Mechanism of insulin sensitization by BMOV (bis maltolato oxo vanadium); unliganded vanadium (VO4) as the active component.

Abstract

Organovanadium compounds have been shown to be insulin sensitizers in vitro and in vivo. One potential biochemical mechanism for insulin sensitization by these compounds is that they inhibit protein tyrosine phosphatases (PTPs) that negatively regulate insulin receptor activation and signaling. In this study, bismaltolato oxovanadium (BMOV), a potent insulin sensitizer, was shown to be a reversible, competitive phosphatase inhibitor that inhibited phosphatase activity in cultured cells and enhanced insulin receptor activation in vivo. NMR and X-ray crystallographic studies of the interaction of BMOV with two different phosphatases, HCPTPA (human low molecular weight cytoplasmic protein tyrosine phosphatase) and PTP1B (protein tyrosine phosphatase 1B), demonstrated uncomplexed vanadium (VO(4)) in the active site. Taken together, these findings support phosphatase inhibition as a mechanism for insulin sensitization by BMOV and other organovanadium compounds and strongly suggest that uncomplexed vanadium is the active component of these compounds.

Full text

Page 1

Journal of Inorganic Biochemistry 96 (2003) 321–330
www.elsevier.com/locate/jinorgbio
M echanism of insulin sensitization by BMOV (bis maltolato oxo
vanadium); unliganded vanadium (VO ) as the active component
4
*
Kevin G. Peters , Mike G. Davis, Brian W. Howard, Matthew Pokross, Vinit Rastogi,
Conrad Diven, Kenneth D. Greis, Elaine Eby-Wilkens, Matthew Maier, Artem Evdokimov,
Shari Soper, Frank Genbauffe
Procter & Gamble Pharmaceuticals, Cardiovascular Research, Health Care Research Center, 8700 Mason-Montgomery Road, Mason, OH 45040,
USA
Received 28 January 2003; received in revised form 28 May 2003; accepted 29 May 2003
Abstract
Organovanadium compounds have been shown to be insulin sensitizers in vitro and in vivo. One potential biochemical mechanism for
insulin sensitization by these compounds is that they inhibit protein tyrosine phosphatases (PTPs) that negatively regulate insulin receptor
activation and signaling. In this study, bismaltolato oxovanadium (BMOV), a potent insulin sensitizer, was shown to be a reversible,
competitive phosphatase inhibitor that inhibited phosphatase activity in cultured cells and enhanced insulin receptor activation in vivo.
NMR and X-ray crystallographic studies of the interaction of BMOV with two different phosphatases, HCPTPA (human low molecular
weight cytoplasmic protein tyrosine phosphatase) and PTP1B (protein tyrosine phosphatase 1B), demonstrated uncomplexed vanadium
(VO ) in the active site. Taken together, these findings support phosphatase inhibition as a mechanism for insulin sensitization by BMOV
4
and other organovanadium compounds and strongly suggest that uncomplexed vanadium is the active component of these compounds.
 2003 Elsevier Inc. All rights reserved.
Keywords: BMOV; Phosphatase; Vanadate; HCPTPA; PTP1B; Insulin; Diabetes
1 . Introduction
Type 2 diabetes [7,8]. Overexpression of PTP1B, LAR or
HCPTPA in cultured cells attenuates insulin receptor
activation and signaling [9–14]. Conversely, reducing
PTP1B activity in vivo using either antisense oligos or
gene targeting results in enhanced insulin receptor sig-
naling and improved insulin sensitivity and glucose toler-
ance [15–17]. These findings show that PTPs, PTP1B in
particular, may be good therapeutic targets for treatment of
insulin resistance states such as type 2 diabetes.
Vanadium compounds have gained attention because of
their insulin mimetic activity [18–20]. In cultured cell
lines, vanadium compounds enhance insulin receptor acti-
vation and downstream signaling [21–26]. In animal
models of diabetes, vanadium compounds improve insulin
sensitivity resulting in decreased levels of plasma glucose
and insulin [27–37]. Importantly, this enhanced insulin
sensitivity occurs in the absence of weight gain that
complicates therapy with exogenous insulin and the recent-
ly developed PPAR-g (peroxisome proliferator-activated
receptor) agonists.
One likely mechanism for the insulin mimetic activity of
vanadium compounds relates to their potent inhibition of
The insulin receptor belongs to a family of growth factor
receptors termed receptor tyrosine kinases (RTKs) [1].
Ligand binding of the insulin receptor results in activation
of the kinase domain leading to autophosphorylation on
specific tyrosine residues [2–6]. Autophosphorylation in
turn further activates the kinase domain and provides
binding sites for the recruitment and subsequent phos-
phorylation of signaling molecules such as IRS-1 (insulin
receptor substrate 1) which drive the cellular responses
important for insulin action. Thus, the regulation of
tyrosine phosphorylation of the insulin receptor likely
plays an important role in insulin action.
Recent findings demonstrate that protein tyrosine phos-
phatases (PTPs) are negative regulators of insulin receptor
signaling. For example, PTP1B and LAR (leukocyte
antigen related-protein) are upregulated in patients with
*Corresponding author. Tel.: 11-513-622-0834; fax: 11-513-622-
1433.
E-mail address: peters.kg@pg.com (K.G. Peters).
0162-0134/03/$ – see front matter
doi:10.1016/S0162-0134(03)00236-8
 2003 Elsevier Inc. All rights reserved.

Page 2

322
K.G. Peters et al. / Journal of Inorganic Biochemistry 96 (2003) 321–330
PTPs [23,38,20]. A variety of vanadium compounds have
been shown to directly inhibit a wide range of PTPs
including PTP1B [39,40]. Recently, organovanadium com-
pounds have been shown to have superior insulin mimetic
activities compared to sodium orthovanadate [41,23,42,43].
Currently the reasons for the superior activity of or-
ganovanadium compounds are not clear but may relate to
better bioavailability of these compounds or more potent
activity at the enzyme active site.
To gain further insight into the insulin mimetic actions
of organovanadium compounds, we have explored the
mechanism by which a novel organovanadium compound,
bismaltolato oxovanadium (BMOV), inhibits phosphatase
activity. Like other organovanadium compounds BMOV
was a reversible, competitive phosphatase inhibitor. Im-
portantly, BMOV inhibited phosphatase activity in cul-
tured cells and enhanced the autophosphorylation of the
insulin receptor in vivo. NMR and X-ray crystallographic
approaches demonstrated that the active component of
BMOV is most likely to be uncomplexed vanadium
(VO ). These studies support the hypothesis that or-
4
ganovanadium compounds exert their insulin mimetic
activities, at least in part, by phosphatase inhibition. In
addition, these studies suggest that the reason for improved
efficacy of organovanadium compounds vs. inorganic
vanadium most likely relates to bioavailability rather than
increased potency at the phosphatase enzyme active site.
PCR
(Stratagene) and sequenced before being further subcloned
for bacterial (HCPTPA; see next paragraph) or mammalian
(PTP1B; see Section 2.3) expression.
The HCPTPA cDNA was cloned into a pET-28a vector
(Novagen) and expressed in E. coli strain BL-21(DE3)
(Stratagene) at 37 8C. Induced cells were lysed by sonica-
tion in 20 mM tris–HCl (pH 7.5) on ice. Soluble protein
was loaded onto a Q Sepharose FF column (Pharmacia),
washed with lysis buffer and eluted with a linear 0–1 M
NaCl gradient. Peak fractions were pooled, adjusted to 1.5
M (NH ) SO and loaded onto a phenyl sepharose 6 FF
4 24
column (Pharmacia). The column was washed with loading
buffer, then eluted with a 1.5–0 M (NH ) SO gradient in
20 mM tris–HCl (pH 7.5). Peak fractions were pooled,
concentrated and further purified on a Superdex 75 column
(Pharmacia) in 25 mM Hepes (pH 8.0) 150 mM NaCl.
productsweresubclonedinto pPCR-Script
4 24
2 .3. Phosphatase assays
Kinetic assays were done using the recombinant phos-
phatases and fluorogenic small molecule substrate 6,8-
difluoro-4-methylumbelliferyl phosphate (DiFMUP, Mo-
lecular Probes) (10 mM) was incubated for 15 min with
nM concentrations of phosphatase in buffer containing 10
mM Na Acetate, 150 mM NaCl, 5 mM DTT, pH 6.
Recombinant HCPTPA was produced as described above
(Section 2.2) and PTP1B was purchased from Biomol
(SE-332). The resulting phosphatase product was mea-
sured at 355/460 nm (ex/em) using aVictorVplate reader
(Wallac). Inhibitors (0.002–40 mM) were pre-incubated
with phosphatase for 10 min prior to addition of DiFMUP
substrate. IC50 curves were generated using Excel-Fit.
Kinetic analysis was performed using 3 concentrations of
inhibitor to calculate velocity over a range of DiFMUP
concentrations (0–400 mM) and Lineweaver–Burke plots
were used to evaluate inhibitor mechanism.
To measure phosphatase activity and inhibition in
cultured cells a cDNA clone encoding PTP1B (see Section
2.2 above) was subcloned into the expression vector
pcDNA3.1 (Invitrogen). HEK 293 h cells (Gibco) were
transfected withthe
lipofectamine2000 (Gibco). Expression was detected by
western at 48 h using an anti-PTP1B antibody (PTP1B-
Ab1; Calbiochem). Phosphatase activity was detected by
treating the cells with 50 mM DiFMUP for 30 min and
measuring fluorescence at 355/460 nm (ex/em) using a
Victor V plate reader. BMOV was pre-incubated with cells
for 30 min before substrate addition.
2 . Materials and methods
2 .1. Materials
Bismaltolato oxovanadium was synthesized according to
published procedures [44]. Briefly, vanadyl sulfate (2.00 g,
12.3 mmol) was dissolved in 100 ml H O and 3-hydroxy-
2-methyl-4-pyrone (maltol) (2.48 g, 19.7 mmol) was added
at once. Using a pH meter, 1 N NaOH was added drop-
wise with stirring into the solution until pH 8.50. The
mixture was refluxed overnight and the product was
crystallized upon cooling to room temperature. After
filtering, the product was vacuum dried and stored in a
desiccator.
2
PTP1Bplasmidusing
2 .2. Molecular cloning of HCPTPA and PTP1B and
production of recombinant HCPTPA
cDNA clones encoding HCPTPA and PTP1B were
generated by PCR from human placenta cDNA purchased
from Clontech. Primers used for HCPTPA were: forward
withXhoI site59-CCGCTCGAGGAAGATGGCGG-
]]]
AACAG-39, reverse with NotI site 59-ATAAGAAT-
GCGGCCGCCTGGAACGTGATTACACACCG-39,
]]]]
for PTP1B: forward with EcoRI site 59-GGAATTC-
ATGGAGATGGAAAAGGAG-39, reverse with NotI site
59-TGCGGCCGCCTATGTGTTGCTGTTG-39.Subsequent
2 .4. Mass spectrometric analysis
and2 .4.1. Perfusion HPLC-coupled mass spectrometry
Screening of intact protein masses after reaction with
various vanadium inhibitors was done by perfusion chro-
matography (POROS II R/H, 300 mm i.d.) coupled to a
]]]

Page 3

K.G. Peters et al. / Journal of Inorganic Biochemistry 96 (2003) 321–330
323
Sciex API 165 single quadrupole Mass Spectrometer in
positive, electrospray ionization mode. After a 1 min hold
at 1% CH CN/0.02%TFA the protein was eluted with a
3
linear gradient to 85% CH CN over 2 min.
3
19, Santa Cruz Biotech). Resulting films were scanned and
quantitated using Quantity-One software (Bio-Rad).
2 .6. NMR studies
15
2 .4.2. Capillary HPLC-coupled, electrospray ionization
tandem mass spectrometry (CAPLC-ESI-MS/MS)
Trypsin digested samples were separated on a Pepmap
C18, 3 mm, 300 mm i.d.350 mm column (LCPackings)
using an LCPackings Ultimate capillary LC system with a
gradient of 2% B to 50% B in 30 min at 4 ml/min where
A50.1% formic acid/2%CH CN; B50.1% formic acid/
98%CH CN. The effluent from the LC was coupled
3
directly to a custom-built micro-ESI interface on a Fin-
Deca
nigan LCQion-trap mass spectrometer. Positive ion
spectra were collected in data dependent mode such that an
MS/MS fragmentation spectrum was obtained for each
peak detected above a threshold 1310 .
Uniformly
Escherichia coli BL21(DE3) strain grown in a minimal
15
media containingN-NH Cl (1 g/l) (Isotec Inc., Miamis-
4
burg, OH) as the sole nitrogen source.
Expression and purification details were the same as
previously published [45]. NMR samples contained |167
mM HCPTPA in 50 mM acetate buffer (pH 5.1). NMR
experiments were recorded at 298 K using a Varian Inova
600 NMR spectrometer. A ligand/protein ratio of 5 was
obtained by diluting 6 mM BMOV and Na VO stock in
50 mM sodium acetate (pH 5.0) containing 10% D O.
151
2D- N/ H HSQC was recorded on the protein in the
absence and presence of ligand. Significant chemical shift
151
changes in theN H-HSQC spectra of the protein on
addition of the ligand indicated the binding of the ligands
to the protein. The sequential resonance assignments in
HCPTPA in the presence of phosphate were achieved by
making use of a series of double and triple resonance
NMR experiments [46].
N-labeled HCPTPA was overexpressed in
3
34
2
5
2 .5. In vivo insulin receptor activation
To assess the effect of BMOV on insulin receptor
activation in-vivo, fasted rats (250–300 g) were infused
with either saline or BMOV for 5 min followed immedi-
ately by a 10-min infusion with either saline or insulin via
a carotid artery catheter. Animals were euthanized and the
heart was removed, flash frozen in liquid nitrogen and
stored at 280 8C until assayed. For analysis of insulin
receptor activation, 250 mg of frozen tissue was homogen-
ized in RIPA buffer: 50 mM Tris (pH 7.5), 150 mM NaCl,
1 mM EDTA, 1% NP-40, 0.25% SDS, 1 mM Na VO , 1
mM NaF, 10 nM Okadaic acid plus 1 complete protease
inhibitor tablet (Roche). Homogenates were centrifuged at
|21,0003g speed for 30 min at 4 8C. Supernatants were
recovered and protein concentrations were determined by
the BCA assay (Pierce). One milligram (1 mg) of extracted
protein was pre-cleared with 25 ml of protein A/G-Plus
agarose beads (Santa Cruz Biotech.) for 1 h at 4 8C. Insulin
receptor beta was immunoprecipitated from the pre-cleared
lysate using 10 mg of an anti-insulin receptor beta antibody
(C-19, Santa Cruz Biotech) at 4 8C overnight. The complex
was precipitated using 25 ml of protein A/G-Plus agarose
beads (Santa Cruz Biotech.) for 1 h at 4 8C. Afterwards the
beads were sedimented at |21,0003g for 1 min, washed
once in cold lysis buffer and bound proteins eluted by
boiling for 5 min in 30 ml of 13 sample buffer (50 mM
Tris–HCl (pH 6.8), 10% glycerol, 2% SDS, 0.1 mM DTT,
0.1% bromphenol blue). The samples were centrifuged for
1 min at maximum speed and 20 ml of the supernatant was
loaded onto an 8% SDS–PAGE gel, transferred to PVDF
membranes and phosphotyrosine western blotting was
performed using anti-phosphotyrosine antibody (PY99,
Santa Cruz Biotech.) diluted 1:1000 in 2.5% bovine serum
albumin in TBS–0.1% Tween-20. Signal was detected
using ECL (Amersham). After exposure the blots were
stripped and re-probed with anti-insulin receptor beta (C-
2 .7. X-ray crystallographic studies
Crystals of PTP1B C215S trap mutant (protein provided
by Dr. Zhong-Yin Zhang Albert Einstein College of
Medicine) were grown in 18–20% PEG 8000, Tris (pH
8.0), 1% BME at 4 8C using hanging drop vapor diffusion.
Crystals appeared in |1–2 weeks and were then used for
soaking experiments. For soaking, BMOVwas dissolved in
water to a stock concentration of |100 mM and added
directly to the crystal drop to a final concentration of |1
mM. Crystals were soaked at room temperature in room air
for 2 h prior to data collection. Data were collected at
beamline 17-ID (or 17-BM) in the facilities of the In-
dustrial Macromolecular Crystallography Association Col-
laborative Access Team (IMCA-CAT) at the Advanced
Photon Source. These facilities are supported by the
companies of the Industrial Macromolecular Crystal-
lography Association through a contract with Illinois
Institute of Technology (IIT), executed through IIT’s
Center for Synchrotron Radiation Research and Instru-
mentation. Crystals were placed in 20% glycerol plus well
solution and immediately frozen at 100 K. The structure
was solved using Molrep and refined using Refmac, from
˚
the CCP4 programs, to 2.2 A resolution [47].
34
3 . Results
3 .1. BMOV is a competitive, reversible PTP inhibitor
Enzyme/substrate competition assays demonstrated that

Page 4

324
K.G. Peters et al. / Journal of Inorganic Biochemistry 96 (2003) 321–330

Fig. 1. BMOV is a nonselective protein tyrosine phosphatase inhibitor. (A) Molecular structure of BMOV (bismaltolatol oxovanadium). (B) Activity of
BMOV at four different structurally diverse recombinant tyrosine phosphatases. Inset shows IC50 values.
BMOV, like other organovanadium compounds, was a
potent, nonselective PTP inhibitor (Fig. 1). To gain further
insight into the mechanism of action of BMOV, enzyme
kinetic analysis was done against two structurally diverse
PTPs, HCPTPA and PTP1B (Fig. 2). In these studies,
BMOV demonstratedclassical
against the fluorogenic substrate DiFMUP (6,8-difluoro-4-
methylumbelliferyl phosphate) for both HCPTPA and
PTP1B (KI50.79 mM and 0.90 mM, respectively). Thus,
like sodium orthovanadate and other organovanadium
compounds, BMOV is a nonselective, competitive PTP
inhibitor.
Other organovanadium compounds such as bpV-phen
(bis peroxo vanadium phenanthroline) oxidize the active
site cysteine of PTP ‘P-loop’ and can function as irrevers-
ible PTP inhibitors [48]. To determine the propensity of
BMOV to oxidize the active site cysteine, HCPTPA was
incubated with either BMOV or bpV-phen and mass
spectrometric analysis was done to determine alterations at
the enzyme active site (Fig. 3). As anticipated, incubation
of HCPTPA with bpV-phen at a 1:10 molar ratio caused a
mass shift of 48 Daltons consistent with the oxidation of
the P-loop cysteine to cysteic acid (SO ) and this modi-
fication was confirmed by LC-ES-MS–MS (data not
shown). At a 1:1000 molar ratio, incubation with bpV-phen
resulted in generation of multiple higher molecular weight
species suggesting the nonselective oxidation of multiple
residues in addition to the P-loop cysteine. Conversely,
incubation of HCPTPA with BMOV even at a molar ratio
of 1:1000 failed to result in a mass shift, suggesting that
BMOV, unlike bpV-phen does not irreversibly modify the
enzyme.
phosphatase activity has been difficult to determine. In an
attempt to measure inhibition of intracellular phosphatase
activity, HEK293 cells were transfected with a vector
directing the overexpression of PTP1B. The transfected
cells overexpressed PTP1B as shown by western blot and
by increased phosphatase activity as measured by the
fluorescence of the cell permeable fluorogenic substrate
DiFMUP (Fig. 4). Despite overexpression of PTP1B,
.50% of the total cellular PTP activity was inhibited by
10 mm BMOV, consistent with its ability to readily cross
the cell membrane and its activity against the isolated,
recombinant enzyme.
Having shown that BMOV could inhibit intracellular
phosphatase activity, its ability to enhance insulin receptor
activation in vivo was tested. Briefly, insulin was adminis-
tered to fasted rats with or without BMOV pretreatment.
Insulin receptors were then isolated from heart tissue by
immunoprecipitation and assayed for activation by an-
tiphosphotyrosine immunoblot. In the absence of exogen-
ous insulin, little if any increase in insulin receptor
activation could be detected following BMOV treatment
(Fig. 5). However, in the presence of insulin, receptor
activation was enhanced after pretreatment with BMOV
compared to animals treated with insulin only.
competitive inhibition
3
3 .3. Uncomplexed vanadium is the active component of
BMOV
In order to understand the mechanism of BMOV action
at the molecular level, NMR studies were done with
15
N-labeled HCPTPA. Analysis of the chemical shift
changes in the NMR spectrum of HCPTPA with either
BMOVor Na VO revealed that the changes corresponded
34
to the residues forming the active site region of the protein,
especiallythe consensus
12CLGNICRS–). Surprisingly, BMOVand Na VO caused
essentially identical chemical shift changes of amide
resonances in HCPTPA (Fig. 6). This suggests that the
3 .2. BMOV inhibits intracellular phosphatase activity
and enhances insulin receptor activation in vivo
P-loopsequence(–
34
Although organovanadium compounds inhibit isolated,
recombinant PTPs, whether or not they inhibit intracellular

Page 5

K.G. Peters et al. / Journal of Inorganic Biochemistry 96 (2003) 321–330
325

Fig. 2. BMOV is a competitive inhibitor of HCPTPA and PTP1B. Michaelis–Menten curves for HCPTPA (A) and PTP1B (C) in the presence of increasing
concentrations of BMOV (see inset). Lineweaver–Burke analysis of data in panels A and C showing plots consistent with competitive inhibition of
HCPTPA (B) and PTP1B (D).
protein experiences a similar chemical environment in the
presence of the bound ligands and that uncomplexed
vanadium (VO ) from BMOVmust be the binding moiety.
4
Confirming this result, when BMOV was soaked into
PTP1B crystals, only VO could be fitted into the differ-
4
ence electron density and no other difference electron
density was seen near the active site to indicate the intact
BMOVmolecule (Fig. 7A). The difference electron density
unambiguously indicates that the geometry of VO in the
active site was that of a trigonal bipyramid with the base
formed by three oxygen atoms and the apices formed by an
oxygen and the hydroxyl group (Og) of serine 215.
Interatomic distances between the vanadium atom and the
oxygens appear to be unequal—the two axial distances are
longer than the three equatorial ones. The vanadate ion was
stabilized in the active site by a complex network of
hydrogen bonds (Fig. 7B). The geometry and hydrogen
binding network of the vanadate ion in the active site are
consistent with previously published structures of vanadate
with wild type PTP1b, Yersinia PTP and chloroperoxidase
[49–51]. Resolution of the structure did not warrant
unrestrained refinement, therefore the distances and angles
of the oxovanadate ion were restrained to the values found
in small molecules.
4
4 . Discussion
In this report, we have explored the mechanism of
phosphatase inhibition and insulin sensitization by BMOV,
a unique organovanadium insulin mimetic. In previous
studies, BMOV has been shown to be an effective antidia-
betic agent in animal models of Type 2 diabetes, but little
was known about its precise mechanism of action [33–