Synthesis of immobilized CMP-sialic acids and their enzymatic transfer with sialyltransferase

Article abstract of DOI:10.1002/(SICI)1521-3773(19981204)37:22<3166::AID-ANIE3166>3.0.CO;2-I

A novel immobilization procedure for glycoproteins is based on the reaction of immobilized CMP-NeuAc (CMP-sialic acids) with sialyltransferase (see scheme). The transfer ability was tested under various conditions; a corresponding oligosaccharide or asial

Full text of DOI:10.1002/(SICI)1521-3773(19981204)37:22<3166::AID-ANIE3166>3.0.CO;2-I

COMMUNICATIONS
Pople, J. Chem. Phys. 1972, 56, 2257; M. S. Gordon, Chem. Phys. Lett.
1980, 76, 163) were carried out on the super-computer NEC-SX4 of
the Höchstleistungs-Rechenzentrum Stuttgart. The starting point was
the structure coordinates of the two compounds (Figure 1a and 2a), in
which the vectors of the electron pairs on N were turned perpendic-
ular to the connecting axes SiN while controlling the interatomic
distances on-line to minimize additional structural changes (Figur-
es 1b and 2b, c). Removal of the molecular Si centers (SiH or SiH2) by
Synthesis of Immobilized CMP-Sialic Acids
and Their Enzymatic Transfer with
Sialyltransferase**
Yasuhiro Kajihara,* Takashi Ebata, and
Hisashi Kodama
H
saturation of the ruptured bonds of the substituent groups
The enzymatic synthesis of oligosaccharides and their
analogues with glycosyltransferases (GTases) has progressed
remarkably as a result of the ability to synthesize the donor
substrates.[1] Sialyltransferase (STase)[2] and fucosyltransfer-
ase[3] reactions have also been used to study the structure and
function of oligosaccharides on glycoproteins and on cell
surfaces by use of modified sugar nucleotides. We have
developed a concise synthetic method for the preparation of
CMP-sialic acid (CMP-N-acetylneuraminic acid, CMP-Neu-
Ac; CMP ˆ cytidinmonophosphate) and its analogues.[4] Since
then we have begun investigating possible applications of
STase reactions with use of synthetic CMP-NeuAc analogues
(Figure 1).
(Figures 1c, d and 2d, e) allows an estimate of the van der Waals
energy changes in the periphery of the molecule. After subtraction of
the total energy from these estimations (Figures 1a and 2a),
approximate sums of the energy contributions due to the long-range
SiN interactions result. The accuracy of the B3LYP procedure has
been statistically evaluated for relative energies to be about 0.4 kcal
1
mol (J. B. Foresmanm, A. Frisch in ªExploring Chemistry with
Electronic Structure Methodsº, 2nd ed., GAUSSIAN Inc., Pittsburg,
1996), and, therefore, deviations of only 1 ± 2 kJmol 1 are assumed for
the basis sets used. In addition, substantial changes in the Si tetahedral
geometry due to N !Si hyperconjugation can be excluded by
comparison with the average bond angle f(CSiC) ˆ 1118 in triphe-
nylsilane (J. Allemand, R. Gerdi, Cryst. Struct. Commun. 1979, 8, 927):
both the smaller average angle of 108o in the compound with the
seven-coordinate Si center (Figure 2) as well as the larger one of 113o
for eightfold coordination (Figure 3) can be straightforwardly ration-
alized by the different central SiH or SiH2 units and the phenyl group
substituents of considerably different bulk.
[7] The total energy potential curve for the model adduct chosen, H3NA,
has been calculated using fourth-order Mùller-Plesset perturbation
theory (MP4) and a correlation consistent basis set with polarized
valence ªdouble-zetaº functions(aug-cc-pVDZ) (D. E. Woon, T. H.
Dunning, Jr., J. Chem. Phys. 1993, 98, 1358; R. A. Kendall, T. H.
Dunning Jr., R. J. Harrison, J. Chem. Phys. 1992, 96, 6796). The same
basis set also was employed in the geometry optimization at the MP2
level.
Figure 1. Schematic representation of the novel immobilization method
for glycoproteins.
[8] a) A. Schäfer, M. Weidenbruch, K. Peters, H. G. von Schnering,
Angew. Chem. 1984, 96, 311; Angew. Chem. Int. Ed. Engl. 1984, 23,
302; b) M. Weidenbruch, B. Flintjer, K. Peters, H. G. von Schnering,
Angew. Chem. 1986, 98, 1090; Angew. Chem. Int. Ed. Engl. 1986, 25,
1129.
[9] H. Bock, J. Meuret, K. Ruppert, Angew. Chem. 1993, 105, 413; Angew.
Chem Int. Ed. Engl. 1993, 32, 414; see also H. Bock, J. Meuret, C.
Näther, K. Ruppert, Organosilicon Chemistry ± From Molecules to
Materials, VCH, Weinheim, 1994, p. 11.
[10] ªThe Crystal as a Supramolecular Entityº in Perspectives in Supra-
molecular Chemistry, Vol. 2 (Ed.: G. R. Desiraju), Wiley, Chichester,
1995.
[11] H. Bock, Z. Havlas, A. Rauschenbach, C. Näther, M. Kleine, Chem.
Commun. 1996, 1529; see also: H. Bock, N. Nagel, A. Seibel, Liebigs
Ann. 1997, 2151.
Many enzymes and lectins are glycoproteins containing N-
or O-linked oligosaccharides. When these proteins have
several oligosaccharides on their surfaces, this layer of
oligosaccharides exhibits dynamic fluctuation.[5] Therefore,
increasing the number of oligosaccharide chains also increases
the surface area covered. With most immobilization meth-
ods[6] used for such glycoproteins, this carbohydrate layer
hinders the approach to the amino or carboxyl groups on the
surface of the protein. Unfortunately, if the reagent attaches
to amino acids located close to the catalytic site, the
immobilized enzyme may lose its activity. Therefore, we
wanted to synthesize a CMP-NeuAc derivative in which the
9''-position is attached to the solid phase, and examine its
sialyltransfer ability as part of a novel immobilization
procedure. The nonreducing end of oligosaccharides are
frequently galactosides which serve as acceptors in STase
[12] G. A. Jeffrey, W. Saenger, Hydrogen Bonding in Biological Structures,
Springer, Berlin, 1991.
[13] N. W. Mitzel, U. Losehand, Angew. Chem. 1997, 109, 2897; Angew.
Chem. Int. Ed. Engl. 1997, 36, 2807.
[*] Dr. Y. Kajihara
Department of System Function, Faculty of Science
Yokohama City University
22-2, Seto, Kanazawa-ku, Yokohama, 236-0027 (Japan)
Fax: (‡81)45-787-2370
E-mail: kajihara@yokohama-cu.ac.jp
Further address: Dr. Y. Kajihara, Dr. T. Ebata, Dr. H. Kodama
Life Science Research Laboratry
Japan Tobacco Inc.
6-2 Umegaoka, Aoba-ku, Yokohama 227-0052 (Japan)
[**] We are grateful to Dr. James C. Paulson (Cytel corporation) for a
generous gift of a-(2 !3)-sialyltransferase. We thank Dr. Hajime
Matsushita and Dr. Koshi Koseki (Life Science Research Laboratory,
Japan Tobacco Inc.) for helpful discussions and encouragement. We
also thank Mr.Masayosi Kusama for the mass spectroscopic analyses
(Japan Tobacco Inc.).
3166
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3722-3166 $ 17.50+.50/0
Angew. Chem. Int. Ed. 1998, 37, No. 22

COMMUNICATIONS
reactions. If STase successfully catalyzes the transfer of
immobilized NeuAc to these terminal galactosides, a novel
and mild immobilization method will have been discovered.
Such a method would have two advantages. First, since the
STase reaction generally occurs at the galactosides most
remote from the protein, the immobilization reaction should
smoothly occur in an area which avoids the steric hindrance of
the carbohydrate layer. Second, since sialylation of galacto-
sides at the nonreducing end of bioactive glycoproteins or
enzymes may not depend on their activity in vitro,[7] sialoside
can be used as a new type of linker between bioactive
glycoproteins and the solid-phase support.
by acetylation. Quantification by the periodate/resorcinol
method[11] after O-deacetylation and saponification of 5
indicated 576 nmol of the immobilized NeuAc derivative
per 1 g of CPG. By our previously developed method[4] the
immobilized NeuAc 5 was converted into immobilized CMP-
NeuAc 7, which was identified by comparison of its 31P
NMR[12] spectrum with that of CMP-NeuAc (Figure 2).
Here we report the solid-phase synthesis of CMP-NeuAc
and its transfer reaction to both a corresponding oligosac-
charide acceptor and an asialoglycoprotein (that is, a non-
sialylated glycloprotein). The chemical synthesis of immobi-
lized CMP-NeuAc is shown in Scheme 1. The NeuAc 4,[8]
which is modified in the 9-position and can be obtained by
conventional methods from 1,[9] was coupled with long-chain
alkylamino controlled pore glass (CPG, Sigma).[10] The
residual amino functionalities on the CPG were then capped
Figure 2. 31P NMR spectra of immobilized CMP-NeuAc 7 (top) and CMP-
NeuAc (bottom). X: impurity.
The transfer assay[13] was performed in a solution of HEPES
buffer
containing
7,
[U-14C]-Gal-b-(1 !4)-GlcNAcb-
OMe([14C]-LacNAc[14]), BSA (bovine serum albumin), and
STase. If the immobilized NeuAc was transferred to [14C]-
LacNAc, sialyl-[14C]-LacNAc formed on the surface of CPG.
After incubation at 378C, isotope counts on CPG were
measured with a liquid scintillation counter.[15] For the
transfer reaction, we used both enzymes a-(2 !3)-STase
(3STase, EC 2.4.99.6, Cytel) and a-(2 !6)-STase (6STase, EC
2.4.99.1, Boeringer-Mannheim). The results of the assays are
presented in Figure 3 and Table 1 . Panel A (3STase) indicates
that the transfer activity of NeuAc in the form of immobilized
CMP-NeuAc 7 was elevated with increasing amounts of the
enzyme. Transfer activities were also raised by increases in
either the acceptor concentration or the incubation time
(panels B and C, 3STase). In the case of 6STase (panels D ± F),
the reaction dynamics are similar to those of 3STase. When
STase was first boiled under reflux for 5 min and then added
to the reaction mixture, no transfer activity was observed
(panel G). These results indicate that the immobilized CMP-
NeuAc 7 can act as the sialyl donor. When the incubation time
was extended to 24 h (panel B), the amount transferred at
500mm [14C]-LacNAc increased to 721 pmol. These data
indicate that more than 721 pmol of CMP-NeuAc is immo-
bilized on 1 mg of CPG. We also compared the relative
transfer rates of CMP-NeuAc and 7. The transfer rate for 7
was estimated to be 3% of that for CMP-NeuAc.[16]
Scheme 1. a) 1. NaOMe, MeOH, 258C; 2. TsCl, py, 258C, 48% (2 steps);
b) NaN3, DMF, 808C, 92%; c) 1. H2S, py/NEt3/H2O (5/1/2), 258C;
2. glutaric anhydride, MeOH, 258C, 83%, (2 steps); d) 1. CPG, 1-ethyl-3-
(3-diaminopropyl)carbodiimide (HCl salt), NEt3, DMF, 258C; 2. Ac2O, py,
DMAP, 258C, 3 ± 5% (estimated by the periodate/resorcinol method);
e) 1. NBS, acetone/H2O (10/1), 258C; 2. 6, 1H-tetrazole, MeCN, 258C;
3. tBuOOH, MeCN, 258C; 4. DBU, THF, 258C; 5. NaOMe, MeOH/H2O
(2/1), 258C. Ts ˆ p-toluenesulfonyl; py ˆ pyridine; DMAP ˆ 4-dimethyl-
aminopyridine;
NBS ˆ N-bromosuccinimide;
DBU ˆ 1,8-diazabicy-
clo[5.4.0]undec-7-ene.
Angew. Chem. Int. Ed. 1998, 37, No. 22
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3722-3167 $ 17.50+.50/0
3167

COMMUNICATIONS
tein. When [14C]-LacNAc was used in the assays (panels A ±
F), the control values were less than 5% of the value
determined for the actually immobilized [14C]-LacNAc. The
control values for the assay in panel H are due to nonspecific
binding of asialoglycoprotein to CPG. Therefore, the differ-
ence in control values between 6 h and 24 h is only 4.4 pmol.
On the other hand, the difference in immobilized asialogly-
coprotein between 6 h and 24 h is 15.2 pmol. Since the value
for the immobilized protein increased independent of the
control value upon simple extention of the incubation time,
the data in panel H show that the immobilized CMP-NeuAc
actively immobilizes asialoglycoprotein. In addition, the
transfer activity is reproducible over several experiments.
Since 58 pmol of a1-acidic asialoglycoprotein was incorpo-
rated into CPG, about 10% of the CMP-NeuAc was
consumed during immobilization. Furthermore, these data
also suggest that about 2.5 mg of the glycoprotein can be
immobilized on 1 g of CPG. Since the a1-acidic asialoglyco-
protein used in this assay is estimated to have one equivalent
of [U-14C]-galactoside on its surface,[18] this glycoprotein may
be a poor sialyl acceptor. Therefore, a glycoprotein containing
several galactosides might be a good sialyl acceptor due to the
multivalency effect, and, as a result, a practical quantity of
glycoprotein might be immobilized.
Our methodology is currently limited to the immobilization
of glycoproteins having galactosides at the nonreducing
terminus. Other glycoproteins having fully sialylated oligo-
saccharides or high mannose-type oligosaccharides are more
difficult to immobilize. In the former case, the immobilization
can be performed after sialidase digestion. As shown in
panel H, the overall efficiency[19] and yield[20] of immobiliza-
tion are still not satisfactory. Work is in progress to solve the
above difficulties in order to enable more efficient immobi-
lization.
Figure 3. Amount [pmol] of immobilized CMP-NeuAc (7) transferred
with a-(2 !3)- and a-(2 !6)-sialyltransferases. The assay conditions are
given in Table 1.
Table 1. Assay conditions for the experiments presented in panels A ± H of
Figure 3.[a]
Panel
7
[14C]-
LacNAc
[mg]
STase [mU]
Volume Incuba-
tion time
[mg]
2,3
2,6
[mL]
[h]
In conclusion, we have described a novel finding in which
immobilized CMP-NeuAc reacts with sialyltransferase. This
sialoside can be used as a new type of linker to covalently
bond a labile biomolecule to a solid-phase support.
[b]
A
B
C
D
E
F
2
1
1
2
1
1
2
1
0.4
±
±
32
32
35
32
32
34
34
27[e]
5
[b]
0.8
0.8
±
3
[b]
1.0
±
[b]
3.8
4
[b]
±
2.2
3
[b]
1.0
±
2.0
[c]
[c]
Received: April 6, 1998 [Z11690IE]
German version: Angew. Chem. 1998, 110, 3324 ± 3326
G
H
0.7
3
[d]
[b]
0.8
±
[a] Transfer assays with CMP-Neu5Ac (7) were carried out at 378C in 1.5-
mL eppendorf tubes containing methyl N-acetyl-[1'-6'-14C]-lactosaminide
([14C]-LacNAc), ca 1.15 Kbq per 100 pmol to 1.15 KBq per 16 nmol by
dilution with LacNAc, STase, HEPES buffer (0.1m, pH 7.0), and BSA (1 mg
per 1 mL of assay mixture). All assays (except in G) were performed
without enzyme system as control experiments. [b] Various amounts.
[c] Enzyme quantities: lane 1 (6.6 mU of 6STase), lane 2 (6.6 mU of boiled
6STase), lane 3 (2.0 mU of 3STase), lane 4 (2.0 mU of boiled 3STase).
[d] [14C]-asialoglycoprotein, ca. 1.0 nmol. [e] The buffer solution contains
0.1% nonidet P-40 and MnCl2 (5mm).
Keywords: enzymes ´ glycoproteins ´ immobilization ´ sialic
acids ´ solid-phase synthesis
[1] C. H. Wong, G. M. Whitesides, Enzymes in Synthetic Organic
Chemistry, Pergamon, Oxford, 1994.
[2] H. J. Gross, R. Brossmer, Eur. J. Biochem. 1988, 177, 583 ± 589.
[3] G. Srivastava, K. J. Kaur, O. Hindsgaul, M. M. Palcic, J. Biol. Chem.
1992, 267, 22356 ± 22361.
[4] Y. Kajihara, T. Ebata, K. Koseki, H. Kodama, H. Matsushita, H.
Hashimoto, J. Org. Chem. 1995, 60, 5732 ± 5735.
[5] R. A. Dwek, Chem. Rev. 1996, 96, 683 ± 720.
[6] K. Mosbach, Methods in Enzymology XLIV, Academic Press, New
York, 1976, 134 ± 148.
[7] A. Varki, Glycobiology 1993, 3, 97 ± 130.
[8] The b isomer: 1H NMR (300 MHz, CD3OD; HOD: d ˆ 4.81): d ˆ
7.49 ± 7.24 (m, 5H; Ph), 5.28 ± 5.21 (m, 2H; H-4,8), 4.60 (dd, J ˆ 2.3,
10.3 Hz; H-7), 3.92 (dd, J ˆ 10.4, 10.4 Hz, 1H; H-5), 3.61 (dd, J ˆ 3.2,
14.5 Hz, 1H; H-9a), 3.51 (s, 3H; Me), 3.08 (dd, J ˆ 7.1, 14.5 Hz, 1H;
In addition, we examined whether 7 is active toward an
asialoglycoprotein acceptor (a1-acidic asialoglycoprotein con-
taining [U-14C]-gal[18]) by use of 3STase. As shown in panel H,
the isotope counts for the asialoglycoprotein increased upon
extention of the incubation time. However, isotope counts
measured in the control assay (without enzyme) indicated as
much as half the level of the actually immobilized glycopro-
3168
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3722-3168 $ 17.50+.50/0
Angew. Chem. Int. Ed. 1998, 37, No. 22

COMMUNICATIONS
H-9b), 2.51 (dd, J ˆ 4.8, 13.8 Hz, 1H; H-3eq), 1.90, 1.88, 1.80, 1.77
Cationic Homoleptic Vanadium(ii), (iv), and (v)
Complexes Arising from Protonolysis of
[V(NEt2)4]
‡
(each s, 3H; Me); HR-FAB-MS: calcd for C29H38N2O13SNa [M‡Na ]:
677.1993; found: 677.1997.
[9] A. Marra, P. SinayÈ, Carbohydr. Res. 1989, 187, 35 ± 42.
[10] The CPG was used for the solid-phase synthesis of glycopeptide: M.
Schuster, P. Wang, J. C. Paulson, C. H. Wong, J. Am. Chem. Soc. 1994,
116, 1135 ± 1136.
[11] G. W. Jourdian, L. Dean, S. Roseman, J. Biol. Chem. 1971, 246, 430 ±
435.
Robert Choukroun,* Pierre Moumboko, Sandrine
Chevalier, Michel Etienne, and Bruno Donnadieu
Although the chemistry of cations containing Group 4
metals is very well documented[1] and a subject of ongoing
investigations, with regard to the modeling of the Ziegler±
Natta polymerization, the chemistry of cationic vanadium
complexes has not been as thoroughly studied as that of their
neutral complexes.[2±4] The recent interest in complexes in
which amido and imido groups are directly bound to a
Group 4 or 5 metal atom has led to new catalysts in alkene
polymerization.[5] Hydridotris(pyrazolyl)borate imidovana-
dium(v) in the presence of a methylaluminoxane (MAO)
activator shows modest ethylene polymerization activity.[6] We
reported that the protonolysis of [Cp2VMe2] or [Cp2Zr(BH4)2]
can lead to a disproportionation redox reaction resulting in
cationic vanadium(iii) or zirconium(iii) complexes.[7,8] In this
context we report here the formation of new cationic
vanadium(ii), (iv), and (v) species when [V(NEt2)4] (1) is
treated with the ammonium salt [NHMe2Ph][BR4] (R ˆ Ph,
C6F5).
[12] Immobilized CMP-NeuAc 7 was suspended in 0.5 mL of D2O, and
the 31P NMR spectrum was measured in a conventional manner:
W. L. Fitch, G. Detre, C. P. Holmes, J. Org. Chem. 1994, 59, 7955 ±
7956.
[13] Prior to the assays, 7 was washed with BSA (2 mg of BSA in 1 mL of
HEPES buffer) in a 1.5-mL eppendorf tube in order to prevent
nonspecific absorption of STase or glycoprotein acceptor to the CPG.
After centrifugation, the supernatant was removed. This treatment
was repeated three times, and then 7 was washed with water. After
lyophylization, CMP-Neu5Ac 7 was used for assays.
[14] S. Sabesan, S. Neira, F. Davidson, J. é. Duus, K. Bock, J. Am. Chem.
Soc. 1994, 116, 1616 ± 1634.
[15] After incubation (panels A ± H) the mixture was filtered, and the CPG
remaining on the filter paper was washed with water twice (3 mL).
The filter paper and CPG were inserted into a scintillation vial for
counting. The radioactivity on the CPG was measured by a liquid
scintillation counter, and the amount of [14C]-LacNAc incorporated
on the CPG was estimated. The transfer rates after subtraction of
control values are summarized in Figure 3 (panels A ± F). The data
given (panels A ± H) are averages values.
[16] The transfer assays were performed by one of the following two
methods. The assays of CMP-NeuAc were carried out at 378C with a
solution of HEPES buffer (total 35 mL) containing CMP-[U-14C]-
NeuAc (12.1 KBqnmol 1, 20mm), LacNAc (1mm), BSA, and 3STase
(0.8 mU). The amounts that were transferred were estimated by the
conventional method.[17] For immobilized CMP-NeuAc, a solution of
HEPES buffer (total 35 mL) containing 7 (1 mg: the concentration of
CMP-NeuAc corresponds to 20mm based on 721 pmol of CMP-NeuAc
on 1 mg of CPG), [14C]-LacNAc (1mm), BSA, and 3STase (0.8 mU) was
incubated at 378C, and the transferred amount was estimated with the
above method.[15]
Treatment of 1 with one equivalent of [NHMe2Ph][BPh4] in
thf at room or low temperature ( 788C) causes precipitation
of the unexpected air-sensitive, dicationic, heteroleptic dial-
kylamidovanadium(iv) complex 2 with one thf molecule of
[V(NEt2)4]
1
[V(NEt2)2(thf)4][BPh4]2 ´ thf
2
[17] Y. Kajihara, H. Kodama, T. Wakabayashi, K.-I. Sato, H. Hashimoto,
Carbohydr. Res. 1993, 247, 179 ± 193.
crystallization (yield: 26%). This crystalline red product was
characterized by X-ray structure analysis (Figure 1).[9] The
environment around the metal is octahedral and the V N and
V O distances (2.054 and 2.11 Š (average), respectively) are
comparable to those in other vanadium compounds contain-
ing the bis(trimethylsilyl)amide ligand and a coordinated thf
molecule.[10] The sum of the angles about each N center is
close to 3608, indicating that the amide ligands are probably
acting as three-electron donors. We also observe that a small,
nonquantifiable amount of an uncharacterized green precip-
itate is present among the crystals of 2, providing evidence
that the reaction is not simple.
[18] a1-Acidic asialoglycoprotein with [U-14C]-galactoside was prepared
from a1-acidic asialo-agalacto-glycoprotein by the action of UDP-[U-
14C]-galactose and b-(1 !4)-galactosyltransferase: Y. Kajihara, T.
Endo, H. Ogasawara, H. Kodama, H. Hashimoto, Carbohydr. Res.
1995, 269, 273 ± 294.
[19] Immobilization of the STase as a side reaction of this method could
not be detected in several self-immobilization assays of STase.
[20] Immobilization yields are estimated to be about 4.5% and 5.8% in the
case of LacNAc and asialoglycoprotein, respectively, based on the
sialylacceptors added.
Reconsidering the formation of 2, we repeated our experi-
ment with 1 and two equivalents of [NHMe2Ph][BR4] (R ˆ Ph
or C6F5) in thf at room temperature. Compound 1 was treated
with two equivalents of [NHMe2Ph][B(C6F5)4] in thf, and
pentane was allowed to diffuse slowly into the solution to give
a mixture of crystalline, highly air-sensitive, red products.
[*] Dr. R. Choukroun, P. Moumboko, S. Chevalier, Dr. M. Etienne,
B. Donnadieu
Laboratoire de Chimie de Coordination du CNRS
UPR 8241, 205 Route de Narbonne
F-31077 Toulouse Cedex 4 (France)
Fax: ( ‡ 33)561-5530-03
E-mail: choukrou@lcctoul.lcc-toulouse.fr
Angew. Chem. Int. Ed. 1998, 37, No. 22
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3722-3169 $ 17.50+.50/0
3169