First nonenzymatic synthesis of Kdo8P through a mechanism similar to that suggested for the enzyme Kdo8P synthase

Article abstract of DOI:10.1021/jo961929y

The mechanism of Kdo8P synthase, the enzyme that catalyzes the unusual condensation of D-arabinose 5-phosphate (A5P) with phosphoenolpyruvate (PEP) to form Kdo8P, remains a fascinating subject for bioorganic research. This paper describes the synthesis of two intramolecular models (1 and 2) bearing an enolpyruvate moiety at C-3 of the arabinose fraction. This means that their open-chain aldehyde forms closely mimic the proposed situation, whereby two substrates A5P and PEP evolve into a ternary complex with the synthase. Examination of 1 (in organic solvent) and 2 (in a water solution) under Lewis acid conditions establishes that they both undergo highly stereospecific intramolecular condensation of the enolpyruvate double bond with the carbonyl of sugar. This results in the required Kdo structure possessing the desired stereochemistry. Mechanistic studies suggest that the observed intramolecular condensation process takes place via a stepwise mechanism involving the formation of a transient oxocarbenium ion intermediate. The results obtained, uniquely demonstrate enzyme-like chemistry in the stereospecific synthesis of the Kdo system. Further investigation is certainly warranted, in order to facilitate the construction of other 3-deoxy-2-ulosonoc acids and sialic acids on the basis of the same general model. This is illustrated here in the case of Kdo. Furthermore, the results support the validity of the mechanism suggested for the Kdo8P synthase action, in particular, the possible role of the enzyme in the catalysis of the initial condensation step.

Full text of DOI:10.1021/jo961929y

794
J . Org. Chem. 1997, 62, 794-804
Articles
F ir st Non en zym a tic Syn th esis of Kd o8P th r ou gh a Mech a n ism
Sim ila r to Th a t Su ggested for th e En zym e Kd o8P Syn th a se
Shoucheng Du, Dorit Plat, Valery Belakhov, and Timor Baasov*
Department of Chemistry, Technion-Israel Institute of Technology, Haifa, 32,000, Israel
Received October 14, 1996X
The mechanism of Kdo8P synthase, the enzyme that catalyzes the unusual condensation of
D-arabinose 5-phosphate (A5P) with phosphoenolpyruvate (PEP) to form Kdo8P, remains a
fascinating subject for bioorganic research. This paper describes the synthesis of two intramolecular
models (1 and 2) bearing an enolpyruvate moiety at C-3 of the arabinose fraction. This means
that their open-chain aldehyde forms closely mimic the proposed situation, whereby two substrates
A5P and PEP evolve into a ternary complex with the synthase. Examination of 1 (in organic solvent)
and 2 (in a water solution) under Lewis acid conditions establishes that they both undergo highly
stereospecific intramolecular condensation of the enolpyruvate double bond with the carbonyl of
sugar. This results in the required Kdo structure possessing the desired stereochemistry.
Mechanistic studies suggest that the observed intramolecular condensation process takes place
via a stepwise mechanism involving the formation of a transient oxocarbenium ion intermediate.
The results obtained, uniquely demonstrate enzyme-like chemistry in the stereospecific synthesis
of the Kdo system. Further investigation is certainly warranted, in order to facilitate the
construction of other 3-deoxy-2-ulosonoc acids and sialic acids on the basis of the same general
model. This is illustrated here in the case of Kdo. Furthermore, the results support the validity
of the mechanism suggested for the Kdo8P synthase action, in particular, the possible role of the
enzyme in the catalysis of the initial condensation step.
In tr od u ction
PEP as a substrate and cleavage of the P-O bond of PEP
occurs. The unusual C-O bond cleavage is very rare for
PEP-utilizing enzymes, and a nonenzymatic analogy of
this transformation in solution has yet to be demon-
strated. Many synthetic methodologies of 3-deoxy-2-
ulosonic acids, both chemical6 and enzymatic,7 have been
published over the last decade. However, the puzzling
question of the preference of the C-O bond cleavage of
PEP over the P-O bond cleavage, in these enzyme-
catalyzed reactions, remains unanswered. In an attempt
to understand this unusual condensation mechanism, we
constructed an appropriate chemical model using this
system to demonstrate similar chemistry in solution. The
formation of eight-carbon ulosonic acid Kdo8 was initially
selected for this purpose. This sugar is synthesized as
an 8-phosphate derivative (Kdo8P) by the coupling of
arabinose 5-phosphate (A5P) with PEP and is catalyzed
by the enzyme Kdo8P synthase.9
Higher 3-deoxy-2-ulosonic acids are widely spread,
natural carbohydrates. They participate in various
important biological processes and contain either seven
carbon atoms (3-deoxy-D-arabino-heptulosonic acid 7-phos-
phate, DAHP),1 eight carbon atoms (3-deoxy-D-manno-
octulosonic acid, Kdo)2 or nine carbon atoms (3-deoxy-D-
glycero-D-galacto-nonulosonic acid, KDN, and sialic acids).3
In these unusual carbohydrates, the anomeric center
possesses a carboxylate group, and the adjacent position
is not oxygenated. Interestingly, the biosynthesis of
these compounds follows a similar general route, involv-
ing the stereospecific condensation of phosphoenolpyru-
vate (PEP) with an appropriate aldose.4 The extraordi-
nary feature of these enzyme-catalyzed reactions is that
the PEP undergoes an unusual C-O bond cleavage.5 This
differs from the majority of cases where enzymes utilize
* To whom correspondence should be addressed. Tel.: +972-4-
8293730. Fax: +972-4-8233735. E-mail: chtimor@tx.technion.ac.il
X Abstract published in Advance ACS Abstracts, J anuary 15, 1997.
(1) Haslam, E. Shikimic Acid. Metabolism and Metabolites; Wiley:
New York, 1993.
(2) (a) Unger, F. M. Adv. Carbohydr. Chem. Biochem. 1981, 38, 323.
(b) Bacterial Lipopolysaccharides: Structure, Synthesis and Biological
Activities; Anderson L., Unger, F. M., Eds.; ACS Symposium Series
No. 231; American Chemical Society; Washington, DC, 1983.
(3) (a) Song, Y.; Kitajima, K.; Inoue, S.; Inoue, Y. J . Biol Chem. 1991,
266, 21929. (b) Sialic Acids. Chemistry, Metabolism and Function. In
Cell Biography Monographs; Schauer, R., Ed.; Springer Verlag: Wien,
New York, 1982; Vol. 10.
(4) (a) Floss, H. G.; Onderka, D. K.; Carroll, M. J . Biol. Chem. 1972,
247, 736. (b) Kohen, A.; Berkovich, R.; Belakhov, V.; Baasov, T. Bioorg.
Med. Chem. Lett. 1993, 3, 1577. (c) Dotson, G. D.; Nanjappan, P.; Reily,
M. D.; Woodard, R. W. Biochemistry 1993, 32, 12392.
(5) (a) DeLeo, A. B.; Dayan, J .; Sprinson, D. B. J . Biol. Chem. 1973,
248, 2344. (b) Hedstrom, L.; Abeles, R. H. Biochem. Biophys. Res.
Comm. 1988, 157, 816. (c) Schauer, R. Adv. Carbohydr. Chem. Biochem.
1982, 40, 131.
(6) For recent chemical syntheses, see: (a) Ramage, R.; MacLeod,
A. M.; Rose, G. W. Tetrahedron 1991, 47, 5625. (b) Dondoni, A.; Marino,
P. J . Org. Chem. 1991, 56, 5294. (c) Giese, B.; Carboni, B.; Gobel, T.;
Muhn, R.; Wetterich, F. Tetrahedron Lett. 1992, 33, 2673. (d) Giese,
B.; Linker, T. Synthesis 1992, 46. (e) Chan, T.-H.; Li, C.-J . J . Chem.
Soc., Chem. Commun. 1992, 747. (f) Libineau, A.; Auge, J .; Lubin, N.
Tetrahedron 1993, 49, 4639. (g) Dondoni, A.; Marra, A.; Merino, P. J .
Am. Chem. Soc. 1994, 116, 3324. (h) Gao, J .; Harter, R.; Gordon, D.
M.; Whitesides, G. M. J . Org. Chem. 1994, 59, 3714.
(7) For recent enzymatic syntheses, see: (a) Auge, C.; Gautheron,
C.; David, S.; Malleron, A.; Cavaye, B.; Bouxom, B. Tetrahedron 1990,
46, 201. (b) Kragl, U.; Gygax, D.; Ghisalba, O.; Wandrey, C. Angew.
Chem., Int. Ed. Engl. 1991, 30, 827. (c) Lin. C.-H.; Sugai, T.; Halcomb,
R. L.; Ichikawa, Y.; Wong, C.-H. J . Am. Chem. Soc. 1992, 114, 10138.
(d) Sugai, T.; Shen, G.-J .; Ichikawa, Y.; Wong, C.-H. J . Am. Chem. Soc.
1993, 115, 413. (e) J ohnson, C. R.; Kozak, J . J . Org. Chem. 1994, 59,
2910.
(8) Isolated and initially reffered to as 2-keto-3-deoxyoctonic acid
(Kdo). See: Levin, D. H.; Racker, E. J . Biol. Chem. 1959, 234, 2532.
(9) Ray, P. H. J . Bacteriol. 1980, 141, 635-644.
S0022-3263(96)01929-9 CCC: $14.00 © 1997 American Chemical Society

Nonenzymatic Synthesis of Kdo8P
J . Org. Chem., Vol. 62, No. 4, 1997 795
Sch em e 1
Most recent suggestions regarding the mechanism of
this enzyme-catalyzed reaction (Scheme 1) involve the
formation of either the acyclic intermediate II (path a)5b
or the cyclic intermediate III (path b).10 The question
as to which of these pathways is correct remains to be
determined conclusively. The path leading to the forma-
tion of the transient oxocarbenium intermediate I is
determined using the known stereochemical course of the
reaction,10 when condensation is a stepwise process. In
this case, the initial attack of the π electrons, stemming
from the si face of the PEP double bond, and directed
toward the re face of the carbonyl of A5P, will result in
the formation of the transient oxocarbenium intermediate
I. In the next step, the cation I may be captured in two
ways. The first is by the addition of water, which will
lead to the formation of acyclic intermediate II (path a).
A second path is the addition of the hydroxyl-3 of A5P
(path b), which will result in the formation of the cyclic
bisphosphate intermediate III. Both intermediates II
and III are expected to undergo rapid elimination of the
inorganic phosphate (Pi), through C-O bond cleavage of
the C-2 phosphate, to produce the product Kdo8P.
From the mechanism in Scheme 1 it seems that the
initial condensation step in the enzyme catalysis largely
involves catalysis by proximity. This is due to the
reactive groups of two substrates (A5P and PEP) being
held proximal to each other within a bonding, critical
distance.11 The synthase is not a metalloenzyme and
does not require the addition of metal cations for catalytic
activity.9 Instead, extra acceleration of the condensation
step might be achieved through the activation of the
aldehyde carbonyl by one or more active site electro-
philes. This postulation, though, is not new or unusual.
Indeed, over the last two decades, many groups have
pointed out the possible relationship between enzyme
catalysis and intramolecular catalysis. As shown re-
cently by Menger,12 the idea is attractive because huge
rate acceleration sometimes occurs when an intermo-
lecular reaction is converted into its intramolecular
counterpart. Also, an enzyme reacting with its bound
substrate resembles an intramolecular organic process.
We have, therefore, remained true to the concept of
catalysis by proximity of reactive groups. In an initial
attempt to evaluate intramolecular models, we installed
the enolpyruvate moiety of PEP onto the C-3 hydroxyl
of arabinose and synthesized the derivatives 1 and 2. One
of the conformations of the open-chain, aldehyde form of
1 and 2 closely resembles the proposed situation, whereby
two substrates, A5P and PEP, evolve into a ternary
complex with the synthase. Using these models, we
describe here the novel, stereospecific synthesis of the
Kdo structure in organic solvents.13 Also outlined is the
first stereospecific, chemical synthesis of Kdo8P in a
water solution by a mechanism similar to that suggested
for the synthase. Our results provide additional insight
into the manner in which PEP and A5P are recognized
by the synthase and the nature of the catalytic events
occurring at the enzyme active site.
Resu lts a n d Discu ssion
Prior to embarking on the multistep synthesis and
examination of compound 2, we constructed a simple
model 1. All initial studies on the subject of intramo-
lecular condensation were undertaken in organic solvents
using this model.
Syn th esis a n d Eva lu a tion of th e Mod el Com -
p ou n d 1. The synthesis of 1 begins with the differen-
tially protected arabinose derivative 3. This derivative
was prepared from D-(-)-arabinose in seven chemical
steps following a previously reported procedure14 (Scheme
2). The key step, introduction of the enolpyruvyl moiety
to the C-3 hydroxyl of 3, was successfully accomplished
in four-stages, a method developed by Ganem.15 Thus,
treatment of alcohol 3 with dimethyl diazomalonate and
rhodium(II) acetate yielded the corresponding malonate
ether. Condensation of this product with dimethylmeth-
yleneammonium iodide and triethylamine was followed
by alkylation with iodomethane. Subsequent decarbox-
ylative elimination of the corresponding ammonium salt
afforded the enolpyruvate ester 4 in an overall 36% yield.
(13) The preliminary results on the synthesis and examination of
compound 1 in organic solvents have recently been published as a
communication: Du, S.; Plat, D.; Baasov, T. Tetrahedron Lett. 1996,
37, 3545.
(14) Kalvoda, L.; Prystas, M.; Sorm, F. Collect. Czech. Chem.
Commun. 1976, 41, 788.
(10) (a) Sheffer-Dee-Noor, S.; Belakhov, V.; Baasov, T. Bioorg. Med.
Chem. Lett. 1993, 3, 1583. (b) Baasov, T.; Sheffer-Dee-Noor, S.; Kohen,
A.; J akob, A.; Belakhov, V. Eur. J . Biochem. 1993, 217, 991.
(11) (a) J encks, W. P. Catalysis in Chemistry and Enzymology;
McGraw-Hill: New York, 1969. (b) Kirby, A. J . Adv. Phys. Org. Chem.
1980, 17, 183. (c) Menger, F. M. Acc. Chem. Res. 1985, 18, 128.
(12) Menger, F. M. Acc. Chem. Res. 1993, 26, 206.
(15) Ganem, B.; Ikota, N.; Muralidharan, V. B.; Wade, W. S.; Young,
S. D.; Yukimoto, Y. J . Am. Chem. Soc. 1982, 104, 6787.

796 J . Org. Chem., Vol. 62, No. 4, 1997
Du et al.
Sch em e 2a
a required 3-deoxy-2-ulosonic acid system, possessing the
desired stereochemistry. The structure of bicyclic 5 was
unambiguously determined by chemical, spectroscopic,
and X-ray diffraction analysis.13 The newly formed
stereogenic center (C-4) was found to have the same (R)-
configuration as natural Kdo. The condensation step
appeared to be highly stereospecific, as no C-4 epimer
was detected among the reaction products.
Although more work is required to determine the true
origin of the high diastereoselectivity observed in this
intramolecular condensation, one rationalization of the
results is shown in eq 1. Presumably, 5 arises via a
a
Key: (a) N2C(CO2Me)2, Rh2(OAc)4, C6H6, reflux; (b)
[CH2)NMe2]+ I-, Et3N, CH2Cl2; (c) MeI, CH2Cl2; (d) DMSO, 100
°C, 0.5 h; (e) HOAc/H2O/THF (65:35:10), 70 °C.
Sch em e 3
stepwise pathway involving the formation of the oxocar-
benium ion intermediate 5i.17 A concerted process,
involving a conformationally rigid, boatlike transition
state such as 5ii, is regarded as less likely. Indeed, an
attempt to quench the cation 5i by the addition of 2 mol
equiv of methanol (0.25 equiv of SnCl4, CH2Cl2, 10 h),
resulted in the mixture of products (eq 2) containing the
Removal of the acetonide moiety with a mild acid (65:
35:10 HOAc:H2O:THF) resulted in the target compound
1, in a 93% isolated yield, as a mixture of anomers.
The intramolecular condensation of the aldehyde with
the olefinic function in 1 was examined with different
Lewis acids, and it was found that SnCl4 in dichlo-
romethane was the best mixture for this purpose. Thus,
treatment of 1 with SnCl4 at 0 °C afforded a mixture of
products 5 (major, Scheme 3) and 6 (minor). However,
when the reaction was carried out using the same
stoichiometry, at room temperature, a predominance of
mixture 6 is formed. Upon close investigation of the
reaction, we found that the first product formed during
the condensation process is the bicyclic 5. The ketal ring
of 5 opens under reaction conditions, resulting in the
formation of the anomeric mixture 6. This result was
further corroborated by performing the above procedure
on the chemically pure 5. The same anomeric mixture 6
was afforded. Further treatment of the condensation
reaction mixture with acid (MeOH, TsOH) induced the
formation of R- and â-methyl furanoside derivatives of
Kdo (R-7 and â-7). The structures of these derivatives
were determined by comparison with the reported fura-
noside structures of Kdo.16 The predominant formation
of the furanoside structures of Kdo is in accordance with
the earlier reported results.16a Furanoside structures
were seen to predominate when Kdo is sequentially
protected, that is, when carboxylic acid protection is
followed by the protection of hydroxyl groups.
anomeric mixture (R and â) of methoxy pyranosides 9
(18%). This was also true for 5 (26%), 6 (32%), 7 (5%),
and 8 (6.3%). Raising the amount of methanol to 4 mol
equiv subsequently increased the percentage of the
anomeric mixture of 9 (27%, 24 h). The reaction, using
methanol (eq 3) as solvent (2.7 equiv of SnCl4, 5 days),
afforded a mixture of 8 (61%) and 9 (35%). Only trace
amounts of 5 and 7 were detected by 1H-NMR in this
mixture. The same series of experiments were carried
out using pure 5 (eqs 4 and 5). The aim of this series
was to verify whether the methoxy pyranosides 9 form
by direct quenching of the cationic intermediate 5i with
methanol, by acid-catalyzed methanolysis of bicyclic 5,
or by methylation of pyranoside derivatives of 6. There
was no discernible formation of methoxy pyranosides 9,
indicating that 9 should be formed only by the direct
quenching of intermediate 5i with methanol. The results
The formation of bicyclic adduct 5 is especially note-
worthy. To the best of our knowledge, this represents
the first example of nonenzymic condensation of the
enolpyruvate moiety to the carbonyl group. It results in
(16) (a) Charon, D.; Szabo, L. J . Chem. Soc., Perkin Trans. 1 1979,
2369. (b) Brade, H.; Zahringer, U.; Rietschel, E. T.; Christian, R.;
Schulz, G.; Unger, F. M. Carbohydr. Res. 1984, 134, 157. (c) See also
ref 2b.
(17) (a) Kresge, A. J .; Leibovitch, M.; Sikorski, J . A. J . Am. Chem.
Soc. 1992, 114, 2618. (b) Baasov, T.; Kohen, A. J . Am. Chem. Soc. 1995,
117, 6165.

Nonenzymatic Synthesis of Kdo8P
J . Org. Chem., Vol. 62, No. 4, 1997 797
Sch em e 4a
group in 2 should be in its carboxylate form. Therefore,
we would expect that a high rate of reaction may be
caused due to stabilization of the positive charge gener-
ated on the transient, oxocarbenium ion intermediate.
Recently reported acid-catalyzed hydrolysis studies of
methyl R-methoxyacrylate17a and Kdo 2-phosphates17b
have illustrated a similar inductive stabilizing effect of
the R-carboxylate group. Therefore, when considering
the possible positive role of the carboxylate function in
the condensation process, we set out to use pH g 5 as
the working pH. (ii) In an attempt to accelerate the
condensation step through the activation of the aldehyde
carbonyl by a Lewis acid, we chose to add ZnCl2 to the
reaction mixture. Selection of Zn2+ as a Lewis acid
seemed promising as zinc is a widely distributed ion and
an essential component of many enzymes and proteins.
However, the combination of the pH and Zn2+ restricted
our work to a limited pH range (between 5 and 6). This
is attributable to the fact that compound 2 precipitates
at pH values >6, perhaps due to the formation of highly
insoluble zinc phosphate salt. (iii) Compound 2 exists
in water solutions as a mixture of two anomeric forms.
The expected product of the intramolecular condensation,
Kdo8P, also exists as a mixture of four isomers (R- and
â-furanosides, 30%, and R- and â-pyranosides, 70%).18 As
the geminal C-3 protons of Kdo8P anomers are well
separated in the region of 1.7-2.5 ppm, close monitoring
a
Key: (a) (TCEO)2P(O)Cl, Py; (b) N2C(CO2Me)2, Rh2(OAc)4,
C6H6, reflux; (c) [CH2)NMe2]+ I-, Et3N, CH2Cl2; (d) MeI, CH2Cl2;
(e) DMSO, 100 °C, 4 h; (f) Zn, HOAc, MeOH; (g) Dowex 50W (H+),
D2O; (h) 1 N KOH.
support the possibility that 5 is formed by a stepwise
mechanism (eq 1), through the formation of a transient
oxocarbenium intermediate 5i, followed by the capture
of the cation by the C-5 hydroxyl of the sugar moiety.
Syn th esis a n d Exa m in a tion of th e Mod el Com -
p ou n d 2. We now had a new synthetic methodology in
hand for the construction of the Kdo skeleton, via
intramolecular condensation in an organic solvent. This
was supported by the promising diastereoselectivity
shown in the condensation step. In seeking to improve
our model and more closely mimic the natural system,
the following questions were addressed. (i) Can the same
intramolecular condensation method be used in an aque-
ous solution, without the participation of protecting
groups on the sugar substrate, for the direct chemical
synthesis of the enzymatic reaction product, Kdo8P? (ii)
What will the timing and stereochemical outcome of such
a reaction be? In order to answer these questions, the
model structure 2 was synthesized. Structure 2 satisfies
most of the requirements of the solution studies. It bears
the additional advantage of holding most of the functional
groups of both enzyme substrates (A5P and PEP).
Therefore, the prospect of examining its inhibitory and
substrate properties in relation to Kdo8P synthase was
very attractive.
Scheme 4 depicts our approach to the synthesis of 2.
First, the known14 diol 10 was phosphorylated, with
reasonable regioselectivity, to afford the protected phos-
phate 11, in a 70% isolated yield. Compound 11 was
subjected to a four-step sequence of reactions, similar to
that employed in the preparation of tosylate 4. The
purpose of this was to introduce the enolpyruvate side
chain on to the C-3 hydroxyl, resulting in enolpyruvyl
phosphate 12, in an overall 34% yield (for the four steps).
In order to furnish the synthesis of target compound 2,
optimal conditions were achieved through stepwise depro-
tection of all the protecting groups. The best results were
obtained using the following deprotection sequence: First,
treatment of 12 with Zn-HOAc removed the trichloro-
ethyl protections from the phosphate group. The isopro-
pylidene protection was then removed by treatment of
the observed phosphate with Dowex (H+) resin in D2O
(36 h). Progress of the reaction was followed by 1H-NMR.
Finally, saponification of the carboxylic ester (KOH, H2O,
4 °C) afforded the target 2 (89% after chromatography)
as a mixture of anomers.
1
of the condensation process by H-NMR19 was expected
to be possible.
Condensation experiments using the model 2 were
performed in water solutions containing 5 mol equiv of
1
ZnCl2. Progress of the reaction was monitored by H-
NMR (Figure 1A). As seen from these spectra, the
characteristic resonances of 2 in the range of 5.1-5.35
ppm gradually decrease with time, and new resonances
in the region of 1.7-2.5 ppm subsequently evolve. Care-
ful analysis of the spectra reveals that the first product
exclusively formed during the reaction is Kdo8P. Under
the reaction conditions, the product suffers Lewis acid-
catalyzed dehydration20 to afford a mixture of bicyclic
products 13 and 14 (in a ratio of 2:1). A trace amount of
furoic acid derivative 15 was detected21 during the later
stage of the reaction (Scheme 5). To further support this
conclusion, pure Kdo8P was synthesized enzymatically22
and was subjected to the same reaction conditions as for
the model 2 (Figure 1B). It is evident from the very
(18) Baasov, T.; J acob, A. J . Am. Chem. Soc. 1990, 112, 4972.
(19) It is noteworthy that during early stages of our investigation,
this procedure of monitoring the progress of the condensation reaction
by 1H-NMR (10 mM compound 2, 50 mM ZnCl2, 50 °C, pH 5, D2O)
was particularly troublesome. Initially, we could not discern any
formation of expected resonances of Kdo8P in the region of 1.7-2.5
ppm, although the starting material (2) was monotonously converted
to a mixture of products (as indicated by the disappearance of the
resonances in the region of 5.1-5.35 ppm). This was very difficult to
assign using the observed spectra. Upon close investigation we found
that, under the reaction conditions, C-3 protons of Kdo8P undergo rapid
exchange with the solvent (D2O) protons to form 3,3-dideuterio-Kdo8P.
This conclusion was further supported when the same reaction was
performed using authentic Kdo8P (10 mM, 50 mM ZnCl2, 50 °C, pH 5,
D2O), yielding a rapid exchange of geminal (C-3) protons (half-life of
10 min).
(20) The acid-catalyzed degradation of Kdo to form the furoic acid
derivatives, as well as a bicyclic structure similar to 14, has been
previously reported: McNicolas, P. A.; Batley, M.; Redmond, J . W.
Carbohydr. Res. 1987, 165, 17.
(21) The presence of furoic acid derivative 15 was determined by
characteristic resonances of the furan ring protons (6.45 ppm, J ) 3.3
Hz, and 6.89 ppm, J ) 3.3 Hz) in 1H-NMR of the reaction mixture and
by comparison of these results with previously reported data for a
similar structure.20
In order to devise appropriate conditions for the
examination of the intramolecular condensation process
in model compound 2, the following arguments were con-
sidered: (i) We anticipated that if 2 was to undergo a
stepwise condensation process similar to the previously
examined compound 1 (eq 1) at pHg5, the carboxylic acid
(22) Bednarski, M. D.; Crans, D. C.; DiCosimo, R.; Simon, E. S.;
Stein, P. D.; Whitesides, G. M. Tetrahedron Lett. 1988, 29, 427.

798 J . Org. Chem., Vol. 62, No. 4, 1997
Du et al.
two-dimensional (1H-1H) COSY analyses and by com-
parison with other known, similar structures.23
The half-life (t1/2) for the conversion of 2 to Kdo8P was
estimated to be about 5 h. In the production of the
mixture 13 and 14, the half-life for the dehydration of
Kdo8P was approximately 15 h. It is worth noting that
a conversion such as this, the formation of 13 and 14 from
Kdo8P, is largely catalyzed by Zn2+ ions. This is indi-
cated by the fact that under the same reaction conditions,
but without the inclusion of Zn2+, Kdo8P was predomi-
nantly stable. The constant 2:1 ratio of these two bicyclic
structures, observed during the entire course of the
reaction, suggests that both are formed from the same
reaction intermediate. This intermediate is most likely
the furanose structure of Kdo8P, which after dehydration
forms the stable oxocarbenium intermediate IV (Scheme
6). This intermediate cation may be captured intramo-
lecularly, either by the addition of a side chain C-6
hydroxyl, leading to the formation of bicyclic 13 (path
a), or by the addition of a C-7 hydroxyl, leading to the
formation of bicyclic 14 (path b). Alternatively, IV can
undergo the elimination of a second molecule of water to
form the stable, furoic acid derivative 15.
Th e In tr a m olecu la r Con d en sa tion Mech a n ism in
a Wa ter Solu tion . The observed results clearly dem-
onstrate that 2 undergoes intramolecular condensation
in a water solution to exclusively form the desired
structure of Kdo8P. As formation of the C-4 epimer of
Kdo8P was not detected among the reaction products, it
appears that this condensation process is highly ste-
reospecific. This was also demonstrated for the model 1
in organic solvents (see above). In order to further
investigate the diastereoselectivity of this reaction, the
same reaction (10 mM compound 2, pH ) 5, 50 °C) was
induced without the addition of ZnCl2, and its progress
1
F igu r e 1. Time course of H-NMR spectra showing (A) the
conversion of 2 to Kdo8P followed by production of bicyclic
products 13 and 14 and (B) the conversion of Kdo8P to a
mixture of 13 and 14. The experiments were performed in H2O
(pH ) 5, 50 °C) in the presence of ZnCl2 (50 mM) together
with compound 2 (10 mM, spectra A) or with Kdo8P (10 mM,
spectra B). Aliquots of the reaction mixture were withdrawn
at various time intervals and immediately passed through a
small (5 mL) column of Dowex 50W (H+ form) in order to
remove Zn2+ ions. The pH of the elute was adjusted to 3.8 (0.1
N NaOH) and rapidly concentrated under a high vacuum. The
residue was dissolved in D2O (0.5 mL), and the 1H-NMR
spectrum was recorded on a Bruker WH-400 (referenced to
HOD at 4.63 ppm).
1
was monitored by H-NMR. Under these conditions the
condensation process was largely retarded (the half-life
of the disappearance of 2 was estimated to be about 22
h). It was very interesting to find that in addition to the
hydrolytic degradation products of 2 (A5P and pyruvate,
both about 32% of total products), a mixture of Kdo8P
and 4-epi-Kdo8P24 was also detected (in a ratio of about
55:45, approximately 68% of total products). However,
the same reaction (10 mM compound 2, 50 °C, no ZnCl2
added) at pH ) 1.6 virtually abolished the condensation
process. After 60 h of incubation, only trace amounts of
Kdo8P and 4-epi-Kdo8P were obtained, alongside the
main products, A5P and pyruvatesthe hydrolytic deg-
radation products of 2.
similar time course obtained in this experiment that
Kdo8P is indeed unstable under the reaction conditions.
It suffers a further transformation to form the mixtures
13 and 14. In addition, the observed resonances in the
region of the geminal C-3 protons (1.7-2.5 ppm, Figure
1A) following 5 h of treatment of compound 2 are
identical to those of authentic Kdo8P (Figure 1B, the
spectrum at zero time). This indicates that at first the
model 2 undergoes intramolecular condensation between
the enolpyruvate double bond and aldehyde functions to
form Kdo8P. The bicyclic structures of 13 and 14 were
assigned using a combination of 1H NMR, 13C NMR, and
The above results clearly demonstrate the dual role of
Zn2+ ions in the condensation process of 2. The addition
of Zn2+ at pH ) 5 accelerates the reaction roughly 5-fold
(24) Purification of this mixture by ion-exchange chromatography
[AG1 × 8 (HCO3- form), eluting with a linear gradient of triethylam-
monium bicarbonate buffer (0-0.7 M, pH 7.5] resulted in a good
separation of A5P and pyruvate, but the Kdo8P and 4-epi-Kdo8P were
still obtained as the same mixture (by 1H-NMR) as in the crude
reaction. Attempts to use other buffer systems or another ion-exchange
resin (DEAE Sephadex) gave similar results. Nevertheless, the struc-
ture of 4-epi-Kdo8P was determined on the basis of coupling constants
of the major R-pyranose anomer in the 1H-NMR of the reaction mixture
and by comparison of the observed data with those previously reported
for a similar structure (Auge, C.; Bouxom, B.; Cavaye, B.; Gautheron,
C. Tetrahedron Lett. 1989, 30, 2217). The resonances of this anomer
in the region of C-3 protons were well separated from that of Kdo8P
anomers and had small H3-H4 couplings (3J ) 2.7 and 3.8 Hz) and
large geminal (H3a-H3e) couplings (2J ) 15.0 Hz), indicative of the
presence of an axial C-4 hydroxyl group. In addition, the close and
upfield signals of C-3 protons (1.70 and 2.06 ppm) were consistent with
the pyranose structure of 4-epi-Kdo8P.
(23) The structure of 13 was assigned on the basis of a very
characteristic broad doublet of H3a′ (bd, 2J ) 14.8 Hz, 3J e 1 Hz),
indicating the pseudoaxial orientation of this proton relative to H4,
with a dihedral angle of about 90°. The same couplings of H3a′ were
recorded for the bicyclic 5, whose structure was unambigously deter-
mined by single-crystal X-ray analysis.13 The structure of 14 was
confirmed by the characteristic coupling constants of H-3 and H-3′ in
the 1H-NMR and by comparison of its 1H- and 13C-NMR data to that
of a similar structure.20

Nonenzymatic Synthesis of Kdo8P
J . Org. Chem., Vol. 62, No. 4, 1997 799
Sch em e 5
Sch em e 6
stabilizing the cation intermediate 2i (eq 6). Inductive
stabilization such as this can alone bring about the
condensation process, even without the addition of a
Lewis acid (formation of a mixture of Kdo8P and 4-epi-
Kdo8P in the absence of ZnCl2). At pH ) 1.6, however,
the observed rate of retardation (as compared to that at
pH ) 5) most likely reflects the electron-withdrawing
effect of the neutral carboxylic acid, which would desta-
bilize the intermediate 2i. These results are in agree-
ment with the previously reported catalytic role of the
R-carboxylate group in the acid-catalyzed hydrolysis of
R-methoxyacrylate17a and Kdo 2-phosphates.17b They
strongly support, therefore, the idea of a stepwise mech-
anism for the condensation of 2 in a water solution, as
illustrated in eq 6.
In ter a ction of m od el 2 w ith Kd o8P Syn th a se. The
acyclic form of 2 bears a strong similarity to the proposed
situation (Scheme 1) whereby two substrates, A5P and
PEP, evolve into a ternary complex with the synthase.
It is, therefore, of considerable interest to verify whether
this acyclic form may be accepted by the enzyme as a
substrate. This would mean that the enzyme could
catalyze intramolecular condensation in 2 and generate
its natural product Kdo8P. Toward this goal, 2 was
incubated (0.1 M Tris-HCl buffer, pH 7.3, 37 °C) with
highly concentrated, homogeneous Kdo8P synthase. The
synthase was at a level of concentration 1000 times
greater than that normally introduced into an assay
experiment. Progress of the reaction was monitored by
31P-NMR over a 24-h period. However, no difference was
detected between the observed spectra and those of the
blank experiment (run parallel, under the same condi-
tions minus the enzyme). Neither a further increase in
enzyme concentration (up to 10 mg of enzyme per
and results in very high diastereoselectivity. The ob-
served erythro selectivity in aldose 2 at the carbon, with
the newly generated hydroxyl group (C-1), relative to the
hydroxyl group at C-2, is the same as that shown in the
SnCl4-assisted, intramolecular condensation in model 1.
Both are in accordance with the chelation-Cram model.25
As to whether the formation of new C-C and C-O
bonds in Kdo8P is a synchronous or stepwise process, the
results observed suggest that the intramolecular con-
densation of 2 in water is similar to that of model
compound 1 in an organic solvent (eq 1). Both follow a
stepwise process that involves the formation of a tran-
sient oxocarbenium intermediate (eqs 1 and 6). Such an
intermediate, in an organic solvent (5i, eq 1), carries an
R-linked carboxylic ester group. This causes destabiliza-
tion of the cation, due to its electron-withdrawing char-
acter. Therefore, in order to observe condensation using
1, we must employ a strong Lewis acid such as SnCl4.
The weaker acids, such as LiClO4 and ZnCl2, do not result
in any condensation products, even at higher tempera-
tures.26 In contrast to model 1, at pH ) 5 the carboxylic
acid in model 2 should be in its carboxylate form, thus
(25) (a) Cram, D. J .; Abd Elhafez, J . Am. Chem. Soc. 1952, 74, 5828.
(b) Reetz, M. T. Angew. Chem. Int. Ed. Engl. 1984, 23, 556.
(26) During the initial steps of our investigation with compound 1,
we screened other Lewis acids, in addition to SnCl4. These included
LiClO4 and ZnCl2, which performed in ether and in THF/H2O as
solvents, and contained up to 5 mol equivs of each, relative to 1. The
reaction temperature was varied from rt up to the reflux of the
corresponding solvent. No significant change was detected in the
starting material 1 in either case, even after 12 h reflux, as determined
by TLC.

800 J . Org. Chem., Vol. 62, No. 4, 1997
Du et al.
experiment) nor the addition of inorganic phosphate27
altered the results. No enzyme-catalyzed acceleration in
the condensation process was detected. Competition
experiments have revealed, however, that 2 is a competi-
tive inhibitor against A5P binding (Ki ) 200 ( 10 µM)
but is uncompetitive against PEP binding (Ki ) 10 ( 1
µM). The binding results indicate that 2 is a moderate
inhibitor of the enzyme, and inhibition patterns suggest
that 2 behaves as a simple A5P-based substrate-analog
inhibitor.28 This allows for the familiar, sequentially
ordered kinetic mechanism of the synthase (PEP binding
is followed by A5P binding).29 It also indicates that 2
specifically interacts with the A5P binding site but has
little, if any, interaction with the PEP binding site. The
observed inability of 2 to serve as a substrate further
supports this explanation.
In summary, two model compounds (1 and 2) were
prepared in order to test the hypothesis that catalysis
by proximity is a predominate mechanism in the initial
condensation step between A5P and PEP, catalyzed by
Kdo8P synthase. We have shown that installation of the
enolpyruvate moiety in precise proximity to the aldehyde
function leads to a rapid, Lewis acid-promoted addition
of the enolpyruvate double bond to the aldehyde carbonyl,
in an intramolecular manner.30 The condensation pro-
cess was highly diastereospecific at the newly formed
stereogenic center (C-4) and took place according to a
stepwise mechanism, via the participation of a transient
oxocarbenium ion intermediate. We have also shown
that the carboxylic acid of the enolpyruvate moiety
affords catalysis and is more effective when in the
carboxylate form than when unionized. This suggests
that the PEP may be recognized at the enzyme active
site, so as to maintain the ionized form of its carboxylate
group during catalysis. Our results are a unique dem-
onstration of enzyme-like chemistry, apparent in the
stereospecific synthesis of the Kdo system (in both
organic solvents and a water solution). This certainly
warrants further investigation in order to facilitate the
construction of other 3-deoxy-2-ulosonic acids and sialic
acids, based on the same general model as shown here
in the case of Kdo. Furthermore, the results support the
validity of the Kdo8P synthase mechanism in Scheme 1,
particularly, the possible role of the enzyme at the initial
condensation step and the feasibility of the formation of
transient intermediate I. Thus, the fact that the catalytic
groups of A5P and PEP are held at bonding distances,
in conjunction with minor, general acid catalysis, could
be sufficient to explain the high rate of acceleration
attributed to the initial condensation step in the Kdo8P
synthase-catalyzed reaction. A more detailed character-
ization of the initial steps of this enzyme catalysis, using
new models that will include the phosphate moiety of
PEP, is the subject of an ongoing study.
Exp er im en ta l Section
(27) The coaddition of inorganic phosphate (up to 5 mM) to the
reaction mixture containing compound 2 and the enzyme, was expected
to cause a noncovalent occupation of the site of the enzyme which is
naturally occupied by the phosphate of PEP and thus may promote
enzyme catalysis.
Gen er a l Meth od s. Kdo8P synthase (specific catalytic
activity 9 U/mg) was isolated from overproducing strain
Escherichia coli DH5R (pJ U1). The plasmid pJ U1, containing
the kdsA gene,32 was provided by Professor J . R. Knowles. The
purification procedure followed the protocol of Ray9 with some
modifications, as previously described.10b A5P and Kdo8P were
prepared enzymatically according to the procedure of White-
sides.22 Methyl R-methoxyacrylate and R-methoxyacrylate
were prepared by a previously reported procedure.31 All other
chemicals were received from Aldrich or from Sigma and used
without further purification, unless noted. In all the synthetic
work described, reactions were performed under dry argon
atmosphere unless otherwise noted. All solvents were dried
over standard drying agents33 and freshly distilled prior to use.
Flash column chromatography34 was performed on silica gel
60 (70-230 mesh). Reactions were monitored by TLC on silica
gel 60 F254 (0.25 mm, Merck) and detected by charring with a
yellow solution containing Na2MoO4‚4H2O (5 g) and cerric
ammonium nitrate (5 g) in 10% H2SO4 (300 mL). Soviet
spray35 solution was used for the detection or qualitative
analysis of the compounds containing phosphate monoesters.
Spectrophotometric measurements were made on a Hewlett-
Packard 8452A diode array spectrophotometer using 1-cm
pathlength cells with a thermostated cell holder and circulat-
ing water bath at desired temperature. Mass spectra were
obtained using a TSQ-70B mass spectrophotometer (Finnigan
Mat) under fast atom bombardment (FAB) in the glycerol
matrices.
(28) For
a similar example of EPSP synthase inhibition, see:
Marzabadi, M. R.; Gruys, K. J .; Pansegrau, P. D.; Walker, M. C.; Yuen,
H. K.; Sikorski, J . A. Biochemistry 1996, 35, 4199.
(29) Kohen, A.; J akob, A.; Baasov, T. Eur. J . Biochem. 1992, 208,
443.
(30) It is noteworthy that, in order to quantify the intramolecular
benefit of the condensation process in the models 1 and 2, we have
attempted to measure the effective molarity11 parameters of these
systems by comparison to the corresponding intermolecular reactions.
As an intramolecular counterpart of model 1, we used 3,5-bistosyl
arabinose 16 [prepared in two steps from 3: TsCl, pyridine, 90%;
HOAc/H2O/THF, 70 °C, 12 h, 70%]. It was reacted with methyl
R-methoxyacrylate31 under the same conditions (SnCl4, CH2Cl2, 0 °C).
There was no discernible formation of the expected condensation
product over a 24-h period, and when the reaction mixture was warmed
to room temperature (with the presence of 3 mol equiv of SnCl4), only
methoxy furanosides of the starting bistosylate (R-17 and â-17) were
isolated. In order to test the intermolecular counterpart of model 2,
D-arabinose was treated with R-methoxyacrylate [5 mol equiv ZnCl2,
pH 5.0] at two different temperatures: 50 and 90 °C. After 50 h, we
had not detected any possible condensation product in either reactions.
During this time the arabinose remained unchanged and R-methoxy-
acrylate decomposed to form pyruvic acid. Thus, although at this stage
of the investigation the observed results do not, unfortunately, allow
us to provide any numerical value to represent the advantage of our
intramolecular models (1 and 2) over their intermolecular counterparts,
the data clearly demonstrate the importance of the proximity factor
in the occurrence of condensation.
1,2-O -I s o p r o p y lid e n e -3-O -[1-(m e t h o x y c a r b o n y l)-
eth en yl]-5-O-(tolu en esu lfon yl)-â-D-a r a bin ofu r a n ose (4).
To a solution of tosylate 314 (5.2 g, 15 mmol) in dry benzene
(130 mL) was added rhodium tetraacetate (100 mg). The green
mixture was heated at reflux, and then dimethyl diazoma-
lonate (8.2 g, 30 mmol) was added dropwise for 30 min. This
(31) Ogata, N.; Nozakura, S.; Murahashi, S. Bull. Chem. Soc. J pn.
1970, 43, 2987.
(32) Woisetschlager, M.; Hogenauer, G. J .Bacteriol. 1986, 168, 437.
(33) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory
Chemicals, 3rd ed.; Pergamon Press: Oxford, 1988.
(34) Still, W. C.; Kahn, M; Mitra, A. J . Org. Chem. 1978, 43, 2923.
(35) Vaskovsky, V. E.; Latyshev, N. A. J . Chromatogr. 1975, 115,
246.

Nonenzymatic Synthesis of Kdo8P
J . Org. Chem., Vol. 62, No. 4, 1997 801
mixture was refluxed for 3.5 h, cooled, and then filtered
through Celite. After thorough washing of the Celite with
ethyl acetate, the washes were combined and evaporated to
afford 13.4 g of a brown oil. Purification by chromatography
on silica gel (EtOAc/hexane 1:1) afforded 5.8 g of the corre-
sponding malonate ester (80.6% yield): 1H-NMR (CDCl3, 400
MHz) δ 1.26 (s, 3H, isopropylidene protons), 1.35 (s, 3H,
isopropylidene protons), 2.42 (s, 3H, CH3 of Ts), 3.80 (s, 3H,
CO2CH3), 3.81 (s, 3H, CO2Me), 4.07 (d, 1H, J ) 2.6 Hz, H-3),
4.12 (dd, 1H, J ) 10.2, 5.8 Hz, H-5), 4.16 (dd, 1H, J ) 10.2,
6.3 Hz, H-5′), 4.24 (ddd, 1H, J ) 6.3, 5.8, 2.6 Hz, H-4), 4.62 (s,
1H, CHCO2Me), 4.65 (d, 1H, J ) 3.9 Hz, H-2), 5.83 (d, 1H, J
) 3.9 Hz, H-1), 7.32 (d, 2H, J ) 8.1 Hz, aromatic protons),
7.77 (d, 2H, J ) 8.1 Hz, aromatic protons); 13C-NMR (CDCl3,
50.3 MHz) δ 21.57 (Me of Ts), 26.11 and 26.79 (Me of
isopropylidene), 53.08 (CO2CH3), 68.07, 77.92, 81.81, 83.74,
84.81, 105.77 (C1), 113.18 (C(CH3)2), 128.05, 129.89, 132.70,
145.03 (aromatic carbons), 166.16 (CO2CH3), 166.22 (CO2CH3).
The above-mentioned malonate (5.8 g, 0.012 mol) was
dissolved in dry dichloromethane (80 mL), to which triethyl-
amine (2.8 mL) and dimethylmethyleneammonium iodide (2.8
g, 0.015 mol) were added, and the mixture was stirred at room
temperature for 2 h. The reaction mixture was diluted with
CH2Cl2, washed with water and saturated NaCl, dried, and
evaporated to give the corresponding Mannich base. This
material, without further purification, was dissolved in dry
dichloromethane (140 mL), and 15 mL of freshly distilled
methyl iodide was added. The reaction mixture was protected
from the light and stirred at room temperature for 24 h. The
mixture was then evaporated to dryness, and the residue
obtained was dissolved in freshly distilled DMSO (6.5 mL).
This mixture was heated at 100 °C for 30 min. The solvent
was evaporated under reduced pressure and the residue
dissolved in dichloromethane. The insoluble impurity (white
powder) was filtered off, the solution was evaporated, and the
compound was purified by chromatography on a silica gel
column (EtOAc/hexane, 1:1) to give pure product 4 (2.34 g, 36%
yield referring to compound 3): 1H-NMR (CDCl3, 400 MHz) δ
1.27 (s, 3H, isopropylidene protons), 1.41 (s, 3H, isopropylidene
protons), 2.42 (s, 3H, Me of Ts), 3.77 (s, 3H, CO2CH3), 4.12
(dd, 1H, J ) 10.1, 5.0 Hz, H-5), 4.23 (dd, 1H, J ) 10.1, 7.2 Hz,
H-5′), 4.31 (ddd, 1H, J ) 7.2, 5.0, 2.6 Hz, H-4), 4.37 (d, J ) 2.6
Hz, H-3), 4.58 (d, 1H, J ) 4.1 Hz, H-2), 4.81 (d, 1H, J ) 3.3
Hz, vinylic proton), 5.47 (d, 1H, J ) 3.3 Hz, vinylic proton),
5.86 (d, 1H, J ) 4.1 Hz, H-1), 7.32 (d, 2H, J ) 8.3 Hz, aromatic
protons), 7.77 (d, 2H, J ) 8.3 Hz, aromatic protons); 13C-NMR
(CDCl3, 50.3 MHz), δ 21.57 (Me of Ts), 26.25 and 26.65 (2Me
of isopropylidene), 52.41 (CO2CH3), 67.89, 80.92, 81.00, 83.95,
97.64 (â-vinylic carbon), 105.85 (C1), 113.18 (C(CH3)2), 128.02,
129.94, 132.64, 145.12 (aromatic carbons), 149.24 (R-vinylic
carbon), 162.86 (CO2Me); negative CIMS m/z 428.0 (M-,
C19H24O9S requires 428.45).
3-O-[1-(Meth oxyca r bon yl)eth en yl]-5-O-(tolu en esu lfo-
n yl)-D-a r a bin ofu r a n ose (1). A stirred solution of acetonide
4 (1.26 g, 2.95 mmol) in 60 mL of a mixture of acetic acid/
water/tetrahydrofuran (65:35:10) was heated at 70 °C for 12
h. The reaction mixture was diluted with EtOAc (60 mL),
washed with saturated Na2CO3, dried (MgSO4), and concen-
trated. The crude product was purified by column chroma-
tography (EtOAc/hexane, 2:1) to afford an anomeric mixture
of the target diol 1 as a clear oil (1.06 g, 93%): 1H-NMR (400M
Hz, CDCl3) data for â-1, δ 2.42 (s, 3H Me of Ts), 3.76 (s, 3H,
CO2CH3), 4.14-4.20 (m, 3H, H-2, H-5 and H-5′), 4.28-4.30
(m, 1H, H-4), 4.32 (bs, 1H, H-3), 4.91 (d, 1H, J ) 3.2 Hz, vinyl
proton), 5.41 (d, 1H, J ) 4.2 Hz, H-1), 5.50 (d, 1H, J ) 3.2 Hz,
vinyl proton), 7.33 (d, 2H, J ) 8.1 Hz, aromatic protons), 7.75
(d, 2H, J ) 8.1 Hz, aromatic protons); R-1, 2.42 (s, 3H, Me of
Ts), 3.76 (s, 3H, CO2CH3), 4.13 (dd, 1H, J ) 8.4, 4.8 Hz, H-5),
4.15-4.19 (m, 2H, H-2 and H-3), 4.26 (t, 1H, J ) 5.2 Hz, H-4),
4.51 (dd, 1H, J ) 8.4, 4.8 Hz H-5′), 4.97 (d, 1H, J ) 3.0 Hz,
vinyl proton), 5.27 (s, 1H, H-1), 5.46 (d, 1H, J ) 3.0 Hz, vinyl
proton), 7.33 (d, 2H, J ) 8.1 Hz, aromatic protons), 7.75 (d,
2H, J ) 8.1 Hz, aromatic protons); 13C-NMR (50.3 MHz, CDCl3)
data for the anomeric mixture of 1, δ 21.46 (Me of Ts), 52.38
and 52.55 (2CO2CH3), 68.77 and 69.33 (2C5), 74.95, 77.99,
78.57, 79.92, 82.95, 83.03, [97.31, 98.58, 98.63, 103.49, (2C1
and 2â-vinylic carbons)], [127.86, 129.89, 132.38, 132.42,
145.16, (aromatic carbons)], 149.96 and 149.29 (R-vinylic
carbons), 163.34 and 163.48 (2CO2CH3); CIMS m/z 389.0 (MH+,
C16H21O9S requires 389.4).
In tr a m olecu la r Con d en sa tion of Mod el 1. P r oced u r e
A. The anomeric mixture of 1 (88 mg 0.227 mmol) in CH2Cl2
(4 mL) was treated with stannic chloride (18.5 mg, 0.072 mmol)
at 0 °C. After 5 h at 0 °C, the reaction mixture was diluted
with CH2Cl2 (10 mL) and EtOAc (20 mL) and neutralized with
diluted NaHCO3. The aqueous layer was extracted with
EtOAc, and the combined organic layers were dried (MgSO4)
and concentrated. The silica gel chromatography of the crude
(EtOAc/hexane, 1.5:1) afforded the bicyclic 5 (60 mg, 68.2%)
along with unreacted diol 1 (13 mg, 14.8%). Data for 5: 1H
NMR (CDCl3, 400 MHz), δ 1.88 (d, 1H, J ) 13.5 Hz, H-3a′),
2.43 (s, 3H, CH3 of Ts), 2.48 (dd, 1H, J ) 13.5, 6.6 Hz, H-3e′),
3.51 (d, 1H, J ) 8.2 Hz, H-6), 3.66 (m, 1H, H-7), 3.83 (s, 3H,
CO2CH3), 4.04 (dd, 1H, J ) 10.8, 5.6 Hz, H-8), 4.19 (dd, 1H, J
) 10.8, 2.2 Hz, H-8′), 4.22 (d, 1H, J ) 6.6 Hz, H-4), 4.84 (s,
1H, H-5), 7.32 (d, 2H, J ) 8.1 Hz, aromatic protons), 7.75 (d,
2H, J ) 8.1 Hz, aromatic protons); Ha′ and He′ refer to the
geminal protons of the furanose anomers of Kdo that have
similar orientations to the axial and equatorial protons,
respectively, in the pyranose anomers; 13C-NMR (CDCl3, 50.3
MHz) δ 21.61 (Me of Ts), 46.46 (C3), 53.09 (CO2CH3), 69.08,
70.73, 71.42, 74.96, 84.05, 104.59 (C2), 128.03, 129.94, 132.40,
145.18, 165.48 (C1); CIMS m/z 389.2 (MH+, C16H21O9S requires
389.4).
P r oced u r e B. To a solution of diol 1 (84.3 mg, 0.217 mmol)
in CH2Cl2 (8 mL) at 0 °C was added a solution of SnCl4 (14
mg, 0.053 mmol) in CH2Cl2 (0.3 mL). The reaction mixture
was allowed to warm at room temperature, and the reaction
progress was monitored by TLC [silica gel, EtOAc/hexane (4:
1), the Rf of 1, 5, and 6 were 0.56, 0.35, and 0.12, respectively].
At the beginning of the reaction, mainly the bicyclic 5 was
formed. This was in time gradually transformed to ketose 6.
After approximately 8 h all of the starting material (1) was
gone, and a repetition of the workup and purification procedure
used in A afforded the bicyclic 5 (17.8 mg, 20.2%) and the
ketose 6 (57.4 mg, 62.3%). Data for 6: 1H NMR (CDCl3, 400
MHz) δ 1.86 (dd, J ) 12.8, 4.7 Hz, H-3e of R-Py), 2.12 (dd, J
) 14.2, 2.2 Hz, H-3e′ of R-Fu), 2.12 (dd, J ) 12.8, 11.4 Hz,
H-3a of R-Py), 2.37-2.39 (m, Me of Ts, and H-3′ of â-Fu), 2.46
(dd, J ) 13.9, 6.1 Hz, H-3′′ of â-Fu), 2.58 (dd, J ) 14.2, 6.7 Hz,
H-3a′ of R-Fu), 3.73 (CO2CH3), 3.75 (CO2CH3), 3.77 (CO2CH3),
3.92 (d, J ) 9.0 Hz), 3.97-4.31 (m), 7.29 (d, J ) 8.0 Hz,
aromatic protons), 7.73-7.76 (m, aromatic protons); CIMS m/z
407.0 (MH+, C16H23O10S requires 407.4).
Met h yl [2-Met h oxy-8-O-(t olu en esu lfon yl)-3-d eoxy-D-
ma n n o-2-octu lofu r an osid]on ate (7). p-Toluenesulfonic acid
(20 mg, 0.1 mmol) was added to a solution of bicyclic 5 (26.5
mg, 0.068 mmol) in dry MeOH (2 mL). After being stirred for
12 h at room temperature, the reaction mixture was diluted
with EtOAc, washed with saturated Na2CO3 solution, dried
over MgSO4, and concentrated under a vacuum. Chromatog-
raphy of the oily residue (EtOAc/hexane, 2:1) furnished 7 (26.4
mg, 92%) as mixture of R- and â-anomers [Rf 0.29, EtOAc/
hexane (4:1)]. This anomeric mixture could only be separated
using preparative TLC plates (silica gel, 5% MeOH/CHCl3, four
runs). Data for R-7: 1H NMR (CDCl3, 400 MHz) δ 2.23 (dd,
1H, J ) 14.7, 1.2 Hz, H-3a′), 2.41 (dd, 1H, J ) 14.7, 6.6 Hz,
H-3e′), 2.44 (s, 3H, CH3 of Ts), 3.33 (s, 3H, OMe), 3.56-3.62
(m, 1H, H-6), 3.80 (s, 3H, CO2CH3), 3.82-3.88 (m, 1H, H-7),
4.18 (dd, 1H, J ) 10.4, 6.1 Hz, H-8), 4.32 (dd, 1H, J ) 10.4,
3.3 Hz, H-8′), 4.34 (d, 1H, J ) 6.5 Hz, H-4), 4.49 (s, 1H, H-5),
7.34 (d, 2H, J ) 8.1 Hz, aromatic protons), 7.79 (d, 2H, J )
8.1 Hz, aromatic protons); 13C-NMR (CDCl3, 50.3 MHz) δ 21.63
(Me of Ts), 45.72 (C3), 51.81 and 53.13 (OCH3 and CO2CH3),
70.65, 71.09, 71.64,73.95, 88.73, 106.98 (C2), 128.04, 129.97,
132.67, 145.15 (aromatic carbons), 170.08 (C1). Data for â-7:
1H NMR(CDCl3, 400 MHz) δ 2.32 (dd, 1H, J ) 13.6, 6.8 Hz,
H-3a′), 2.54 (dd, 1H, J ) 13.6, 7.0 Hz, H-3e′), 2.44 (s, 3H, CH3
of Ts), 3.29 (s, 3H, OMe), 3.55-3.61 (m, 1H, H-6), 3.81 (s, 3H,
CO2CH3), 3.82-3.85 (m, 1H, H-7), 4.19 (dd, 1H, J ) 10.5, 6.1
Hz, H-8), 4.26 (t, 1H, J ) 5.0 Hz, H-5), 4.34 (dd, 1H, J ) 10.5,
2.0 Hz, H-8′), 4.49 (ddd, 1H, J ) 7.0, 6.8, 5.0 Hz, H-4), 7.34 (d,

802 J . Org. Chem., Vol. 62, No. 4, 1997
Du et al.
2H, J ) 8.1 Hz, aromatic protons), 7.79 (d, 2H, J ) 8.1 Hz,
aromatic protons); 13C-NMR (CDCl3, 50.3 MHz) δ 21.64 (Me
of Ts), 44.58 (C3), 51.95, 52.86 (OCH3 and CO2CH3)), 71.09,
71.75, 72.04, 88.73 (C4, C5, C6, C7, C8), 106.05 (C2), 128.04,
130.00, 132.77, 145.24 (aromatic carbons), 169.31 (C1); CIMS
m/z 421.0 (MH+, C17H25O10S requires 421.4).
After 24 h, an additional amount of SnCl4 (50 mg, 0.19 mmol)
was added, and the reaction was continued until all the
starting material had been consumed (total 5 days). After the
usual workup, as in the original procedure, and subsequent
chromatography on a silica gel column [EtOAc/hexane (2:1)],
8 (18.4 mg, 60.6%), R-9 (11.0 mg, 34.8%), and trace amounts
of bicylic 5 and furanosides 7 were also isolated.
Qu en ch in g Exp er im en ts w ith MeOH. (i) In tr a m o-
lecu la r Con d en sa tion of 1 in th e P r esen ce of 2 Mol
Equ iv MeOH. To a solution of 1 (54.5 mg, 0.14 mmol) in dry
CH2Cl2 (4.5 mL) was added freshly distilled dry MeOH (9 mg,
0.28 mmol) in CH2Cl2 (0.5 mL). The mixture was cooled at 0
°C, and SnCl4 (9.2 mg, 0.037 mmol) in CH2Cl2 (0.5 mL) was
introduced dropwise. The reaction progress was monitored by
TLC [silica gel, EtOAc/hexane (4:1), Rf 0.56 (1), 0.35 (5), 0.12
(6), 0.29 (R-7 + â-7), 0.70 (R-8), 0.77 (â-8), 0.12 (R-9), and 0.18
(â-9)], and the mixture was stirred at room temperature until
all the starting material had been consumed (10 h). After an
usual workup, as in the original procedure, the 1H-NMR
spectrum of the crude showed the following distribution of
products (as was determined by the integration of the char-
acteristic protons of each isomer): 5 (26%), 6 (32%), 7 (5%), 8
(6.3%), 9 (18%). This mixture was completely separated by
twice repeating the chromatography on silica gel columns with
EtOAc/hexane (2:1) as an eluent, in conjunction with one
preparative TLC (silica gel, 5% MeOH/CHCl3). Data for R-8:
1H-NMR (CDCl3, 400 MHz) δ 2.43 (s, 3H, Me of Ts), 3.32 (s,
3H, OMe), 3.76 (s, 3H, CO2CH3), 4.09-4.13 (m, 2H, H-2, H-5),
4.21 (dd, 1H, J ) 11.2, 3.3 Hz, H-5′), 4.37-4.40 (m, 2H, H-3,
H-4), 4.85 (s, 1H, H-1), 4.88 (d, 1H, J ) 2.6 Hz, vinyl proton),
5.46 (d, 1H, J ) 2.6 Hz, vinyl proton), 7.33 (dd, 2H, J ) 8.1
Hz, aromatic protons), 7.78 (dd, 2H, J ) 8.1 Hz, aromatic
protons); 13C-NMR (CDCl3, 50.3 MHz) δ 21.62 (Me of Ts), 52.39
(CO2CH3), 54.95 (OMe), 68.56 (C5), 78.97, 79.66, 84.05, 98.96
(â-vinylic carbon), 109.37 (C1), 128.00, 129.94, 132.94, 145.12
(aromatic carbons), 150.24 (R-vinylic carbon), 163.28 (CO2CH3).
1,2-O-Isopr opyliden e-5-[bis(2,2,2-tr ich lor oeth oxy)ph os-
p h or yl]-â-D-a r a bin ofu r a n ose (11). To a solution of the diol
1014
(45.9 mg, 0.241 mmol) in dry pyridine (1 mL) was added
bis(2,2,2-trichloroethyl)phosphochloridate (92 mg, 0.241 mmol)
in 20 mg portions every 1 h. The mixture was stirred at room
temperature, and progress was monitored by TLC. After more
than 90% of the starting material had been consumed (5 h),
the solvent was removed by evaporation under a high vacuum
and the residue was purified on a silica gel column [EtOAc/
hexane (2:1)] to afford 11 (90 mg, 70%) as an oil:
1H-NMR
(CDCl3, 200 MHz) δ 1.31 (s, 3H, isopropylidene protons), 1.52
(s, 3H, isopropylidene protons), 4.1-4.4 (m, 4H, H-3, H-4, H-5,
H-5′), 4.56 (d, 1H, J ) 3.79 Hz, H-2), 4.63 (d, 2H, J ) 1.49 Hz,
CCl3
CH ), 4.66 (d, 2H, J ) 1.40 Hz, CCl CH ), 5.93 (d, 1H, J
2
3
2
) 3.79 Hz, H-1); proton-decoupled 31P-NMR (CDCl3, 81.03
MHz) δ -6.32 (s); CIMS m/z 530.6 (MH+, C12H18O8Cl6P
requires 531.2).
1,2-O -I s o p r o p y lid e n e -3-O -[1-(m e t h o x y c a r b o n y l)-
et h en yl]-5-[b is(2,2,2-t r ich lor oet h oxy)p h osp h or yl]-â-D-
a r a bin ofu r a n ose (12). To a solution of 11 (1.4 g, 2.63 mmol)
in dry benzene (50 mL)was added rhodium tetraacetate (18.5
mg). The green mixture was heated at reflux, and then
dimethyl diazomalonate (2.6 g, 18.05 mmol) was added drop-
wise for 30 min. This mixture was refluxed for 3 h and then
cooled and filtered through Celite. After thorough washing
of Celite with ethyl acetate, the washes were combined and
evaporated. Purification by chromatography on silica gel
(EtOAc/hexane, 2:1) afforded 1.78 g (54% yield) of the corre-
sponding malonate ester: 1H-NMR (CDCl3, 200 MHz) δ 1.31
(s, 3H, isopropylidene protons), 1.51 (s, 3H, isopropylidene
protons), 3.80 (s, 6H, CO2CH3), 4.07 (d, 1H, J ) 2.6 Hz, H-3),
4.0-4.49 (m, 4H, H-3, H-4, and 2H-5), 4.63-4.70 (m, 5H, H-2,
2CCl3CH2), 5.90 (d, 1H, J ) 3.91 Hz, H-1); proton-decoupled
31P-NMR (CDCl3, 81.03 MHz) δ -4.24 (s).
The above malonate (1.64 g, 2.47 mmol) was dissolved in
dry CH2Cl2 (30 mL) to which triethylamine (0.5 mL) and
dimethylmethyleneammonium iodide (0.55 g, 2.97 mmol) were
added, and the mixture was stirred at room temperature for
3 h. The reaction mixture was diluted with CH2Cl2 (60 mL)
and washed with 10% Na2CO3. The organic fraction was then
dried and evaporated to give the corresponding Mannich base
(1.66 g, 93%) as an oil: 1H-NMR (CDCl3, 200 MHz), δ 1.26 (s,
3H, isopropylidene protons), 1.49 (s, 3H, isopropylidene pro-
tons), 2.24 [s, 6H, N(CH3)2], 2.82-2.97 (m, 2H, CH2N), 3.76
(s, 3H, CO2CH3), 3.78 (s, 3H, CO2CH3), 4.37-4.41 (m, 4H, H-3,
H-4 and 2H-5), 4.58 (d, 1H, J ) 3.84 Hz, H-2), 4.65 (m, 4H,
2CCl3CH2), 5.86 (d, 1H, J ) 3.76 Hz, H-1); proton-decoupled
31P-NMR (CDCl3, 81.03 MHz) δ -4.24 (s).
This Mannich base was not further purified but was
dissolved in dry dichloromethane (20 mL), and 9 mL of freshly
distilled methyliodide was added. The reaction mixture was
protected from the light and stirred at room temperature for
24 h. The mixture was evaporated to dryness, and the residue
obtained was dissolved in freshly distilled DMSO (3.0 mL).
This mixture was heated at 100 °C for 4 h. The solvent was
evaporated under reduced pressure, and the residue was
purified by chromatography on a silica gel column (EtOAc/
hexane 1:1) to give 0.96 g of pure 12 (34% yield referring to
compound 11): 1H-NMR (CDCl3, 200 MHz) δ 1.30 (s, 3H,
isopropylidene protons), 1.53 (s, 3H, isopropylidene protons),
3.77 (s, 3H, CO2CH3), 4.38-4.39 (m, 4H, H-3, H-4, H-5, H-5′),
4.60-4.63 (m, 5H, H-2, 2CCl3CH2), 4.87 (d, J ) 3.1 Hz, vinylic
proton), 5.50 (d, J ) 3.1 Hz, vinylic proton), 5.90 (d, 1H, J )
3.9 Hz, H-1); 13C-NMR (CDCl3, 50.3 MHz) δ 26.25 (CH3 of
isopropylidene), 27.14 (CH3 of isopropylidene), 52.55 (CH3 of
CO2CH3), 60.38 (CCl3), 67.36 (d, J ) 5.2 Hz, C5), 77.23 (bs,
CH2-CCl3), 80.91 (C2), 81.51 (d, J ) 7.4 Hz, C4), 83.99 (C3),
97.79 (â-vinylic carbon), 105.76 (C1), 113.49 (C(CH3)2), 149.28
(R-vinylic carbon), 162.81 (CO2CH3); proton-decoupled 31P-
1
Data for â-8: H-NMR (CDCl3, 400 MHz) δ 2.43 (s, 3H, Me of
Ts), 3.39 (s, 3H, OMe), 3.76 (s, 3H, CO2CH3), 4.08 (dd, 1H, J
) 10.1, 6.0 Hz, H-5), 4.14 (dd, 1H, J ) 10.1 and 5.4 Hz, H-5′),
4.21-4.26 (m, 2H, H-2 and H-4), 4.29 (dd, 1H, J ) 4.0, 3.6
Hz, H-3), 4.93 (d, 1H, J ) 4.7 Hz, H-1), 5.04 (d, 1H, J ) 2.6
Hz, vinyl proton), 5.48 (d, 1H, J ) 2.6 Hz, vinyl proton), 7.33
(dd, 2H, J ) 8.1 Hz, aromatic protons), 7.78 (dd, 2H, J ) 8.1
Hz, aromatic protons); 13C-NMR (CDCl3, 50.3 MHz) δ 21.61
(Me-Ts), 52.37 (CO2CH3), 54.90 (OMe), 69.38 (C5), 76.75, 79.15,
83.95, 99.40 (â-vinylic carbon), 103.19 (C1), 127.98, 129.91,
132.86, 145.04 (aromatic carbons), 149.77 (R-vinyl carbon),
163.28 (CO2CH3); CIMS m/z 403.0 (MH+, C17H23O9S requires
403.4).
Data for R-9: 1H-NMR (CDCl3, 500 MHz) δ 1.83 (dd, 1H, J 1
) J 2 ) 12.5 Hz, H-3a), 2.06 (dd, 1H, J ) 12.5, 5.0 Hz, H-3e),
2.44 (s, 3H, Me of Ts), 3.13 (s, 3H, OMe), 3.61 (d, 1H, J ) 8.5
Hz, H-6), 3.77 (s, 3H, CO2CH3), 4.00 (dd, 1H, J ) 12.5, 5.0 Hz,
H-4), 4.02 (bs, 1H, H-5), 4.18 (ddd, 1H, J ) 8.5, 4.5, 4.5 Hz,
H-7), 4.26-4.27 (m, 2H, H-8 and H-8′), 7.34 (d, 2H, J ) 8.0
Hz, aromatic protons), 7.79 (d, 2H, J ) 8.0 Hz, aromatic
protons); 13C-NMR (CDCl3, 50 MHz) δ 21.63 (Me of Ts), 34.61
(C-3), 51.30 (OMe), 52.74 (CO2CH3), 65.69 (C-8), 66.25 (C-4),
67.80 (C-5), 70.78, 70.96, 99.42 (C-2), 128.00, 129.97, 132.65,
145.15 (aromatic carbons), 168.87 (C-1). Data for â-9: 1H-NMR
(CDCl3, 400 MHz) δ 1.89 (dd, 1H, J 1 ) J 2 ) 12.5 Hz, H-3a),
2.37 (dd, 1H, J ) 12.5, 4.6 Hz, H-3e), 2.43 (s, 3H, Me of Ts),
3.24 (s, 3H, OMe), 3.62 (d, 1H, J ) 8.3 Hz, H-6), 3.67-3.70
(m, 1H, H-4), 3.79 (s, 3H, CO2CH3), 3.99 (bs, 1H, H-5), 4.15
(dd, 1H, J ) 8.3, 6.0 Hz, H-7), 4.19 (dd, 1H, J ) 10.0, 6.0 Hz,
H-8), 4.31 (d, 1H, J )10.0 Hz, H-8′), 7.33 (d, 2H, J ) 8.0 Hz,
aromatic protons), 7.80 (d, 2H, J ) 8.0 Hz, aromatic protons);
13C-NMR (CDCl3, 50.3 MHz) δ 21.61 (Me of Ts), 34.72 (C-3),
51.64 (OMe), 52.55 (CO2CH3), 65.89 (C8), 66.99 (C5), 68.43
(C4), 71.61, 73.56, 99.62 (C2), 128.00, 129.91, 132.76, 145.10
(aromatic carbons), 168.66 (C1); FAB mass spectrum m/z 420.9
(MH+, C17H25O10S requires 421.4).
(ii) In tr a m olecu la r Con d en sa tion of 1 in MeOH. The
diol 1 (29.3 mg, 0.076 mmol) was dissolved in freshly distilled
dry MeOH (2 mL). SnCl4 (5.0 mg, 0.019 mmol) was was then
added dropwise at 0 °C. The mixture was stirred at room
temperature, and the reaction progress was monitored by TLC.

Nonenzymatic Synthesis of Kdo8P
J . Org. Chem., Vol. 62, No. 4, 1997 803
NMR (CDCl3, 81.03 MHz), δ -4.17 (s); negative CIMS m/z
613.0 [(M - H)-, C16H20O10Cl6P requires 613.3).
(dd, 1H, J ) 3.4, 1.5 Hz, H-6), 4.52 (dd, 1H, J ) 4.7, 1.5 Hz,
H-5), 4.80 (dd, 1H, J ) 6.5, 4.7 Hz, H-4); 13C-NMR (D2O, 100.6
MHz) δ 46.20 (C3), 68.20 (C8), 79.06 (C6), 85.60 (C4), 89.10
(C7), 92.72 (C5), 109.16 (C2), 178.43 (C1). FAB mass spectrum
for the mixture 13 and 14 m/z 415.0 (MH+, C8H11O10PK3
requires 415.4). The furoic acid derivative 15 was identified
by the presence of the following characteristic resonances: 1H-
NMR (D2O, 400 MHz) δ 6.45 (d, 1H, J ) 3.3 Hz, H-4), 6.89 (d,
1H, J ) 3.3 Hz, H-3).
A time course of the ZnCl2-catalyzed dehydration of Kdo8P
was performed in the same manner as described for 2, which
provided the same mixture of the bicyclic 13 and 14 (in the
ratio of 2:1, respectively), along with a trace amount of furoic
acid derivative 15.
3-O-[1-(Met h oxyca r b on yl)et h en yl]-5-O-p h osp h on o-D-
a r a bin ofu r a n ose (2r a n d 2â). To a solution of 12 (198.6
mg, 0.322 mmol) in dry MeOH (5 mL) were added activated
Zn powder (0.4 g) and acetic acid (1 mL). The mixture was
stirred at room temperature for 2 h and then filtered through
Celite. After thorough washing of Celite with MeOH, the
washes were combined and evaporated to dryness. The
observed residue was dissolved in D2O (5 mL), and sufficient
Dowex 50W (H+ form) resin (which had been thoroughly
washed with D2O) was added with stirring to bring the solution
to pH ) 1. Progress of the reaction was monitored by 1H-NMR
(the disappearance of isopropylidene methyl resonances), and
after 24 h the mixture was filtered. The pH of the resulting
filtrate was adjusted to 14 with KOH (1 N), and the resulting
base solution was stirred at room temperature for 2 h. After
this time, the pH was adjusted to 7.3 with Dowex 50W resin
(H+ form), the solution was filtered, and the product was
purified through ion-exchange chromatography on AG1 × 8
The experiments carried out without the presence of ZnCl2
took place as follows: Compound 2 (52 mg, 0.125 mmol) was
dissolved in water (1.3 mL), and the pH was adjusted to 5.0
with diluted H2SO4. The solution was heated at 50 °C. The
1
reaction progress was monitored using H-NMR by removing
aliquots (0.3 mL) at various time intervals, concentrating the
solution under high vacuum, and then dissolving the residue
in D2O (0.4 mL). Analysis of the observed 1H-NMR spectra
revealed that the reaction mixture contaned Kdo8P, 4-epi-
Kdo8P, A5P, and pyruvate.
-
(HCO3 form), eluting with a linear gradient of triethylam-
monium bicarbonate buffer (0-0.7 M, pH ) 7.5). Fractions
were analyzed for inorganic phosphate after digestion with
HClO4. The active fractions were then combined and concen-
trated until dry. The residue was dissolved in water, passed
through a column of Dowex 50W (K+ form), and concentrated
again by lyophilization to give the highly purified phosphate
2 (85 mg, 89%) as a mixture of anomers. Data for â-2: 1H-
NMR (potassium salt, pD ) 7.1, D2O, 400 MHz) δ 3.70-3.82
(m, 2H, H-5, H-5′), 3.99-4.01 (m, 2H, H-2 and H-3), 4.35-
4.37 (m, 1H, H-4), 4.58 (d, 1H, J ) 2.90 Hz, vinylic proton),
5.03 (d, 1H, J ) 2.90 Hz, vinylic proton), 5.23 (d, 1H, J ) 3.9
Hz, H-1); 13C-NMR (D2O, 100.6 MHz) δ 67.05 (C5), 76.51 (C2),
83.94 (C3), 82.50 (d, J ) 7.85 Hz, C4), 96.10 (â-vinylic carbon),
99.63 (C1), 162.86 (R-vinylic carbon), 172.79 (CO2-). Data for
R-2: 1H-NMR (potassium salt, pD ) 7.1, D2O, 400 MHz) δ
3.70-3.82 (m, 2H, H-5, H-5′), 3.98 (s, 1H, H-2), 4.20 (d, 1H, J
) 3.29 Hz, H-3), 4.35-4.37 (m, 1H, H-4), 4.52 (d, 1H, J ) 2.90
Hz, vinylic proton), 5.05 (d, 1H, J ) 2.90 Hz, vinylic proton),
5.20 (s, 1H, H-1); 13C-NMR (D2O, 100.6 MHz) δ 66.51 (C5),
80.74 (C2), 84.94 (C3), 85.40 (d, J ) 8.95 Hz, C4), 96.10 (â-
vinylic carbon), 104.88 (C1), 156.35(R-vinylic carbon), 162.86
(CO2-). Proton-decoupled 31P-NMR (D2O, 162 MHz) δ 4.52 (s);
proton-coupled 31P-NMR (D2O, 162 MHz) δ 4.47-4.35 (m); FAB
mass spectrum m/z 415.3 (MH+, C8H11O10PK3 requires 415.4).
In tr a m olecu la r Con d en sa tion of Mod el 2. In a typical
condensation experiment, the phosphate 2 (52 mg, 0.125 mmol)
was dissolved in water (1.3 mL), and ZnCl2 (80 mg, 0.592
mmol, in 80 µL of H2O) was added at room temperature. The
pH of this solution was carefully adjusted to pH 5.0 with
diluted H2SO4. The mixture was heated to 50 °C, and the
In ter m olecu la r Con d en sa tion Exp er im en ts. As an
intermolecular counterpart for the model 1 we used 3,5-bis-
(toluenesulfonyl)-D-arabinofuranose (16), which was prepared
from 3 by following procedure. To a solution of acetonide 3
(511 mg, 1.49 mmol) in dry pyridine (2.5 mL) at 0 °C were
added, p-toluenesulfonyl chloride (730 mg, 2.97 mmol) and
4-(dimethylamino)pyridine (50 mg). After the mixture was
stirred at room temperature for 2 days, it was diluted with
EtOAc (10 mL), washed with diluted H2SO4 and saturated
NaCl, dried (MgSO4), and concentrated. Column chromatog-
raphy (EtOAc/hexane 1:1) afforded the target bistosyl ac-
etonide (629.6 mg, 92%, based on unreacted starting material
recovered) along with unreacted acetonide 3 (37.5 mg, 7.3%):
1H-NMR (CDCl3, 400 MHz) δ 1.22 (s, 3H, isopropylydene
protons), 1.31 (s, 3H, isopropylydene protons), 2.42 (s, 3H, Me
of Ts), 2.45 (s, 3H, Me of Ts), 3.96 (dd, 1H, J ) 10.1, 5.8 Hz,
H-5), 4.01 (dd, 1H, J ) 10.1, 7.6 Hz, H-5′), 4.14 (ddd, 1H, J )
7.6, 5.8, 1.5 Hz, H-4), 4.65 (d, 1H, J ) 3.8 Hz, H-2), 4.74 (d,
1H, J ) 1.5 Hz, H-3), 5.83 (d, 1H, J ) 3.8 Hz, H-1), 7.32 (d,
2H, J ) 8.2 Hz, aromatic protons), 7.35 (d, 2H, J ) 8.2 Hz,
aromatic protons), 7.71 (d, 2H, J ) 8.2 Hz, aromatic protons),
7.77 (d, 2H, J ) 8.2 Hz, aromatic protons); 13C-NMR (CDCl3,
100.6 M Hz) δ 21.62 (Me of Ts), 21.71 (Me of Ts), 25.71 (Me of
isopropylidene), 26.35 (Me of isopropylidene), 67.32 (C-5),
81.79, 81.87, 84.25, 105.81 (C-1), 113.19 (C(CH3)2), 127.98,
129.90, 130.23, 132.26, 145.14, 145.78 (aromatic carbons). The
bistosyl acetonide from the above (300 mg, 0.60 mmol) was
dissolved in a mixture of acetic acid/water/tetrahydrofuran (8:
2:1, 20 mL). After the mixture was stirred at 70 °C for 12 h,
and subjected to the same workup and column chromatogra-
phy as for the diol 1, the bistosyl 16 was obtained as a clear
oil (192 mg, 70%). Data for R-16: 1H-NMR (CDCl3, 400 MHz)
δ 2.42 (s, 3H, Me of Ts), 4.05-4.09 (m, 2H), 4.18-4.21 (m, 1H),
4.36 (dd, 1H, J ) 8.5, 4.3 Hz, H-5), 4.51 (dd, 1H, J ) 4.8, 2.3
Hz), 5.25 (s, 1H, H-1), 7.31-7.34 (m, 4H, aromatic protons),
7.70-7.76 (m, 4H, aromatic protons). Data for â-16: 1H-NMR
(CDCl3, 400 MHz) δ 2.41 (s, 3H, Me of Ts), 4.01 (dd, 1H, J )
11.0, 4.4 Hz, H-5), 4.05-4.09 (m, 1H, H-2), 4.11 (dd, 1H, J )
11.0, 3.6 Hz, H-5′), 4.18-4.21 (m, 1H, H-4), 4.64 (t, 1H, J )
4.3 Hz, H-3), 5.28 (d, J ) 4.3 Hz, H-1), 7.31-7.34 (m, 4H,
aromatic protons), 7.70-7.76 (m, 4H, aromatic protons); 13C-
NMR data for the anomeric mixture of 16 (CDCl3, 100.6 MHz)
δ 21.67 (2Me of Ts), 67.90 (C-5), 69.06 (C-5′), 75.49, 77.70,
79.27, 79.98, 83.61, 96.20 (C-1), 102.33 (C-1′), 127.90, 128.01,
129.90, 130.11, 130.17, 132.12, 132.29, 145.22, 145.70 (aro-
matic carbons).
1
reaction progress was monitored by H-NMR in the following
manner: Aliquots (0.3 mL) of the reaction mixture were
withdrawn at various time intervals and immediately passed
through a small (5 mL) column of Dowex 50W (H+ form) in
order to remove Zn2+ ions. The column was then washed with
distilled water (15 mL), and the fractions were analyzed for
inorganic phosphate after digestion with HClO4. The active
fractions were combined and the pH carefully adjusted to 3.8
with NaOH (0.1 N). The solution was then rapidly concen-
trated under a high vacuum. The residue was dissolved in
D2O (0.4 mL), and the 1H-NMR spectrum was recorded.
Analysis of the observed 1H-NMR spectra revealed, in the
presence of Kdo8P, bicyclic structures 13 and 14 and a trace
amount of furoic acid derivative 15. For the quantitative
production of bicyclic structures 13 and 14 alone, the experi-
ment was performed as above but the incubation at 50 °C was
continued for 30 h. Data of 13: 1H-NMR (D2O, 400 MHz) δ
2.10 (bd, 1H, J ) 14.8 Hz, H-3a′), 2.26 (dd, 1H, J ) 14.8, 5.5
Hz, H-3e′), 3.77-3.98 (m, 3H, H-7, H-8, H-8′), 4.10 (dd, 1H, J
) 6.5, 1.8 Hz, H-6), 4.54 (dd, 1H, J ) 4.8, 1.8 Hz, H-5), 4.73
(dd, 1H, J ) 4.8, 5.5 Hz, H-4); 13C-NMR (D2O, 100.6 MHz) δ
45.12 (C3), 67.83 (C8), 85.60 (C4), 87.87 (C7), 80.44 (C6), 94.94
(C5), 108.55 (C2), 179.06 (C1). Data of 14: 1H-NMR (D2O, 400
MHz) δ 2.19 (dd, 1H, J ) 14.7, 6.5 Hz, H-3a′), 2.31 (dd, 1H, J
) 14.7, 2.3 Hz, H-3e′), 3.77-3.98 (m, 3H, H-7, H-8, H-8′), 4.11
In ter m olecu la r Con d en sa tion in Or ga n ic Solven t. To
a solution of bistosyl 16 (34.8 mg, 0.076 mmol) and methyl
R-methoxyacrylate31 (8.8 mg, 0.076 mmol) in CH2Cl2 (3 mL)
at 0 °C was added SnCl4 (5.0 mg, 0.019 mmol, in 0.3 mL of
CH2Cl2). The mixture was stirred at room temperature for 2
days. Using the same workup and purification as for intramo-

804 J . Org. Chem., Vol. 62, No. 4, 1997
Du et al.
lecular condensation (procedure A), pure R- and â-methoxy
furanosides of 17 (23.3 mg, 65%) along with unreacted 16 (10.5
mg, 30%) were afforded. Data for R-17: 1H-NMR (CDCl3, 500
MHz) δ 2.45 (s, 3H, Me of Ts), 2.47 (s, 3H, Me of Ts), 2.73 (d
, 1H, J ) 6.5 Hz, OH-2), 3.31 (s, 3H, OMe), 4.10 (dd, 1H, J )
11.5, 4.0 Hz, H-5), 4.16 (dd, 1H, J ) 11.5, 3.0 Hz, H-5′), 4.18
(dd, 1H, J ) 6.5, 2.5 Hz, H-2), 4.23 (ddd, 1H, J ) 5.5, 3.0, 4.0
Hz, H-4), 4.45 (dd, 1H, J ) 5.5, 2.5 Hz, H-3), 4.79 (s, 1H, H-1),
7.35 (d, 2H, J ) 8.0 Hz, aromatic protons), 7.37 (d, 2H, J )
8.5 Hz, aromatic protons), 7.74 (d, 2H, J ) 8.5 Hz, aromatic
protons), 7.78 (d, 2H, J ) 8.0 Hz, aromatic protons); 13C-NMR
(CDCl3, 100.6 MHz) δ 21.71 (Me of Ts), 21.77 (Me of Ts), 55.21
(OMe), 67.54 (C-5), 78.76, 79.98, 83.76, 108.53 (C-1), 127.96,
128.05, 129.90, 130.20, 132.17, 132.54, 145.09, 145.79 (aro-
matic carbons). Data for â-17: 1H-NMR (CDCl3, 500 MHz) δ
2.46 (s, 6H, Me of Ts), 2.58 (d, 1H, J ) 8.5 Hz, OH-2), 3.34 (s,
3H, OMe), 4.02 (dd, 1H, J ) 10.5, 6.0 Hz, H-5), 4.09 (dd, 1H,
J ) 10.5, 4.0 Hz, H-5′), 4.18 (ddd, 1H, J ) 6.0, 6.0, 4.0 Hz,
H-4), 4.26 (ddd, 1H, J ) 8.5, 6.0, 4.5 Hz, H-2), 4.58 (t, 1H, J )
6.0 Hz, H-3), 4.81 (d, 1H, J ) 4.5 Hz, H-1), 7.36 (dd, 4H, J )
8.0 Hz, aromatic protons), 7.78 (dd, 4H, J ) 8.0 Hz, aromatic
protons); 13C-NMR (CDCl3, 100.6 MHz) δ 21.66 (Me of Ts),
21.71 (Me of Ts), 55.56 (OMe), 69.13 (C-5), 76.24, 78.17, 84.00,
101.96 (C-1), 127.99, 128.16, 129.92, 130.08, 132.50, 132.88,
145.05, 145.58 (aromatic carbons).
Exa m in a tion of Su bstr a te a n d In h ibitor y Activities
of Mod el 2 w ith Kd o8P Syn th a se. In order to examine the
substrate activity of 2, we used the proton-decoupled 31P-NMR
assay in which the conversion of starting material could be
clearly monitored. Reaction mixtures contained a 0.1 M Tris-
HCl buffer (prepared in D2O, pD ) 7.3, 37 °C), bovine serum
albumin (1.5 mg/mL, for the stabilization of the enzyme), 10
mM substrate (2), and 225 units (approximately 25 mg) of
Kdo8P synthase in a total volume of 0.6 mL. A control
experiment, containing all of the above but without the
enzyme, was run parallel to the above. The resonance of 2
(4.5 ppm) did not change significantly during the first 3 h, but
some new resonances (about 25% of the starting material) were
evident after 24 h of incubation, as judged by the integration
of signals. Although the identity of these new resonances were
not precisely determined, a very similar time course for the
formation of these new peaks was also seen in the correspond-
ing control experiment. Thus, the model 2 failed to show any
detectable enzyme-catalyzed acceleration in the expected
condensation process, after 24 h. Similar results were ob-
tained when the above experiments (with and without the
enzyme) were performed in the presence of 5 mM inorganic
phosphate.
In order to determine the Ki value of 2 against A5P binding,
the reaction solutions were prepared as described above but
with constant A5P (600 µM, ) 23Km) and variable PEP (9.8-
29.5 µM). The concentrations of 2 in these experiments were
in the range of 0-118 µM. Similarly, for the determination
of the Ki value of 2 against PEP binding, the constant
concentration of PEP (200 µM, ) 20Km) and variable A5P (15-
54.7 µM) were used, while concentrations of the inhibitor (2)
were in the range of 0-90 µM. Rate measurements were made
as described above. A 5 s delay was allowed following
initiation of the reaction. The initial rate was then determined
by least-squares fitting of the first 10% of the progress curve
to a straight line (between 20 and 80 s, depending on the initial
concentration of PEP). All samples were assayed in triplicate,
and analogous results were obtained in two to four different
experiments. The Ki values were calculated from the second-
ary replots of the slopes from initial double-reciprocal plots
(1/v vs 1/[S]) versus inhibitor concentration.38 These values
were found to be 200 ( 10 µM and 10 ( 1 µM, against A5P
and PEP, respectively.
In ter m olecu la r Con d en sa tion in Wa ter Solu tion . To
a solution of D-arabinose (208 mg, 1.39 mmol) and R-meth-
oxyacrylic acid31 (140 mg, 1.37 mmol) in H2O (15 mL), was
added ZnCl2 (933 mg, 6.90 mmol). The pH of the mixture was
adjusted to 5.0 using KOH (0.1 N). This solution was divided
in two equal volumes, with one being incubated at 50 °C and
the other at 90 °C. The reaction progress in both experiments
1
was monitored by H-NMR in the same manner as described
above for model 2 in the presence of ZnCl2. We did not detect
any possible condensation product in either experiment, even
after 50 h. During this time the arabinose was remained
unchanged, while the R-methoxyacrylate decomposed to pyru-
vic acid.
En zym e Assa ys. Unless otherwise stated, the enzyme
activity was assayed in a 0.8 mL reaction buffer consisting of
0.1 M Tris-acetate, pH ) 7.3, PEP (0.2 mM, ) 20Km), and A5P
(0.5 mM, ) 20Km). All solutions except the enzyme were
filtered through Millipore type-HA filters (0.45 µΜ) before use.
Following equilibration at 37 °C for 2 min, Kdo8P synthase
(20 µL, at a final concentration of about 30 nM) was added.
The decrease in the absorbance difference between 232 and
350 nm (as internal reference) was monitored with time (MS-
DOS UV/vis software). This method36 is based on the absor-
Ack n ow led gm en t. This research was supported by
the United States-Israel Binational Science Foundation
(BSF), J erusalem, Israel (Grant No. 94-00371), and by
the Center for Absorption in Science, Ministry of Im-
migrant Absorption and the Ministry of Science and
Technology, State of Israel (Grant No. 3611191, Gilady
Program).
Su p p or tin g In for m a tion Ava ila ble: Copies of 1H-NMR,
13C-NMR, and 31P-NMR spectra for new compounds 1, 2, 4, 5,
7-9 (R and â), 11, 12, 16, and 17 (R and â) and 1H-NMR
spectra of the mixtures of 13 and 14 and Kdo8P and 4-epi-
Kdo8P (condensation products of 2 at pH ) 5 and pH ) 1.6)
(38 pages). This material is contained in libraries on micro-
fiche, immediately follows this article in the microfilm version
of the journal, and can be ordered from the ACS; see any
current masthead page for ordering information.
bance difference at 232 nm between PEP (ꢀ ) 2840 M-1 cm-1
)
and the other substrates and products (ꢀ < 60 M-1 cm-1) under
the assay conditions. The initial rate was calculated from a
linear, least-squares fit of the first 30 s of the progress curve.
The concentrations of PEP, A5P, and 2 were determined
precisely by quantitative assaying of the Pi released by alkaline
phosphatase.37 In each case, to ensure complete hydrolysis of
the phosphate monoester, the aliquots of the incubation
mixture with alkaline phosphatase were tested by 31P NMR.
One unit of the enzyme activity is the amount that catalyzes
the consumption of 1 µmol PEP/min at 37 °C. During the
enzyme purification, the enzyme activity was assayed by the
thiobarbituric acid assay.9
J O961929Y
(36) Schoner, R.; Herrmann, K. M. J . Biol. Chem. 1976, 251, 5440-
5447.
(38) Segel, I. H. Biochemical Calculations, 2nd ed.; Wiley: New
(37) Ames, B. N. Methods Enzymol. 1966, 8, 115-118.
York, 1976.