Radical-induced oxidation of metformin
H. Khouri
1
, F. Collin
1
, D. Bonnefont-Rousselot
2,3
, A. Legrand
2
, D. Jore
1
and M. Garde
`
s-Albert
1
1
Laboratoire de Chimie Physique UMR 8601-CNRS, Universite
´
Paris 5, France;
2
Laboratoire de Biochimie Me
´
tabolique et Clinique,
Faculte
´
de Pharmacie, Paris 5, France;
3
Laboratoire de Biochimie B, Ho
ˆ
pital de la Salpe
ˆ

determined for the second order r ate co nstant
k(
Æ
OH + metformin). Superoxide free radicals and hydro-
gen peroxide do not initiate any oxidation on metformin in
our in vitro experiments.
Keywords: metformin; hydroxyl radical; antioxidant; radio-
lysis.
Metformin (MTF) (1 ,1-dimethylbiguanide, see structure in
Fig. 1) is one of the most used oral antihyperglycaemic
agents. It normalizes plasma glucose concentration without
any stimulation of insulin production. It has been demon-
strated that elevated glucose levels induce oxidative stress
in diabetes, i.e. an imbalance between the production of
oxidant species, particularly radical species, and the anti-
oxidant defences [1]. This might partly explain the elevated
risk factors for diabetic patients to develop cardiovascular
complications [2,3]. This imbalance c an be detected by
oxidative stress markers such as those of lipid peroxidation
and protein oxidation.
Previous in vivo and in vitro studies have demonstrated
several antioxidant proper ties of m etfo rmin such as the
inhibition of the formation of advanced glycation end
products (AGEs) [4,5] that a re thought to be responsible
for further diabetic complications, and the decrease in
the formation of methylglyoxal, one of the precursors of
AGEs [6].
Metformin improves liver antioxidant potential in rats
fed a high-fructose d iet [ 7]. It has been observed that the
administration of metformin in diabetic p atients ameliorates

radicals thus generated have been used to initiate one
electron oxidation reaction(s) on metformin dissolved in
water. In a previous work, we have identified the oxida-
tion end-products of
Æ
OH-induced oxidation of m etformin
[19]. Four products have been characterized (Fig. 1):
methylbiguanide (MBG), a dimer o f M TF (diMTF), a
hydroperoxide of MTF (MTFOOH) and 4-amino-2-
imino-1-methyl-1,2-dihydro-1,3,5-triazine (4,2,1-AIMT).
The generation of these oxidation end-products was
shown to be dependent on the e xperimental conditions:
MTFOOH is only produced under aerated conditions,
while diMTF occurs only in nonaerated solutions,
saturated with nitrogen protoxide. The two other products,
MBG and 4,2,1-AIMT, have been found in both aerated
Correspondence t o H. Khouri or F. Collin, L aboratoire de Chimie
Physique, CNRS U M R 8601 universite
´
s Pa ris 5, 45 rue de s Sain ts-
Pe
`
res, 75270 Paris Cedex 06, Fr ance. F a x: + 33 1 42862213,
Tel.: + 33 1 42862173, E- mail: h or

Abbreviations: 4,2,1-AIMT, 4-amino-2-imino-1-methyl-1,2-dihydro-
1,3,5-triazine; AGE, a dvanced glycation end p roducts; GSH , glut a -
thione; MBG, methylbiguanide; MTF, metformin; ROS, reactive
oxygen species; TBARS, thiobarbituric acid reactive substance.
(Received 1 7 Au gust 2004, accepted 1 4 O ctober 2004)

Æl ¼
f([metformin])}, allowing us to discuss the possible compe-
tition of hydroxyl radicals between metformin and radio-
lytically generated hydrogen peroxide.
Materials and methods
Chemicals
All chemicals were purchased from Sigma (St Louis,
MO, USA) except when mentioned. Metformin solutions
(4–500 lmolÆL
)1
)werepreparedin10mmolÆL
)1
phosphate
buffer NaH
2
PO
4
Æ2H
2
O (purchased from Prolabo, Manche-
ster, UK) at pH 7. Ultra pure water (Maxima Ultra-pure
Water, ELGA, resistivity 18.2 MW) was used to pre pare the
solutions. Irradiations were carried out in test tubes that
have been previously cleaned with hot TFD4 d etergent
(Franklab S.A., France), rinsed thoroughly with ultra pure
water,andthenheatedat400°C for 4 h to avoid any
pollution by remaining organic compounds.
Gamma radiolysis
Radiolysis corresponds to the c hemical transformations of a
solvent due to the absorption of ionizing radiations, which

¼ 2204 LÆmol
)1
Æ
cm
)1
)at25°C, and a radiolytic yield of G(Fe
3+
) ¼
1.62 lmolÆJ
)1
. Different radiation doses, ranging from 52
to 627 Gy, were delivered to 5 mL of the solution depend-
ing on the time of the exposure to the c-ray source: the
longer the time of the exposure, the higher the radiation
dose. For each experimental set, 5 mL of non-irradiated
solution was taken as a control.
Water radiolysis by c-rays generates the free radical
species e
–
aq
,
Æ
OH,
Æ
H, and the molecular species H
2
and
H
2
O

O scavenges hydrated electrons and
converts them into hydroxyl radicals: as a result,
Æ
OH is
produced with a final G-value of 0.56 lmolÆJ
)1
,thatistwice
as high as the G-value in an aerated medium [20]. To
selectively obtain superoxide anions, s odium formate (Pro-
labo) was added to the solution at a concentration of
0.1 molÆL
)1
in order to c onvert all radicals (
Æ
OH,
Æ
Hand
e
–
aq
)intoO
2
Æ–
radicals with a final G-value of 0.62 lmolÆJ
)1
[20].
Analysis
Detection of the oxidation products was achieved by
spectrophotometric measurements with an UV/visible spec-
trophotometer (Beckman DU 70). Samples were scanned

ence ¼ phosphate buffer, 10 mmolÆL
)1
, pH 7) a re presen-
ted in Fig. 2A, as a function of the radiation dose. The
non-irradiated solution shows a main absorption band at
232 nm c orresponding to the absorption of metformin [22].
As the radiation dose increased, the absorption at this
wavelength decreased (illustrating the consumption of
metformin) and two new bands were detected at 208 nm
(intensified) and 258 nm, probably due to the generation of
oxidation products. Differential absorption spectra (refer-
ence ¼ non-irradiated metformin solution) allows us to
better show the same phenomenon (Fig. 2B). The arrows in
Fig. 2. UV/visible absorption spectra of
metformin (450 lmo lÆL
)1
) as a function of the
radiation dose (52–627 Gy) in aerated medium.
(A) Absolute absorption spectra (refer-
ence, phosphate buffer, 10 mmolÆL
)1
,pH7).
(B) Differential absorption spectra (refer-
ence, non-irradiated metformin solution).
Optical path-length: l ¼ 0.2 cm, dose rate:
I ¼ 10.45 GyÆmin
)1
. The arrows in dicate th e
decrease (disappearance) and the increase
(appearance) in t he a bsorbance values as a

of the curves [DAbs
k
¼ f(dose)], corresponding to GÆDe
k
Æl
(where G is the radiolytic yield, De
k
the molar extinction
coefficient and l the optical path-length) at 232 and 258 nm,
respectively, have b een reported as a function of the initial
concentration of metformin (Fig. 4A,B). These dilution
curves give the evolution of GÆDe
k
(corrected for optical
path-length l ¼ 1 cm) with the initial concentration of
metformin. Both dilution curves at 232 nm and 258 nm
exhibit the same profile, namely increasing values of GÆDe
k
at low metformin concentration (from 4 to 200 lmolÆL
)1
)
followed by plateau values of GÆDe
k
at high metformin
concentration (200–500 lmolÆL
)1
). Hence , these dilution
curves exhibit two key areas. At the plateau, the value G.De
k
at 232 nm or 258 nm reaches a steady state, meaning that

k
as a function of the initial
concentration of metformin), [phosphate buffer] = 10 mmolÆL
)1
,pH7,
aerated medium. (A) 232 nm, (B) 258 nm – values are corrected for an
optical path-length of 1 cm. Uncertainties (RSD) have been calculated
as being equal to 4%, at the 95% confidence level (2 r, n ¼ 3).
4748 H. Khouri et al. (Eur. J. Biochem. 271) Ó FEBS 2004
metformin and either phosphate buffer or hydrogen
peroxide radiolytically generated, towards the action of
Æ
OH/O
ÆÀ
2
radicals.
To verify this assumption, the effect of various phosphate
buffer concentrations (0.05, 0.5 and 5 mmolÆL
)1
)atpH7
was studied in the presence of 50 lmolÆL
)1
of metformin.
After irradiation, solutions were analyzed by absorption
spectrophotometry at 232 nm. Any change was observed i n
the consumption of metformin, indicating that phosphate
buffer did not compete at all with metformin towards
Æ
OH/O
ÆÀ

O
2
from
0.05 to 5 mmolÆL
)1
regardless of the radiation dose. At
5 mmolÆL
)1
H
2
O
2
, i t can be seen in Fig. 5 that m etformin
was n ot consumed as the radiation dose increased, i.e.
metformin no longer reacted with the radiolytically gener-
ated free radicals.
Action of O
ÆÀ
2
radicals
In order to study the effect of superoxide radicals as
initiators of metformin oxidation, metformin solutions at
different concentrations ranging from 50 to 100 lmolÆL
)1
were irradiated in the presence o f sodium formate
(0.1 mol ÆL
)1
). Under these conditions (0.1 molÆL
)1
of

of metformin oxidation in nonaerated medium, different
solutions of metformin (4–500 lmolÆL
)1
)weresaturated
with nitrogen protoxide (N
2
O). Under these conditions,
Æ
OH radicals are the main radical s pecies produced from
water r adiolysis with a radiolytic yield of 5.6 · 10
)7
molÆJ
)1
(see Materials and methods). The apparition of m etformin
oxidation product(s) was followed by absorption spectro-
photometry at 258 nm.
In Fig. 6, for a metformin concentration of 500 lmolÆ
L
)1
, differential absorbances at 232 n m (Fig. 6A) and
258 nm ( Fig. 6B) have b een reported as a function of the
radiation dose (from 52 to 627 Gy). The formation of
oxidized product(s) exhibit the same profile as under a erated
conditions (Fig. 3 A,B), confirming that
Æ
OH radicals are
responsible for the initiation of metformin oxidation.
Several metformin concentrations were studied under the
same experimental conditions (nonaerated and N
2

,
Fig. 4A and 3.2 ± 0.2 · 10
)4
, Fig. 4B). These observa-
tions can be explained by the fact that
Æ
OH radicals have a
formation yield under N
2
O atmosphere (0.56 lmolÆJ
)1
)
twice as high as those of
Æ
OH radicals formed under aerated
medium (0.28 lmolÆJ
)1
). However, the exact G-values of
metformin oxidation products formation are not actually
known.
Fig. 5. Differential absorbance at 232 nm as a
function of the radiation dose for me tformin
solutions (50 lmolÆL
)1
) with or w ithout H
2
O
2
(0.05, 0.5 and 5 mmolÆL
)1

)
2
and NH
2
.
Hydroxyl radicals can also add to the C¼NH double bonds
(giving nitrogen-centred free radicals). It can be noted that,
because of the conjugation of the nitrogen electron pair [of
NH
2
,NHandN(CH
3
)
2
]withtheC¼NH double bonds, the
charge transfe r process of
Æ
OH abstracting an electron from
the nitrogen electron pair seems rather unfavourable.
Scheme 1 summarizes the radical-induced oxidation of
metformin. MTF
Æ
symbolizes the
Æ
OH-induced radical of
metformin. Once metformin radicals are produced, they
might undergo various reactions leading to different oxida-
tion products [19]. In the presence of oxygen, metformin
radical may react with oxygen molecules leading to peroxy
radicals which could be reduced (maybe by superoxide

2
O
2
Þ¼kð
Æ
OH þ H
2
O
2
Þ½
Æ
OH½H
2
O
2

0
ð1Þ
vð
Æ
OH þ MTFÞ¼kð
Æ
OH þ MTFÞ½
Æ
OH½MTF
0
ð2Þ
It is well known that the rate constant of
Æ
OH radicals with

)1
,pH7,
N
2
O-saturated solutions. (A) 232 nm, (B) 258 nm – values are correc-
ted for an optical path-length of 1 cm. Uncertainties (RSD) have been
calculated as being equal to 17% (A) and 8% (B), at the 95% con-
fidence le vel ( 2 r, n ¼ 3).
4750 H. Khouri et al. (Eur. J. Biochem. 271) Ó FEBS 2004
hydrogen peroxide concentration (5 mmolÆL
)1
), a quasi-
total inhibition of metformin (50 lmolÆL
)1
) oxidation
(Fig. 5) has been observed, involving a reaction rate of it
Æ
OH radicals with H
2
O
2
at least 10 times higher than those
of
Æ
OH radicals with metformin [relation 3].
vð
Æ
OH þ H
2
O

LÆmol
)1
Æs
)1
[24]. Accordingly, metformin exhibits a relat-
ively weak radical scavenging capacity against
Æ
OH radicals
in vitro.
In the radiolysis solutions, H
2
O
2
could come from
different pathways: (i) from
Æ
OH radical recombination (in
the spurs) giving H
2
O
2
with a G-value of 0.7 · 10
)7
molÆJ
)1
(this production being independent of the presence of
metformin); (ii) from O
ÆÀ
2
(in equilibrium with HO

2
with a G-value of 3.4 · 10
)7
molÆJ
)1
.
H
2
O
2
concentration in the radiolysis solution is propor-
tional to G(H
2
O
2
) and to the radiation dose ([H
2
O
2
] ¼
G(H
2
O
2
) · dose). For example, at 50 Gy (which is a dose
where G-value can be determined), t he following H
2
O
2
concentration can be calculated: 3 .5 lmolÆL

OH + metformin)] be of the same order of magnitude
[i.e. % 10
7
LÆmol
)1
Æs
)1
]. In agreement with these consider-
ations, it can be proposed that the decrease of GÆDe
k
values
at low m etformin concentration ( 4–200 lmolÆL
)1
)(Figs4
and 7) would come from the competition of
Æ
OH radicals
between metformin and radiolytically generated hydrogen
peroxide.
Conclusion
We have investigated the antioxidant properties of metfor-
min against
Æ
OH and O
ÆÀ
2
-free radicals produced by water
gamma radiolysis. Metformin aqueous solutions (from 4 to
500 lmolÆL
)1

Our results obtained with an in vitro model allow
assuming that metformin, at a molecular level, is not a very
good scavenger of reactive oxygen species. Consequently, it
seems that metformin would certainly exert its in vivo
antioxidant activity by different pathways other than the
simple free radical scavenging action, such as increasing
the antioxidant enzyme activities [8,11,25], decreasing the
markers of lipid peroxidation [10,11] and inhibiting the
formation of AGEs [4,5].
Acknowledgements
Authors s how gratitude towards Dr N. Wiernsperger (LIPHA S.A.,
Lyon, France) for h is support to this work. As well our thanks to
Dr Averbeck of the Institut Curie – P aris for c irradiation f acil ities.
References
1. Betteridge, D.J. (2002) What is oxidative stress? Metabolism 49 ,
3–8.
2. Chu, N.V., Kong, A.P.S., Kim, D.P., Armstrong, D., Baxi, S.,
Deutch, R., Caulfield, M., Mudaliar,S.R.,Reitz,R.,Henry,R.R.
& Reaven, P.D. (2002) Differential effects of metformin and tro-
glitazone on cardiovascular risk factors in patients with type 2
diabetes. Diabetes Care 25, 542–549.
3. Rosen,P.,Nawroth,P.P.,King,G.,Moller,W.,Tritchler,H.J.&
Packer, L. (2001) The role of oxidative stress in the onset and
progression of diabetes and its complications: a summary of a
Scheme 1. Radical-induced oxidation scheme
of metformin.
Ó FEBS 2004 Radical-induced oxidation of metformin (Eur. J. Biochem. 271) 4751
congress series sponsored by UNESCO-MCBN, the American
Diabetes Association and the German Diabetes Society. Diabetes
Metab. Res. Rev. 17, 189–212.

¨
p, T. (1999) Effect of
gliclazide versus metformin on the clinical profile and lipid per-
oxidation markers in type 2 diabetes. Metabolism 48, 897–903.
11. Pavlovic, D., Kocic, R., Kocic, G., Jevtovic, T., Radenkovic, S.,
Mikic, D., Stojanovic, M. & Djordjevic, P.B. (2000) Effect of four-
week metformin treatment on plasma and erythrocyte anti-
oxidative d efense enzymes in newly diagnosed obese patients with
type 2 diabetes. Diabetes Obes. Metab. 2, 251–256.
12. Wiernsperger, N. (2000) Metformin: intrinsic vasculoprotective
properties. Diab etes Technol. Ther. 2, 259–272.
13. UK Prospective Diabetes Study (UKPDS) Group. (1998) Effect of
intensive blood-glucose control with metformin on complications
in over weight patients with type 2 diab etes ( UKPDS 34). Lancet
352, 854–865.
14. Derosa, G., Mugellini, A., Ciccarelli,L.,Crescenzi,G.&Fogari,
R. (2003) Comparison of glycaemic control and cardiovascular
risk profile in patients wit h type 2 diabetes during treatment wit h
either repaglinide or metformin. Diabetes Res. Clin. Pract. 60 ,
161–169.
15. Garde
`
s-Albert, M. & Jore, D. (1998) La radiolyse: une me
´
thode
efficace d’approche des me
´
canismes radicalaires antioxydants.
J. Chim. Phys. 95, 763–766.
16. Bonnefont-Rousselot, D. (1999) Oxydation des lipoprote

21. Fricke, H. & Morse, S. (1927) The chemical action of Roentgen
rays on dilute ferrosulfonate solutions as a measure of dose. Am. J.
Roentgenol. Radium. Ther. 18, 430–432.
22. Beckmann, R. (1968) The fate o f biguanides in man. Ann. NY.
Acad. Sci. 148, 820–832.
23. Bielski, B.H.J., C abelli, D.E., Arudi, R.L. & Ross, A.B. (1985)
Reactivity of HO
2
/O
À
2
radicals in aqueou s solution. J. Phys.
Chem. Ref. Data 14, 1041–1100.
24. Buxton, G.V., Greenstock, C.L., Helman, W.P. & Ross, A.B.
(1988) Critical review of rate constants for reactions of hydrated
electrons, hydrogen atoms and hydroxyl radicals in aqueous
solution. J. Phys. Chem. Ref. Data 17, 513–887.
25. Ewiss, S.A. & Abdel-Rahman, M.S. (1995) Effect of metformin on
glutathione and m agnesium in n ormal an d stre tozotocin- indu ced
diabetic rats. J. Appl. T oxicol. 15, 387–390.
4752 H. Khouri et al. (Eur. J. Biochem. 271) Ó FEBS 2004