4••] used coupled plasmon-waveguide resonance (CPWR) spectroscopy tostudy interactions and conformational change in therhodopsin–transducin system. The CPWR spectrum has ahigher information content than the traditional SPR measurement, which typically records only a change in theresonance angle. CPWR may have some important advantages in the study of anisotropic membrane systems.

SPR in proteomics and drug discoveryThe general versatility of SPR methods, the ease of automa-tion, and the lack of labeling requirements make it a promisingtool for large-scale screening for binding events, both for smallmolecules in drug discovery efforts or for macromolecules inlarge-scale ligand fishing experiments. The improved sensitivity of new biosensor instruments routinely permits thedetection of small molecular weight (<500 Da) analytes, evenfor low-affinity interactions (KD > 1 mM). Adamczyk et al. [43]used SPR as an immunoassay to monitor binding of thyroxinanalogs to an immobilized monoclonal antibody. The La JollaPharmaceutical Company has used SPR in the clinical

development of a new drug for the treatment of the autoim-mune lupus [44]. A review by Myszka and Rich [45] discussesrecent progress in SPR in drug discovery efforts. This reportalso describes a prototype microarray chip, a sensor surfacewith 64 individual immobilization sites in a single flow cell(the standard BIAcore chip has four independent flow cells).SPR-based arrays offer the opportunity to move towards larg-er-scale matrices of receptor–ligand interactions, the type ofanalysis that will be required to build complete proteome-interaction maps. A great number of different protein biochiptechnologies are being pursued [46•,47–49]. One promisingapproach is the combination of protein-chip-based technolo-gies with mass spectrometry (MS) (see for example [50]).Several groups have integrated SPR and MS for affinity-basedcapture and characterization of ligands [51,52]. Nelson et al.[53] discuss performing matrix-assisted laser desorption/ionisa-tion (MALDI)-MS directly on ligand-bound biosensorsurfaces. A fully in-line system combining an SPR biosensorand electrospray tandem MS has been described [54•]. TheSPR platform allows the detection, capture and subsequentdigestion and delivery of nanomolar to femtomolar levels ofligand for MS analysis. SPR/MS is a rapid and powerfulapproach for identifying binding partners from complex mixtures of components. Combining this technology withlarger microarrays makes it a feasible approach for large-scaleligand-fishing experiments or interaction proteomics analyses.

ConclusionsWith the introduction of a number of new SPR instrumentsand a series of novel sensor surfaces and chemistries, theimpact of SPR biosensors on molecular interaction studieswill continue to grow. The ability to form stable membranesurfaces on biosensor chips will greatly simplify bindinganalyses in membrane systems, making this important class ofbiological systems far more accessible to quantitative analysis.With improved experimental design and data analysis methods, it is now easier to obtain high-quality data for thekinetic and thermodynamic parameters of intermolecularinteractions. These data promise additional insights into themechanisms of molecular binding events, which will beimportant in rational drug design of inhibitors of macromole-cular interactions. SPR and other protein-chip-basedtechnologies are beginning to show promise in larger-scaleinteraction studies, both for small-molecule analysis andmacromolecular ligand-fishing experiments. Great potentialexists for interfacing SPR and MS in a broader microarrayapproach for characterizing proteome-wide interaction maps.

AcknowledgementsI acknowledge support from the EP Abraham Fund. I am grateful to E Garman, L Hewitt, R Lewis and M Noble for critical review of themanuscript and H Gould and B Sutton for helpful discussions.

References and recommended readingPapers of particular interest, published within the annual period of review,have been highlighted as:

• of special interest••of outstanding interest

1. Mullett WM, Lai EPC, Yeung JM: Surface plasmon resonance-based immunoasays. Methods 2000, 22:77-91.

Surface plasmon resonance McDonnell 575

2. Rich RL, Myszka DG: Survey of the 1999 surface plasmon• resonance biosensor literature. J Mol Recog 2000, 13:388-407.A very thorough review of SPR publications from 1999, with over 500papers classified by subject area.

3. Kukanskis K, Elkind J, Melendez J, Murphy T, Garner H: Detection ofDNA hybridization using the Texas Instruments, Inc. TISPR-1surface plasmon resonance biosensor. Anal Biochem 1999, 274:7-17.

4. Salamon Z, Brown MF, Tollin G: Plasmon resonance spectroscopy:•• probing molecular interactions within membranes. Trends

Biochem Sci 1999, 24:213-219.Describes CPWR, a variation of SPR. This technique is particularly well suited for analysis of membrane systems and may have a number of advantages over traditional SPR methods.

5. Malmqvist M: BIAcore: an affinity biosensor system for characterizationof biomolecular interactions. Biochem Soc Trans 1999, 27:335-340.

6. Nieba L, Nieba-Axmann SE, Persson A, Hamalainen M, Edebratt F,Hansson A, Lidholm J, Magnusson K, Karlsson AF, Pluckthun A:BIAcore analysis of histidine-tagged proteins using a chelatingNTA sensor chip. Anal Biochem 1997, 252:217-228.

7. Davis TM, Wilson WD: Determination of the refractive indexincrements of small molecules for correction of surface plasmonresonance data. Anal Biochem 2000, 284:348-353.

8. Quinn JG, O’Neill S, Doyle A, McAtamney C, Diamond D,MacCraith BD, O’Kennedy R: Development and application ofsurface plasmon resonance-based biosensors for the detectionof cell-ligand interactions. Anal Biochem 2000, 281:135-143.

9. Myszka DG, Jonsen MD, Graves BJ: Equilibrium analysis of highaffinity interactions using BIAcore. Anal Biochem 1998, 265:326-330.

10. Mavaddatt N, Mason DW, Atkinson PD, Evans EJ, Gilbert RJC, Stuart DI,Fennelly JA, Barclay NA, Davis SJ, Brown MH: Signalling lymphocyticactivation molecule (CDw150) is homophilic but self-associateswith very low affinity. J Biol Chem 2000, 275:28100-28109.

11. Myszka DG: Improving biosensor analysis. J Mol Recog 1999,•• 12:279-284.A very useful practical guide to improving SPR experimental design, datacollection and data analysis.

12. Andersson K, Gulich S, Hamalainen M, Nygren PA, Hober S,Malmquist M: Kinetic characterization of the interaction of theZ-fragment of protein A with mouse-IgG3 in a volume in chemicalspace. Proteins 1999, 37:494-498.

13. Baerga-Ortiz A, Rezaie AR, Komives EA: Electrostatic dependenceof the thrombin-thrombomodulin interaction. J Mol Biol 2000,296:651-658.

14. Laich A, Sim RB: Complement C4bC2 complex formation: aninvestigation by surface plasmon resonance. Biochem BiophysActa 2001, 1544:96-112.

15. Ha JH, Capp MW, Hohenwalter MD, Baskerville M, Record MT:Thermodynamic stoichiometries of participation of water, cationsand anions in specific and non-specific binding of lac repressor toDNA. J Mol Biol 1992, 228:252-264.

16. Oda M, Nakamura H: Thermodynamic and kinetic analyses forunderstanding sequence-specific DNA recognition. Genes Cells2000, 5:319-326.

17. Atkins P: Chapter 10. In The Elements of Physical Chemistry, edn 3.Oxford: Oxford University Press; 2001.

18. Onuchic JN, Nymeyer H, Garcia AE, Chahine J, Socci ND: Theenergy landscape theory of protein folding: insights into foldingmechanisms and scenarios. Adv Protein Chem 2001, 53:88-153.

19. Tsai CJ, Kumar S, Ma B, Nussinov R: Folding funnels, bindingfunnels and protein function. Protein Sci 1999, 8:1181-1190.

20. Frisch C, Fersht AR, Schreiber G: Experimental assignment of the•• structure of the transition state for the association of barnase and

barstar. J Mol Biol 2001, 308:69-77.A detailed kinetic and thermodynamic characterization of the initial interac-tion between barnase and barstar using stopped-flow fluorescence. Animportant first step in attempts to understand transition states in molecularbinding events.

21. Myszka DG: Kinetic, equilibrium and thermodynamic analysis ofmacromolecular interactions with BIAcore. Methods Enzymol2000, 323:325-340.

22. Roos H, Karlsson R, Nilshans H, Persson A: Thermodynamicanalysis of protein interactions with biosensor technology. J MolRecog 1998, 11:204-210.

23. Myszka DG, Sweet RW, Hensley P, Brigham-Burke M, Kwong PD,• Hendrickson WA, Wyatt R, Sodroski J, Doyle ML: Energetics of the

HIV gp120-CD4 binding reaction. Proc Natl Acad Sci USA 2000,97:9026-9031.

A thorough and detailed study of the CD4–gp120 binding event, SPR incombination with other structural and biophysical methods provide insightsinto the mechanism of this interaction.

24. De Crescenzo G, Grothe S, Lortie R, Debanne MT, O’Connor-McCourt M: Real-time studies on the interaction of transforminggrowth factor αα with the epidermal growth factor receptorextracellular domain reveal a conformational change model.Biochem 2000, 39:9466-9476.

25. Gunnarsson M, Stigbrand T, Jensen PEH: Conformational variantsof human αα2-macroglobulin are reflected in a C-terminal ‘switchregion’. Eur J Biochem 2000, 267:4081-4087.

26. Cook JPD, Henry AJ, McDonnell JM, Owens RJ, Sutton BJ, Gould HJ:Identification of contact residues in the IgE binding site of humanFcεεRI-αα. Biochemistry 1997, 36:15579-15588.

27. Lipshultz CA, Li Y, Smith-Gill S: Experimental design for analysis ofcomplex kinetics using surface plasmon resonance. Methods2000, 20:310-318.

28. Leferink AEG, van Zoelen EJJ, van Vugt MJH, Grothe S, vanRotterdam W, van de Poll MLM, O’Connor-McCourt MD:Superantagonistic activation of ErbB-1 by EGF-related growthfactors with enhanced association and dissociation rateconstants. J Biol Chem 2000, 275:26748-26753.

29. McDonnell JM, Calvert R, Beavil RL, Beavil AJ, Sutton BJ, Gould HJ,Cowburn D: The structure of the IgE Cεε2 domain and its role instabilizing the complex with its high-affinity receptor FcεεRI. NatStruct Biol 2001, 8:437-441.

30. Myszka DG, Morton TA: CLAMP: a biosensor kinetic data analysisprogram. Trends Biochem Sci 1998, 23:149-150.

31. Rich RL, Myszka DG: Advances in surface plasmon resonancebiosensor analysis. Curr Opin Biotechnol 2000, 11:54-61.

32. Goldstein B, Coombs D, He X, Pineda AR, Wofsy C: The influenceof transport on the kinetics of binding to surface receptors:application to cells and BIAcore. J Mol Recog 1999, 12:293-299.

33. Nieba L, Krebber A, Plückthun A: Competition BIAcore for measuringtrue affinities: large differences from values determined frombinding kinetics. Anal Biochem 1996, 234:155-165.

34. Ladbury JE, Lemmon MA, Zhou M, Green J, Botfield MC, Schlessinger J:Measurement of the binding of tyrosyl phosphopeptides to SH2domains: a reappraisal. Proc Natl Acad Sci USA 1995, 92:3199-203.

35. Brockman JM, Nelson BP, Corn RM: Surface plasmon resonanceimaging measurements of ultrathin organic films. Annu Rev PhysChem 2000, 51:41-63.

36. Hickel W, Kamp D, Knoll W: Surface plasmon microscopy. Nature1989, 339:186.

37. Erb EM, Chen X, Allen S, Roberts CJ, Tendler SJB, Davies MC,Forsen S: Characterization of the surfaces generated by liposomebinding to the modified matrix of a surface plasmon resonancesensor chip. Anal Biochem 2000, 280:29-35.

38. Danelian E, Karlen A, Karlsson R, Winiwarter S, Hansson A, Lofas S,Lennernas H, Hamalainen MD: SPR biosensor studies of the directinteraction between 27 drugs and a liposome surface: correlationwith fraction absorbed in humans. J Med Chem 2000, 43:2083-2086.

39. Celia H, Wilson-Kubalek E, Milligan RA, Teyton L: Structure and• function of a membrane-bound murine MHC class I molecule.

Proc Natl Acad Sci USA 1999, 96:5634-5639.A lipid monolayer is established and binding studies performed between theT cell receptor and a membrane-oriented MHC molecule. Crystallography isperformed on the two-dimensional crystals from the membrane-anchoredMHC and confirm the correct orientation of the molecule.

40. Slepak VZ: Application of surface plasmon resonance for analysisof protein–protein interactions in the G protein-mediated signaltransduction pathway. J Mol Recog 2000, 13:20-26.

576 Analytical techniques

41. Satpaev DK, Slepak VZ: Analysis of protein–protein interactions inphototransduction cascade using surface plasmon resonance.Methods Enzymol 2000, 316:20-40.

42. Heyse S, Ernst OP, Dienes Z, Hofmann KP, Vogel H: Incorporation ofrhodopsin in laterally structured support membranes: observationof transducin activation with spatially and time-resolved surfaceplasmon resonance. Biochemistry 1998, 37:507-522.

43. Adamczyk M, Moore JA, Yu Z: Application of surface plasmonresonance towards studies of low-molecular-weight antigen-antibody binding interactions. Methods 2000, 20:319-328.

44. McKay D, Davies MJ: BIAcore, La Jolla sense new drugs. TrendsBiotechnol 2001, 19:130.

45. Myszka DG, Rich RL: Implementing surface plasmon resonancebiosensors in drug discovery. Pharm Sci Tech Today 2000, 3:310-317.

46. MacBeath G, Schreiber SL: Printing proteins as microarrays for• high-throughput function determination. Science 2000,

298:1760-1763.Describes a non-SPR protein-chip microarray methodology for high-through-put screening of protein interactions.

47. Rudert F: Genomics and proteomics tools for the clinic. Curr OpinMol Ther 2000, 2:633-642.

48. Weinberger SR, Morris TS, Pawlak M: Recent trends in proteinbiochip technology. Pharmacogenomics 2000, 1:395-416.

49. Figeys D, Pinto D: Proteomics on a chip: promising developments.Electrophoresis 2001, 22:208-216.

50. Thulasiraman V, McCutchen-Maloney SL, Motin VL, Garcia E:Detection and identification of virulence factors in Yersinia pestisusing SELDI ProteinChip system. BioTechniques 2001,30:428-432.

51. Williams C, Addona TA: The integration of SPR biosensors withmass spectrometry: possible applications for proteome analysis.Trends Biotechnol 2000, 18:45-48.

52. Nelson RW, Nedelkov D, Tubbs KA: Biosensor chip massspectrometry: a chip-based proteomics approach. Electrophoresis2000, 21:1155-1163.

53. Nedelkov D, Nelson RW: Practical consideration in BIA/MS:optimizing the biosensor-mass spectrometry interface. J MolRecog 2000, 13:140-145.

54. Natsume T, Nakayama H, Jansson O, Isobe T, Takio K, Mikoshiba K:• Combination of biomolecular interaction analysis and mass

spectrometric amino acid sequencing. Anal Chem 2000,72:4193-4198.

This paper describes an approach for combining SPR and electrospray tan-dem MS used to obtain unambiguous sequence information for proteinsbound to the sensor chip. This is potentially a very powerful approach foridentification of novel interaction partners.

Surface plasmon resonance McDonnell 577

ORIGINAL ARTICLE

Hesheng Ou Æ Limei Yan Æ Devkumar Mustafi

Marvin W. Makinen Æ Matthew J. Brady

The vanadyl (VO2+) chelate bis(acetylacetonato)oxovanadium(IV)potentiates tyrosine phosphorylation of the insulin receptor

Received: 10 July 2005 / Accepted: 20 September 2005 / Published online: 19 October 2005� SBIC 2005

Abstract We have compared the insulin-like activity ofbis(acetylacetonato)oxovanadium(IV) [VO(acac)2], bis(maltolato)oxovanadium(IV) [VO(malto)2], and bis(1-N-oxide-pyridine-2-thiolato)oxovanadium(IV) [VO(OPT)2] indifferentiated 3T3-L1 adipocytes. The insulin-likeinfluence of VO(malto)2 and VO(OPT)2 was decreasedcompared with that of VO(acac)2. Also, serum albuminenhanced the insulin-like activity of all three chelatesmore than serum transferrin. Each of the three VO2+

chelates increased the tyrosine phosphorylation ofproteins in response to insulin, including the b-subunitof the insulin receptor (IRb) and the insulin receptorsubstrate-1 (IRS1). However, VO(acac)2 exhibited thegreatest synergism with insulin and was thereforefurther investigated. Treatment of 3T3-L1 adipocyteswith 0.25 mM VO(acac)2 in the presence of 0.25 mMserum albumin synergistically increased glycogenaccumulation stimulated by 0.1 and 1 nM insulin, andincreased the phosphorylation of IRb, IRS1, proteinkinase B, and glycogen synthase kinase-3b. Wortman-nin suppressed all of these classical insulin-signalingactivities exerted by VO(acac)2 or insulin, except fortyrosine phosphorylation of IRb and IRS1. Addi-tionally, VO(acac)2 enhanced insulin signaling andmetabolic action in insulin-resistant 3T3-L1 adipocytes.Cumulatively, these results provide evidence thatVO(acac)2 exerts its insulin-enhancing properties bydirectly potentiating the tyrosine phosphorylation ofthe insulin receptor, resulting in the initiation of insulinmetabolic signaling cascades in 3T3-L1 adipocytes.

Keywords Adipocyte Æ Diabetes Æ Insulin receptor andinsulin receptor substrate Æ Insulin signaling Æ Tyrosinephosphorylation Æ Vanadyl

Abbreviations BSA: Bovine serum albumin Æ DMSO:Dimethyl sulfoxide Æ ENDOR: Electron–nucleardouble resonance Æ EPR: Electron paramagneticresonance Æ GSK: Glycogen synthase kinase Æ HEPES:N-(2-Hydroxyethyl])piperazine-N¢-(2-ethanesulfonicacid), sodium salt Æ IRb: b-subunit of the insulinreceptor Æ IRS1: Insulin receptor substrate-1 Æ KRBH:Kreb’s Ringer’s buffer Æ PBS: Phosphate-bufferedisotonic saline Æ PI: Phosphatidyl inositol Æ PKB:Protein kinase B Æ VO(acac)2: Bis(acetylacetonato)oxovanadium(IV) Æ VO(malto)2: Bis(maltolato)oxovanadium(IV) Æ VO(OPT)2: Bis(1-N-oxide-pyridine-2-thiolato)oxovanadium(IV)

Introduction

Studies have now convincingly demonstrated blood glu-cose lowering effects in diabetic humans [1–4] and labo-ratory animals [5–7] and enhanced lipogenesis [8, 9] andglucose uptake [10] in whole cells by vanadium com-pounds. Vanadium, occurring betweenCa2+ andZn2+ inthe first transition-metal series, exhibits complex chemis-try because of multiple oxidation states [11–13]. Thisfactor has led to confusion in identifying the antidiabeticagent, for vanadium compounds of oxidation states IVand V are both associated with insulin-enhancing activ-ity—but through different mechanisms. Furthermore,vanadyl (VO2+) compounds, of oxidation state IV, underphysiological conditions are subject to oxidation by avariety of oxidants, including molecular oxygen [14]; andvanadate compounds, of oxidation state V are thought toundergo reduction to state IV in the cell [14–16]. For thesereasons, the mechanisms by which vanadium compoundsmediate antidiabetic effects in vivo are poorly describedand incompletely understood.

H. Ou Æ L. Yan Æ M. J. Brady (&)Department of Medicine and Committee on Molecular Metabolismand Nutrition, The University of Chicago, MC1027, 5841 S.Maryland Ave, Chicago, IL 60637, USAE-mail: mbrady@medicine.bsd.uchicago.eduTel.: +1-773-7022346Fax: +1-773-8340486

D. Mustafi Æ M. W. MakinenDepartment of Biochemistry and Molecular Biology,The University of Chicago, Chicago, IL 60637, USA

J Biol Inorg Chem (2005) 10: 874–886DOI 10.1007/s00775-005-0037-x

The chemical bonding structures of the three organicchelates of VO2+ that have received the greatestattention through laboratory studies are illustrated inFig. 1. These VO2+ chelates exhibit significantly enhan-ced insulin-mimetic activity in diabetic laboratoryanimals or adipocyte cells over that of inorganic VO2+

introduced as VOSO4 [5–10]. Although these observa-tions suggest that the increase is at least in part due tothe increased lipophilic character of the chelate, theyalso indicate that the structure of the organic chelatingligand and its electronic structure, i.e., bonding inter-actions with the VO2+ moiety, are important factorsgoverning the reactivity of VO2+ chelates with biologi-cal macromolecules.

Organic chelates of VO2+, of which the bis(maltola-to)oxovanadium(IV) [VO(malto)2] compound is pres-ently the most widely studied example in the literature[17–22], offer the most likely vehicle for drug designbecause the structural properties of the organic ligandschelating the central V4+ ion could be syntheticallymolded to increase specificity of action. Since inorganicVO2+, existing as the penta-aquo vanadyl cation[VO(H2O)5]

2+ in solution [23, 24], is itself associatedwith insulin-like activity [25, 26], an important propertyof the chelating ligand is its intrinsic binding affinity withwhich the central VO2+ moiety is retained in the com-plex, particularly in the presence of biological macro-molecules such as serum transport proteins. Theseaspects have not received uniform attention, and theinsulin-like activity of a variety of VO2+ chelates hasbeen measured in whole animal and cellular systemswith little regard to the relative proportions of intactchelate, inorganic VO2+, or partial, hemichelated spe-cies into which the compound may have dissociated.Since the insulin-mimetic properties of each chemicalspecies are likely to differ according to specific activityand structural basis and a variety of cellular, enzyme,and whole animal assay methods for measuring antidi-abetic activity are employed, it is often not evident whichchemical species of vanadium is responsible for the ob-served effects.

We have argued that the organic chelating ligands ofthe VO2+ compounds illustrated in Fig. 1 are likely tofacilitate binding to proteins in the blood stream [10].

Binding to serum albumin, for instance, as themajor serum transport protein, may stabilize VO2+

against oxidation or result in formation of a specific[protein/VO2+ chelate] adduct that is recognized at themembrane surface of target cells. In support of theseideas, we have observed that bis(acetylacetonato)oxo-vanadium(IV) [VO(acac)2] (Fig. 1a) forms an adductof 1:1 stoichiometry with serum albumin and thatits insulin-mimetic effect, measured as the uptake of2-deoxy-D-[1-14C]glucose by differentiated, cultured3T3-L1 adipocytes, is greatly enhanced in the presenceof albumin [10]. To investigate the structural origins ofthe insulin-enhancing activity of VO2+ chelates further,we compared the stability and structures in solution ofthe three organic chelates of VO2+ illustrated in Fig. 1,the influence of albumin and transferrin on their insu-lin-enhancing activity in cultured 3T3-L1 adipocytes,and their interactions with components of the insulin-signaling pathway. Our results lead to the conclusionthat VO(acac)2 most strongly promotes the autophos-phorylation and tyrosine kinase activity of the insulinreceptor, enhancing the phosphorylation of tyrosineresidues on insulin receptor substrate-1 (IRS1) in adose-dependent manner that is synergistic with addedinsulin. Since our observations run counter to theconclusions of the group of Shechter [8, 9, 27] that theinfluence of inorganic VO2+ and of VO2+ chelates liesdownstream from the insulin receptor involving acytosolic protein tyrosine kinase, it is important toresolve the origin of these apparently contradictoryresults and to identify target enzymes of VO2+ chelatesto assign the molecular basis of their insulin-enhancingactivity.

Materials and methods

Materials

Cell culture reagents and calf serum were supplied byMediatech (Herndon, VA, USA). Fetal bovine serumwas obtained from Hyclone (Logan, UT, USA). Porcineinsulin and all other chemicals were from Sigma (St.Louis, MO, USA). D-[U-14C]glucose (317 mCi/mmol)

Fig. 1 Comparison of the chemical bonding structures of organicchelates of VO2+ with insulin-enhancing activity. a Bis(acetyl-acetonato)oxovanadium(IV) [VO(acac)2], b bis(maltolato)

oxovanadium(IV) [VO(malto)2], c bis(1-N-oxide-pyridine-2-thio-lato)oxovanadium(IV) [VO(OPT)2]

875

was obtained from MP Biomedicals (Irvine, CA, USA),while UDP-[U-3H]glucose (60 Ci/mmol) was suppliedby American Radiolabeled Chemicals (St. Louis).Enhanced chemiluminescence reagents and Protein ASepharose beads were purchased from Amersham Bio-sciences (Piscataway, NJ, USA), and GF/A filters weresupplied by Whatman (Haverhill, MA, USA). Com-mercial sources of antibodies were as follows:anti-phospho-GSK-3b (Ser9), where GSK-b is glycogensynthase kinase-3b, and anti-phospho-PKB (Thr308),where PKB is protein kinase B, were purchased fromCell Signaling Technology (Beverly, MA, USA), anti-phosphotyrosine (clone 4G10) was supplied by UpstateCell Signaling Solutions (Lake Placid, NY, USA), anti-insulin receptor b-subunit (C-19) was from Santa CruzBiotechnology (Santa Cruz, CA, USA), and anti-IRS1polyclonal was generated by injection of rabbits with aGST-fusion construct comprising the N-terminus ofmouse IRS1. Horseradish peroxidase-conjugated goatantirabbit and goat antimouse immunoglobulin G wereobtained from Bio-Rad (Hercules, CA, USA).

Vanadyl sulfate hydrate andVO(acac)2 were purchasedfrom Sigma-Aldrich (Milwaukee, WI, USA). 1-N-Oxide-2-thiolato-pyridine (99%) was obtained from ResearchChemicals (Heysham, Lancashire, UK) and used withoutfurther purification for preparation of crystalline bis(1-N-oxide-pyridine-2-thiolato)oxovanadium(IV) [VO(OPT)2]according to the procedure of Sakurai et al. [7]. CrystallineVO(malto)2 was a gift from C. Orvig and K. H. Thomp-son of the University of British Columbia. Fatty acid freebovine serum albumin (BSA) and bovine apotransferrin(98%) were from Sigma. All other reagents were as pre-viously described [10].

Cell culture

3T3-L1 fibroblasts were maintained and differentiatedinto adipocytes, as previously reported [10, 28]. Cellswere used 5–10 days after completion of the differenti-ation protocol, when more than 95% of the cells con-tained lipid droplets. Prior to experiments, the cells werewashed twice with Kreb’s Ringer’s buffer (KRBH)/0.5%BSA/5 mM glucose and incubated in the same mediumfor 3 h. The cells were then washed twice with KRBHlacking BSA, and then placed in KRBH containing aVO2+ chelate in 1:1 molar ratio with BSA in the absenceor presence of the indicated concentration of insulin.After treatment of adipocytes under these conditions for5–15 min, the cells were used for glycogen synthesisassay, or placed onto ice and washed three times withPhosphate-buffered isotonic saline (PBS).

Preparation of cell lysates

After washing with PBS, the cells were scraped directlyinto homogenization buffer [50 mM (N-(2-hydroxyethyl)piperazine-N¢-(2-ethanesulfonic acid), sodium salt(HEPES), pH 7.5, 150 mM NaCl, 10% glycerol, 0.5%

Triton X-100, 2 mM EDTA, 10 mM NaF, with theprotease inhibitors aprotinin (10 lg/ml) and benzami-dine (10 lM) added just before use] for immunoblottingexperiments. Lysates were centrifuged for 10 min at10,000g at 4 �C, and the supernatants were transferredto new microfuge tubes. Immunoblotting was performedas previously described [29].

Immunoprecipitations

Cell lysates were prepared as just described, and pre-cleared for 15 min with empty Protein A Sepharosebeads to remove any nonspecific protein binding to theimmunocomplex. A 10-ll aliquot of the indicatedantiserum was added to the supernatant, and the samplewas mixed for 1 h at 4 �C. A 50-ll aliquot of a 50%protein A sepharose bead slurry was added, and thesamples were mixed for 30 min. The sample was cen-trifuged for 2 min at 800g and 4 �C, and the pelletedantibody–bead complex was then washed three timeswith 1 ml of homogenization buffer. Bound proteinwas eluted from the beads using 80 ll of 1X Laemmlisample buffer and boiling for 2 min. Samples were sep-arated by sodium dodecyl sulfate polyacrylamide gelelectrophoresis, transferred to nitrocellulose, and ana-lyzed by anti-phosphotyrosine immunoblotting [29].

Metabolic assays

Glycogen synthesis assays were performed as previouslyreported [28, 30]. Briefly, after serum starvation, 3T3-L1adipocytes in 12-well dishes were incubated as describedearlier in the presence of the indicated vanadyl com-pound with or without insulin for an additional 15 min.Subsequently, the cells were incubated with 5 mMD-[U-14C]glucose (2 lCi per well) for 45 min at 37 �C.The cells were then washed three times with ice-cold PBSand solubilized in 30% (w/v) KOH. Cellular glycogenwas precipitated with 70% ethanol and radiolabeledglucose incorporation into glycogen was determined byliquid scintillation counting.

Glycogen synthase activity was determined asdescribed previously [28] with some modifications.Cells in 12-well dishes were serum-deprived, thenincubated with the indicated vanadyl compound withor without insulin for 15 min at 37 �C. After washingwith ice-cold PBS three times, the cells were scrapedinto 500 ll of glycogen synthase assay buffer [50 mMtris(hydroxymethyl)aminomethane)–HCl, pH 7.8, 10 mMEDTA, 100 mM NaF, 0.5% Triton X-100] and centri-fuged (10,000g for 10 min). To measure glycogensynthase activity, a 50-ll aliquot of the supernatant(50–100 lg protein) was added to an equal volumeof buffer lacking detergent, containing 10 mM UDP-[U-3H]glucose (0.1 lCi/lmol) and 16 mg/ml glycogen,in the presence or absence of 20 mM glucose-6-phos-phate. After 15-min incubation at 37 �C, the assay

876

tubes were chilled for 15 min in an ice bath. A 90-llaliquot of the reaction mixture was then spotted on2.4-cm GF/A Whatman filter disks, and glycogen wasprecipitated by immersion in 70% ethanol at 4 �C. FreeUDP-[U-3H]glucose was removed by washing the filtersthree times for 10 min each time in 70% ethanol, thelast two washings occurring at room temperature. Thefilters were air-dried, and incorporation of radioactivitywas determined by liquid scintillation counting.

Electron paramagnetic resonance and electron–nucleardouble resonance

For comparative studies of the stability of VO2+ chelates,stock solutions were prepared by dissolving the crystal-line compound in a small volume of dimethyl sulfoxide(DMSO) followed by dilution to the desired concen-tration with N2-purged KRBH buffered with 10 mMHEPES to pH 7.4. The final solutions contained no morethan 10% (v/v) DMSO. No evidence of chelate dis-placement by DMSO was detectable by electron para-magnetic resonance (EPR) or electron–nuclear doubleresonance (ENDOR). Also, no evidence of binding ofDMSO to inorganic VO2+ in solution or to the axialcoordination site in VO2+ chelates was found on the basisof ENDOR spectra. Aliquots of these solutions were re-moved under a N2 atmosphere as a function of time andfrozen immediately and kept in liquid N2 until used forcollection of EPR and ENDOR spectra.

EPR and ENDOR spectra were recorded with an X-band Bruker ESP 300E spectrometer equipped with acylindrical TM110 ENDOR cavity and an OxfordInstruments ESR910 liquid helium crysotat and BrukerENDOR digital accessories, as previously described [31,32]. The spectrometer was equipped with a completecomputer interface (Bruker ESP3220 data system) forspectrometer control and data acquisition and proces-sing. Typical experimental conditions for ENDORmeasurements were as follows: temperature 20 K,microwave frequency 9.45 GHz, incident microwavepower 64 lW (full power 640 mW at 0 dB), rf power50–70 W, rf modulation frequency 12.5 kHz, and rfmodulation depth 10–20 kHz. The static laboratorymagnetic field was not modulated for ENDOR.

Results

Comparative insulin-like activity of VO2+ chelates

In efforts to identify the VO2+ species that potentiateinsulin signaling in 3T3-L1 adipocytes, we comparedthe patterns of tyrosine phosphorylated proteins stim-ulated by VO(acac)2, VO(malto)2, and VO(OPT)2, inthe presence and absence of insulin. As seen in Fig. 2,treatment of cells with each of the VO2+ chelates inthe presence or absence of insulin resulted in an arrayof tyrosine phosphorylated proteins. However, it is

evident that the intensity of only two bands migratingat 90 and 180 kDa increased in a dose-dependentmanner with addition of insulin. These proteins werepresumed to be the b-subunit of the insulin receptor(IRb) and IRS1, respectively, and this supposition wassubsequently confirmed (see later). In Fig. 2a, it isevident that the intensity of tyrosine phosphorylation,particularly of the insulin-sensitive proteins at 90 and180 kDa, was significantly greater with VO(acac)2 thanwith VO(malto)2 or VO(OPT)2.

In earlier studies, the group of Shechter [9, 27] con-cluded that the insulin-mimetic action of VO2+ andVO(acac)2 involves only activation of a postreceptorpathway. To confirm that VO(acac)2 treatment of 3T3-L1 adipocytes increased tyrosine phosphorylation ofIRS1 and IRb, immunoprecipitations were performed,using a specific antibody for each protein. 3T3-L1adipocytes were treated in the presence or absence of10 nM insulin with and without 0.25 mM VO(acac)2 for5 min, lysates were prepared and subjected to anti-IRS1or anti-IRb immunoprecipitation, and samples werethen analyzed by anti-phosphotyrosine immunoblotting.Insulin increased the tyrosine phosphorylation of the180-kDa IRS1 protein and the 90-kDa IRb subunit(Fig. 2b, c), confirming the identity of these proteins inthe anti-phosphotyrosine immunoblot shown in Fig. 2a.These results demonstrated that VO(acac)2 increased thetyrosine phosphorylation of the insulin receptor and itsimmediate downstream target IRS1 in 3T3-L1 adipo-cytes.

In Fig. 3, we compare the influence of serum albu-min and serum transferrin on the insulin-mimeticresponse of each VO2+ compound. While the intensityof the immunoblots was uniformly greater for eachVO2+ chelate in the presence of serum albumin than inthe presence of transferrin, the pattern of tyrosinephosphorylated proteins remained the same. However,close inspection of the immunoblots showed that theordering of intensities was dependent on the proteinadded to the assay mixture. In the presence of albumin,the relative intensities followed the order: VO(acac)2�VO(OPT)2>VO(malto)2@VOSO4>control. On theother hand, in the presence of transferrin, the orderingof intensities was VO(acac)2�VO(OPT)2>VO(mal-to)2@VOSO4@control, with virtually no differentiatingfeature among the last three systems. Furthermore, theintensity of the immunoblots was responsive toincreasing concentrations of insulin only for VO(acac)2and VO(OPT)2 in the presence of serum transferrin,while the intensity of the immunoblots elicited byVOSO4 and VO(malto)2 remained unaltered and didnot differ from that of the control. These observationsindicated that the intensities of the immunoblots foreach VO2+ chelate were sensitive to insulin only in thepresence of albumin.

Figure 4 compares the intensity of tyrosine phos-phorylation of signaling proteins promoted by VO(acac)2 in the presence and absence of 10 nM insulin. Inthe absence of insulin, lanes 1–6 showed that there is a

877

monotonic increase in the intensity of IRb withincreasing concentration of VO(acac)2 in the incubationmedium. In the presence of 10 nM insulin, as shownthrough lanes 7–12, there was similarly a monotonicincrease in the intensity of IRb with increasing concen-trations of VO(acac)2. However, in the presence ofinsulin the bands belonging to IRb were more intensethan in the absence of insulin. These observations indi-cated that VO(acac)2 acts synergistically with insulin,increasing the tyrosine phosphorylation of the insulinreceptor potentiating the same enzymatically controlledpathway.

Comparative spectroscopic studies of the relativestability of VO2+ chelates in solution

We presumed that the insulin-enhancing action oforganic chelates of VO2+ was likely to be a function of

their relative chemical stability regulated by the bindingaffinity of the organic ligand for VO2+ and the tendencyof the VO2+ chelate to be oxidized. Therefore, we exam-ined the stability of the organic VO2+ chelates in KRBHmedium used for incubation of adipocytes in metabolicassays. To determine the chemical stability with time, weselected a combined EPR and ENDOR approach tomonitor the presence of intact VO2+ chelate. Figure 5illustrates progress curves showing a time-dependentdecrease in the intensity of EPR and ENDOR absorp-tions of VO(acac)2, VO(malto)2, and VO(OPT)2 inKRBH medium. Analysis showed that the data corre-sponded to a single-exponential decay with half-lives(t1/2) of (4, 6, and 4,000 days or more for VO(malto)2,VO(OPT)2, and VO(acac)2, respectively. Since the EPRintensity monitors the concentration of the paramag-netic VO2+ ion while the ENDOR intensity is spe-cific for the content of covalent hydrogens of ligandsattached to the paramagnetic VO2+ ion, the parallel

Fig. 2 Comparison of effects ofvanadyl chelates on tyrosinephosphorylation in 3T3-L1adipocytes. Cells were serum-starved for 3 h, and thenincubated in Kreb’s Ringer’sbuffer (KRBH) + 0.25 mMbovine serum albumin, with theindicated concentration ofinsulin, in the presence of0.25 mM VO(acac)2,VO(malto)2, or VO(OPT)2 for5 min. a Cellular lysatesseparated by sodium dodecylsulfate polyacrylamide gelelectrophoresis and transferredto nitrocellulose forimmunoblotting with anti-phosphotyrosine antibody.Alternatively, lysates weresubjected toimmunoprecipitation usinganti-insulin receptor substrate-1(anti-IRS1) (b) or the b-subunitof the anti-insulin receptor(anti-IRb) (c) antibodies, andsamples were then analyzed byanti-phosphotyrosineimmunoblotting. All results arerepresentative of two to fourindependent experiments

878

decrease in intensity reflects loss of intact VO2+ chelate.The results in Fig. 5, thus, demonstrate that the stabilityof VO(acac)2 in solution as an intact complex farexceeded that of the other two VO2+ chelates under

conditions comparable to those employed for measuringglucose uptake by adipocytes. Other investigators havealso observed diminution of the EPR absorption ofVO2+ chelates as a function of time comparable to the

Fig. 3 Comparison of effects ofvanadyl chelates on tyrosinephosphorylation in 3T3-L1adipocytes in the presence ofserum albumin or transferrin.Insulin in indicated amountsand 0.25 mM VO2+ chelateswere added in the presence ofeither 0.25 mM serum albuminor 0.25 mM serum transferrin.Arrows point to IRS1 and IRbcomponents identified byspecific antibody reaction. It isevident that the intensity oftyrosine phosphorylatedproteins upon addition of0.25 mM VOSO4 in thepresence or absence of insulindid not differ discernably frombasal conditions

Fig. 4 Dose response of VO(acac)2-stimulated tyrosine phosphory-lation in the presence and absence of insulin. 3T3-L1 adipocytes wereserum-starved for 2.5 h prior to addition of 10 nM insulin to half ofthe wells, and varying concentrations of VO(acac)2. After 5 min, cel-lular lysates were prepared and analyzed by anti-phosphotyrosine

immunoblotting. Concentrations of VO(acac)2 used: lanes 1 and 70 mM; lanes 2 and 8 0.01 mM; lanes 3 and 9 0.05 mM; lanes 4 and10 0.1 mM; lanes 5 and 11 0.25 mM; and lanes 6 and 12 0.5 mM.The immunoblot shown is representative of three independentexperiments

879

results described here for VO(malto)2 and VO(OPT)2[33]. Since the insulin-enhancing activity of VO(acac)2far exceeded that of the other two chelates and wassynergistic with added insulin, as shown through Figs. 2and 4, we used only VO(acac)2 in further investigationsof its influence on other components of the insulin-sig-naling pathway in 3T3-L1 adipocytes in view of itssuperior chemical stability.

Influence of VO(acac)2 on phosphorylationof insulin-signaling proteins

In our earlier investigation of the insulin-like action oforganic chelates of VO2+, we demonstrated thatVO(acac)2 is superior to other VO2+ chelates enhancingglucose uptake in 3T3-L1 adipocytes [10]. In Fig. 2, it isseen that VO(acac)2 increased the intensity of tyrosinephosphorylation of IRS1 in the absence and presence of10 nM insulin. Therefore, to examine the downstreameffects of enhanced IRS1 phosphorylation, we deter-mined the phosphorylation states of PKB and GSK-3bby immunoblotting. In Fig. 6a, it is seen that immuno-blotting indicative of serine/threonine phosphorylationof PKB and GSK-3b was significantly more intense inthe presence of VO(acac)2 for all levels of added insulin.In contrast, inclusion of VOSO4 had little discernable

effect under all conditions tested (Fig. 6a). Figure 6bshows that the intensity of immunoblots indicative ofphosphorylation of GSK-3b in the absence of insulin(lanes 1–6) is dependent on the concentration of VO(acac)2 in the incubation medium. In the presence of10 nM insulin (lanes 7–12), there is also an increase inthe intensity of immunoblots dependent on the concen-tration of VO(acac)2; however, the influence of increas-ing concentrations of VO(acac)2 in the presence of10 nM insulin quickly saturates the response and syn-ergistic activity is not observed as readily as in Fig. 6a.Cumulatively, these observations demonstrate that theinsulin-mimetic action of VO(acac)2 potentiating thephosphorylation of PKB and GSK-3b is synergistic withinsulin and is not replicated by VOSO4.

Potentiation of glycogen synthesis and glycogensynthase activation by VO(acac)2

A primary metabolic effect of insulin in adipocytesand muscle is to increase incorporation of cellularglucose into glycogen through the coordinated trans-location of GLUT4 and the activation of glycogensynthase. As shown in Fig. 7, the influence of VO(acac)2 on glucose incorporation into glycogen in theabsence of insulin was significantly greater than the

Fig. 5 Comparison of the time-dependent decrease in intensity ofelectron paramagnetic resonance (EPR) absorption and in theintensity of electron–nuclear double resonance (ENDOR) absorp-tions of covalent hydrogens specific for the ligand of VO2+ chelatesin KRBH. Aliquots of solutions of VO(acac)2, VO(malto)2, andVO(OPT)2, prepared as described in the ‘‘Materials and methods’’and kept at ambient temperature (22 �C), were removed at theindicated times under N2 with a Hamilton air-tight syringeequipped with a Teflon catheter, transferred directly into EPRsample tubes, and immediately frozen and stored in liquid N2. EPR

and ENDOR spectra were collected at (0 K as previously described[28, 29]. The peak-to-peak amplitude of the �3/2^ component ofthe EPR absorption spectrum and of the A^ features of theENDOR spectrum with H0 at the �3/2^ setting [10, 18] are plottedas IEPR and IENDOR, respectively, normalized to the correspondingamplitudes at time zero. The data were found to adhere to first-order exponential decay with use of the program Origin (MicrocalSoftware, Northampton, MA, USA), from which the half-liveswere estimated as 4, 6, and 4,000 days or more for VO(malto)2,VO(OPT)2, and VO(acac)2, respectively

880

basal rate and also greater than that for VOSO4.Furthermore, the action of VO(acac)2 was synergisticwith submaximal insulin in promoting incorporationof glucose into glycogen (Fig. 7a). In the presence of10 nM insulin, the capacity of the adipocytes for gly-cogen synthesis was saturated, and the contribution ofthe chelate or of VOSO4 above the influence of thehormone was no longer detectable from that observedat lower insulin concentrations.

Since glycogen synthase activity is normally stimu-lated by insulin, we examined whether the influence ofthe VO2+ chelate in augmenting glycogen synthesisderives from changes in activity of this rate-limitingenzyme. As illustrated in Fig. 7b, the glycogen syn-thase activity ratio was increased with VO(acac)2 inthe incubation medium both in the absence and in thepresence of insulin. However, this result does notdistinguish between VO(acac)2 acting upstream ofglycogen synthase vs. direct activation of glycogensynthase.

To address this issue, 3T3-L1 adipocytes werepretreated for 15 min with 200 nM of the phosphati-dyl inositol (PI) 3¢-kinase inhibitor wortmannin priorto stimulation with insulin or VO(acac)2. Wortmanninblocked insulin and VO(acac)2 stimulated phosphory-lation of GSK-3b (Fig. 8a), as well as the activation ofglycogen synthase (Fig. 8b). In contrast, wortmanninhad no influence on insulin or VO(acac)2 induced

tyrosine phosphorylation (Fig. 8c). Cumulatively, theseresults strongly suggest that VO(acac)2 enhanced gly-cogen synthase activation through the insulin receptorand initiation of classical, PI 3¢-kinase dependentinsulin signaling cascade rather than through directinfluence on glycogen synthase or the kinase orphosphatase enzymes that control its phosphorylationstate.

Influence of VO(acac)2 on insulin-resistant 3T3-L1adipocytes

In type II diabetes, insulin resistance is characterized bya decrease in both sensitivity and responsiveness of tis-sues to the circulating hormone [41–43]. The results ofprevious investigations have indicated that 18-h treat-ment of 3T3-L1 adipocytes with low-dose (1 nM) insulinresults in the induction of insulin resistance [30]. We nextexamined the influence of VO(acac)2 on glucose meta-bolism in insulin-resistant 3T3-L1 adipocytes.

In Fig. 9a, it is seen that the extent of insulin-stimu-lated tyrosine phosphorylation of IRb and IRS1 ininsulin-resistant adipocytes was markedly reducedcompared with that in control cells. Incubation of theinsulin-resistant adipocytes with 0.25 mM VO(acac)2restored the phosphorylation of insulin receptor andIRS1 to levels observed in normal cells, even in the

Fig. 6 Influence of VO(acac)2 on the phosphorylation states ofprotein kinase B (PKB) and glycogen synthase kinase-3b (GSK-3b).a 3T3-L1 adipocytes were serum-starved for 3 h, and treated withthe indicated concentration of insulin, in the presence of 0.25 mMVO(acac)2 or VOSO4 for 5 min. Lysates were prepared and thephosphorylation states of PKB and GSK-3b were determined byimmunoblotting with phospho-specific antibodies. Bands corre-sponding to phospho-PKB and phospho-GSK-3b (pGSK-3b) are

indicated by arrows. b Serum-starved 3T3-L1 adipocytes werestimulated for 5 min with varying amounts of VO(acac)2, in theabsence (basal) or presence of 10 nM insulin. Lysates wereprepared and analyzed by immunoblotting using anti-pGSK-3bantibody. Concentrations of VO(acac)2 used: lanes 1 and 7 0 mM;lanes 2 and 8 0.01 mM; lanes 3 and 9 0.05 mM; lanes 4 and 100.1 mM; lanes 5 and 11 0.25 mM; and lanes 6 and 12 0.5 mM. Allresults are representative of three independent experiments

881

absence of added insulin. Similarly, Fig. 9b illustratesthat insulin-mediated phosphorylation of GSK-3b wassignificantly reduced in insulin-resistant 3T3-L1 adipo-cytes compared with that in control cells. However,VO(acac)2 restored phosphorylation of GSK-3b ininsulin-resistant cells to near maximal levels.

Figure 10a correspondingly compares the influence ofVO(acac)2 and VOSO4 on the incorporation of glucoseinto glycogen in control and insulin-resistant 3T3-L1adipocytes. In insulin-resistant cells, the effect of insulinon glycogen synthesis was noticeably decreased com-pared with that in normal cells. Treatment of insulin-resistant adipocytes with VO(acac)2 in the presence andabsence of insulin restored glycogen synthesis activity tonearly control levels. Interestingly, while VO(acac)2acted synergistically with insulin in control and insulin-

resistant cells, VOSO4 had little effect on insulin actionin either type of 3T3-L1 adipocyte (Fig. 10).

Discussion

Chemical and structural stability of VO2+ chelatesin solution

Schieven et al. [44] have shown that 10–25 lMVO(malto)2acts as an inhibitor of protein tyrosine phosphatasesaltering signal transduction during induction of B cellapoptosis. Peters et al. [22] concluded on the basis ofNMR and X-ray studies that the competitive inhibitionof VO(malto)2 against protein tyrosine phosphatases canbe attributed to the unliganded VO2+ cation extractedfrom the intact complex. On this basis, the influenceof VO(malto)2 on glucose metabolism appears not toderive from the intact metal-chelated complex but ratherfrom VO2+ binding to macromolecular components ofinsulin signal transduction pathways more tightly thanto the maltolato ligands. Our observations of low insu-lin-enhancing activity by VO(malto)2 in Figs. 2 and 3comparable to that observed for inorganic VO2+ are inaccord with these conclusions, suggesting that albuminand transferrin similarly extract VO2+ from a significantfraction of the intact, chelated complex. In contrast, theintensity of tyrosine phosphorylated proteins elicited byVO(OPT)2 in Figs. 2 and 3 was more marked than forVO(malto)2. By EPR studies the structural intactness ofVO(OPT)2 appears not to be disrupted in the presence ofalbumin as severely as is that of VO(malto)2 (D. Mustafi,M.W. Makinen, unpublished observations).

We conclude, therefore, that the relative ordering ofintensities of tyrosine phosphorylated proteins in Fig. 2parallels the chemical stability of the VO2+ chelates, asevidenced by the results in Fig. 5, and their abilityto remain intact in the midst of proteins. Therefore,the insulin-enhancing activity of VO(malto)2 in thesemetabolic assays must be ascribed to a mixture ofeffects resulting from albumin:VO2+complexes, albu-min: VO(malto)2 complexes, and possibly hemichelateforms of the complex bound to serum albumin. Theinsulin-enhancing activity of VO(OPT)2 is due to analbumin-bound form of the VO2+ chelate and the freecomplex as a minimum number of species. On theother hand, VO(acac)2 binds to serum albumin as anintact chelate in 1:1 stoichiometry [10] and elicits themost intense pattern of tyrosine phosphorylated pro-teins. Because the insulin-mimetic action of VO(acac)2in these assays is best ascribed to the albumin: VO(acac)2 adduct as the predominant species, we emplo-yed only VO(acac)2 for further mechanistic studiesbecause of its inherently greater stability and because itis associated with the highest activity measuredaccording to the intensity of phospho-tyrosine immu-noblots.

Fig. 7 Effect of VO(acac)2 on insulin-stimulated glycogen synthesisand glycogen synthase (GS) activity. Replicate 12-wells of 3T3-L1adipocytes were serum-starved for 3 h. The cells were then stimu-lated in triplicate with the indicated concentration of insulin, in thepresence of 0.25 mMVO(acac)2 or VOSO4 for 15 min. aCell lysateswere prepared, and GS activity was measured in vitro, in theabsence and presence of 10 mM glucose-6-phosphate. b For mea-surement of glycogen synthesis, 2 lCi of 14C-glucose was added toall wells after the 15-min stimulation with insulin and chelates. After30 min, the cells were washed and glucose incorporation intoglycogen was determination. VO(acac)2 significantly increased GSactivation by insulin, resulting in synergistic enhancement ofinsulin-stimulated glycogen synthesis

882

Molecular locus of the insulin-enhancing actionof VO(acac)2

Suppression of insulin-induced signaling by the clas-sical inhibitor wortmannin occurs at the level of PI 3¢-kinase, the first component of the enzyme cascade totransduce the effects of IRS1 when activated throughthe kinase activity of the insulin receptor [37]. Thus,although increased phosphorylation of PKB andGSK-3b and increased glycogen synthase activity wereobserved upon addition of VO(acac)2 to 3T3-L1adipocytes, as observed through Figs. 6 and 7, inhi-bition of these effects by wortmannin, as demonstratedin Fig. 8, rules out significant direct action of the

VO2+ chelate on these components of the insulin-signaling pathway. In view of the increased tyrosinephosphorylation of immunoprecipitated IRb and IRS1in Fig. 2, we conclude that the insulin-enhancing ac-tion of VO(acac)2 is associated with insulin receptoractivation. However, it not certain whether VO(acac)2directly stimulates insulin receptor tyrosine kinaseactivity, or whether it acts indirectly through activa-tion of other tyrosine kinases or inhibition of tyrosinephosphatases. Determination of the molecular path-ways through which VO(acac)2 exerts its insulin-likeeffects will require further study.

The group of Shechter [8, 9] ascribe the greater potencyof VO(acac)2 over that of VOSO4 in stimulating

Fig. 8 Wortmannin (wort)inhibition of the influence ofVO(acac)2 on GSK-3b and GS.Replicate 12-well dishes of3T3-L1 adipocytes were serum-starved for 2.5 h. The cells werepretreated for 15 min in theabsence and presence of200 nM wort, prior to theaddition of 10 nMinsulin ± 0.25 mM VO(acac)2or VOSO4, as indicated. a After5 min, lysates prepared andanalyzed by anti-pGSK-3bimmunoblotting. b After15 min of stimulation, GSactivity was measured in vitro,in the absence and presence of10 mM glucose-6-phosphate. cAfter 5 min, lysates wereprepared and analyzed by anti-phosphotyrosineimmunoblotting. GS activityresults are representative ofthree independent experiments,each performed in duplicate,while immunoblots arerepresentative of three to fourindependent experiments

883

lipogenesis purely to an insulin-independent pathway andincreased stability of the VO2+ chelate against oxidation.Since they employed only primary rat adipocytes in theirexperiments [8, 9, 27, 45, 46], one potential source ofthis apparent contradiction is that primary adipocytesdiffer metabolically from cultured 3T3-L1 adipocyteswith respect to the effects of vanadium compounds: It iswell known that the main metabolic action of primaryadipocytes is to convert glucose into lipids and fatty acids,while the primary metabolic flux in 3T3-L1 adipocytes isthe conversion of glucose into glycogen [47]. It is, inour estimation, unlikely that these contradictory inter-pretations of experimental results have their origin indifferences in metabolic fluxes between the two cell lines.The group of Shechter [9] demonstrated that inorganicVO2+ is noncompetitively inhibitory against purifiedinsulin receptor protein tyrosine kinase and that it isinhibitory to insulin receptor autophosphorylation. Inthese studies, the group of Shechter investigated theeffects only of inorganic VO2+ added as VOSO4 andinorganic vanadate on the insulin receptor. At no point intheir experiments did they directly test the influence ofVO(acac)2 on insulin receptor tyrosine kinase activity.

As demonstrated in Figs. 2 and 3, VOSO4 added to3T3-L1 adipocytes did not elicit tyrosine phosphoryla-tion of IRb or of IRS1 above basal conditions in thepresence or absence of insulin, in contrast to the cleardose-dependent and insulin-synergistic influence ofVO(acac)2 shown in Fig. 4. Moreover, through Fig. 8 wehave demonstrated not only that VO(acac)2-enhancedglycogen synthase action and VO(acac)2-enhanced

phosphorylation of GSK-3b were wortmannin-sensitive,but also that wortmannin in the same adipocyte cells didnot inhibit insulin-activated or VO(acac)2-facilitatedtyrosine phosphorylation of IRb or of IRS1. Further-more, the reduced levels of tyrosine phosphorylation ofIRb and IRS1 elicited by VO(malto)2 and VO(OPT)2, asillustrated in Figs. 2 and 3, are fully consistent with theirdecreased chemical stability as intact VO2+ chelates, asshown through Fig. 5. These results collectively arguestrongly that VO(acac)2 promotes the tyrosine phosphor-ylation of the insulin receptor resulting in initiation ofseveral insulin-senstive signaling pathways. The emphasisby the group of Shechter [9, 27, 33, 45, 46] that the insulin-like effects of vanadium compounds occur entirelythrough a cytosolic, insulin-receptor-independent pro-tein tyrosine kinase enzyme was made without testingdirectly VO2+ chelates in which the organic ligandexhibits high affinity for the VO2+ ion, such as VO(acac)2.

Peroxovanadates (of oxidation state V) are associatedwith greatly increased phosphorylation of the insulinreceptor owing to their capacity for irreversible oxida-tion of the active-site cysteine residue of protein phos-phatases [48, 49]. There is, however, no resemblance ofthe insulin-like action of VO(acac)2 facilitating tyrosinephosphorylation of IRb and IRS1 described throughthese investigations with the action of peroxovanadates.Vanadates (of oxidation state V) are known to exhibit astrong tendency to bind to sulfhydryl groups, whileVO2+ (of oxidation state IV) shows low affinity forsulfur-donor ligands. For instance, the low affinity of

Fig. 9 Restoration of IRS1,IRb, and GSK-3bphosphorylation by VO(acac)2in insulin-resistant 3T3-L1adipocytes. 3T3-L1 adipocyteswere preincubated in theabsence (basal) or presence of1 nM insulin for 15 h. The nextday, the cells were serum-starved for 3 h, prior to a 5-mintreatment with 0.25 mMVO(acac)2 in the absence orpresence of 1 or 10 nM insulin.Lysates were prepared, andanalyzed by anti-phosphotyrosine (a) or anti-pGSK-3b (b) immunoblotting.Bands corresponding to IRS1,IRb and pGSK-3b are indicatedby arrows. All results arerepresentative of two to fourindependent experiments

884

VO2+ for sulfhydryl groups is readily demonstratedthrough the study by Cornman et al. [50] showing thatVO2+ added to a synthetic polypeptide analog of theactive site of protein phosphatase 1B was bound to ahistidine and a serine residue rather than to the cysteineresidue. Although VO2+ extracted from VO(malto)2 hasbeen implicated as a reversible inhibitor of proteinphosphatases [22], the stability of VO(acac)2, as shownin these studies, is significantly greater. It is, therefore,unclear if potentiation of tyrosine phosphorylation ofthe insulin receptor facilitated by VO(acac)2 occurs onlythrough inhibition of phosphatase action, and furtherstudies will be needed to examine the relative contribu-tions of kinase activation and phosphatase inhibition tothe enhancement of tyrosine phosphorylation of theinsulin receptor.

Need for development of specific insulin-mimeticcompounds

Type II diabetes mellitus is a complex metabolic diseaseinvolving defects in both insulin secretion and insulinaction and is influenced by a broad range of genetic andenvironmental factors. Although there is still consider-able lack of consensus about the primary defect, thepathogenesis of type II non-insulin-dependent diabetesmellitus involves progressive development of insulinresistance and a defect in insulin secretion. Overt diseasemarked by hyperglycemia begins when insulin outputfrom the pancreas fails to meet the requirement forinsulin as a result of insulin resistance [41, 42]. Therequirement for insulin at the cell surface begins withbinding of insulin to the insulin receptor, initiatingautophosphorylation of the b-subunits through activa-tion of its intrinsic tyrosine kinases. Subsequent trans-phosphorylation of IRS proteins and activation ofdownstream signaling molecules continue the spectrumof biological responses associated with the physiologyof this important polypeptide hormone. While mostpatients with type II diabetes initially respond to insulinsecretogues, agents that facilitate pancreatic output ofthe hormone, a significant fraction show defects ininsulin signaling. Thus, development of orally activeagents that enhance or mimic the action of insulin couldlead to additional routes for therapy; also, such agentscould be advantageous in the treatment of patients withtype I diabetes who depend on exogenous insulin formetabolic control.

It is evident that the functioning of the receptortyrosine kinase activity is essential for the biologicaleffects of insulin, and development of orally active, smallmolecules that could facilitate the action of insulin at thelevel of the insulin receptor enhancing its autophosph-orylative and tyrosine kinase activities could be ofpharmaco-therapeutic value. In our studies we havedemonstrated that the action of VO(acac)2 was syner-gistic with insulin and that the synergism was observablegenerally only under low concentrations of insulin. Thisresult helps to explain the observations of others [5, 6]that VO2+ compounds are associated with insulin-enhancing activity only in diabetic laboratory animals(in which there is a deficit of circulating insulin hormoneor systemic insulin resistance). Our observations thathigh concentrations of insulin mask the synergismindicate that the insulin-like action of VO2+ compoundsis not restricted to only diabetic tissue. In this regardVO2+ chelates with insulin-enhancing activity can beviewed as potential reagents for reducing the level ofinsulin to maintain euglycemia. Development of thetherapeutic potential of VO2+ chelates, such as VO(acac)2, therefore, may be advantageous as insulin-enhancing agents also in type I diabetes. To this end,identification of structural and kinetic factors thatensure specificity in insulin-enhancing effects will beimportant. It is, therefore, of interest to note that theorigin of virtually all of the tyrosine phosphorylated

Fig. 10 Effect of VO(acac)2 on glycogen synthesis and GS activa-tion in insulin-resistant 3T3-L1 adipocytes. 3T3-L1 adipocytes weremade insulin-resistant as in Fig. 9, serum-starved for 3 h, and thenstimulated in the absence or presence of 10 nM insulin. Results arecompared for 0.25 mM VO(acac)2 (A) and VOSO4 (S). a After15 min, 2 lCi of 14C-glucose was added to all wells. The cells wereplaced on ice 30 min later, washed three times with cold phosphate-buffered isotonic saline and glucose incorporation into glycogen wasdetermined. b The cells were stimulated for 15 min with the agents,and then cellular lysates were prepared. GS activity was measured invitro. VO(acac)2 significantly stimulated glycogen synthesis andincreased GS activity in insulin-resistant cells, in both the absenceand the presence of insulin. The results are the average of threeindependent experiments, performed in triplicate or duplicate

885

proteins that are observed through immunoblottingexperiments, such as in Figs. 2 and 3, including control,basal conditions [51, 52], have not been identified.Assignment of their cellular origins and the dependenceof their expression on insulin binding to its receptor inthe membrane may become a sensitive probe to guidedesign of specificity of therapeutic reagents.

Acknowledgements This work was supported by grants of theNational Institutes of Health (DK57599 and DK20959). M.J.B.is a recipient of a Career Development Award of the AmericanDiabetes Association.

References

1. Goldfine AB, Simonson DC, Folli F, Patti ME, Kahn CR(1995) Mol Cell Biochem 153:217–231

2. Goldfine AB, Simonson DC, Folli F, Patti ME, Kahn CR(1995) J Clin Endocrinol Metab 80:3311–3320

3. Cohen N, Halberstam M, Shlimovich P, Chang CJ, ShamoonH, Rossetti L (1995) J Clin Invest 95:2501–2509

4. Boden G, Chen X, Ruiz J, van Rossum G, Turco S (1996)Metabolism 45:1130–1135

5. Yuen VG, Orvig C, McNeill JH (1995) Can J Physiol Phar-macol 73:55–64

6. Reul BA, Amin SS, Buchet JP, Ongemba LN, Crans DC,Brichard SM (1999) Br J Pharm 126:467–477

7. Sakurai H, Sano H, Takino T, Yasui H (2000) J Inorg Biochem80:99–105

8. Elberg G, Li J, Shechter Y (1994) J Biol Chem 269:9521–95279. Li J, Elberg G, Crans DC, Shechter Y (1996) Biochemistry

35:8314–831810. Makinen MW, Brady MJ (2002) J Biol Chem 277:12215–1222011. Sakurai H, Kojima Y, Yoshikawa Y, Kawabe K, Yasui H

(2002) Coord Chem Rev 226:187–19812. Crans DC, Smee JJ, Gaidamauskas E, Yang L (2004) Chem

Rev 104:849–90213. Thompson KH, Orvig C (2004) In: Sigel H, Sigel A (eds) Metal

ions in biological systems. vol 41. Marcel Dekker, New York,pp 221–252

14. Macara IG, Kustin K, Cantley LCJ (1980) Biochim BiophysActa 629:95–106

15. Sakurai H, Shimomura S, Ishizu K (1981) Inorg Chim Acta55:L67–L69

16. Willsky GR, White DA, McCabe BC (1984) J Biol Chem259:13273–13281

17. Dai S, Yuen VG, Orvig C, McNeill JH (1993) Pharm Commun3:311–321

18. Yuen VG, Pederson RA, Dai S, Orvig C, McNeill JH (1996)Can J Physiol Pharmacol 74:1001–1009

19. Yuen VG, Orvig C, Thompson KH, McNeill JH (1993) Can JPhysiol Pharmacol 71:270–276

20. Caravan P, Gelmini L, Glover N, Herring FG, Li H, McNeillJH, Rettig SJ, Setyawati IA, Shuter E, Sun Y, Tracey AS, YuenVG, Orvig C (1995) J Am Chem Soc 117:12759–12770

21. Hanson GR, Sun Y, Orvig C (1996) Inorg Chem 35:6507–651222. Peters KG, Davis MG, Howard BW, Pokross M, Rastogi V,

Diven C, Greis KD, Eby-Wilkens E, Maier M, Evdokimov A,Soper S, Gengauffe F (2003) J Inorg Biochem 96:321–330

23. Mustafi D, Makinen MW (1988) Inorg Chem 27:3360–336824. Makinen MW, Mustafi D (1995) In: Sigel H, Sigel A (eds)

Metal ions in biological systems, vol 31. Marcel Dekker, NewYork, 1995, pp 89–127

25. Pandey SK, Anand-Srivastava MB, Srivastava AK (1998)Biochemistry 37:7006–7014

26. Pandey SK, Theberge JF, Bernier M, Srivastava AK (1999)Biochemistry 38:14667–14675

27. Shechter Y, Shisheva A, Lazae R, Libman J, Shanzer A (1992)Biochemistry 31:2063–2068

28. Brady MJ, Kartha PM, Aysola AA, Saltiel AR (1999) J BiolChem 274:27497–27504

29. Lazar DF, Wiese RJ, Brady MJ, Mastick CC, Waters SB,Yamauchi K, Pessin JE, Cuatrecasas P, Saltiel AR (1995) J BiolChem 270:20801–20807

30. Thomson M. J., Williams M. G., Frost SC (1997) J Biol Chem272:7759–7764

31. Jiang FS, Makinen MW (1995) Inorg Chem 34:1736–174432. Mustafi D, Nakagawa Y (1994) Proc Natl Acad Sci USA

91:11323–1132733. Goldwasser I, Qian S, Gershonov E, Fridkin M, Shechter Y

(2000) Mol Pharmacol 58:738–74634. White MF (1998) Mol Cell Biochem 182:3–1135. Inoue G, Cheatham B, Emkey R, Kahn CR (1998) J Biol Chem

273:11,548–11,55536. White MF (2002) Am J Physiol Endocrinol Metab 283:E413–

E42237. Misuuchi Y, Johnson SW, Moonblatt S, Testa JR (1998) J Cell

Biochem 70:433–44138. Jope RS, Johnson GV (2004) Trends Biochem Sci 29:95–10239. Frame S, Cohen P (2001) Biochem J 359:1–1640. Brady MJ (2003) Methods Mol Med 83:155–16141. Saltiel AR (2001) Cell 104:517–52942. Lawrence JC Jr, Roach PJ (1997) Diabetes 46:541–54743. Thomson MJ, Williams MG, Frost SC (1997) J Biol Chem

272:7759–776444. Schieven GL, Wahl AF, Grosmaire L, Ledbetter JA (1995) J

Biol Chem 70:20824–2083145. Hindi S, Grossman DP, Goldwaser I, Shechter Y, Fridkin M

(2002) Lett Pept Sci 9:235–25446. Elberg G, Li J, Shechter Y (1994) J Biol Chem 269:9521–952747. Greenberg CC, Meredith KN, Yan L, Brady MJ (2003) J Biol

Chem 278:30835–3084248. Kadota S, Fantus IG, Deragon G, Guyda HJ, Hersh B, Posner

BI (1987) Biochem Biophys Res Commun 147:259–26649. Fantus IG, Kadota S, Deragon G, Foster B, Posner BI (1989)

Biochemistry 28:8864–887150. Cornman CR, Zovinka EP, Meixner M H (1995) Inorg Chem

34:5099–510051. Levenson RM, Blackshear PJ (1989) J Biol Chem 264:19884–

1999352. Sadowski HB, Wheeler TT, Young DA (1992) J Biol Chem

266:4722–4731

886