Synthesis and evaluation of dual site inhibitors of 3-deoxy-d-arabino- heptulosonate 7-phosphate synthase

Article abstract of DOI:10.1016/j.bmcl.2011.03.071

3-Deoxy-d-arabino-heptulosonate 7-phosphate (DAH7P) synthase catalyses the first step of the shikimate pathway for the biosynthesis of aromatic compounds. Enzymes of this pathway have been identified as potential targets for drug design. The reaction catalysed by DAH7P synthase is an aldol condensation between phosphoenolpyruvate (PEP) and d-erythrose 4-phosphate (E4P). In this study inhibitors of DAH7P synthase were prepared which were designed to fit into the binding sites of both PEP and E4P substrates simultaneously. Inhibitors, known to target the PEP binding site, were extended using a C4 linker to include an appropriately placed phosphate group in order to access the phosphate-binding site of E4P. A small increase in inhibition was observed with this modification, and the inhibition results have been rationalised by induced-fit docking.

Full text of DOI:10.1016/j.bmcl.2011.03.071

Bioorganic & Medicinal Chemistry Letters 21 (2011) 5092–5097
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and evaluation of dual site inhibitors
of 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase
Scott R. Walker a,b, Wanting Jiao a, Emily J. Parker a,
⇑
a Biomolecular Interaction Centre and Department of Chemistry, University of Canterbury, Christchurch 8140, New Zealand
b Cancer Therapeutics CRC, Monash University, Melbourne, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
3-Deoxy-D-arabino-heptulosonate 7-phosphate (DAH7P) synthase catalyses the first step of the shikimate
Received 9 February 2011
Revised 16 March 2011
Accepted 17 March 2011
Available online 12 April 2011
pathway for the biosynthesis of aromatic compounds. Enzymes of this pathway have been identified as
potential targets for drug design. The reaction catalysed by DAH7P synthase is an aldol condensation
between phosphoenolpyruvate (PEP) and
D-erythrose 4-phosphate (E4P). In this study inhibitors of
DAH7P synthase were prepared which were designed to fit into the binding sites of both PEP and E4P
substrates simultaneously. Inhibitors, known to target the PEP binding site, were extended using a C4 lin-
ker to include an appropriately placed phosphate group in order to access the phosphate-binding site of
E4P. A small increase in inhibition was observed with this modification, and the inhibition results have
been rationalised by induced-fit docking.
Keywords:
Shikimate pathway
DAH7P synthase
Intermediate mimic
Mechanism-based inhibitor design
Aromatic amino acid biosynthesis
Ó 2011 Elsevier Ltd. All rights reserved.
The shikimate pathway is the biosynthetic pathway responsible
for the production of aromatic compounds.1 These compounds
include the three aromatic amino acids, tyrosine, phenylalanine
and tryptophan, folic acid, an essential cofactor for many enzy-
matic processes, and salicylate, used for the biosynthesis of the sid-
erophores through which bacteria acquire iron.2 The pathway is
present in plants, bacteria, fungi and apicomplexan parasites, but
it is absent in animals,3 and in a number of organisms the pathway
has been shown to be necessary for growth and virulence. Conse-
quently, the enzymes of the shikimate pathway have attracted
considerable interest as targets for drug design.4,5
to be promoted via Lewis acid activation of the E4P aldehyde group
by a divalent metal ion, followed by attack of PEP to give an
oxocarbenium intermediate 4 with trigonal planar geometry at
OPO32-
E4P (2)
HO
2-
OPO3
O
OH
2-
PO3
DAH7P synthase
phosphate
O
O
-
HO
HO
CO2
M2+
-
CO2
OH
DAH7P (3)
PEP (1)
The first committed step in the shikimate pathway is catalysed
by the enzyme 3-deoxy-D-arabino-heptulosonate 7-phosphate
(DAH7P) synthase (EC 2.5.1.54). This enzyme catalyses the aldol-
like condensation reaction between phosphoenolpyruvate (PEP,
1) and D-erythrose 4-phosphate (E4P, 2) to generate 3-deoxy-D-
arabino-heptulosonate 7-phosphate (DAH7P, 3) and inorganic
phosphate (Fig. 1).
Many of the details of the reaction mechanism catalysed by
DAH7P synthase are known, and crystal structures of the proteins
from several sources have been solved, revealing several features
of the reaction intermediates that could be targeted in inhibitor de-
sign.6–11 The aldol-like reaction is stereospecific with respect to
both substrates, with the si face of PEP attacking the re face of
the E4P carbonyl.12–14 The initial attack of PEP on E4P is believed
2-
OPO3
2-
OPO3
2-
OPO3
OH
OH
2-
HO
HO
OH
PO3
O
HO
HO
2-
HO
PO3
-
CO2
O
-
CO2
HO
-
CO2
O
O
H
open chain
DAH7P
6
O
H
H
phosphohemiketal
intermediate
oxocarbenium
intermediate
4
5
Figure 1. The reaction catalysed by DAH7P synthase. The overall metal dependent
transformation of PEP 1 and E4P 2 to DAH7P 3 and phosphate is shown, and key
stages of the likely mechanism are indicated.
⇑
Corresponding author. Tel.: +64 3 364 2871; fax: +64 3 364 2110.
E-mail address: Emily.parker@canterbury.ac.nz (E.J. Parker).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.03.071

S. R. Walker et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5092–5097
5093
the former C2 of PEP (Fig. 1).8,15 Attack of water on the electrophilic
oxocarbenium ion gives rise to tetrahedral phosphohemiketal
intermediate 5, and elimination of phosphate leads to the open
chain form (6) of the product DAH7P.
CO2-
O
P
CO2-
CO2-
O
O
O
O
O
O
P
O
HO
O
P
O
HO
Previously, a series of PEP analogues was synthesised and inves-
tigated as inhibitors to examine the ability of inhibitors to target
the PEP site of the DAH7P synthase from Escherichia coli.16 Com-
pounds bearing both carboxylate and phosphonate or phosphate
groups separated by a similar distance to those found in PEP were
found to be effective competitive inhibitors with respect to PEP.
The best inhibitors of the enzyme were compounds that included
a trigonal centre adjacent to the carboxylate moiety, such as ally-
licphosphonate 7 and vinylphosphonate 8. These compounds were
considered to be mimicking the intermediate oxocarbenium ion 4.
In this study, we have elaborated these single-site inhibitors to
include an appropriately spaced distal phosphate moiety, posi-
tioned to be capable of interacting with the binding pocket for
the corresponding phosphate moiety of co-substrate E4P (Fig. 2).
This alteration was made in order to allow the inhibitors to bind
to both PEP and E4P binding sites simultaneously, utilising the
principle of multivalency.17 We report the synthesis ofthe ex-
tended inhibitors based on allylicphosphonate 7 and vinyl phos-
phonate 8 (Fig. 3)16, and study of their interaction with E. coli
DAH7P synthase by both enzyme inhibition measurements and
molecular modelling.
The synthesis of the allylicphosphonate 9 began with the syn-
thesis of aldehyde 12 in two steps from 1,5-pentanediol. Monopro-
tection of 1,5-pentanediol with tert-butyldiphenylsilyl chloride
followed by oxidation with 2-iodoxybenzoic acid gave aldehyde
12.18,19 The aldehyde 12 was treated with the lithium enolate of
selenide 1320 to give a separable mixture of syn and anti adducts
14 (Scheme 1). The relative stereochemistry of each adduct was
established by comparison of the 13C NMR spectra with the exam-
ples reported by Yang et al.21 On mild oxidation both syn and anti-
phenylselenides 14 underwent clean elimination to the corre-
sponding acrylate 15. Acetylation of the allylic alcohol 15 gave allyl
7,Ki =270 50µM
9
-
O
P
CO2
O
O
P
O
HO
O
P
O
HO
µM
K
8, i =4.7 0.7
10
Figure 3. Parent PEP-competitive inhibitors allylicphosphonte7 and vinyl phos-
phonate 8, and their corresponding extended inhibitors 9 and 10.
acetate 16, and heating with triethylphosphite gave allylicphosph-
onate 17 as an inseparable 7:3 mixture of (Z) and (E)-isomers. The
identity of the major and minor isomers of phosphonate 17 was as-
signed by 1H NOESY NMR, with only the minor (E)-product giving a
clear NOESY signal from resonance transfer between the vinyl pro-
ton and thephosphonate methylene protons (see Supplementary
data, Figure S1). Desilylation and phosphorylation of 17 gave the
diethyl phosphate ester 19, and deprotection with trimethylsilyl
bromide followed by aqueous potassium hydroxide gave the de-
sired allylicphosphonate 9 after anion-exchange chromatography.
The yield of the unprotected allylicphosphonate was disappoint-
ingly low, and was due to cleavage of the phosphate ester during
deprotection.
The synthesis of vinyl phosphonate 10 was achieved in nine
steps from 1,5-pentanediol via aldehyde 12 (scheme 2). Using a
modified Darzen’s condensation22,23 the aldehyde was elaborated
into chloroepoxide 21, which on treatment with magnesium iodide
underwent ring opening with subsequent expulsion of chloride to
give the corresponding iodoketone, which on reductive workup
give the
a-ketoester 22. The preparation of a-ketoester 22 from
aldehyde 12 could be readily carried out as one sequence, without
chromatographic purification of the intermediates. Treatment of
OH
a
CO2Et
SePh
TBDPSO
TBDPSO
O
TBDPSO
TBDPSO
14
12
b
OAc
OH
CO2Et
c
CO2Et
16
15
d
O
CO2Et
OEt
P
CO2Et
TBDPSO
EtO
EtO
O
O
e, f
Z:E
OEt
OEt
P
17
Z
:
E
OEt
P
O
19
7:3
7:3
O
g, h
O
P
O
CO2
O
PhSe
CO2Et
O
P
Z:E
OH
7:3
O
13
9
Figure 2. The active site of phenylalanine-regulated DAH7PS from E. coli (PDB
1N8F).1 A sulfate ion bound the active site interacts with Thr100 and Arg99, which
is indicative of the E4P phosphate position when it is bound in the active site. The
enzyme is displayed with white carbons and the active site bound PEP is shown
with green carbon atoms. H-bonds are displayed as black dashed lines.
Scheme 1. Synthesis of allylicphosphonate 9. Reagents and conditions: (a) 13, LDA,
THF, À78 °C (72%), (b) aq H2O2, Et2O (94%), (c) AcCl, py, DCM, 0 °C (93%), (d) P(OEt)3,
80 °C (82%), (e) TBAF, THF (90%), (f) (EtO)2POCl, iPrNEt2, DMAP, DCM (83%), (g)
TMSBr, DCM, (h) aq KOH (11%, two steps).

5094
S. R. Walker et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5092–5097
of the distal phosphate to the electron-deficient alkene of phospho-
O
a
CO2iPr
Cl
nate 10. The role of the distal phosphate functionality in this isom-
erisation is supported by the observation that no isomerisation
was observed in the related synthesis of the shortened vinyl phos-
phonate 8. Investigation of the low yield of the deprotection se-
quence revealed that the major reaction product was indeed the
bromide 26, formed by non-regioselective dealkylation of the
phosphate group.
The ability of the extended compounds to inhibit E. coli DAH7P
synthase was determined by kinetic assay (Table 1). The diastereo-
meric mixture of extended allylicphosphonate 9 was found to give
competitive inhibition with respect to PEP, with an inhibition con-
TBDPSO
TBDPSO
O
12
21
b
CO2iPr
TBDPSO
c
CO2iPr
EtO
EtO
TBDPSO
P
O
O
22
23
d, e
O
P
O
CO2
O
O
stant Ki of 154 20 lM. The inhibition constant is an improvement
O
O
P
EtO
CO2iPr
over that found for the parent PEP-like inhibitor 7, demonstrating
the advantages of a dual-site approach for DAH7P synthase inhibi-
tion (Table 1). To understand the binding modes and contributions
of each isomer in the mixture to the inhibition, the two isomers
were modelled into the E. coli DAH7P synthase active site sepa-
rately using induced-fit modelling which allows movement of
active site residue side chains on inhibitor binding (Figs. 4a and
b). The (Z)-isomer of allylicphosphonate 9 showed a reversed bind-
ing mode, with the PEP portion of the molecule binding in the E4P
site and the E4P phosphate group binding in the PEP site (Figure
4b). This suggests that the (Z)-geometry of this compound at the
double bond does not fit well with the shape of the active site bind-
ing pocket. As a result, the PEP portion of the compound was
placed in the more accommodating E4P site. However, the (E)-iso-
mer of phosphonate 9 showed the expected binding mode, with
the PEP-like and E4P-phosphate like portions placed in the desig-
nated binding site (Fig. 4a). The carboxylate group on the PEP side
of the (E)-isomer shows interactions with Arg92 and Lys186; the
phosphonate group on the PEP side interacts with Lys186,
Arg234 and Arg165. These interactions are equivalent to those ob-
served in the original crystal structure between the bound PEP
molecule and the DAH7P synthase active site (Fig. 2). The E4P-like
phosphate group of the (E)-9 forms interactions with residues
Arg165, Lys273, Arg99 and Thr100. The (E)-isomer of ally-
licphosphonate 9 is a better mimic of the reaction intermediate 4
and therefore is likely to largely account for the inhibition of
E. coli DAH7P synthase observed for the isomeric mixture.
EtO
P
O
HO
f, g
O
EtO
10
P
O
EtO
25
CO2
Br
O
Cl
P
O
HO
Cl
CO2iPr
26
20
Scheme 2. Synthesis of vinyl phosphonate 10. Reagents and conditions: (a) 20,
iPrOK, iPrOH, Et2O, 0 °C, (b) MgI2, Et2O, PhMe, À60 °C then aq NaHSO3 (26% from
12), (c) tetraethyl methylene bis-phosphonate, LDA, THF, À78 °C to À20°C, (31%),
(d) TBAF, THF (75%) (e) (EtO)2POCl, iPrNEt2, DMAP, DCM (94%), (f) TMSBr, DCM, (g)
aq KOH (3%, two steps).
the
a-ketoester 22 with tetraethyl methylenebisphosphonate un-
der Horner–Wadsworth–Emmons conditions gave vinyl phospho-
nate 23 as a separable 8:1 mixture of E/Z isomers. Assignment of
the stereochemistry of each isomer was readily achieved by 1H
NOESY experiments, with the undesired minor product (Z)-23 dis-
playing a NOE correlation between the vinyl and allylic protons.
Desilylation of (E)-23 with TBAF gave alcohol 24, which was phos-
phorylated with diethyl chlorophosphate to give phosphoester 25.
Deprotection of phosphoester 25 with trimethylsilyl bromide, fol-
lowed by saponification of the carboxyl ester gave a mixture of
products, from which the desired compound 10 could be isolated
in low yield by anion exchange chromatography. Under the reac-
tion conditions the phosphonate 10 isomerised to an inseparable
1:1 mixture of E/Z-isomers as assigned by NOESY experiments.
This isomerisation is thought to arise due to a reversible addition
The vinyl phosphonate 10 was obtained as a 1:1 mixture of
(Z)- and (E)-isomers. This mixture was determined to competi-
tively inhibit E. coli DAH7P synthase with respect to PEP. The vinyl
phosphate 10 was a potent inhibitor of DAH7P synthase, with an
Table 1
Inhibition constants of the parent compounds and the extended compounds
Parent Compounda
Ki (l
M) Parent compound
Extended compounds
Ki (l
M) Extended compoundb
CO2
O
O
P
O
CO2
O
O
P
OH
270 50
154 20
O
P
Z:E
O
OH
7
7:3
O
9
CO2
O
P
O
CO2
O
O
O
P
O
HO
O
4.7 0.7
5.3 0.6
3.6 0.4
P
O
8
HO
Z:E
1:1
10
CO2
Br
O
P
O
HO
26
a
Parent compounds were tested as described in Walker et al.16
Inhibitors were assayed on E. coli DAH7P synthase using a direct assay system following the loss of PEP at 232 nm. All compounds were found to be competitive inhibitors
b
with respect to PEP.

S. R. Walker et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5092–5097
5095
Figure 4. Poses (in stereoview) from induced fit docking calculations for compounds (a) (E)-9 and (b) (Z)-9. Atoms on the enzyme active site are displayed with white carbons
and the inhibitor molecules are displayed with green carbons. Hydrogen bonds are showed as black dashed lines.
inhibition constant Ki of 5.3 0.6
similar to the unelaborated parent compound 8, however this com-
pound consisted solely of the desired (E)-isomer, and the corre-
l
M. This inhibition constant is
of E. coli DAH7P synthase. It too proved to be an inhibitor of
E. coli DAH7P synthase, with a Ki of 3.6 0.4 M. The potency of
l
bromide 26 was initially surprising, since it lacks the terminal
phosphate group designed to enhance the binding of these com-
pounds to DAH7P synthase. Modelling of this compound showed
that the PEP portion of bromide 26 (Fig. 5c) is in the same position
as the PEP molecule found in the crystal structure (Fig. 2). As ex-
pected, the bromide compound 26 is not picking up distal interac-
tions at the E4P-phosphate binding site. However, the Ki value for
this compound is not significantly different from that of the ex-
tended vinyl phosphonate compound 10, or the parent compound
8. The similar Ki values may be due to the bromide compound 26
consisting of only the PEP-like (E)-isomer of the vinyl phosphonate
group, whereas compound 10 also contains the interfering (Z)-iso-
mer. The presence of the hydrocarbon tail therefore does not ap-
pear to impede the ability of the PEP-mimicking portion of
bromide 26 to occupy the PEP binding site of E. coli DAH7P syn-
thase. Indeed, the modelling results indicate that the side chain
of Lys97 can move away from the PEP site, creating a binding pock-
et for the brominated hydrocarbon tail.This added hydrophobic
interaction between the hydrocarbon linker of bromide 26 and
the enzyme active site may also contribute to the potency of this
inhibitor. Bromide 26 is the most potent inhibitor for E. coli DAH7P
synthase yet described.
sponding (Z)-isomer is
a much weaker inhibitor of DAH7P
synthase.16 The inseparable mixture of isomers in 10 may obscure
the anticipated increase in binding strength expected for the orig-
inally designed (E)-isomer of the vinyl phosphonate 10.
To understand the binding mode and potency of each isomer in
the compound 10 mixture, the (E)- and (Z)-isomers of compound
10 were modelled into the active site of E. coli DAH7P synthase
separately (Fig. 5a and b) Both isomers showed the expected bind-
ing mode with the PEP and E4P portions positioned in the desired
sites. The (E)-isomer fits well in the active site (Fig. 5a), with the
PEP phosphate-like phosphonate group interacting with Arg234
and Arg165. The side chains of Lys186 and Arg92 interact with
the carboxylate group. The E4P-like phosphate group forms inter-
actions with Arg165, Arg99 and Thr100. In the (Z)-isomer model,
the PEP-like portion does not fit deep in the PEP site, presumably
due to the change in relative geometry of the phosphonate and car-
boxylate groups (Fig. 5b). As a result, interactions between the
inhibitor carboxylate group and Arg92 are lost in compound (Z)-
10; a carboxylate-Arg92 binding interaction is observed for PEP
in the original crystal structure (Fig. 2).1 Therefore, the modelling
results suggest that the (E)-isomer of compound 10 binds to the
enzyme active site better than the corresponding (Z)-isomer.
The bromide 26, produced by non-regioselective dealkylation
during the deprotection step was also evaluated as an inhibitor
In summary, the extension of small PEP-site inhibitors to larger
molecules capable of interacting with both the PEP site and the E4P
phosphate site in DAH7P synthase was investigated. For ally-

5096
S. R. Walker et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5092–5097
Figure 5. Poses (in stereoview) from induced fit docking calculations for compounds (a) (E)-10, (b) (Z)-10, and (c) 26. Atoms on the enzyme active site are displayed with
white carbons and the inhibitor molecules are displayed with green carbons. The H-bonds are showed as black dashed lines.
licphosphonate 8, extension of the inhibitor into the E4P phosphate
pocket produced an increase in binding affinity compared to the
parent PEP-like compound. Whereas the improvement in binding
affinity for the alternative vinyl phosphonate scaffold was ob-
scured by isomerisation issues, induced fit modelling studies indi-
cate that the (E)-isomer is capable of interacting with the E4P
phosphate pocket as designed. These results indicate that the
extension of PEP-like molecules into other areas of the DAH7P syn-
thase active site can result in increased potency, and this strategy
may be used to develop effective inhibitors for this key biosyn-
thetic enzyme.
Acknowledgements
SRW was supported by a University of Canterbury Doctoral
Scholarship and WJ is a recipient of a New Zealand Top Achiever
Doctoral Scholarship.

S. R. Walker et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5092–5097
5097
9. Shumilin, I. A.; Bauerle, R.; Wu, J.; Woodard, R. W.; Kretsinger, R. H. J. Mol. Biol.
2004, 341, 455.
Supplementary data
10. Schofield, L. R.; Anderson, B. F.; Patchett, M. L.; Norris, G. E.; Jameson, G. B.;
Parker, E. J. Biochemistry 2005, 44, 11950.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2011.03.071.
11. Webby, C. J.; Patchett, M. L.; Parker, E. J. Biochem. J. 2005, 390, 223.
12. DeLeo, A. B.; Sprinson, D. B. Biochem. Biophys. Res. Commun. 1968, 32, 873.
13. Floss, H. G.; Onderka, D. K.; Carroll, M. J. Biol. Chem. 1972, 247, 736.
14. Onderka, D. K.; Floss, H. G. J. Am. Chem. Soc. 1969, 91, 5894.
15. Ahn, M.; Pietersma, A. L.; Schofield, L. R.; Parker, E. J. Org. Biomol. Chem. 2005, 3,
4046.
References and notes
1. Bentley, R. Crit. Rev. Biochem. Mol. Biol. 1990, 25, 307.
16. Walker, S. R.; Cumming, H.; Parker, E. J. Org. Biomol. Chem. 2009, 7, 3031.
17. Mammen, M.; Choi, S. K.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37,
2755.
18. Nemoto, H.; Shiraki, M.; Fukumoto, K. J. Org. Chem. 1996, 61, 1347.
19. Barrett, A. G. M.; Flygare, J. A. J. Org. Chem. 1991, 56, 638.
20. Arrica, M. A.; Wirth, T. Eur. J. Org. Chem. 2005, 395.
21. Yang, D.; Gao, Q.; Zheng, B.-F.; Zhu, N.-Y. J. Org. Chem. 2004, 69, 8821.
22. Coutrot, P.; Grison, C.; Tabyaoui, M.; Czernecki, S.; Valery, J. M. J. Chem. Soc.,
Chem. Commun. 1988, 1515.
2. Walsh, C. T.; Liu, J.; Rusnak, F.; Sakaitani, M. Chem. Rev. 1990, 90, 1105.
3. Roberts, C. W.; Roberts, F.; Lyons, R. E.; Kirisits, M. J.; Mui, E. J.; Finnerty, J.;
Johnson, J. J.; Ferguson, D. J. P.; Coggins, J. R.; Krell, T.; Coombs, G. H.; Milhous,
W. K.; Kyle, D. E.; Tzipori, S.; Barnwell, J.; Dame, J. B.; Carlton, J.; McLeod, R. J.
Infect. Dis. 2002, 185, S25.
4. Coggins, J. R.; Abell, C.; Evan, L. B.; Frederickson, M.; Robinson, D. A.; Roszak, A.
W.; Lapthorn, A. P. Biochem. Soc. Trans. 2003, 31, 548.
5. Ducati, R. G.; Basso, L. A.; Santos, D. S. Curr. Drug Targets 2007, 8, 423.
6. Shumilin, I. A.; Bauerle, R.; Kretsinger, R. H. Biochemistry 2003, 42, 3766.
7. Shumilin, I. A.; Kretsinger, R. H.; Bauerle, R. H. Struct. London 1999, 7, 865.
8. König, V.; Pfeil, A.; Braus, G. H.; Schneider, T. R. J. Mol. Biol. 2004, 337, 675.
23. Coutrot, P.; Grison, C.; Tabyaoui, M.; Tabyaoui, B.; Dumarcay, S. Synlett 1999,
792.