Anti-thyroid drugs and thyroid hormone synthesis: Effect of methimazole derivatives on peroxidase-catalyzed reactions

Article abstract of DOI:10.1021/ja054497u

Syntheses and characterization of the selenium analogue (MSeI) of anti-thyroid drug methimazole and a series of organoselenium compounds bearing N-methylimidazole pharmacophore are described. In contrast to the sulfur compound that exists predominantly in its thione form, the selenium analogue exists in a selenol form, which spontaneously oxidizes in air to produce the corresponding diselenide. The reduction of the diselenide by GSH or NaBH ₄ affords the biologically active selenol, which effectively inhibits the lactoperoxidase (LPO) activity in vitro. The monoselenides having N-methylimidazole moiety are found to be much less active than the selenol, suggesting that the presence of a selenol moiety is important for the LPO inhibition. The kinetic and mechanistic studies reveal that MSeI inhibits the LPO activity by reducing the H₂O₂, providing a novel method to reversibly inhibit the enzyme. Although MSeI strongly inhibits LPO, the enzyme's activity can be completely recovered by increasing the H ₂O₂ concentration. On the other hand, the inhibition by methimazole (MMI), the sulfur analogue, cannot be reversed by increasing the H₂O₂ concentration, leading to a complete inactivation of the enzyme. The reversible inhibition of LPO by some of the selenium derivatives is correlated with their glutathione peroxidase (GPx) activity, and the high GPx activity of the selenium compounds as compared with their sulfur analogues suggests that the selenium derivatives may protect the thyroid gland from oxidative damage.

Full text of DOI:10.1021/ja054497u

Published on Web 10/08/2005
Anti-Thyroid Drugs and Thyroid Hormone Synthesis: Effect
of Methimazole Derivatives on Peroxidase-Catalyzed
Reactions
Gouriprasanna Roy and G. Mugesh*
Contribution from the Department of Inorganic & Physical Chemistry, Indian Institute of
Science, Bangalore 560 012, India
Received July 7, 2005; E-mail: mugesh@ipc.iisc.ernet.in
Abstract: Syntheses and characterization of the selenium analogue (MSeI) of anti-thyroid drug methimazole
and a series of organoselenium compounds bearing N-methylimidazole pharmacophore are described. In
contrast to the sulfur compound that exists predominantly in its thione form, the selenium analogue exists
in a selenol form, which spontaneously oxidizes in air to produce the corresponding diselenide. The reduction
4
of the diselenide by GSH or NaBH affords the biologically active selenol, which effectively inhibits the
lactoperoxidase (LPO) activity in vitro. The monoselenides having N-methylimidazole moiety are found to
be much less active than the selenol, suggesting that the presence of a selenol moiety is important for the
LPO inhibition. The kinetic and mechanistic studies reveal that MSeI inhibits the LPO activity by reducing
the H
the enzyme’s activity can be completely recovered by increasing the H
hand, the inhibition by methimazole (MMI), the sulfur analogue, cannot be reversed by increasing the H
2
O
2
, providing a novel method to reversibly inhibit the enzyme. Although MSeI strongly inhibits LPO,
2 2
O
concentration. On the other
2 2
O
concentration, leading to a complete inactivation of the enzyme. The reversible inhibition of LPO by some
of the selenium derivatives is correlated with their glutathione peroxidase (GPx) activity, and the high GPx
activity of the selenium compounds as compared with their sulfur analogues suggests that the selenium
derivatives may protect the thyroid gland from oxidative damage.
Introduction
Thyroxine (T4), the main secretory hormone of the thyroid
gland, is produced from thyroglobulin by thyroid peroxidase
TPO)/hydrogen peroxide/iodide system. The synthesis of T4
Scheme 1. Proposed Mechanism for the Deiodination of
Thyroxine by ID-I and Inhbition of ID-I by n-Propyl-2-thiouracil
(PTU) and Gold Thioglucose (GTG)
(
by TPO involves two independent steps: iodination of tyrosine
1
and phenolic coupling of the resulting iodotyrosine residues.
The prohormone T4 is then converted to its biologically active
form T3 by a selenium-containing iodothyronine deiodinase
(ID-I), which is present in highest amounts in liver, kidney,
thyroid and pituitary. The 5′-deiodination catalyzed by ID-I is
a ping-pong, bisubstrate reaction in which the selenol (or
-
selenolate) group of the enzyme (E-SeH or E-Se ) first reacts
with thyroxine (T4) to form a selenenyl iodide (E-SeI)
intermediate. Subsequent reaction of the selenenyl iodide with
an as yet unidentified intracellular cofactor (1,4-dithiothreitol
(
DTT, Cleland’s reagent) in vitro) completes the catalytic cycle
2
and regenerates the selenol (Scheme 1).
Although the deiodination reactions are essential for the
function of the thyroid gland, the activation of thyroid stimulat-
ing hormone (TSH) receptor by autoantibodies leads to an
overproduction of thyroid hormones. In addition, these antibod-
3
ies stimulate ID-I and probably other deiodinases to produce
relatively more T3. As these antibodies are not under the
pituitary feedback control system, there is no negative influence
on the thyroid activity, and therefore, the uncontrolled produc-
tion of thyroid hormones leads to a condition called “hyper-
thyroidism”. Under these conditions, the overproduction of T4
and T3 can be controlled by specific inhibitors, which either
block the thyroid hormone biosynthesis or reduce the conversion
of T4 to T3. A unique class of such inhibitors is the thiourea
(
1) (a) Doerge, D. R. Xenobiotica 1985, 25, 761-767. (b) Taurog, A. Thyroid
Hormone Synthesis. In Werner’s The Thyroid; Braverman, L. E.; Utiger,
R. D., Eds. 1991, 51-97. (c) Taurog, A.; Doriss, M. L.; Doerge, D. R.
Arch. Biochem. Biophys. 1994, 315, 82-89. (d) Doerge, D. R.; Taurog,
A.; Doriss, M. L. Arch. Biochem. Biophys. 1994, 315, 90-99. (e) Taurog,
A.; Doriss, M. L.; Doerge, D. R. Arch. Biochem. Biophys. 1996, 330, 24-
3
2.
10.1021/ja054497u CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 15207-15217
9
15207

A R T I C L E S
Roy and Mugesh
Scheme 2. Synthetic Route to Compound 8a
drugs, methimazole (1, MMI), 6-n-propyl-2-thiouracil (3, PTU),
and 6-methyl-2-thiouracil (5, MTU).
a
Reagents and conditions: i) n-BuLi, THF, -78 °C, 40 min; ii) Se
powder, rt; iii) H2O, 1 N HCl.
mechanism by which the selenium compounds exert their
inhibitory action. We have shown, in a preliminary communica-
tion, that the unexpected behavior of the selenium compounds
as compared to that of their sulfur analogues may be due to
7
their facile oxidation to the corresponding diselenides. Our
initial attempts to isolate 2 were unsuccessful, and the final
stable compound in the synthesis was characterized to be the
diselenide (8). In view of the current interest in anti-thyroid
drugs and their mechanism, we extended our approach to the
synthesis and biological activities of a number of sulfur and
selenium derivatives bearing the methimazole pharmacophore.
In this article, we provide experimental evidence that the
replacement the sulfur atom in methimazole by selenium leads
to a completely different mechanism. In addition, we describe
the effect of a range of sulfur and selenium compounds bearing
methimazole moiety on peroxidase-catalyzed oxidation reac-
tions. We also describe the effect of substituents attached to
the imidazole moiety on the selenol-selone tautomerism by
theoretical calculations.
Although these compounds are the most commonly employed
drugs in the treatment of hyperthyroidism, the detailed mech-
anism of their action is still not clear. According to the initially
proposed mechanism, these drugs may divert oxidized iodides
away from thyroglobulin by forming stable electron donor-
acceptor complexes with diiodine, which can effectively reduce
4
the thyroid hormone biosynthesis. It has also been proposed
that these drugs may block the thyroid hormone synthesis by
5
coordinating to the metal center of thyroid peroxidase (TPO).
After the discovery that the ID-I is responsible for the activation
of thyroxine, it has been reported that PTU, but not MMI, reacts
with the selenenyl iodide intermediate (E-SeI) of ID-I to form
a selenenyl sulfide as a dead end product, thereby blocking the
Results and Discussion
2
conversion of T4 to T3 during the monodeiodination reaction.
The synthesis of selones can be achieved by a variety of
methods, which include reactions with electrophilic selenium,
The mechanism of anti-thyroid activity is further complicated
by the fact that the gold-containing drugs such as gold
thioglucose (GTG) inhibit the deiodinase activity by reacting
8
nucleophilic selenium and carbon diselenide. The synthesis of
2 was approached by the electrophilic selenium route, which
utilizes the high reactivity of elemental selenium toward vinyl
anions (Scheme 2). The low-temperature metalation of 1-me-
thylimidazole (9) by n-BuLi afforded the lithiated species 10,
which upon treatment with elemental selenium produced the
corresponding lithium selenolate 11. The addition of a 1 N HCl
solution followed by an aqueous workup afforded a viscous
liquid, which slowly solidified to form an orange solid. This
method was originally employed by Guziec et al. for the
synthesis of selone 2, and in this particular case, the authors
have observed the formation of two types of compounds that
differ considerably from each other in their melting points.6d
However, all our attempts to isolate the selone (2) were
unsuccessful, and this compound readily oxidized by air to
produce the corresponding diselenide (8). This is in agreement
with the report of Guziec et al. that the recrystallization of 2
2
with the selenol group of the native enzyme (Scheme 1).
Recently, the selenium analogues 2 (MSeI), 4 (PSeU) and 6
(MSeU) attracted considerable attention because these com-
pounds are expected to be more nucleophilic than their sulfur
analogues and the formation of an -Se-Se- bond may occur
more readily than the formation of an -Se-S- bond with the
6
ID-I enzyme. However, the data derived from the inhibition
of TPO by selenium compounds show that these compounds
may inhibit the TPO activity by a different mechanism.
Therefore, further studies are required to understand the
(
2) (a) Behne, D.; Kyriakopoulos, A.; Meinhold, H.; K o¨ hrle, J. Biochem.
Biophys. Res. Commun. 1990, 173, 1143-1149. (b) Berry, M. J.; Banu,
L.; Larsen, P. R. Nature 1991, 349, 438-440. (c) Berry, M. J.; Kieffer, J.
D.; Harney, J. W.; Larsen, P. R. J. Biol. Chem. 1991, 266, 14155-14158.
(
d) K o¨ hrle, J. Exp. Clin. Endocrinol. 1994, 102, 63-89. (e) Larsen, P. R.;
Berry, M. J. Annu. ReV. Nutr. 1995, 15, 323-352. (f) St. Germain, D. L.;
Galton, V. A. Thyroid 1997, 7, 655-668. (g) K o¨ hrle, J. Biochimie 1999,
6d
8
1, 527-553. (h) du Mont, W.-W.; Mugesh, G.; Wismach, C.; Jones, P.
affords orange crystals having a melting point of 142 °C, which
is identical with that of 8 in our synthesis. This indicates that
the diselenide (8) is the final stable compound in the preparation.
The structure of this compound was confirmed by a variety of
G. Angew. Chem., Int. Ed. 2001, 40, 2486-2489. (i) Bianco, A. C.;
Salvatore, D.; Gereben, B.; Berry, M. J.; Larsen, P. R. Endocr. ReV. 2002,
2
3, 38-89. (j) K o¨ hrle, J. Methods Enzymol. 2002, 347, 125-167.
(3) In addition to ID-I, two other deiodinases (ID-II and ID-III) have been
shown to have selenium in their active centers (For more details, see refs
77
2
e-j).
techniques including Se NMR studies. The proposed diselenide
(
4) (a) Buxeraud, J.; Absil, A. C.; Claude, J.; Raby, C.; Catanzano, G.; Beck,
C. Eur. J. Med. Chem. 1985, 20, 43-50. (b) Raby, C.; Lagorce, J. F.;
Jambut-Absil, A. C.; Buxeraud, J.; Catanzano, G. Endocrinology 1990, 126,
structure was finally confirmed by single-crystal X-ray studies.7
The tautomeric behavior of MMI has been subjected to many
1
683-1691.
9
investigations, which show that MMI exists almost exclusively
(
5) Bassosi, R.; Niccolai, N.; Rossi, C. Biophys. Chem. 1978, 8, 61-69.
6) (a) Visser, T. J.; Kaptein, E.; Aboul-Enein, H. Y. Biochem. Biophys. Res.
Commun. 1992, 189, 1362-1367. (b) Aboul-Enein, H. Y.; Awad, A. A.;
Al-Andis, N. M. J. Enzymol. Inhib. 1993, 7, 147-150. (c) Taurog, A.;
Dorris, M. L.; Guziec, L. J.; Guziec, F. S., Jr. Biochem. Pharmacol. 1994,
(
as the thione tautomer (1a, Figure 1). Recent studies have shown
(7) Roy, G.; Nethaji, M.; Mugesh, G. J. Am. Chem. Soc. 2004, 126, 2712-
2713.
(8) Murai, T.; Kato, S.; Selenocarbonyls. In Topics in Current Chemistry; Wirth,
T.; Ed.; Springer-Verlag, Berlin, 2000; Vol. 208, pp 177-199.
4
8, 1447-1453. (d) Guziec, L. J.; Guziec, F. S., Jr. J. Org. Chem. 1994,
5
9, 4691-4692. (e) Taurog, A.; Dorris, M. L.; Hu, W.-X.; Guziec, F. S.,
Jr. Biochem. Pharmacol. 1995, 49, 701-709.
15208 J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005

Anti-Thyroid Drugs and Thyroid Hormone Synthesis
A R T I C L E S
Figure 1. Proposed tautomeric structures of MMI and MSeI.
that the thiourea-based drugs PTU and MTU also exist as the
1
0
thione tautomers. The stability of the thione tautomers may
prevent these compounds from being oxidized to their corre-
sponding disulfides, which may account for their high anti-
thyroidal activity. Laurence et al. have shown that the thione
tautomer of MMI is responsible for its complexation with
diiodine and the iodine complex of the thione tautomer 1a is
-1
favored by 13.2 kJ‚mol compared to that of the thiol tautomer
1
1a
1
b. Therefore, the facile oxidation of 2 to the corresponding
diselenide (8) requires the compound to be in its selenol form
2b) and not in the selone form (2a). Although compound 2
can exist in both selenol and selone forms in solution, the Se
NMR spectrum recorded immediately after the workup of the
reaction showed a signal at 4 ppm, which can be ascribed to
the selenol (2b) tautomer. In the presence of air, the selenol
slowly oxidizes to the corresponding diselenide, which shifts
the equilibrium to the right (Figure 1), and this process continues
until all the selone-selenol mixture is converted to diselenide
(
7
7
Figure 2. Optimized geometries of 1 and 2. The conversion of thiol to
thione is more favored than the conversion of selenol to selone. The
structures were optimized at the B3LYP level of theory using 6-311++G-
(d,p) basis set.
8
. In contrast, the sulfur analogue, which exists predominantly
in its thione tautomer form (1a), was found to be very stable
and could not be converted to the corresponding disulfide (7)
even by using oxidizing agents such as O2, H2O2 etc. It should
be mentioned that MMI is readily oxidized by the TPO system
to form the disulfide. Although the oxidation of this compound
1
2
by iodine has been postulated to be a possible mechanism,
the chemical way through which MMI is transformed into
disulfide 7 in vivo is unknown.
Figure 3. Variation of relative stabilization energy of 2b for the change
in the N1-C1-Se-H dihedral angle. The single-point energies were
calculated using B3LYP/6-311++G(d,p) level of theory.
The theoretical investigations on selones are highly limited
to the compounds having simple substituents, mainly due to
in gas phase. These studies show that the formation of the
diselenide (8) from 2b is energetically more favored than the
formation of the disulfide (7) from the corresponding thiol (1b)
1
3
the requirement of large basis sets for the calculations. The
relatively larger size and more polarizability of selenium as
compared with those of sulfur have led to the assumption that
the compounds with selone moiety are less stable than their
sulfur analogues. Because the inhibition of TPO by anti-thyroid
drugs depends on the redox state of sulfur or selenium, we
performed detailed quantum chemical calculations on 1 and 2
(see Supporting Information, Scheme S1). Interestingly, the
conversion of thiol to thione is more favored than the conversion
of selenol to the corresponding selone. This can be rationalized
by comparing the relative position of hydrogen on sulfur or
selenium with respect to N1 in their most stable conformations.
In the selenol (2b, Figure 2), the H atom is located away from
N1, leading to an increase in the energy barrier for the selenol-
selone conversion. In contrast, the H atom is located in the close
proximity of N1 in the thiol (1b, Figure 2), which may favor
the thiol-thione conversion. The calculations performed at the
B3LYP level of theory using 6-311++G(d,p) show that the
stucture with a N1-C1-Se-H dihedral angle of 107° is the
most stable conformation. The variation in the relative stabiliza-
tion energy of 2b for changing the position of H atom with
respect to the N1-C1-Se plane is shown in Figure 3. The
position of this hydrogen with respect to the two nitrogen atoms
in the imidazole ring also confirms the delocalized charge as
shown in 2c. As the thione form is calculated to be much more
stable than the corresponding thiol, it is quite unlikely that the
thiol form contributes to the anti-thyroid activity of MMI.
Although the selone (2a) was also calculated to be more stable
than the selenol (2b), the facile oxidation of the selenol to the
(
9) (a) Kjellin, G.; Sandstr o¨ m, J. Acta Chem. Scand. 1969, 23, 2888-2899.
(
b) Faur e´ , R.; Vincent, E. J.; Assef, G.; Kister, J.; Metzger, J. Org. Magn.
Reson. 1977, 9, 688-694. (c) Bojarska-Olejnik, E.; Stefaniak, L.; Witanows-
ki, M.; Hamdi, B. T.; Webb, G. A. Magn. Reson. Chem., 1985, 23, 166-
1
69. (d) Balestrero, R. S.; Forkey, D. M.; Russell, J. G. Magn. Reson. Chem.
1
986, 24, 651-655.
(
10) Antoniadis, C. D.; Corban, G. J.; Hadjikakou, S. K.; Hadjiliadis, N.; Kubicki,
M.; Warner, S. Butler, I. S. Eur. J. Inorg. Chem. 2003, 8, 1635-1640.
11) (a) Laurence, C.; El Ghomari, M. J.; Le Questel, J.-Y.; Berthelot, M.;
Mokhlisse, R. J. Chem. Soc., Perkin Trans. 2 1998, 1545-1551. (b)
Although there is no definite evidence for the coordination of the thione
moiety of the anti-thyroid drugs to the iron center, this model is given on
the basis of the fact that iron-sulfur coordination is very common in
metalloenzymes and hydrogen bonding of heme ligands with a distal
histidine appears to be common in peroxidases. (For details, see ref 11a.)
12) Aragoni, M. C.; Arca, M.; Demartin, F.; Devillanova, F. A.; Garau, A.;
Isaia, F.; Lippolis, V.; Verani, G. J. Am. Chem. Soc. 2002, 124, 4538-
(
(
(
4
539.
13) (a) Bock, H.; Aygen, S.; Rosmus, P.; Solouki, B.; Weissflog, E. Chem.
Ber. 1984, 117, 187-202. (b) Collin, S.; Back, T. G.; Rauk, A. J. Am.
Chem. Soc. 1985, 107, 6589-6592. (c) Dapprich, S.; Frenking, G. Chem.
Phys. Lett. 1993, 205, 337-342. (d) Kwiatkowski, J. S.; Leszczynski, J.
Mol. Phys. 1994, 81, 119-131. (e) Ha, T.-K.; Puebla, C. Chem. Phys. 1994,
1
81, 47-55. (f) Jemmis, E. D.; Giju, K. T.; Leszczynski, J. J. Phys. Chem.
A 1997, 101, 7389-7395.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005 15209

A R T I C L E S
Roy and Mugesh
corresponding diselenide disfavors the selenol-selone conver-
sion by 13.42 kcal/mol.
According to the previous studies on tautomeric rearrange-
ments,14 the electron delocalization in selenourea should be
larger than that in thiourea and urea. On the other hand, the
lower electronegativity of Se (2.4) as compared with those of
S (2.5) and O (3.5) suggests that the electron delocalization in
selenourea should be less than that of thiourea and urea.
However, Glendening and Hrabal reported that the polarizability
of the C-X (X ) O, S, Se) bond rather than electronegativity
of X plays an important role in allowing the chalcogen atom to
accommodate more charge density.15 In accordance with this,
the C-Se bond length of 1.833 Å in selone is found to be much
longer than that observed in compound 15 (1.775 Å), which
does not have any nitrogen atom in the five-membered ring.
77
This also confirms the tautomeric behavior of 2. The Se NMR
chemical shift of 30 ppm for 2 with respect to Me2Se suggests
the existence of this compound in a selenolate form (2c). To
further understand the nature of selenium moiety under in vivo
conditions, we included the solvent effects in the calculations
1
6
by using Tomasi’s polarizable continuum model (PCM). The
structure of selone (2a) was optimized in water (ꢀ ) 78.39)
using the PC model at the B3LYP/6-311++G(d,p) level, and
7
7
the Se NMR chemical shift was calculated by using GIAO
1
7
method at the same level of theory as with 6-311++G(2d,p)
77
basis set. These calculations predict a Se NMR chemical shift
of 67 ppm, which is shifted upfield as compared with that of
selenourea (350 ppm), confirming the existence of 2 in its
selenolate form.
Figure 4. Optimized geometries of 2c, 12-15. The structures were
optimized at the B3LYP level of theory using 6-311++G(d,p) basis set.
The reactivity and reaction patterns of selones, in general,
vary considerably, depending upon the substituents adjacent to
the selenocarbonyl group. Therefore, the heteroatom-substituted
Table 1. Theoretical Data for 2, 8, 12, 14, and 15 Obtained by
DFT Calculations at B3LYP/6-311++G(d,p) Level along with the
8
77
selones are more polar than selenoaldehydes and selenoketones.
GIAO Se NMR Chemical Shifts
In view of this, we have undertaken further studies to understand
the effect of substituents on selenocarbonyl moiety. The
replacement of the H atom in 2a with a methyl substituent
77Se chemical shift
C
−
Se bond length
no.
compd
(Å)
(ppm)b
1
2
3
4
5
6
7
8
2a
2b
2c
1.835
1.917
1.835
1.902a
1.908
1.811
1.839
1.774
30
-101
30
386
659
(compound 14, Figure 4) also leads to the stabilization of this
compound in its selenolate form as evidenced by a large upfield
shift in the 77Se NMR chemical shift (46 ppm). In contrast, an
isosteric replacement of the -NH- moiety in 2a with a -CH2-
group stabilizes the compound (12) in its selone form (12a),
indicating that the presence of a nitrogen atom in the 5-position
is important for the conversion of selone to the more reactive
selenolate. The calculated 77Se NMR chemical shift showed a
dramatic downfield shift for 12a (659 ppm) as compared with
that for 2a, confirming the stability of compound 12 in its selone
form (Table 1). On the other hand, the optimized geometry of
the corresponding selenol (12b) and the 77Se NMR chemical
shift calculated for this species (-150 ppm) suggest that the
8
12a
12b
14
15
-150
46
2100
a
Calculated using B3LYP/6-31G(d) level. b NMR values were calculated
using B3LYP/6-311++G(2d,p) level and referenced to Me Se.
2
the replacement the N-H moiety in 1 with the N-Me group
has been shown to abolish the inhibitory effect of 1 in in vivo
experiments. In agreement with this, the calculations on 13
show that the thione form is highly stabilized by the methyl
substituents.
18
-SeH group in this compound exists truly as a selenol moiety
and not in a tautomeric form. This clearly indicates that the
presence of the unsubstituted nitrogen in 2b is responsible for
the formation of the zwitterion 2c. It should be mentioned that
Although compound 2 readily oxidizes to diselenide 8, the
corresponding selenol (2b) can be conveniently obtained by
reducing the diselenide by NaBH or glutathione (GSH). The
4
reaction of 8 with NaBH4 followed by aqueous workup afforded
(
14) (a) Sakaizumi, T.; Yasukawa, A.; Miyamoto, H.; Ohashi, O.; Yamaguchi,
I. Bull. Chem. Soc. Jpn. 1986, 59, 1614-1616. (b) Leszczynski, J.;
Kwiatkowski, J. S.; Leszczynska, D. J. Am. Chem. Soc. 1992, 114, 10089-
the selenol as yellow solid, which was found to be stable under
inert atmosphere and could be employed for in vitro biological
assays without any noticeable oxidation. The treatment of
diselenide 8 with 2 equiv of GSH has also produced the selenol
1
0091. (c) Prasad, B. V.; Uppal, P.; Bassi, P. S. Chem. Phys. Lett. 1997,
76, 31-38. (d) Moudgil, R.; Bharatam, P. V.; Kaur, R.; Kaur, D. Proc.
2
Ind. Acad. Sci. (Chem. Sci.) 2002, 114, 223-230.
(
15) Glendening, E. D.; Hrabal, J. A., II. J. Am. Chem. Soc. 1997, 119, 12940-
1
2946.
(
16) Miertus, S.; Scrocco, E.; Tomasi, J. J. Chem. Phys. 1981, 55, 117-129.
(18) Visser, T. J.; Van Overmeeren, E.; Fekkes, D.; Docter, R.; Hennemann,
(
17) Nakanishi, W.; Hayashi, S. J. J. Phys. Chem. A 1999, 103, 6074-6081.
G. FEBS Lett. 1979, 103, 314-318.
15210 J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005

Anti-Thyroid Drugs and Thyroid Hormone Synthesis
A R T I C L E S
Scheme 3. Synthetic Routes to Compound 19-24a
in nearly a quantitative yield. Interestingly, the formation of
selone (2a) was not observed in any of these processes,
supporting the theoretical calculations that the selenol is
deprotonated (2c) by the nitrogen atoms present in the imidazole
1
13
ring. The H and C NMR data and the large upfield shift in
77
the Se NMR chemical shift for the selenol (-5 ppm) supports
7
7
this assumption. In agreement with the theoretical data, the
-
Se NMR experiments show that the selenol is more dissociated
in water (-53 ppm) than in organic solvent (-5 ppm) as
evidenced by a large upfield shift when changing the solvent
from CDCl3 to D2O. The facile reduction of 8 by GSH suggests
that this compound may exist in its selenol form under in vivo
conditions, because GSH is present in the thyroid gland in high
concentrations. It has been shown that GSH is an important
antioxidant in thyroid gland and the peroxide scavenging activity
a
of MMI is considerably increased by GSH.19
Reagents and conditions: i) CH3I, 0 °C, 4 h; ii) PhCH2Br, 0 °C, 4 h;
iii) R,R′-dibromo-m-xylene, 0 °C, 4 h.
Interestingly, the solvents employed for the workup appear
to have some effect on the nature of products in the reaction
depicted in Scheme 2. When CH2Cl2 was used for the workup
procedure, an unexpected species with a 77Se NMR chemical
shift of 280 ppm was obtained as one of the major products.
The isolation of this species by column chromatography and
mass spectral studies showed the formation of a dimeric
selenium compound having molecular mass (m/z) of 335.9. The
suggests that the selenium moiety exists as a selenolate rather
than selone. The reaction of lithium chalcogenolates with
PhCH Br and R,R’-dibromo-m-xylene produced the correspond-
ing benzylic derivatives. It has been shown that aryl benzyl and
aryl allyl selenides are generally unstable, but the stability of
these compounds could be enhanced by intramolecular Se‚‚‚N
interactions. The expected H‚‚‚ Se coupling observed in the
H NMR spectra of the selenium compounds confirms that the
substitution takes place at the selenium center and not at the
nitrogen.
The third class of compounds (25, 26) that we used in our
study belongs to the group of selenazoles, which do not have
the methimazole moiety but contain a readily cleavable Se-N
bond. Compound 25 was synthesized from commercially
available benzanilide by following literature method and
compound 26 was obtained unexpectedly during our attempts
to synthesize diselenide 27. Although the formation of com-
pound 27 could not be detected during the synthesis, we presume
that the formation of the five-membered heterocycle proceeds
by a spontaneous disproportionation of the diselenide 27
2
1
3
22
1
77
C NMR studies also did not give any conclusive evidence,
1
but these studies suggested two possible structures (16 and 17)
1
with a bridging methylene moiety. However, the H NMR
spectrum of this compound showed a signal at 4.67 ppm (in
77
CD3OD) for the -CH2- with Se coupling, which is consistent
20
with structure 17. The unexpected formation of this compound
is probably due to the high reactivity of the selenol (or
selenolate), which reacts with the solvent CH2Cl2 to produce
1
7.
(Scheme 4). This phenomenon can be rationalized by correlating
As previously mentioned, the N-methylation of MMI leads
the stability of this compound with the presence of the oxazoline
nitrogen, which interacts noncovalently with the divalent
selenium. When one of the selenium atoms in 27 participates
in such interactions, the ortho donor group (N in this case) must
be approximately collinear with the Se-X substituent on
selenium (Se-Se in this case), because the donor atom interacts
with the σ* orbital of the Se-Se bond. The presence of a
substituent at the 6-position prevents such an interaction through
steric hindrance. As the second half of the molecule bearing
the arylselenium moiety would have to approximately occupy
the position of the 6-substituent, a strain is imposed in the
molecule, which leads to the unexpected disproportionation. This
is in agreement with the recent reports that organoselenium
compounds bearing two benzamide groups ortho to selenium
undergo disproportionation reactions, leading to the formation
to a complete loss of its TPO inhibitory effect. Recent studies
on the hepatotoxicity of MMI and the corresponding S-
methylated derivative (19) in glutathione-depleted mice show
that the presence of a free thiol or thione group is important for
the hepatotoxicity of MMI.21 To probe the role of selenol in
the inhibition, we synthesized compounds 19-24, which do not
have any thione/thiol or selone/selenol moiety. These com-
pounds were synthesized by following the low-temperature
lithiation and chalcogen (S, Se) insertion reactions (Scheme 3).
The reaction of lithium chalcogenolates with MeI afforded the
corresponding methyl derivative in good yield. Although the
tautomeric behavior of chalcogenolates having imidazole moiety
are expected to produce both N-substituted and Se-substituted
compounds, the facile formation of the Se-substituted derivatives
2
3
of selenazoles. However, a number of diselenides having two
identical or different substituents in the ortho positions have
(
19) Kim, H.; Lee, T.-H.; Hwang, Y. S.; Bang, M. A.; Kim, K. H.; Suh, J. M.;
Chung, H. K.; Yu, D.-Y.; Lee, K.-K.; Kwon, O.-Y.; Ro, H. K.; Shong, M.
Mol. Pharmacol. 2001, 60, 972-980.
24
been shown to be stable. This indicates that the presence of
(
20) 1H NMR (CD
OD) δ: 3.99 (s, 6H), 4.68 (s, 2H), 7.71 (s, 2H), 7.78 (s,
3
2
H); 13C NMR (CD
OD) δ: 23.7, 35.9, 121.9, 126.2, 133.6; HRMS m/z
3
+ 77
(
TOF) calcd for C
9
H N
12 4
Se [M + H] 336.9470, found 336.9464; Se
(22) (a) Mugesh, G.; Panda, A.; Singh, H. B.; Butcher, R. J. Chem. Eur. J.
1999, 5, 1411-1421. (b) Mugesh, G.; Singh, H. B. Acc. Chem. Res. 2001,
35, 226-236.
NMR δ: 292 (CD
3
OD); 294 (CDCl ).
3
(
21) Mizutani, T.; Yoshida, K.; Murakami, M.; Shirai, M.; Kawazoe, S. Chem.
Res. Toxicol. 2000, 13, 170-176.
(23) Kersting, B.; Delion, M. Z. Naturforsch. 1999, 54b, 1042-1047.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005 15211

A R T I C L E S
Roy and Mugesh
Scheme 4. Formation of Selenazole 26 by Disproportionation
Reactiona
Figure 5. The optimized structure of 26, calculated with the B3LYP/
6
-31G(d) method, showing Se‚‚‚N noncovalent interactions. Se‚‚‚N 2.547
Å, C1-Se 1.870 Å, Se-N1 1.949 Å, ∠ C1-Se-N1 84.8° (X-ray data:
Se‚‚‚N 2.601 Å, C1-Se 1.860 Å, Se-N1 1.912 Å, ∠ C1-Se-N1 84.7°).
7
Table 3. Inhibition of LPO Activity by 1-3 and 5
no.
compda
IC50 (µ
M)a
1
2
3
4
MMI (1)
MSeI(2)
PTU (3)
MTU (5)
7.0 ( 1.1
16.4 ( 1.5
45.0 ( 2.1
47.8 ( 0.1
a
Reagents and conditions: i) LDA, benzene, TMEDA, THF, 4 h; ii) Se
powder, THF, 12 h; iii) H2O, O2.
Table 2. Experimental and Theoretical 77Se NMR Chemical
Shifts for Some of the Selenium Compounds
a
a
Concentration of the compound causing 50% inhibition. Each IC50 value
was calculated from at least three independent experiments.
no.
compd
77Se (
δ
) ppm (Expt)
77Se (
δ
) ppm (Calcd)b
1
2
3
4
5
6
8
397
117
282
959
820
386c
131
20
22
25
26
32
the -OH and carbonyl groups in this compound participate in
intermolecular hydrogen bonding, leading to the formation of
a dimeric structure.
299
951c
c
752
1101
1026c
The enzyme inhibition experiments were carried out with Fe-
containing lactoperoxidase (LPO) since it is readily available
in purified form. Furthermore, LPO has been shown to behave
very similarly to TPO with respect to iodination of thyroglo-
a
77Se NMR chemical shifts δ relative to Me2Se. b
B3LYP/6-311G++(d,p)
Se NMR chemical shift calculations, respectively. c B3LYP/6-31G(d) and
and B3LYP/6-311++G(2d,p) basis sets were used for optimization and
77
7
7
B3LYP/6-311++G(d,p) basis sets were used for optimization and Se
NMR chemical shift calculations, respectively.
26
bulin, the natural substrate, and other iodide acceptors.
Edelhoch et al. have reported the inactivation of LPO by
thiourea-based drugs using the LPO-N-acetyltyrosylamide as-
2
7
an oxazoline moiety that can be hydrolyzed easily is responsible
for the cyclization.
say. We have employed 2,2′-azio-bis-3-ethyl-benthiazoline-
2
8
6-sulfonic acid (ABTS) and H O as substrates to determine
2
2
However, compound 26 turned out to be an interesting ebselen
analogue due to its high stability and solubility in water. Another
interesting feature of this compound is the presence of nonco-
valent Se‚‚‚N interactions between selenium and the nitrogen
50
the half-maximal inhibitory concentration (IC ) of test com-
pounds. The IC50 values for the inhibition of LPO-catalyzed
oxidation of ABTS by 1-3 and 5 are summarized in Table 3.
To obtain reliable IC50 values for compound 2 and to make
a direct comparison with those of the sulfur analogue, it is
important to carry out the inhibition experiments with the
completely reduced species. We have observed that the selenium
compound obtained directly from the reaction (Scheme 2) does
not give any reproducible results. Therefore, we carried out the
experiments with the reduced species (selenol, 2b), which was
obtained by reducing the diselenide (8) with NaBH4 in an
aqueous solution. As expected, MMI inhibited the LPO activity
with an IC50 value of 7.0 µM, which is much lower than those
observed with PTU and MTU. The selenium analogue (2) also
inhibited LPO, and the IC50 value was found to be almost 4-5
times lower than those of PTU and MTU. The higher activity
of MMI as compared with those of PTU and MTU is in
agreement with the previous studies on the inhibition of TPO.
Since the activation of the iron center in TPO must proceed
through an interaction of Fe(III) with H2O2, TPO inactivation
may occur through a competitive coordination of the drug to
iron, assisted by hydrogen bonding with a histidine residue of
2
5
atom present in the five-membered oxazoline ring. These
interactions would enhance the Se-N bond cleavage by thiols
77
as previously shown for some ebselen analogues. The Se NMR
chemical shifts for this compound along with other selenium
compounds used in this study are summarized in Table 2. The
B3LYP optimized structure of 26 and its comparison with the
crystal structure are shown in Figure 5. The Se‚‚‚N distance
calculated for 26 is found to be slightly shorter than that
observed in the crystal structure, probably due to the presence
of intermolecular interactions in the crystal lattice. For example,
(
24) (a) Wirth, T. Angew. Chem., Int. Ed. Engl. 1995, 34, 1726-1728. (b) Wirth,
T.; Fragale, G. Chem. Eur. J. 1997, 3, 1894-1902. (b) Fragale, G.;
Neuburger, M.; Wirth, T. J. Chem. Soc., Chem. Commun. 1998, 1867-
1
868. (c) D e´ ziel, R.; Goulet, S.; Grenier, L.; Bordeleau, J.; Bernier, J. J.
Org. Chem. 1993, 58, 3619-3621. (d) D e´ ziel, R.; Malenfant, E. J. Org.
Chem. 1995, 60, 4660-4662. (c) D e´ ziel, R.; Malenfant, E.; B e´ langer, G.
J. Org. Chem. 1996, 61, 1875-1876. (e) D e´ ziel, R.; Malenfant, E.; Thibault,
C.; Fr e´ chette, S.; Gravel, M. Tetrahedron Lett. 1997, 38, 4753-4756.
25) The nature of these interactions is not discussed here. For excellent articles
on Se‚‚‚X (X ) N, O, etc.) interactions, see: (a) Iwaoka, M.; Tomoda, S.
J. Am. Chem. Soc. 1996, 118, 8077-8084. (b) Wirth, T.; Fragale, G.;
Spichty, M. J. Am. Chem. Soc. 1998, 120, 3376-3381. (c) Spichty, M.;
Fragale, G.; Wirth, T. J. Am. Chem. Soc. 2000, 122, 10914-10916. (d)
Iwaoka, M.; Komatsu, H.; Katsuda, T.; Tomoda, S. J. Am. Chem. Soc.
(
(26) Taurog, A.; Dorris, M. L.; Lamas, L. Endocrinology 1974, 94, 1286-1294.
(27) Edelhoch, H.; Irace, G.; Johnson, M. L.; Michot, J. L.; Nunez, J. J. Biol.
Chem. 1979, 254, 11822-11830.
2
002, 124, 1902-1909. (e) Iwaoka, M.; Komatsu, H.; Katsuda, T.; Tomoda,
S. J. Am. Chem. Soc. 2004, 126, 5309-5317. (f) Iwaoka, M.; Katsuda, T.;
Komatsu, H.; Tomoda, S. J. Org. Chem. 2005, 70, 321-327.
(28) Childs, R. E.; Bardsley, W. G. Biochem. J. 1975, 145, 93-103.
15212 J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005

Anti-Thyroid Drugs and Thyroid Hormone Synthesis
A R T I C L E S
Table 4. Inhibition of LPO by PTU, MTU and Methimazole
Analogues
no.
compd
t
50 (h)
no.
compd
t
50 (h)
1
2
3
4
5
6
PTU (3)
MTU (5)
8
19
20
21
3.8
4.6
2.2
32.6
0.1
7
8
9
10
11
22
23
24
25
26
3.3
26.0
0.6
0.6
2.7
17.0
Conditions: LPO: 6.5 nM; H2O2:22.9 µM; test compound: 40 µM.
Figure 6. A hypothetical model for the coordination of thiourea drugs to
the Fe-center of TPO.
a
Table 5. Inhibition of LPO by Selenoxides 28, 29 and 32
the TPO enzyme (Figure 6).11b Under these conditions, MMI
no.
compd
t50 (h)
1
2
3
28
29
32
27.0
19.5
45.7
might compete more successfully than PTU with H2O2, because
the hydrogen-bond (hard) basicity pkHB value of MMI (2.11) is
much higher than that of PTU (∼1.32). Similar to PTU, the
methyl derivative 5 is also expected to be a weak inhibitor of
TPO. On the other hand, compound 2, which exists predomi-
nantly in its selenol form, does not have the ability to coordinate
to the iron center; therefore, this compound must inhibit the
LPO activity by a different mechanism.
a
Conditions: LPO: 6.5 nM; H2O2: 22.9 µM; test compound: 40 µM.
2
4 underwent oxidation at the selenium center to produce the
monooxo derivative (29), which underwent further oxidation
to produce the corresponding dioxo derivative 30. In contrast
to the monooxo derivative, the Se(IV) species (30) was found
to be unstable and decomposed over a period of 2 h to give red
selenium. Ebselen (25) and selenazole 26 were also oxidized
to the corresponding selenoxides 31 and 32, respectively, as
stable products.
Taurog et al. have shown, in their pioneering work, that MMI
and related derivatives irreversibly inhibit LPO and TPO, leading
29
to a complete inactivation of the enzymes. Doerge and other
have shown that mammalian peroxidases including LPO may
activate the anti-thyroid drugs through S-oxygenation to produce
the corresponding sulfoxides or sulfenic acids.30 They have also
shown that the suicide inactivation of LPO and TPO by MMI
proceeds through the S-oxygenation of the thione moiety to form
a reactive sulfenic acid, which binds covalently to the prosthetic
heme and irreversibly blocks enzyme activity.31 Given the higher
reactivity of selenium compounds as compared with the sulfur
derivatives toward oxidation, it is possible that the facile
oxidation of the selenium compounds may lead to an efficient
inhibition of LPO activity. With this in mind, we treated all the
compounds in this study with H2O2 before adding LPO and
ABTS. The LPO activity was measured several times by
increasing the time for the reaction of the test compounds with
H2O2. The reaction time required to inhibit 50% of the LPO
activity are expressed in terms of t50 values, which are
summarized in Table 4.
Although the selenium compounds were readily oxidized to
the corresponding selenoxides, this nonenzymatic oxidation may
not play any significant role in the inhibition of LPO except
for compound 20 and 25, which showed t50 values of 6 and 35
min, respectively. However, given the time interval used for
the LPO inhibition assay, the oxidation of these compounds may
have only minor effect on the inhibition. This is in agreement
All the selenium compounds were oxidized much faster that
than their sulfur analogues (19, 21, 23), which is responsible
for the lower t50 values of the selenium derivatives. The isolation
and characterization of the oxidized products reveal that the
selenium compounds were oxidized by H2O2 to produce the
corresponding selenoxides. For example, compound 22 was
oxidized to selenoxide 28, which was stable enough to be
purified by column chromatography. Addition of an excess
amount of H2O2 to 28 produced an additional signal at 1293
ppm, which could not be characterized. Similarly, compound
6e
with the report of Taurog et al. that the nonenzymatic oxidation
of MSeI by H O does not interfere appreciably with the
2
2
guaiacol and iodination assays. In contrast to the selenides,
selenoxides do not show any appreciable inhibition (Table 5),
confirming that the lower oxidation state of the selenium center
in the selenides is responsible for the weak inhibition. In contrast
to the s-oxygenation of the sulfur compounds by LPO, these
selenium derivatives do not seem to undergo any enzymatic
(
29) (a) Englar, H.; Taurog, A.; Luthy, C.; Dorris, M. L. Endocrinology 1983,
1
12, 86-95. (b) Taurog, A. Endocrinology 1976, 98, 1031-1046. (c)
Nogimori, T.; Braverman, L. E.; Taurog, A. Fang, S.-L.; Wright, G.;
Emerson, C. H. Endocrinology 1986, 118, 1598-1605. (d) Taurog, A.;
Dorris, M. L.; Guziec, F. S., Jr. Endocrinology 1989, 124, 30-39 and
references therein.
77
oxidation by LPO/H2O2 system as evidenced by Se NMR
(
30) (a) Doerge, D. R. Arch. Biochem. Biophys. 1986, 244, 678-685. (b)
Kobayashi, S.; Nakano, M.; Goto, T.; Kimura, T.; Schaap, A. P. Biochem.
Biophys. Res. Commun. 1986, 135, 166-171.
studies. The rates of oxidation of these selenides to the
corresponding selenoxides in the presence of LPO were found
to be almost identical with the rate in the absence of LPO.
In the presence of GSH, the selenoxides exhibit redox shuttles
between oxidized and reduced forms. For example, the selena-
(
31) (a) Doerge, D. R. Biochemistry 1986, 25, 4724-4728. (b) Doerge, D. R.;
Pitz, G. L.; Root, D. P. Biochem. Pharmacol. 1987, 36, 972-974. (c)
Doerge, D. R.; Takazawa, R. S. Chem. Res. Toxicol. 1990, 3, 98-101. (d)
Doerge, D. R.; Cooray, N. M.; Brewster, M. E. Biochemistry 1991, 30,
8
960-8964.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005 15213

A R T I C L E S
Roy and Mugesh
Scheme 5. Hypothetical Model Representing the Inhibition of LPOa
by 2
Figure 7. Plot of initial rates (Vo) for the LPO-catalyzed oxidation of ABTS
vs concentration of H2O2. (a) Control activity, (b) 40 µM of 2b, (c) 40 µM
of 8, (d) 80 µM of PTU, (e) 80 µM of MTU, (f) 40 µM of MMI.
Conditions: LPO: 6.5 nM; H2O2: 22.9 µM.
zole 26 was oxidized to selenoxide 32 by H2O2 and the addition
of GSH to 32 reduced this species back to 26 in nearly
quantitative yield.
a
The porphyrin core inside the circle represents the active center of LPO.
AA: amino acid residues.
H2O2 for the oxidation of selenenic acid (vide infra) to other
oxidized products at lower concentrations of the peroxide.
These observations strongly support the assumption that
MSeI, in contrast to MMI, does not interfere with the enzyme
directly but inhibits the LPO activity by reducing the H2O2,
which is required for the oxidation of the iron center in LPO
(Scheme 5). When coupled with a suitable thiol such as GSH,
In the absence of H2O2, the Se-N bond in 26 is cleaved by
GSH to produce the corresponding selenol (33). In contrast to
ebselen, the formation of the corresponding selenenyl sulfide
compound 2 may constitute a redox cycle involving a catalytic
reduction of H2O2 (glutathione peroxidase (GPx) activity).33 In
this way, compound 2 mimics the action of GPx, a seleno-
enzyme that protects the cellular components from oxidative
damage by reducing H2O2 with the help of GSH.34 Recently,
the GPx enzyme present in thyroid gland has been shown to
inhibit the iodination reactions by degrading the intracellular
H2O2. In fact, the key compound 2 exhibited interesting GPx
activity, leading to an assumption that some of the anti-thyroid
drugs may act as antioxidants in addition to their inhibition
behavior. The cyclic selenazole 26 also showed higher GPx
activity, but this activity does not seem to correlate with its
LPO inhibitory activity. The sulfur compounds (19, 21, 23) and
selenides (20, 22, 24), however, did not show any significant
GPx activity, confirming that the inhibition of LPO by these
compounds is completely different from that of 2. Similar to
the selenol-mediated inhibition, the inhibition by diselenide 8
could also be reversed by increasing the H2O2 concentration
(Figure 7, c). Although compound 8 did not give any new 77Se
(34) was not detected, indicating that the presence of oxazoline
ring at 6-position destabilizes the selenenyl sulfide species. The
7
7
Se NMR signal at 820 ppm for the selenazole 26 did not
disappear completely even with an excess amount of GSH.
3
5
However, the reduction of 32 to 26 may proceed through a ring
3
2
opening, followed by a cyclization as proposed for ebselen.
The oxidation of selenium center in ebselen and 26 may not
have any direct effect on the LPO activity, because this oxidation
is very slow as compared with the enzymatic oxidation of
ABTS. Remarkably, MSeI (2) inhibited the enzyme within few
seconds even at lower concentrations, which can be ascribed
to the facile oxidation of the reactive selenol group in 2 (MSeI)
by H2O2. Because MMI also inhibits the enzyme very efficiently,
we have carried out further experiments to prove that the
mechanisms by which MMI and MSeI exert their inhibitory
action are different. The initial rates (Vo) derived from various
concentrations of H2O2 were plotted against the concentration
of H2O2. The LPO activity was completely inhibited by 40 µM
MMI, and the enzyme’s activity could not be recovered by
increasing the H2O2 concentration (Figure 7, f). The LPO
activity could not be recovered even at lower concentration of
MMI (10 µM) and higher concentration of H2O2 (230 µM). This
suggests that MMI does not act on H2O2 but acts on the enzyme
itself, leading to an irreversible inhibition as previously pro-
posed. On the other hand, 2 also inhibited the LPO activity as
efficiently as MMI, but in this case, the enzyme’s activity could
be completely recovered by increasing H2O2 concentration
(
33) For GPx activity of selenium compounds, see: (a) Reich, H. J.; Jasperse,
C. P. J. Am. Chem. Soc. 1987, 109, 5549-5551. (b) Wilson, S. R.; Zucker,
P. A.; Huang, R.-R. C.; Spector, A. J. Am. Chem. Soc. 1989, 111, 5936-
5
939. (c) Tomoda, S.; Iwaoka, M. J. Am. Chem. Soc. 1994, 116, 6, 2557-
561. (d) Back, T. G.; Dyck, B. P. J. Am. Chem. Soc. 1997, 119, 2079-
2
2083. (e) Wirth, T. Molecules 1998, 3, 164-166. (f) Mugesh, G.; Singh,
H. B. Chem. Soc. ReV. 2000, 29, 347-357. (g) Mugesh, G.; du Mont, W.-
W.; Sies, H. Chem. ReV. 2001, 101, 2125-2179. (h) Mugesh, G.; Panda,
A.; Singh, H. B.; Punekar, N. S.; Butcher, R. J. J. Am. Chem. Soc. 2001,
1
23, 839-850. (i) Jauslin, M.; Wirth, T.; Meier, T.; Schoumacher, F. Hum.
Mol. Gen. 2002, 11, 3055-3063. (j) Back, T. G.; Moussa, Z. J. Am. Chem.
Soc. 2002, 124, 12104-12105. (k) Back, T. G.; Moussa, Z. J. Am. Chem.
Soc. 2003, 125, 13455-13460. (l) Sun, Y.; Li, T.; Chen, H.; Zhang, K.;
Zheng, K.; Mu, Y.; Yan, G.; Li, W.; Shen, J.; Luo, G. J. Biol. Chem. 2004,
279, 37235-37240.
(34) (a) Selenium in Biology and Human Health; Burk, R. F., Ed.; Springer-
Verlag: New York, 1994. (b) Tappel, A. L. Curr. Top. Cell Regul. 1984,
(Figure 7, b). The sigmoidal behavior of the graph for this
24, 87-97. (c) Floh e´ , L. Curr. Top. Cell Regul. 1985, 27, 473-478. (d)
compound (Figure 7, b) is probably due to the utilization of
Epp, O.; Ladenstein, R.; Wendel, A. Eur. J. Biochem. 1983, 133, 51-69.
35) (a) Bj o¨ rkman, U.; Ekholm, R. Mol. Cell. Endocrinol. 1995, 111, 99-107.
(b) Ekholm, R.; Bj o¨ rkman, U. Endocrinology 1997, 138, 2871-2878.
(
(32) Fischer, H.; Dereu, N. Bull. Soc. Chim. Belg. 1987, 96, 757-768.
1
5214 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005
9

Anti-Thyroid Drugs and Thyroid Hormone Synthesis
A R T I C L E S
Scheme 6. Reaction of Selenol 2 with Iodoacetic Acid, Producing
Selenide 35
a catalytic reduction of H2O2 and thereby mimics the glutathione
peroxidase (GPx) activity in vitro. These studies reveal that the
degradation of the intracellular H2O2 by the selenium analogues
of anti-thyroid drugs may be beneficial to the thyroid gland as
these compounds may act as antioxidants and protect thyroid
cells from oxidative damage. Because the drugs with an action
essentially on H2O2 can reversibly inhibit thyroid peroxidase,
such drugs with a more controlled action could be of great
importance in the treatment of hyperthyroidism.
NMR signal with one equiv of H2O2, addition of an excess
amount of H2O2 to 8 produced a new signal at 1045 ppm, which
36
cannot be ascribed to the selenenic acid and this signal appears
to be different from that obtained from the reaction of 2 with
H2O2, which showed a signal at 1207 ppm.
Experimental Section
General Procedure. Lactoperoxidase from bovine milk and ABTS
were purchased from Fluka Chemical Co. n-Butyllithium was purched
from Acros Chemical Co. (Belgium). All other chemicals were of the
highest purity available. All chemical reactions were carried out under
nitrogen or argon using standard vacuum-line techniques. Solvents were
purified by standard procedures and were freshly distilled prior to use.
As expected, the plot of initial rates (Vo) vs concentration of
MSeI shows that the rate of the reaction decreases with
increasing concentration of MSeI (see Supporting Information,
Figure S15). In all these cases, the LPO activity could be
recovered by increasing the hydrogen peroxide concentration.
These experimental observations support the conclusions made
by Taurog et al. that MSeI, unlike MMI, cannot act as an
irreversible inhibitor of TPO. These observations also support
the in vivo experiments, which showed that MMI is at least 50
times more potent than MSeI as an inhibitor of organic iodine
formation in the thyroid. Crucially, the treatment of 2 with the
selenol specific reagent, iodoacetic acid, abolished the inhibitory
potency of 2, confirming that the oxidation of the selenol group
by H2O2 is responsible for the inhibition. In contrast, the sulfur
analogue MMI was found to be less sensitive to the iodoacetic
acid treatment, and this also confirms that the thiol form of MMI
is not only less predominant in solution but also less reactive
as compared with the thione form (see Supporting Information,
Figure S16). The decrease in the inhibitory activity of 2 upon
iodoacetic acid treatment can be ascribed to the formation of
selenide 35 as a dead-end product. The formation of 35 in the
reaction of 2 with iodoacetic acid was further confirmed by its
independent synthesis (Scheme 6).
1
13
77
H (400 MHz), C (100 MHz) and Se (76.3 MHz) NMR spectra
were obtained on a Bruker Avance 400 NMR Spectrometer. Chemical
1
13
4
shifts are cited with respect to SiMe as internal ( H and C) and
77
Me
Synthesis of Compound 2. To a solution of compound 8 (65.0 mg,
.2 mmol) in water was added NaBH (15.4 mg, 0.4 mmol) at room
2
Se ( Se) as external standard.
0
4
temperature. After stirring for 10 min, the product was extracted with
CH Cl , and the organic layer was concentrated to give the expected
2
2
selenol as yellow solid. This was used for the biological studies without
1
any further purification. Yield: 80 mg (82%). H NMR (CDCl
3
): δ )
3
1
.69. (s, 3H), 6.89 (d, J ) 2 Hz, 1H), 6.95 (d, J ) 2 Hz, 1H), 8.75 (br,
H); 13C NMR (CDCl
/MeOH): δ ) -32.
77
3
): δ ) 36.1, 118.6, 121.2, 148.8; Se NMR
(CDCl
3
Synthesis of Compound 8. To a cooled (-78 °C) solution of
-methylimidazole (0.48 mL, 6.09 mmol) in freshly distilled THF (50
1
mL) was added via syringe n-butyllithium (3.8 mL, 1.6 M in hexanes).
The mixture was stirred at this temperature for 35 min and then allowed
to come to room temperature at which time elemental selenium (0.72
g, 9.12 mmol) was added. The resulting mixture was stirred at room
temperature for an additional 12 h. The mixture was quenched with
cold water and neutralized with 1 N HCl. The aqueous mixture was
Conclusion
extracted with CHCl
and dried over Na SO
The product was recrystallized from CHCl
the desired compound. Yield 0.5 g (50%); mp 142-144 °C; H NMR
3
, and the organic layer was washed with brine
. Removal of solvent afforded an orange solid.
to give orange crystals of
Experimental and theoretical studies show that the selenium
analogue of methimazole (MSeI) exists predominantly in its
selenol form, whereas the sulfur compound exists in its thione
form and the oxidation of the selenol to the corresponding
diselenide is energetically more favored than the conversion of
thione/thiol to the corresponding disulfide. Although MSeI
readily oxidizes to produce the diselenide, the oxidized form
can be easily reduced by reducing agents such as NaBH4 or
glutathione (GSH). The 77Se NMR studies show that the selenol
form of MSeI dissociates in solution to form a more reactive
selenolate, which could be trapped by selenol-specific reagents
such as iodoacetic acid. In its reduced form, MSeI effectively
and reversibly inhibits the iron-containing lactoperoxidase
2
4
3
1
(
CDCl
3
) δ: 3.52 (s, 3H), 7.00 (d, 1H), 7.09 (d,1H); 13C NMR (CDCl
Se
) δ: 397.
3
)
δ: 35.2, 124.7, 131.3, 133.0; HRMS m/z (TOF) calcd for C
H N
8 10 4
2
+
77
[M + H] 322.9314, found 322.9319; Se NMR (CDCl
3
Synthesis of Compound 19. To a cooled (-78 °C) solution of
-methylimidazole (1.0 g, 12.18 mmol) in freshly distilled dry THF
1
(100 mL) was added n-butyllithium (9.1 mL, 1.6 M in hexanes) via
syringe dropwise. The mixture was stirred at -78 °C for 35 min and
then allowed to come to r.t at which time elemental sulfur (0.43 g,
13.4 mmol) was added. The solution was heated at reflux 12 h and
then methyl iodide (1.5 mL, 24.36 mmol) was added dropwise at 0 °C
and the stirring was continued for an additional 4 h at r.t. The resulting
turbid solution was filtered through Celite and the solvent was
(LPO). In contrast to methimazole, MSeI does not interfere with
evaporated. The resulting yellow oil was dissolved in CHCl
3
and again
the enzyme directly, but it inhibits LPO by reducing the H2O2
that is required for the oxidation of the iron center in LPO. In
the presence of GSH, MSeI constitutes a redox cycle involving
filtered through Celite and concentrated to give expected product as
1
viscous oil. Yield: 1.24 g (79%). H NMR (CDCl
3
): δ ) 2.58 (s,
3
1
H), 3.59 (s, 3H), 6.91 (s, 1H), 7.03 (s, 1H); 13C NMR (CDCl
6.2, 33.0, 122.1, 128.9, 142.9; HRMS m/z(TOF) calcd for C
[M+H] 129.0486, found: 129.0506.
3
): δ )
5 8 2
H N S
(
36) Although reactions of diselenides with H
2
O
2
generally afford the corre-
does
+
sponding selenenic acid in the first step, the reaction of 8 with H
2
O
2
not give any isolable selenenic acid. This reaction produced a white solid
for which the following NMR and mass spectral data were obtained [ H
Synthesis of Compound 20. To a cooled (-78 °C) solution of
1-methylimidazole (1.0 g, 12.18 mmol) in freshly distilled THF (100
mL) was added n-butyllithium (9.1 mL, 1.6 M in hexanes) via syringe
dropwise. The mixture was stirred at -78 °C for 35 min and then
allowed to come to r.t. at which time elemental selenium (1.44 g, 18.27
1
1
3
NMR (D
2
O) δ: 3.54 (s, 3H), 7.06 (s, 1H), 8.32 (s, 1H);7 C NMR (D
2
O)
O):
to produce
2
O): δ ) 1316] that could not be
7
δ: 35.3, 119.3, 122.7, 134.76; LRMS m/z (TOF): 243; Se NMR (D
2
2 2
δ ) 1045]. However, this species reacts further with H O
7
7
another compound [ Se NMR (D
characterized.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005 15215

A R T I C L E S
Roy and Mugesh
1
mmol) was added. The resulting mixture was stirred at r.t. for 12 h.
compound 20. Yield 3.9 g (76%). H NMR (CDCl
3
): δ ) 3.30 (s,
To the solution was cooled to 0 °C and then was added Methyl iodide
6H), 4.08 (s, 4H), 6.80 (s, 1H), 6.92 (s, 2H), 6.94-6.96 (m, 2H), 7.08-
(m, H); 7.15 (s, 2H); 13C NMR (CDCl
): δ ) 32.8, 34.3, 123.0, 127.5,
128.7, 128.9, 130.6, 134.7, 139.0; HRMS m/z (TOF) calcd for C16
(1.5 mL, 24.36 mmol) dropwise. The reaction mixture was allowed to
3
come to room temperature and the stirring was continued for an
additional 4 h. The resulting turbid solution was filtered through Celite
and concentrated under reduced pressure. The resulting orange color
18 4
H N -
+
77
Se
2
[M+H] 426.9940, found: 426.9944; Se NMR (CDCl ): δ )
3
280.
oil was dissolved in CH
2
Cl
2
and again filtered through Celite and
Synthesis of Compound 26. To a solution of 0.82 g (3 mmol) of
2,6-bis(4,4-dimethyl-2-oxazoline-2-yl)benzene in 20 mL of dry benzene
were added 0.89 mL (9 mmol) of TMEDA and 9.0 mmol of LDA.
The mixture was stirred for 4 h and the resulting precipitate was
dissolved in dry THF (30 mL). The solution was cooled to -15 °C,
and then elemental selenium (0.24 g, 3 mmol) was added. The stirring
was continued for 12 h at r.t., and then the mixture was poured into a
concentrated to give the expected product as orange oil. Yield: 1.05 g
1
(
1
3
60%). H NMR (CDCl ): δ ) 2.45 (s, 3H), 3.63 (s, 3H), 6.97 (s,
1
3
H), 7.08 (s, 1H); C NMR (CDCl
3
): δ ) 8.0, 34.0, 122.7, 129.9,
+
1
35.7; HRMS m/z(TOF) calcd for C
H
5 8
2
N Se [M+H] 176.9931,
7
7
found: 176.9928; Se NMR (CDCl ): δ ) 117.
3
Synthesis of Compound 21. To a cooled (-78 °C) solution of
1
-methylimidazole (1.0 g, 12.18 mmol) in freshly distilled dry THF
3
beaker containing saturated solution of NaHCO . Oxygen was bubbled
(
100 mL) was added n-butyllithium (9.1 mL, 1.6 M in hexanes) via
for 15 min and the solution was extracted with ether. The organic phase
was dried over anhydrous sodium sulfate, and the solvent was removed
in Vacuo to afford dark oil. The unusually cleaved product 26 was
separated by column chromatography as a stable compound on silica
gel with PE/ethyl acetate (1:1) as eluent. The product was recrystallized
from diethyl ether to give yellow crystals. Yield 0.44 g (40%); mp
164-166 °C; 1H NMR (CDCl3) δ: 1.46 (s, 6H), 1.63 (s, 6H), 3.86
(d, 2H), 4.33 (s, 2H), 6.0 (t, 1H), 7.52 (t, 1H), 7.91 (d, 1H), 8.11 (d,
1H); 13C NMR (CDCl3) δ: 167.82, 162.69, 141.87, 131.33, 130.48,
129.10, 126.30, 121.13, 81.66, 77.26, 71.28, 67.26, 62.69, 28.77, 25.99,
syringe dropwise. The mixture was stirred at -78 °C for 35 min and
then allowed to come to r.t at which time elemental sulfur (0.43 g,
1
3.4 mmol) was added. This mixture was reflux for 12 h. To the above
solution was added benzyl chloride (3.0 g, 24.35 mmol) dropwise at 0
C. The reaction mixture was allowed to come to room temperature
°
and the stirring was continued for an additional 5 h and then was
followed the above procedure (for compound 19). Yield: 0.65 g (42%).
1
H NMR (CDCl ): δ ) 3.22 (s, 3H), 4.14 (s, 2H), 6.85 (s, 1H), 7.09-
3
1
3
7
1
.12 (m, 3H), 7.22-7.24 (m, 3H); C NMR (CDCl ): δ )33.0, 40.0,
3
25.93; 77Se NMR (CDCl3, Me2Se) δ: 822; HRMS m/z(TOF) calcd
22.4, 127.4, 128.5, 128.8, 129.8, 137.9, 140.5; HRMS m/z(TOF) calcd
+
+
for C11
H N
12 2
S [M+H] 205.0799, found: 205.0807.
for C16
H
20
N
2
O
3
Se [M+Na] 391.0537, found: 391.0481.
Synthesis of Compound 22. To a cooled (-78 °C) solution of
-methylimidazole (1.0 g, 12.18 mmol) in freshly distilled THF (100
Synthesis of Compound 28. Hydrogen peroxide (135 µl, 30%
solution) was added to a solution of 22 (0.3 g, 1.19 mmol) in CH Cl
1
2
2
.
mL) was added via syringe n-butyllithium (9.1 mL, 1.6 M in hexanes).
The mixture was stirred at -78 °C for 35 min and then allowed to
come to r.t. at which time elemental selenium (1.44 g, 18.27 mmol)
was added. The resulting mixture was stirred at r.t. for 12 h. To the
above solution was added benzyl chloride (3.0 g, 24.35 mmol) dropwise
at 0 °C. The reaction mixture was allowed to come to room temperature
and the stirring was continued for an additional 5 h. The workup was
The reaction mixture was stirred about 4 h and evaporated to dryness.
The crude product was chromatographed on a silica gel column.
1
Yield: 0.240 g (75%). H NMR (CDCl
3
): δ ) 3.37 (s, 3H), 4.31 (d,
J ) 11.2 Hz 1H), 4.42 (d, J ) 11.2 Hz 1H), 6.86 (s, 1H), 7.03 (d, J )
7.6 Hz 2H), 7.17 (s, 1H), 7.26-7.35 (m, 3H); 13C NMR (CDCl
): δ
3
)32.6, 57.2, 125.45, 128.6, 128.9, 129.9, 130.0, 130.2, 138.2; HRMS
+
m/z(TOF) calcd for C11
Se NMR (CDCl ): δ ) 926.
3
H
12
N
2
OSe [M+H] 269.0193, found: 269.0181;
77
carried out by a similar method given for compound 20. Yield: 0.647
1
gm (42%). H NMR (CDCl
1
3
): δ ) 3.22 (s, 3H), 4.16 (s, 2H), 6.89 (s,
Synthesis of Compound 32. To a solution of selenazole 26 (0.1 g,
H), 7.05-7.07 (m, 2H), 7.15 (s, 1H), 7.19-7.22 (m, 3H); 13C NMR
1
.19 mmol) in MeOH was added a 30% solution of hydrogen peroxide
(
CDCl
3
): δ )33.0, 34.0, 123.0, 127.0, 128.5, 128.7, 130.6, 134.6, 138.8;
(30 µl). The reaction mixture was stirred for 12 h and the solvent was
+
HRMS m/z(TOF) calcd for C11
2
H N
12 2
Se [M+Na] 275.0063, found:
evaporated to give compound 32 as white solid in nearly quantitative
7
7
75.0078; Se NMR (CDCl
3
): δ ) 282.
1
yield. H NMR (CDCl ): δ ) 1.41 (s, 3H), 1.47 (s, 3H), 1.69 (s, 3H),
3
Synthesis of Compound 23. To a cooled (-78 °C) solution of
-methylimidazole (1.54 g, 18.82 mmol) in freshly distilled THF (100
1.73 (s, 3H), 3.58 (d, J ) 11.6 Hz 1H), 4.00 (d, J ) 11.6 Hz, 1H),
4.44-4.28 (m, 2H) 7.75 (t, J ) 7.2 Hz, 1H), 7.97 (d, J ) 7.6 Hz, 1H),
1
8.02 (d, J ) 7.6 Hz 1H); 13C NMR (CDCl ): δ ) 25.2, 26.3, 28.0,
mL) was added via syringe n-butyllithium (15.1 mL, 1.6 M in hexanes)
The mixture was stirred at -78 °C for 35 min and then allowed to
come to r.t at which time elemental sulfur (0.72 g, 22.5 mmol) was
added. This mixture was heated at reflux overnight under nitrogen. To
the above solution was added R, R′-dibromo-m-xylene (2.48 g, 9.4
mmol) in ether (25 mL) dropwise and stirring was continued for an
additional 1 h at 0 °C followed by 4 h. The workup procedure used
3
28.5, 63.3, 67.8, 68.2, 81.0, 125.3, 130.2, 131.2, 133.4, 134.0, 143.8,
+
160.2, 168.3; HRMS m/z(TOF) calcd for C H N O Se [M+Na]
1
6
20
2
4
407.0486, found: 407.0506. 77Se NMR (CDCl ): δ ) 1102.
3
Synthesis of Compound 35. To a solution of selenol 2 (75.0 mg,
3
0.47 mmol) in CH OH was added iodoacetic acid (87.0 mg, 0.47 mmol)
at r.t. After stirring 30 min, the solvent was evaporated to dryness to
1
was similar to that of compound 19. Yield 5.2 g (84%). H NMR
1
give the expected compound as orange oil. Yield: 77 mg (75%). H
(
6
3
CDCl
3
): δ ) 3.26 (s, 6H), 4.07 (s, 4H), 6.85 (s, 2H), 6.89 (s, 1H),
NMR (CDCl
3
): δ ) 3.48 (s, 2H), 3.72 (s, 3H), 7.28 (s, 1H), 7.41 (s,
H); C NMR (CDCl /MeOH): δ ) 29.9, 36.9, 121.9, 125.7, 132.9,
70.8; Se NMR (CDCl /MeOH): δ ) 220.
GPx assay: The GPx activity was followed spectrophotometrically
.99-7.01 (m, 2H), 7.08-7.11 (m, 3H); 13C NMR (CDCl
3
): δ ) 33.1,
13
1
1
3
9.7, 122.5, 127.8, 128.7, 129.2, 129.7, 138.1, 140.4; HRMS m/z(TOF)-
77
3
+
calcd for C16
H N S
18 4 2
[M+H] 331.1051, found: 331.1064.
Synthesis of Compound 24. To a cooled (-78 °C) solution of
-methylimidazole (1.0 g, 12.18 mmol) in freshly distilled THF (100
37
at 340 nm as described by Roveri et al. with minor modifications.
The test mixture contained GSH (1 mM), EDTA (1 mM), glutathione
disulfide reductase (0.6 unit/ml), and NADPH (0.2 mM) in 0.1 M
potassium phosphate buffer, pH 7.3. GPx samples were added to the
test mixture at room temperature and the reaction was started by the
1
mL) was added via syringe n-butyllithium (9.1 mL, 1.6 M in hexanes).
The mixture was stirred at -78 °C for 35 min and then slowly allowed
to come to r.t. Elemental selenium (1.44 g, 18.27 mmol) was added to
the above reaction mixture and the stirring was continued for 12 h at
r.t. To the above solution was added R, R′-dibromo-m-xylene (1.6 g, 6
mmol) in ether (15 mL) dropwise at 0 °C and stirring was continued
for an additional 1 h. at this temperature. The reaction mixture was
allowed to come to r.t. and the stirring was continued for an additional
2 2
addition of H O (1 mM). The initial reduction rates were calculated
from the rate of NADPH oxidation at 340 nm. Each initial rate was
measured at least 3 times and calculated from the first 5-10% of the
(
37) Roveri, A.; Maiorino, M.; Ursini, F. Methods Enzymol. 1994, 233, 202-
4
h and then the workup procedure was followed similar to that of
212.
1
5216 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005
9

Anti-Thyroid Drugs and Thyroid Hormone Synthesis
A R T I C L E S
1
1
reaction by using 6.22 mM- cm- as the extinction coefficient for
Geometries were fully optimized at B3LYP level of theory using the
6-31G(d) basis sets. All stationary points were characterized as minima
by corresponding Hessian indices. The NMR calculations were done
at B3LYP/6-311+G(d,p) level on B3LYP/6-31G(d) level optimized
geometries using the GIAO method.17 Orbital interactions were analyzed
using the Natural Bond Orbital (NBO) method at the B3LYP/6-31G-
(d) level, and charges were calculated from Natural Population Analysis
(NPA).40 Single-point energies with zero-point energies (ZPE) were
calculated at B3LYP/6-311++G(d,p) level in vacuo. The solvent effect
NADPH. For the peroxidase activity, the rates were corrected for the
background reaction between H O and GSH.
2 2
LPO Assay: The LPO inhibition experiments were performed in
phosphate buffer (pH 7) at 25 °C. The spectral measurements were
carried out in a Perkin-Elmer spectrophotometer. Assay of LPO enzyme
activity was followed by catalysis of the oxidation of ABTS. The initial
rate was calculated by following UV absorption increase at 411 nm.
Enzyme activity after the addition of various inhibitors was expressed
as the percentage of that observed in the absence of inhibitors. The
peroxide concentration was always present in excess with respect
enzyme. The inhibition plots were obtained by using Origin 6.1 software
and these plots are used for the calculation of the IC50 values. The t50
values were obtained by treating the test compounds with H O before
2 2
adding LPO and ABTS to the reaction mixture. Each initial rate was
calculated by increasing the time for the reaction between the test
2 2
compound and H O and the time required to reduced the catalytic
activity of enzyme to half its value represents the t50 value of a particular
was included in the calculations at the same level using Tomasi’s
16
polarizable continuum model (PCM) in the water solution.
Acknowledgment. This study was supported by the Depart-
ment of Science and Technology (DST), New Delhi, India.
Supporting Information Available: Details of theoretical
calculations (coordinates for the optimized structures and NBO
analysis), LPO inhibition plots and complete ref 38. This
material is available free of charge via the Internet at
http://pubs.acs.org.
compound.
Computational Methods
All calculations were performed using Gaussian98 suite of quantum
JA054497U
3
8
chemical programs. The hybrid Becke 3-Lee-Yang-Parr (B3LYP)
exchange correlation functional was applied for DFT calculations.39
(
39) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789. (b)
Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(
38) Frisch, M. J. et al. Gaussian98; Gaussian, Inc.: Pittsburgh, PA, 1998. The
(40) Glendening, E. D.; Reed, J. E.; Carpenter, J. E.; Weinhold, F. NBO Program
Manual, version 3.1; University of Wisconsin, 1992.
full reference is given in Supporting Information.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 43, 2005 15217