New insights into the mechanism of CDP-D-tyvelose 2-epimerase: An enzyme-catalyzing epimerization at an unactivated stereocenter

Article abstract of DOI:10.1021/ja0022021

Tyvelose is a 3,6-dideoxyhexose found in the O-antigen of Yersinia pseudotuberculosis IVA and is the only member of this class of sugars to be produced directly from another 3,6-dideoxyhexose, paratose. The C-2 epimerization required for this conversion h

Full text of DOI:10.1021/ja0022021

VOLUME 122, NUMBER 43
NOVEMBER 1, 2000
© Copyright 2000 by the
American Chemical Society
New Insights into the Mechanism of CDP-D-Tyvelose 2-Epimerase:
An Enzyme-Catalyzing Epimerization at an Unactivated Stereocenter
Tina M. Hallis, Zongbao Zhao, and Hung-wen Liu*
Contribution from the Department of Chemistry, UniVersity of Minnesota, Minneapolis, Minnesota 55455
ReceiVed June 20, 2000
Abstract: Tyvelose is a 3,6-dideoxyhexose found in the O-antigen of Yersinia pseudotuberculosis IVA and is
the only member of this class of sugars to be produced directly from another 3,6-dideoxyhexose, paratose.
The C-2 epimerization required for this conversion has been proposed to be catalyzed by CDP-D-tyvelose
2-epimerase. This enzyme is intriguing since it belongs to a group of epimerases, including the well-studied
UDP-D-galactose 4-epimerase, that can invert unactivated stereocenters. To study the mechanism of this enzyme,
we have cloned and expressed the tyV gene that encodes CDP-D-tyvelose 2-epimerase. The purified tetrameric
protein contains approximately one equivalent of bound NAD+ per monomer and a small fraction of NADH.
Four possible mechanisms involving NAD+ can be proposed for this enzyme; two involve oxidation at C-2 of
the substrate, while the other two require oxidation at C-4. In a previous contribution, we presented preliminary
data that supported a retro-aldol-type mechanism initiated by C-4 oxidation. However, this mechanism was
refuted by further investigations, which revealed that the 4-fluoro analogue of CDP-D-paratose could be turned
over by the enzyme. More importantly, the direct transfer of a deuterium from C-2 of the labeled substrate to
the enzyme-bound NAD+ was observed by mass spectrometry. These results suggest that epimerization is in
fact initiated by oxidation at C-2, followed by the transfer of the hydride from the transiently formed NADH
to the opposite side of the 2-hexulose intermediate.
Epimerization is a common biological transformation per-
formed by a great variety of enzymes that operate on a diverse
array of substrates including carbohydrates, amino acids, and
fatty acids. Nature has in fact evolved epimerases and racemases
which operate by distinct mechanisms to catalyze this important
transformation.1 For instance, many enzymes involved in the
interconversion of L- and D-amino acids rely on using pyridoxal
phosphate (PLP) to transform the sp3 reaction center to an sp2
carbon via an enamine intermediate. It is the non-stereospecific
reprotonation from either side of this planar intermediate that
leads to racemization.1,2 The need for any cofactor is apparently
obviated by a number of other amino acid racemases. In these
cases, the reaction typically involves two basic groups in the
active-site: one for deprotonation and the other for reprotonation
of the resulting intermediate from the opposite side.1,2 A similar
mechanism is found for acyl-CoA R-epimerases which catalyze
the epimerization of R-branched-acylthioesters.3 However, in
the case of fatty acid â-epimerase, the inversion is achieved by
the breaking and reforming of the â-carbon-oxygen bond of
* To whom correspondence and reprint requests should be addressed.
Current address: Division of Medicinal Chemistry, College of Pharmacy,
University of Texas, Austin, TX 78712. Fax: 512-471-2746. E-mail:
h.w.liu@mail.utexas.edu.
(1) Tanner, M. E.; Kenyon, G. L. In ComprehensiVe Biological Catalysis;
Sinnott, M., Ed.; Academic Press: San Diego, 1998; Vol. II, pp 7-40.
(2) Soda, K.; Tanaka, H.; Tanizawa, K. In Vitamin B6 Pyridoxal
Phosphate; Dolphin, D., Poulson, R., Avramovic, O., Eds.; Wiley and
Sons: New York, 1986; Part B, pp 223-251.
(3) (a) Leadley, P. F.; Fuller, J. Q. Biochem. J. 1983, 213, 635-642. (b)
Fuller, J. Q.; Leadley, P. F. Biochem. J. 1983, 213, 643-650. (c) Schmidt,
D. E.; Fingerhut, R.; Conzelmann, E. Eur. J. Biochem. 1994, 222, 313-
323. (d) Schmidt, D. E.; Albers, C.; Fingerhut, R.; Conzelmann, E. Eur. J.
Biochem. 1995, 231, 815-822.
10.1021/ja0022021 CCC: $19.00 © 2000 American Chemical Society
Published on Web 10/11/2000

10494 J. Am. Chem. Soc., Vol. 122, No. 43, 2000
Hallis et al.
Scheme 1
Scheme 2
mediate 4 is converted to CDP-D-paratose (7) by CDP-D-
paratose synthase (Prt) in a NADPH-dependent reduction.11 The
subsequent inversion of the C-2 hydroxyl group catalyzed by
CDP-D-tyvelose 2-epimerase (Tyv) converts 7 to CDP-D-
tyvelose (8).
the substrate which is a CoA-bound â-hydroxylthioester.1,4 Such
a dehydration/rehydration mechanism is analogous to that
observed with crotonases.5 Interestingly, just like the acyl-CoA
R-epimerase and the crotonases, many of the sugar epimerases
act on stereocenters adjacent to the keto moiety of their
substrates. Epimerization of their substrate, the ketohexoses, is
achieved by the deprotonation of the activated R-H and
reprotonation at R-C from the opposite side.1 However, the most
intriguing carbohydrate epimerases are those that execute their
actions at unactivated stereocenters in their substrates. In this
category there are as yet only three enzymes that have been
studied in detail. These include UDP-galactose 4-epimerase,6
L-ribulose-5-phosphate 4-epimerase,7 and UDP-N-acetylglu-
cosamine 2-epimerase.8 This work describes a new member to
this exclusive group of carbohydrate epimerases, CDP-D-
tyvelose 2-epimerase,9 a novel enzyme found in the biosynthetic
pathway of 3,6-dideoxyhexoses.
The biosynthetic pathway of 3,6-dideoxyhexoses starting from
glucose-1-phosphate (1) is depicted in Scheme 1,10 illustrating
that four different 3,6-dideoxyhexoses are produced from one
common intermediate, 4. Out of these four sugars (drawn as
the cytidylyldiphosphate derivatives)sascarylose (5), abequose
(6), paratose (7), and tyvelose (8)styvelose is the only 3,6-
dideoxyhexose derived from another 3,6-dideoxyhexose, para-
tose. As delineated in Scheme 1, in the final stage of the
biosynthesis of CDP-D-tyvelose (8), the ∆3,4-glucoseen inter-
The epimerization performed by CDP-D-tyvelose 2-epimerase
is intriguing since it occurs at a stereogenic center that is not
activated by an adjacent keto or other electron-withdrawing
groups. Examination of the translated sequence of the tyV gene
from Yersinia pseudotuberculosis IVA led to the identification
of a possible NAD-binding motif near the N terminus,12
suggesting that Tyv may be a pyridine nucleotide-dependent
catalyst.9 The highly studied UDP-galactose 4-epimerase is an
example of an epimerase whose catalysis is also NAD+-
dependent.6 The NAD+ cofactor in this enzyme plays a key
role in the formation of a 4-ketosugar intermediate. Rotation
of the hexulose about the Pâ of UDP and the glycosyl oxygen
bond followed by NADH reduction allows the return of the
hydride to either face of the pyranose ring, resulting in
epimerization at C-4. It is possible that an analogous mechanism
could be operative with CDP-D-tyvelose 2-epimerase, as il-
lustrated by Mechanism A in Scheme 2. However, it is difficult
to envision that a similar rotation in the case of Tyv could
properly align C-2 of the putative 2-ketohexose intermediate 9
to the coenzyme in order to allow the internal hydride return to
C-2. Therefore, three other possible mechanisms have been
proposed for the epimerization catalyzed by CDP-D-tyvelose
2-epimerase.9
As depicted in Scheme 2, Mechanism B also incorporates a
C-2 oxidation step as in Mechanism A. In this case, however,
the acidity of the C-3 proton is exploited to allow keto-enol
tautomerization to give 10. Subsequent reduction at C-3 should
be suprafacial, whereas protonation at C-2 could occur from
either face of the pyranose, resulting in the return of H-2 to
C-3 and the racemization of the C-2 configuration. As depicted
in Scheme 3, Mechanism C involves a reversible dehydration
process made possible by an initial oxidation at C-4 to generate
11. The addition of water to either face of the enone 12 gives
a mixture of C-2 epimers. The final step is the reduction at C-4
to afford 8 and to regenerate the NAD+ coenzyme. The fourth
mechanism, D, relies on the oxidation at C-4 to activate the
hexose to 11 for retro-aldol ring cleavage at the C-2/C-3 bond.
Rotation about the C-1/C-2 bond of 13 followed by aldol ring
reclosure and then reduction at C-4 produces the two epimers
7 and 8. Our preliminary evidence, presented in a previous
contribution,9 appeared to support Mechanism D, a retro-aldol
process, for this epimerization. In this work, we provide a full
(4) (a) Hiltunen, J. K.; Palosaari, P. M.; Kunau, W.-H. J. Biol. Chem.
1989, 264, 13536-13540. (b) Smeland, T. E.; Li, J.; Chu, C.; Cuebas, D.;
Schulz, H. Biochem. Biophys. Res. Commun. 1989, 160, 988-992. (c) Yang,
S.-F.; Elzinga. M. J. Biol. Chem. 1993, 268, 6588-6592.
(5) (a) D’Oridine, R. L.; Tonge, P. J.; Carey, P. R.; Anderson, V. E.
Biochemistry 1994, 33, 12635-12643. (b) D’Oridine, R. L.; Bahnson, B.
J.; Tonge, P. J.; Anderson, V. E. Biochemistry 1994, 33, 14733-14742.
(6) (a) Frey, P. A. In Complex Pyridine Nucleotide-Dependent Trans-
formations; Dolphin, D., Avramovic, O., Poulson, R., Eds.; John Wiley &
Sons: New York, 1987; Vol. B, pp 462-510. (b) Bauer, A. J.; Rayment,
I.; Frey, P. A.; Holden, H. M. Proteins 1992, 12, 372-381. (c) Burke, J.
R.; Frey, P. A. Biochemistry 1993, 32, 13220-13230. (d) Thoden, J. B.;
Frey, P. A.; Holden, H. M. Biochemistry 1996, 35, 5137-5144. (e) Frey,
P. A. FASEB 1996, 10, 461-470.
(7) Johnson, A. E.; Tanner, M. E. Biochemistry 1998, 37, 5746-5754.
(8) (a) Morgan, P. M.; Sala, R. F.; Tanner, M. E. J. Am. Chem. Soc.
1997, 119, 10269-10277, (b) Sala, R. F.; Morgan, P. M.; Tanner, M. E. J.
Am. Chem. Soc. 1996, 118, 3033-3034.
(9) Hallis, T. M.; Liu, H.-w. J. Am. Chem. Soc. 1999, 121, 6765-6766.
(10) (a) Liu, H.-w.; Thorson, J. S. Annu. ReV. Microbiol. 1994, 48, 223-
256. (b) Kirschning, A.; Bechthold, A. F.-W.; Rohr, J. In Bioorganic
Chemistry: Deoxysugars, Polyketides & Related Classes: Synthesis,
Biosynthesis, Enzymes; Rohr, J., Ed.; Springer: Berlin, 1997; pp 1-84. (c)
Johnson, D. A.; Liu, H.-w. In ComprehensiVe Chemistry of Natural Products
Chemistry, Barton, D., Nakanishi, K., Meth-Cohn, O.; Eds.; Elsevier
Science: Oxford; 1999; pp 311-365. (d) Hallis, T. M.; Liu, H.-w. Acc.
Chem. Res. 1999, 32, 579-588.
(11) Hallis, T. M.; Lei, Y.; Que, N. L. S.; Liu, H.-w. Biochemistry 1998,
37, 4935-4945.
(12) (a) Verma, N.; Reeves, P. J. Bacteriol. 1989, 171, 5694-5701. (b)
Hobbs, M.; Reeves, P. R. Biochim. Biophys. Acta 1995, 1245, 273-277.

Mechanism of CDP-D-TyVelose 2-Epimerase
J. Am. Chem. Soc., Vol. 122, No. 43, 2000 10495
tyV proved to be difficult using primers that contained added restriction
sites necessary for cloning. Therefore, initial PCR was performed with
primers designed to exactly match the sequence at each end of the gene
without added restriction sites. The start primer, 5′-CGG TTG GTT
GTG CCA ATA T-3′, was complementary to the sequence 63 base
pairs upstream of the start codon of tyV. The halt primer, 5′-GGT ACC
CGC AAT TAA TT-3′, was complementary to the sequence 26 base
pairs downstream of the halt codon. The resulting PCR product was
used in subsequent amplification reactions that contained primers with
the appropriate restriction sites. The start primer, 5′-GCG CGA ATT
CTA AGG AAT ATA AAT GAA ACT GCT GAT TAC TGG TGG
TTG T-3′, contained an EcoR1 restriction site (in bold), the ribosomal
binding sequence and the AT rich region found just prior to the start
codon, and the codons for the first eight amino acid residues. Underlined
base pairs were modified to match codons of higher bias in Escherichia
coli. The halt primer, 5′-GCG CGC TCG AGA ACA GTT TCA ACC
CAA T-3′, introduced a XhoI restriction site (in bold) immediately
downstream of the last codon. The stop codon was omitted, and the
necessary codons for a histidine tag were incorporated. The PCR-
amplified DNA fragment was purified, digested with EcoR1 and XhoI,
and ligated into the EcoR1/XhoI sites of the transcription vector, pET-
24+ (Novagen) to give the recombinant plasmid pTHEp-10. This
plasmid was used to transform E. coli HB101. Positive clones were
identified by digestion of the plasmid DNA with EcoR1 and XhoI, and
visualization of the excised insert, by staining an agarose gel with
ethidium bromide after electrophoresis. The plasmid DNA from positive
clones was used to transform E. coli BL21(DE3). The general methods
and protocols for recombinant DNA manipulations were as described
by Sambrook et al.15
Growth of Escherichia coli (pTHEp-10) Cells. An overnight culture
of E. coli BL21(DE3)/pTHEp-10, grown in Luria-Bertani (LB) medium
supplemented with kanamycin (35 µg/mL) at 37 °C was used to
inoculate six 1 L cultures of the same medium and antibiotic. These
cultures were incubated at 37 °C until the OD600 reached 0.6, followed
by induction with 0.2 mM isopropyl â-D-thiogalactoside (IPTG) and 4
h of additional incubation at 37 °C. The cells were harvested by
centrifugation (6000g, 10 min) and stored at -20 °C.
Enzyme Purification. Thawed cells were resuspended in 100 mL
of binding buffer consisting of 50 mM potassium phosphate buffer (pH
8.0), 300 mM NaCl, 10% glycerol, and 10 mM imidazole. The cells
were disrupted by sonication in five 1-min bursts with a 1-min break
between each blast. Cell debris was removed by centrifugation (11000g,
20 min), and the supernatant was mixed by slow agitation with 2 mL
of packed Ni-NTA agarose resin (Qiagen) for 2 h at 4 °C. The resin
was collected by centrifugation (6000g, 3 min) and resuspended in 10
mL of binding buffer containing 10 mM imidazole. The slurry was
poured into a small column (1 cm diameter), followed by washing with
60 mL of binding buffer that contained 20 mM imidazole. Elution of
nonbinding proteins was monitored at A280, and once the absorbance
reached the background level, the target protein was eluted with binding
buffer containing 250 mM imidazole. The desired fractions, as detected
at A280 and confirmed by SDS-PAGE, were pooled and dialyzed against
1 L of 20 mM Tris-HCl (pH 7.5) containing 10% glycerol. The purified
enzyme was stored at -80 °C.
Scheme 3
account of our study on this enzyme that ultimately indicates
mechanism A and not D is operative in CDP-D-tyvelose
2-epimerase.
Experimental Section
General. Protein concentrations were determined according to
Bradford13 using bovine serum albumin as the standard. The NMR
spectra were acquired on a Varian Unity 300 or 500 spectrometer, and
chemical shifts (δ in ppm) are given relative to those for Me4Si (for
1H and 13C), CFCl3 (external, for 19F), and aqueous 85% H3PO4
(external, for 31P), with coupling constants reported in hertz (Hz).
Electronic absorption spectra were recorded on a Beckman DU650
spectrophotometer. N-terminal sequencing of purified CDP-D-tyvelose
2-epimerase, synthesis of oligonucleotide primers for polymerase chain
reaction (PCR), and amino acid analysis were all performed by the
Microchemical Facility in the Institute of Human Genetics at the
University of Minnesota. Electrospray mass spectrometry was carried
out by the Mass Spectrometry Consortium for the Life Sciences at the
University of Minnesota-St. Paul, and other high-resolution mass
spectrometry was performed either by the Resource for Biomedical
and Bio-organic Mass Spectrometry at Washington University or by
the Department of Chemistry Mass Spectrometry Facility at the
University of Minnesota-Minneapolis. Flash chromatography was
performed on Lagand Chemical silica gel (230-400 mesh) by elution
with the specified solvents. Analytical thin-layer chromatography (TLC)
was carried out on Polygram Sil G/UV254 plates (0.25 mm). TLC spots
were visualized by heating the plate previously stained with a solution
of phosphomolybdic acid (5% in EtOH). All chemicals were products
of Aldrich Co. (Milwaukee, WI). Bio-Gel P2 resin was purchased from
Bio-Rad Laboratories (Hercules, CA). The bacterial strain Y. pseudo-
tuberculosis IVA was a generous gift from Professor Robert Brubaker
at Michigan State University.
Polyacrylamide Gel Electrophoresis. The purity of the gene product
was assessed by SDS-polyacrylamide gel electrophoresis. Electrophore-
sis was carried out in the discontinuous buffer system of Laemmli,16
and the separating gel and stacking gel were 12 and 4% polyacrylamide,
respectively. Prior to electrophoresis, protein samples were boiled for
5 min in 62.5 mM Tris-HCl buffer (pH 6.8) containing 10% glycerol,
2% SDS, 5% â-mercaptoethanol, and 0.0025% bromophenol blue.
Electrophoresis of the heated samples was run in 25 mM Tris-HCl,
192 mM glycine, and 0.1% SDS (pH 8.3) at 25 mA. Gels were stained
with Coomassie blue and destained with acetic acid/ethanol/water (15:
20:165 by volume).
Gene Amplification and Cloning. The sequence of the tyV gene
(previously known as the rfbE gene)14 in Y. pseudotuberculosis IVA
has been reported,12b allowing the design of oligonucleotide primers
to be used in the amplification of the tyV gene from the genomic DNA
by PCR. The genomic DNA of Y. pseudotuberculosis IVA was isolated
by alkali lysis of the cells and ethanol precipitation. Amplification of
Molecular Mass Determination. The native molecular mass of
CDP-D-tyvelose 2-epimerase was determined by gel filtration performed
(13) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
(14) Reeves, P. R.; Hobbs, M.; Valvano, M. A.; Skurnik, M.; Whitfield,
C.; Coplin, D.; Kido, N.; Klena, J.; Maskell, D.; Raetz, C. R. H.; Rick, P.
D. Trends Microbiol. 1996, 4, 495-503.
(15) Sambrook, J.; Fritsch, E. F.; Maniatis, T. In Molecular Cloning: A
Laboratory Manual; Ford, N., Nolan, C., Ferguson, M., Eds.; Cold Spring
Harbor Press: Cold Spring Harbor, NY, 1989.
(16) Laemmli, U. K. Nature 1970, 227, 680-685.

10496 J. Am. Chem. Soc., Vol. 122, No. 43, 2000
Hallis et al.
on a Pharmacia FPLC equipped with a Superdex 200 HR 10/30 column.
The proteins were eluted with 50 mM potassium phosphate (pH 7.0)
and 0.15 M NaCl at a flow rate of 0.7 mL/min. The system was
calibrated with protein standards (Sigma): â-amylase (200 kDa), alcohol
dehydrogenase (150 kDa), bovine serum albumin (66 kDa), egg albumin
(45 kDa), carbonic anhydrase (29 kDa), and cytochrome c (12.4 kDa).
The data were analyzed by the method of Andrews.17 The subunit mo-
lecular mass was estimated by SDS-PAGE as described by Laemmli.16
Protein standards included R-lactalbumin (14 kDa), trypsin inhibitor
(20 kDa), trypsinogen (24 kDa), carbonic anhydrase (29 kDa),
glyceraldehyde-3-phosphate dehydrogenase (35 kDa), egg albumin (45
kDa), and bovine serum albumin (66 kDa).
Enzyme Assay. CDP-D-tyvelose 2-epimerase activity was deter-
mined by a discontinuous HPLC assay using a Spherisorb S5 SAX
column (0.46 × 25 cm) with a linear gradient from 50 mM potassium
phosphate (pH 3.5) to 140 mM potassium phosphate buffer (pH 3.5)
over 20 min. The substrate and product ratios were calculated from
the integration of the corresponding peaks from the HPLC chromato-
gram. A typical reaction contained 10 µM CDP-D-paratose and ∼0.1
µg of CDP-D-tyvelose 2-epimerase in 100 µL of 50 mM potassium
phosphate (pH 7.5). The solution was incubated at room temperature
for 5 min, followed by heating at 85 °C for 3 min to inactivate the
enzyme. Assays with inhibitors were carried out in a similar manner
except that 45 µM CDP-D-paratose and 0.27 µg of epimerase were
used with a variety of inhibitor concentrations.
Equilibrium Constant. The equilibrium constant, Keq, was calcu-
lated from the ratio of product to substrate at equilibrium. The product
and substrate ratios were determined for both the forward and reverse
directions by integration of the corresponding peaks from the HPLC
chromatogram. For the forward direction, 8.6 nmol of CDP-D-paratose
was mixed with ∼1 mg of CDP-D-tyvelose 2-epimerase in 100 µL of
50 mM potassium phosphate buffer (pH 7.5). The incubation was
allowed to reach equilibrium as determined by the constant product/
substrate ratio, which was monitored by SAX HPLC described under
“Enzyme Assay”. The same procedure was repeated for the reverse
direction with 7 nmol of CDP-D-tyvelose and ∼1 mg of enzyme in
100 µL of 50 mM potassium phosphate buffer (pH 7.5).
Determination of Kinetic Parameters. Assays were carried out as
described above. The kinetic parameters were deduced by fitting the
data to the Michaelis-Menten equation (eq 1) where Km is the
Michaelis-Menten constant, V is the maximum rate, and S is the
substrate concentration. Data for deuterium isotope effects were fitted
to eq 2 where Fi is the fraction of deuterium in the deuterated substrate,
EV and EV/K are the isotope effects minus 1 on V and V/Km, respectively.
solution was calculated from the change of fluorescence emission at
460 nm due to NADH formation: [F2 - F1][concentration of NAD+
internal standard]/[F3 - F2]. For the determination of NADP+, 12 µL
of 0.2 M glucose-6-phosphate and 3 µL of 1 M MgSO4 were combined
with 500 µL of the neutralized supernatant from the original HClO4
enzyme extraction. The fluorescence of this solution was measured as
described above. Glucose-6-phosphate dehydrogenase from yeast (4
units) was used to convert NADP+ to NADPH. The internal standard
was 10 µL of 80 µM NADP+.
Determination of bound NADH and NADPH was performed by first
boiling the enzyme for 3 min to release the coenzyme. After removing
the precipitated protein by centrifugation, the fluorescence of the
supernatant (500 µL) was measured directly as described above. The
fluorescence was recorded again after addition of 10 µL of 80 µM
NAD(P)H as an internal standard.
Production and Characterization of CDP-D-tyvelose. The substrate
for CDP-D-tyvelose 2-epimerase was prepared from CDP-D-glucose (2)
using four enzymes previously overexpressed in our laboratory.11,19-21
A 1-mL reaction contained 29 µmol of 2 and 6 mg of CDP-D-glucose
4,6-dehydratase (Eod, S.A. ) 125 µmol h-1 mg-1 22
) in 50 mM potassium
phosphate buffer, pH 7.5, was incubated for 2 h at room temperature.
The extend of conversion to the product, CDP-6-deoxy-D-glycero-L-
threo-4-hexulose (3), was determined spectrophotometrically at 320
nm under alkaline conditions.22,23 To this solution was added 40 µmol
of NADH, 12 mg of CDP-6-deoxy-D-glycero-L-threo-4-hexulose 3-de-
hydrase (E1, S.A. ) 100 nmol min-1 mg-1) and 0.6 mg of E1 reductase
(E3, S.A. ) 40 µmol min-1 mg-1), followed by incubation for 2 h at
room temperature.19,24 The extent of conversion of 3 to 4 was analyzed
by SAX HPLC using conditions described in “Enzyme Assay”. The
product, CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose (4, ∼29
µmol), without purification was incubated with 40 µmol of NADPH
and 2 mg of CDP-D-paratose synthase (S.A. ) 90 µmol min-1 mg-1
)
in 1 mL of 50 mM potassium phosphate buffer (pH 7.5) for 2 h at
room temperature.11 The conversion was monitored by the consumption
of NADPH at 340 nm, and once the conversion was complete, the
protein was removed with a Centricon 10 microconcentrator (Amicon).
The resulting CDP-D-paratose (7) was purified by FPLC using a MonoQ
HR (10/10) column with a linear gradient from 0 to 150 mM NH4HCO3
over 30 min, followed by a 5 min wash with 500 mM NH4HCO3. The
1
purified product was lyophilized, and its identity was verified by H
NMR.11 CDP-D-paratose (7, 1.3 µmol) was then incubated with 80 µg
of CDP-D-tyvelose 2-epimerase (S.A. ) 250 nmol min-1 mg-1) in 200
µL of 50 mM potassium phosphate buffer (pH 7.5) for 1.5 h at room
temperature, and the conversion to CDP-D-tyvelose (8) was followed
by SAX HPLC as described under “Enzyme Assay”. The desired
product was purified by FPLC using the identical conditions for the
isolation of CDP-D-paratose. The purified CDP-D-tyvelose (8) was
VS
Km + S
V )
V )
(1)
(2)
1
1
lyophilized, and the identity was confirmed by H NMR. H NMR
(2H2O, 300 MHz) δ 1.06 (3H, d, J ) 6.3 Hz, 5-Me), 1.70 (1H, ddd, J
) 3.0, 11.2, 14.0 Hz, 3ax-H), 1.84 (1H, ddd, J ) 3.6, 4.6, 14.0 Hz,
3eq-H), 3.40 (1H, ddd, J ) 4.6, 9.6, 11.2 Hz, 4-H), 3.70 (1H, dq, J )
6.3, 9.6 Hz, 5-H), 3.84 (1H, bs, 2-H), 4.00-4.20 (5H, m, 2′-H, 3′-H,
4′-H, 5′-Hs), 5.12 (1H, d, J ) 7.5 Hz, 1-H), 5.81 (1H, d, J ) 4.2 Hz,
1′-H), 5.95 (1H, d, J ) 7.5 Hz, 5′′-H), 7.82 (1H, d, J ) 7.5 Hz, 6′′-H).
13C NMR (2H2O, 75 MHz) δ 16.8, 32.5, 64.4 (d, J ) 4.5 Hz), 66.7,
67.6 (d, J ) 9.8 Hz), 69.0, 70.5, 74.2, 82.6 (d, J ) 9.8 Hz), 89.2, 94.9
(d, J ) 7.5 Hz), 96.5, 141.3, 160.2, 166.1.
VS
Km(1 + FiEV/K) + S(1 + FiEV)
Quantitation of Enzyme Bound NAD(P)+ and NAD(P)H. Quan-
titation of bound NAD(P)+ and NAD(P)H was determined as described
by Klingenberg.18 Bound NAD(P)+ was extracted from 250 µL of
enzyme (0.75 mg) using 1.25 mL of 0.6 N HClO4. The denatured
protein was removed by centrifugation, and 0.75 mL of the supernatant
was mixed with 0.15 mL of 1 N K2HPO4. Concentrated KOH (10 N)
was used to neutralize the mixture, and the resulting precipitate was
removed by centrifugation. For determination of NAD+, 0.25 mL of
the supernatant was mixed with 0.25 mL of 0.1 N Na4P2O7 buffer
containing 45 mM semicarbazide-HCl (pH 8.8) and 10 µL of ethanol.
The fluorescence (F1) was measured with an excitation wavelength of
365 nm and an emission wavelength of 460 nm. The slit widths for
excitation and emission were 5 and 15 nm, respectively. Alcohol
dehydrogenase from yeast (2 units) was used to convert the NAD+ to
NADH, and the fluorescence was measured again (F2). Another
fluorescence reading (F3) was taken after addition of 10 µL of 80 µM
NAD+ as an internal standard. The concentration of NAD+ in this assay
Synthesis of CDP-D-[2-2H]paratose. The 2-2H-labeled substrate for
CDP-D-tyvelose 2-epimerase was enzymatically synthesized from D-[2-
2H]glucose. The 4-mL incubation mixture contained 20 µmol of D-[2-
(19) Thorson, J. S.; Lo, S. F.; Ploux, O.; He, X.; Liu, H.-w. J. Bacteriol.
1994, 176, 5483-5493.
(20) Lei, Y.; Ploux, O.; Liu, H.-w. Biochemistry 1995, 34, 4643-4654.
(21) Ploux, O.; Lei, Y.; Vatanen, K.; Liu, H.-w. Biochemistry 1995, 34,
4159-4168.
(22) (a) Rubenstein, P. A.; Strominger, J. L. J. Biol. Chem. 1974, 249,
3776-3788. (b) Yu, Y.; Russell, R. N.; Liu, L.-d.; Thorson, J. S.; Liu,
H.-w. J. Biol. Chem. 1992, 267, 5868-5875.
(23) He, X.; Thorson, J. S.; Liu, H.-w. Biochemistry 1996, 35, 4721-
4731.
(17) Andrews, P. Biochem. J. 1964, 91, 222-233.
(18) Klingenberg, M. In Methods of Enzymatic Analysis; Bergmeyer, H.
U., Ed.; Verlag Chemie: New York, 1984; Vol. 4, pp 2045-2059.
(24) Johnson, D. A.; Gassner, G. T.; Bandarian, V.; Ruzicka, F. J.; Ballou,
D. P.; Reed, G. H.; Liu, H.-w. Biochemistry 1996, 35, 15846-15856.

Mechanism of CDP-D-TyVelose 2-Epimerase
J. Am. Chem. Soc., Vol. 122, No. 43, 2000 10497
2H]glucose (Cambridge Isotope Laboratories), 150 µmol of MgCl2, 72
µmol of ATP, and 60 units of hexokinase in 50 mM potassium
phosphate buffer (pH 7.5). After an incubation of 1 h at room
temperature, 33 µmol of CTP, 25 units of inorganic pyrophosphatase,
70 units of rabbit muscle phosphoglucomutase, 1 µmol of glucose 1,6-
diphosphate, and 8 mg of R-D-glucose-1-phosphate cytidylyltransferase
(Ep, S.A. ) 10 µmol h-1 mg-1) were added, and the incubation was
continued for 2 h at room temperature. To this mixture was then added
370 units of Eod, and incubation was allowed to proceed for another 2
h at room temperature. After the conversion to 3 was complete, 6 mg
of E1, 1 mg of E3, and 30 µmol of NADH were added with another 2
h of incubation at room temperature. The final transformation to make
CDP-D-paratose (7) was accomplished by the addition of 25 µmol of
NADH and 0.5 mg of CDP-D-paratose synthase to the above reaction
mixture, followed by 1 h of incubation at room temperature. The
proteins were removed by a Centricon 10 microconcentrator (Amicon),
and the resulting CDP-D-[2-2H]paratose was purified by FPLC as
described under “Production and Characterization of CDP-D-tyvelose”.
After lyophilization, the sample was analyzed by 1H NMR to confirm
and quantify the incorporation of deuterium at C-2.
“Synthesis of CDP-D-[4-2H]paratose”. The isolated CDP-D-paratose and
CDP-D-tyvelose were further purified by FPLC as described under
“Production and Characterization of CDP-D-tyvelose”. The position of
1
the deuterium in the purified samples was determined by H NMR.
Reduction of Enzyme Bound NAD+ with NaBH4. A 500-µL
reaction containing 1.5 mg of CDP-D-tyvelose 2-epimerase in 20 mM
Tris-HCl buffer, pH 7.5 was treated with approximately 1 mg of solid
NaBH4. After the foaming subsided, the solution was dialyzed against
500 mL of 20 mM Tris-HCl buffer (pH 7.5) overnight. Reduction of
the bound NAD+ was confirmed by observing the increase at 350 nm
after treatment with NaBH4.
Incubation of CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose
and Enzyme Reduced with NaBH4. Three incubations were per-
formed. The first sample contained approximately 12 nmol of CDP-
3,6-dideoxy-D-glycero-D-glycero-4-hexulose (4) and 148 µg of reduced
CDP-D-tyvelose 2-epimerase in 150 µL of 20 mM Tris-HCl buffer (pH
7.5), while the second contained only enzyme and the third contained
only 4. After incubation for 2 h at room temperature, all reactions were
spun through a Microcon microconcentrator to remove protein. The
reactions were then analyzed by SAX HPLC as described in “Enzyme
Assay”. Another sample containing a mixture of CDP-D-tyvelose and
CDP-D-paratose was used as a control for comparison during the HPLC
analysis.
Investigation of CDP-D-glucose and CDP-6-deoxy-D-glucose as
Substrates. A 100-µL incubation containing 20 µg of CDP-D-tyvelose
2-epimerase, 20 nmol of CDP-D-glucose (2), and 50 mM potassium
phosphate buffer, pH 7.5 was carried out in parallel with a control
reaction without enzyme. After incubation for 20 min at room
temperature, the enzyme was denatured by heating at 85 °C for 3 min.
A similar incubation was also performed with CDP-6-deoxy-D-glucose.
The formation of new products was analyzed by SAX HPLC under
conditions described in “Enzyme Assay”.
Synthesis of CDP-D-[4-2H]paratose. Labeling of CDP-D-paratose
at C-4 with deuterium was accomplished by using (4S)-[4-2H]NADPH25
in the reduction of CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose
(4) by CDP-D-paratose synthase. To make the labeled coenzyme, a
mixture of 20 µmol of NADP+, 100 µmol of D-[1-2H]glucose
(Cambridge Isotope Laboratories, Inc.), and 12 units of glucose
dehydrogenase from Bacillus megaterum in 3 mL of 50 mM potassium
phosphate buffer (pH 7.5) was incubated at room temperature for 1 h.
The progress of the reaction was monitored at 340 nm. After the glucose
dehydrogenase was removed with a Centricon 10 microconcentrator
(Amicon), the filtrate was mixed with 9 µmol of 4 and 0.5 mg of CDP-
D-paratose synthase, and the reaction was incubated for 1 h at room
temperature. The deuterated CDP-D-paratose was purified by HPLC
with a Spherisorb S10 SAX semiprep column (1.0 × 25 cm) using a
gradient from 50 mM to 118 mM potassium phosphate, pH 3.5,
followed by a 5 min wash with 500 mM potassium phosphate, pH 3.5.
The flow rate was 3 mL/min, and the detector was set at 270 nm. The
collected sample was neutralized with 5 N KOH and lyophilized. The
salts were removed by dissolving the sugar product in 70% ethanol,
pipetting the supernatant from the settled precipitate, and subjecting
the supernatant to rotary evaporation to remove the ethanol. The sample
was dissolved in 2H2O and lyophilized again to ensure the removal of
Large Scale Incubation of CDP-6-deoxy-D-glucose. A reaction
containing 15 µmol of CDP-6-deoxy-D-glycero-L-threo-4-hexulose (3),
20 µmol of NADPH, and 3 mg of CDP-D-paratose synthase in 1 mL
of 50 mM potassium phosphate buffer, pH 7.5 was incubated at room
temperature. The formation of CDP-6-deoxy-D-glucose was followed
by the consumption of NADPH at A340. After the conversion was
complete, the reaction was mixed with 1.2 mg of CDP-D-tyvelose
2-epimerase, and incubation was continued at room temperature for 1
h. The protein was removed with a Centricon 10 microconcentrator,
and the product was purified by HPLC with a Spherisorb S10 SAX
semiprep column (1.0 × 25 cm) as described under “Synthesis of CDP-
D-[4-2H]paratose”. The purified product, CDP-6-deoxy-D-mannose, was
1
all ethanol before analysis by H NMR to confirm and quantify the
incorporation of deuterium at C-4.
Incubation in Buffer Made with 2H2O. A 2-mL reaction contained
approximately 0.3 mg of CDP-D-tyvelose 2-epimerase, 3 µmol of CDP-
D-paratose, and 20 mM Tris-HCl buffer, pD 7.5, made with >99%
deuterated water (Cambridge Isotope Laboratories). The reaction was
incubated at room temperature for 2 h, and the protein was removed
with a Centricon 10 microconcentrator. The product/substrate mixture
was purified by FPLC as described under “Production and Character-
ization of CDP-D-tyvelose”. The collected peaks were lyophilized and
1
1
lyophilized and analyzed by H NMR. H NMR (2H2O, 300 MHz) δ
0.96 (3H, d, J ) 6 Hz, 5-Me), 3.13 (1H, dd, J ) 9.6, 9.6 Hz, 4-H),
3.54 (1H, dd, J ) 3.4, 9.6 Hz, 3-H), 3.58 (1H, m, 5-H), 3.72 (1H, dd,
J ) 1.8, 3.4 Hz, 2-H), 3.82-4.05 (5H, m, 2′-H, 3′-H, 4′-H, 5′-Hs),
5.12 (1H, dd, J ) 1.8, 7.5 Hz, 1-H), 5.68 (1H, d, J ) 3.9 Hz, 1′-H),
5.85 (1H, d, J ) 7.5 Hz, 5′′-H), 7.70 (1H, d, J ) 7.5 Hz, 6′′-H).
Synthesis of CDP-4-deoxy-4-fluoro-D-paratose (14). Methyl 2-O-
Benzyl-3,6-dideoxy-R-D-ribo-hexopyranoside (17). A solution of
methyl 3,6-dideoxy-R-D-ribo-hexopyranoside 1626 (6.0 g, 37 mmol) and
tributyltin oxide (40 mL, 75 mmol) in dry toluene (250 mL) was
refluxed for 23 h under N2 in a flask equipped with a Dean-Stark
separator. After the toluene was removed by distillation, the residue
was mixed with benzyl bromide (27 mL, 220 mmol), heated at 95 °C
for 24 h, cooled to room temperature, and purified by flash chroma-
tography on silica gel. Tributyltin oxide and excess benzyl bromide
were removed first with hexanes, followed by elution with hexanes:
ethyl acetate (5:1 to 1:1), to give the benzyl ether 17 (4.2 g, 45% yield)
as a colorless oil.1H NMR (CDCl3, 300 MHz) δ 1.21 (3H, d, J ) 6.3
Hz, 5-Me), 1.74 (1H, br s, OH), 1.80 (1H, m, 3-Hax), 2.16 (1H, ddd, J
) 11.4, 4.8, 4.8 Hz, 3-Heq), 3.21 (1H, ddd, J ) 11.4, 9.3, 4.8 Hz, 4-H),
3.40 (3H, s, OMe), 3.51 (2H, m, 2-H, 5-H), 4.55 and 4.62 (2H, ABq,
J ) 12.3 Hz, -CH2Ph), 4.60 (1H, d, J ) 2.4 Hz, 1-H), 7.32 (5H, m,
PhHs). 13C NMR (CDCl3, 50 MHz) δ 17.4 (C-6), 33.5 (C-3), 54.8
(OMe), 68.5 (C-5), 71.0 (C-4)*, 71.1 (benzylic-C)*, 73.9 (C-2), 93.1
1
analyzed for deuterium incorporation by H NMR.
Incubation in Buffer Made with H218O. To a lyophilized sample
of 0.45 µmol of CDP-D-paratose and 0.1 µmol of NAD+ were added
0.08 mg of CDP-D-tyvelose 2-epimerase (20 µL), 20 µL of 1 M Tris-
HCl buffer (pH 7.6), and 1 g of H218O (Isotec Inc., 96.8%). The mixture
was incubated at room temperature for 1 h, and a control incubation
was performed concurrently in regular buffer. The product and substrate
from each incubation were purified by FPLC as described under
“Production and Characterization of CDP-D-tyvelose”. The collected
peaks were lyophilized and analyzed by high-resolution FAB MS.
Incubation with CDP-D-[2-2H]paratose. The reaction contained 3.6
µmol of CDP-D-[2-2H]paratose and approximately 0.3 mg of CDP-D-
tyvelose 2-epimerase in 0.8 mL of 50 mM potassium phosphate buffer,
pH 7.5. The progress of the reaction was monitored by SAX HPLC as
described under “Enzyme Assay”. After 2 h of incubation at room
temperature, the sample was purified by SAX HPLC as described under
(25) Podschun, B. Biochem. Biophys. Res. Commun. 1992, 182, 609-
616.
(26) Classon, B.; Garegg, P. J.; Samuelsson, B. Can. J. Chem. 1981,
59, 339-343.

10498 J. Am. Chem. Soc., Vol. 122, No. 43, 2000
Hallis et al.
(C-1), 127.8, 128.4, 128.5, 138.1. HRMS (CI) calcd for C14H24NO4
(M + NH4+) 270.1705; found 270.1695.
anhydrous Na2SO4. The solvents were removed in vacuo, and the
residue was purified by flash chromatography on silica gel (eluent:
hexanes:ethyl acetate, 5:1 to 1:1) to give the product 21 (83 mg, 44%
yield based on the converted alcohol) as a colorless oil, together with
Methyl 2-O-Benzyl-4-fluoro-3,4,6-trideoxy-R-D-ribo-hexopyrano-
side (18). To a solution of benzyl ether 17 (2.38 g, 9.4 mmol) in dry
CH2Cl2 (100 mL) was added 1.6 mL of (diethylamino)sulfur trifluoride
(DAST) dropwise at 0 °C. After being stirred for 1 h at room
temperature, the mixture was diluted with CH2Cl2, washed with
saturated aqueous NaHCO3, and dried over anhydrous Na2SO4. The
solvent was removed in vacuo, and the residue was purified by flash
chromatography on silica gel (eluent: hexanes:ethyl acetate, 10:1 to
5:1) to give the 4-fluorosugar 18 (0.85 g, 35% yield) as a colorless oil.
1H NMR (CDCl3, 300 MHz) δ 1.24 (3H, dd, J ) 6.3, 0.9 Hz, 5-Me),
2.01 (1H, m, 3-Hax), 2.31 (1H, m, 3-Heq), 3.42 (3H, s, OMe), 3.50 (1H,
m, 2-H), 3.71 (1H, m, 5-H), 4.02 (1H, dddd, J ) 48.6, 11.1, 9.3, 4.8
Hz, 4-H), 4.55 and 4.63 (2H, ABq, J ) 12.6 Hz, benzylic-Hs), 4.60
(1H, d, J ) 3.9 Hz, 1-H), 7.33 (5H, m, PhHs). 13C NMR (CDCl3, 75
MHz) δ 17.1 (C-6), 30.9 (d, J ) 19.1 Hz, C-3), 55.0 (OMe), 65.8 (d,
J ) 24.4 Hz, C-5), 71.2 (benzylic-C), 73.4 (d, J ) 10.9 Hz, C-2), 90.3
(d, J ) 180.2 Hz, C-4), 96.9 (d, J ) 1.6 Hz, C-1), 127.8, 127.9, 128.5,
137.9. 19F NMR (CDCl3, 188 MHz) δ -185.3 (ddd, J ) 48.7, 10.0,
4.5 Hz). HRMS (CI) calcd for C14H23FNO3 (M + NH4+) 272.1662;
found 272.1647.
Methyl 4-Fluoro-3,4,6-trideoxy-R-D-ribo-hexopyranoside (19). A
solution of compound 18 (0.85 g, 3.3 mmol) in methanol (40 mL)
containing 10% Pd/C catalyst (100 mg) was hydrogenated for 2 h under
a H2 atmosphere. The mixture was filtered through Celite, concentrated,
and coevaporated with benzene (20 mL) to give the product 19 (0.52
g, 93% yield) as a colorless oil that was used without further
purification. 1H NMR (CDCl3, 300 MHz) δ 1.26 (3H, dd, J ) 6.3, 1.5
Hz, 5-Me), 1.81 (1H, m, 3-Hax), 1.99 (1H, br s, OH), 2.35 (1H, m,
3-Heq), 3.44 (3H, s, OMe), 3.67 (2H, m, 2-H, 5-H), 4.05 (1H, dddd, J
) 48.9, 11.1, 9.3, 4.8 Hz, 4-H), 4.58 (1H, dd, J ) 4.2, 3.9 Hz, 1-H).
13C NMR (CDCl3, 75 MHz) δ 17.1 (C-6), 34.5 (d, J ) 18.5 Hz, C-3),
55.2 (OMe), 65.9 (d, J ) 24.9 Hz, C-5), 67.3 (d, J ) 11.1 Hz, C-2),
90.0 (d, J ) 180.4 Hz, C-4), 98.2 (d, J ) 1.6 Hz, C-1).19F NMR (CDCl3,
282 MHz) δ -195.3 (m). HRMS (CI) calcd for C7H17FNO3 (M +
NH4+) 182.1192; found 182.1199.
1,2-Di-O-Acetyl-4-fluoro-3,4,6-trideoxy-R-D-ribo-hexopyranose (20).
To a solution of 19 (0.52 g, 3.1 mmol) in acetic anhydride (5 mL) was
added 4 drops of concentrated sulfuric acid at 0 °C. After stirring at
room temperature for 12 h, the mixture was diluted with ethyl acetate
and H2O, neutralized with NaHCO3, washed with brine, and dried over
anhydrous Na2SO4. The solvents were removed by rotary evaporation,
and the residue was chromatographed on silica gel (eluent: hexanes:
ethyl acetate, 10:1 to 5:1) to afford the diacetyl ester 20 (0.53 g, 72%
yield, R/â > 9/1) as a colorless oil. Analytical data for the major isomer
were as follows.1H NMR (CDCl3, 300 MHz) δ 1.25 (3H, d, J ) 6.3
Hz, 5-Me), 2.00 (3H, s, OAc), 2.02 (1H, m, 3-Hax), 2.13 (3H, s, OAc),
2.36 (1H, m, 3-Heq), 3.82 (1H, m, 5-H), 4.17 (1H, dddd, J ) 48.0,
11.1, 9.3, 4.8 Hz, 4-H), 4.90 (1H, m, 2-H), 6.10 (1H, dd, J ) 3.9, 3.6
Hz, 1-H). 13C NMR (CDCl3, 75 MHz) δ 17.2 (C-6), 20.7 (OAc), 21.0
(OAc), 30.2 (d, J ) 20.6 Hz, C-3), 66.8 (d, J ) 12.1 Hz, C-2), 68.4
(d, J ) 24.9 Hz, C-5), 88.2 (d, J ) 1.6 Hz, C-1), 89.1 (d, J ) 180.5
Hz, C-4), 169.2 (CdO), 169.8 (CdO). 19F NMR (CDCl3, 282 MHz) δ
-196.2 (ddd, J ) 48.2, 9.9, 4.5 Hz). HRMS (CI) calcd for C10H19-
FNO5 (M + NH4+) 252.1247; found 252.1241.
Dibenzyl 2-O-Acetyl-4-fluoro-3,4,6-trideoxy-R-D-ribo-hexopyranyl
Phosphate (21). A mixture of diacetyl ester 20 (0.26 g, 1.1 mmol)
and tributyltin methoxide (0.4 mL) in dry CH2Cl2 (15 mL) was refluxed
for 56 h under N2. The mixture was concentrated in vacuo and
chromatographed on silica gel (eluent: hexanes:ethyl acetate, 5:1 to
2:1) to afford the C-1 deprotected sugar (0.10 g, 86% yield based on
converted starting material) along with recovered diacetyl ester 20 (0.12
g). The C-1 deprotected sugar (0.10 g, 0.52 mmol) was dissolved in
dry THF (4 mL) at -78 °C. To this solution was added n-BuLi in
hexanes (1.6 M, 0.4 mL) dropwise. After stirring for 20 min, a solution
of dibenzyl phosphorochloridate (prepared from 1 mmol of dibenzyl
phosphite and 1 mmol of N-chlorosuccinimide in benzene) in dry THF
(2 mL) was added. The resulting solution was stirred at -78 °C for 30
min, and then at 0 °C for 2 h. The mixture was concentrated at room
temperature, diluted with ether, washed with H2O, and dried over
1
the C-1 deprotected sugar (21 mg). H NMR (CDCl3, 300 MHz) δ
1.20 (3H, dd, J ) 6.3, 0.9 Hz, 5-Me), 1.92 (3H, s, OAc), 2.01 (1H, m,
3-Hax), 2.34 (1H, m, 3-Heq), 3.85 (1H, m, 5-H), 4.15 (1H, dddd, J )
48.3, 11.4, 9.9, 5.1 Hz, 4-H), 4.79-4.87 (1H, m, 2-H), 5.10 (4H, m,
2× PhHs), 5.72 (1H, m, 1-H), 7.36 (10H, m, PhHs). 13C NMR (CDCl3,
75 MHz) δ 16.9 (C-6), 29.6 (d, J ) 20.6 Hz, C-3), 67.4 (dd, J ) 12.4,
7.9 Hz, C-2), 68.0 (d, J ) 24.9 Hz, C-5), 69.4 (dd, J ) 5.5, 4.2 Hz,
benzylic-C), 88.9 (d, J ) 181.0 Hz, C-4), 93.2 (dd, J ) 5.8, 1.8 Hz,
C-1), 127.8, 127.9, 128.6, 128.7, 135.5, 135.6, 169.9.19F NMR (CDCl3,
282 MHz) δ -186.6 (dd, J ) 48.1, 4.9 Hz). 31P NMR (CDCl3, 121
MHz) δ -1.5. HRMS (FAB) calcd for C22H27FNO7P (M + H+)
453.1502; found 453.1490.
CDP-4-deoxy-4-fluoro-D-paratose (14). A suspension of 21 (103
mg) and 10% Pd/C catalyst (40 mg) in a 1:1 MeOH/EtOAc solution
(10 mL) was stirred under a H2 atmosphere at room temperature for
20 min. The mixture was filtered through Celite, and the filtrate was
neutralized with 3 drops of triethylamine and dried under reduced
pressure to give a colorless syrupy residue. The residue was stirred
with 2 N LiOH solution (10 mL) overnight and then with Dowex-
50W resin (H+ form) for 30 min. The resin was removed by filtration,
and the aqueous solution was neutralized with triethylamine and
lyophilized to give the coupling precursor (112 mg) as a white
amorphous solid. A mixture of the coupling precursor (67 mg) and
cytidine 5′-monophosphomorpholidate (274 mg) was coevaporated with
dry pyridine (3 × 1.5 mL) to form a white powder, to which was added
dry pyridine (2.0 mL) and 1H-tetrazole (48 mg). After the mixture was
stirred for 40 h at room temperature, the solvents were removed under
vacuum and the residue was dissolved in H2O. Purification was
accomplished with a Bio-Rad P2 column (2 cm × 1.2 m) using 50
mM NH4HCO3 as the mobile phase. The desired fractions were detected
by UV absorption at 267 nm and lyophilized to give the coupling
product 14 (59 mg, 54% yield) as a white powder.1H NMR (2H2O,
300 MHz) δ 1.02 (3H, d, J ) 7.2 Hz, 5-Me), 1.65 (1H, m, 3-Hax), 2.07
(1H, m, 3-Heq), 3.58 (1H, m, 2-H), 3.74 (1H, m, 5-H), 3.91-4.15 (6H,
m, 4-H, 2′-H, 3′-H, 4′-H, 5′-Hs), 5.23 (1H, m, 1-H), 5.72 (1H, d, J )
3.6 Hz, 1′-H), 5.90 (1H, d, J ) 7.5 Hz, 5′′-H), 7.77 (1H, d, J ) 7.8
Hz, 6′′-H). 13C NMR (2H2O, 75 MHz) δ 16.1, 32.1 (d, J ) 18.8 Hz),
64.2 (d, J ) 5.4 Hz), 67.1 (d, J ) 5.4 Hz), 68.4, 74.2, 82.3 (d, J ) 9.6
Hz), 89.5, 89.8 (d, J ) 176 Hz), 94.0 (d, J ) 5.4 Hz), 96.4, 141.2,
157.5, 160.2, 166.1.19F NMR (2H2O, 282 MHz) δ -185.3 (dd, J )
48.2, 5.6 Hz). 31P NMR (2H2O, 121 MHz) δ -10.5 (d, J ) 21.4 Hz),
-12.1 (d, J ) 21.4 Hz). HRMS (ESI) calcd for C15H23FN3P2O13 (M -
H-) 534.0690; found 534.0688.
Investigation of CDP-4-deoxy-4-fluoro-D-paratose (14) as a
Substrate. CDP-4-deoxy-4-fluoro-D-paratose (14, 7.4 nmol) was mixed
with 0.1 mg of CDP-D-tyvelose 2-epimerase in 100 µL of 20 mM Tris-
HCl buffer (pH 7.5). After incubation at room temperature for 1 h, the
sample was analyzed by HPLC using a SAX column as described under
“Enzyme Assay”. A large incubation was performed where 37 µmol
of 14 and 4.5 mg of CDP-D-tyvelose 2-epimerase were incubated in
1.3 mL of 20 mM Tris-HCl buffer (pH 7.5) for approximately 8 h at
room temperature. The product, CDP-4-deoxy-4-fluoro-D-tyvelose, was
purified by FPLC as described under “Production and Characterization
1
1
of CDP-D-tyvelose” and analyzed by H NMR. H NMR (2H2O, 300
MHz) δ 0.94 (3H, d, J ) 6.3 Hz, 5-Me), 1.71 (1H, dddd, J ) 3.0,
13.2, 13.2, 13.2 Hz, 3-Hax), 1.87 (1H, m, 3-Heq), 3.73 (2H, m, 2-H,
5-H), 3.84-4.02 (5H, m, 2′-H, 3′-H, 4′-H, 5′-Hs), 4.18 (1H, dddd, J )
4.8, 9.3, 9.3, 49 Hz, 4-H), 4.96 (1H, m, 1-H), 5.65 (1H, d, J ) 3.9 Hz,
1′-H), 5.80 (1H, d, J ) 7.5 Hz, 5′′-H), 7.62 (1H, d, J ) 7.5 Hz, 6′′-H).
13C NMR (2H2O, 75 MHz) δ 16.2, 30.3 (d, J ) 20.4 Hz), 64.5 (d, J )
5.6 Hz), 68.2 (d, J ) 25 Hz), 68.9, 73.9, 82.3 (d, J ) 9.3 Hz), 89.0 (d,
J ) 171 Hz), 89.2, 94.7 (d, J ) 6.1 Hz), 96.4, 141.2, 157, 160.4, 165.9.
19F NMR (2H2O, 282 MHz) δ -190.4 (dd, J ) 5, 48 Hz). 31P NMR
(2H2O, 121 MHz) δ -10.8 (d, J ) 50 Hz), -13.0 (d, J ) 50 Hz).
HRMS (FAB): calcd for C15H23FN3O13P2 534.0690; found 534.0703.
Metal Analysis of CDP-D-tyvelose 2-Epimerase by Inductively
Coupled Plasma (ICP) Spectroscopy. CDP-D-tyvelose 2-epimerase

Mechanism of CDP-D-TyVelose 2-Epimerase
J. Am. Chem. Soc., Vol. 122, No. 43, 2000 10499
(3 mg) was dialyzed against 1 L of 20 mM Tris-HCl buffer (pH 7.5)
containing 10% glycerol overnight. The enzyme (∼2 mg) was diluted
to 5 mL with 10% HCl. A control containing an equal amount of
dialysis buffer was also diluted to 5 mL with 10% HCl. ICP analysis
was performed on both samples by the Research Analytical Laboratory
in the Department of Soil, Water, and Climate at the University of
Minnesota.
Treatment of CDP-D-tyvelose 2-Epimerase with 20 mM EDTA.
A solution of 0.3 mg of epimerase in 1 mL of 100 mM potassium
phosphate buffer (pH 7.5) containing 20 mM EDTA was left at room
temperature for 3 h and then stored at 4 °C. A control without EDTA
was also prepared. The activity of the epimerase with and without
EDTA was directly assayed after 1 day, 2 days, and 5 days of
incubation.
Treatment of CDP-D-tyvelose 2-Epimerase with 1,10-Phenan-
throline. CDP-D-tyvelose 2-epimerase (3 mg) was dialyzed for 4 days
against 500 mL of 50 mM potassium phosphate buffer (pH 7.5) that
contained 5 mM 1,10-phenathroline and 1.5 g of chelex. A control
was run in parallel in which CDP-D-tyvelose 2-epimerase (3 mg) was
dialyzed against 50 mM potassium phosphate buffer (pH 7.5) with 1.5
g of chelex but without 1,10-phenanthroline. The activity was monitored
as described in “Enzyme Assay”. At the end of fourth day, the samples
were submitted for ICP analysis.
Figure 1. SDS-PAGE of purified CDP-D-tyvelose 2-epimerase. (1)
Molecular weight standards, (2) CDP-D-tyvelose 2-epimerase.
Stopped-Flow Studies. Stopped-flow mixing studies were performed
on a Hi-Tech SF-61 DX2 double mixing stopped-flow system
configured for fluorescence detection. During fluorescence detection,
the excitation wavelength was set at 330 nm, and a cutoff filter of 360
nm was used, while wavelengths from 370 to 700 nm were monitored
for emission. All runs were performed at 4 °C unless otherwise
indicated. KinetAsyst 2, version 2.1a, software from Hi-Tech was used
for fitting the data.
Prolonged Incubations with Deuterated CDP-D-paratose. Two
incubations were carried out and each contained 9.3 mg of CDP-D-
tyvelose 2-epimerase and either 3.3 µmol of C-2 deuterated CDP-D-
paratose or 2.3 µmol of C-4 deuterated CDP-D-paratose. After
incubation at room temperature overnight, the bound cofactors were
released by boiling for 3 min. The denatured protein was removed by
centrifugation, and the resulting supernatant containing the NADH was
incubated with 10 units of L-lactate dehydrogenase from rabbit muscle
and 1.4 µmol of pyruvate for 20 min at room temperature. The
conversion of NADH to NAD+ was monitored at 340 nm. The NAD+
was purified by FPLC using a MonoQ HR (10/10) column and a linear
gradient from 0 to 40 mM NH4HCO3 during the first 12 min and from
40 to 250 mM NH4HCO3 during the next 12 min, followed by a 5-min
wash with 500 mM NH4HCO3. The collected fractions were lyophilized
and submitted for electrospray mass spectrometry.
Figure 2. Electronic absorption spectrum of CDP-D-tyvelose 2-epi-
merase at 0.8 mg/mL in 20 mM Tris-HCl buffer, pH 7.5.
ever, after treatment of the crude cell extract with Ni-NTA resin
to isolate the His-tagged protein, a single protein corresponding
to the size of CDP-D-tyvelose 2-epimerase was collected (Figure
1). The yield was ∼25 mg of purified epimerase from 6 L of
culture. Attempts to improve the expression level by varying
the cloning vector and expression conditions were unsuccessful.
The identity of the purified protein was confirmed by N-terminal
sequence analysis, revealing that its first 10 amino acid residue
(M-K-L-L-I-T-G-G-C) matched those of the translated tyV
sequence. The subunit molecular mass of 36 kDa, as ap-
proximated by SDS-PAGE, correlates well to the predicted value
of 38 987 Da calculated from the translated sequence plus the
His-tag. A molecular mass of 147 kDa, estimated by gel
filtration, indicated that the native CDP-D-tyvelose 2-epimerase
exists as a tetramer.
Results
Cloning and Expression of the tyW Gene. Production of
recombinant CDP-D-tyvelose 2-epimerase was made possible
by the cloning and expression of the tyV gene (previously known
as the rfbE gene)14 from Y. pseudotuberculosis IVA.12b Initial
attempts to amplify the desired gene by PCR using primers that
incorporated restriction sites were unsuccessful. Therefore,
primers corresponding to the flanking sequences at either end
of the tyV gene were used to first amplify the desired gene from
the genomic DNA of Y. pseudotuberculosis IVA by PCR. The
resulting PCR product was then used as a template for the
second round of PCR with primers that contained the restriction
sites needed for cloning. The naturally occurring ribosomal
binding site and the AT rich region preceding the start codon
were included in the start primer, allowing the PCR product to
be cloned directly into a transcription vector, pET-24(+), which
carries a T7/lac promoter.
Quantitation of Bound NAD(P)+ and NAD(P)H. Quantita-
tive analysis of NAD+ and NADH bound to CDP-D-tyvelose
2-epimerase was repeated twice on enzyme obtained from two
separate purifications. In the first sample, 64% of the monomers
contained bound NAD+, while 32% had NADH. In the second
sample, there were 69% NAD+ and 22% NADH. Thus, these
results clearly indicate one cofactor binding site per monomer,
and the majority of coenzymes exist in the oxidized state.
Electronic Absorption Spectrum of CDP-D-tyvelose 2-Epi-
merase. As shown in Figure 2, the electronic absorption
spectrum of CDP-D-tyvelose 2-epimerase displays a λmax at 275
Purification and Characterization of CDP-D-tyvelose
2-Epimerase. SDS-PAGE analysis of the crude cell extract from
the induced BL21(DE3) cells containing the pTHEp-10 construct
indicated that the target protein was not overexpressed. How-

10500 J. Am. Chem. Soc., Vol. 122, No. 43, 2000
Hallis et al.
nm. A broad absorption band extending up to 450 nm is likely
due to contributions from bound NADH and from a charge-
transfer band between the bound NAD+ and an amino acid
residue in the active site. Such charge-transfer bands, known
as “Racker Bands”,27 have also been observed in the electronic
absorption spectrum of UDP-galactose 4-epimerase.28
Equilibrium Constant. The reaction catalyzed by CDP-D-
tyvelose 2-epimerase is a reversible process. Separate reactions
containing either CDP-D-paratose (7) or CDP-D-tyvelose (8) led
to the same distribution of product and substrate upon reaching
equilibrium. The determined Keq of 1.22 at pH 7.5 reflects a
nearly equal preference for CDP-D-tyvelose and CDP-D-
paratose.
Isolation and Characterization of the CDP-D-tyvelose
2-Epimerase Product. Since the catalysis of CDP-D-tyvelose
2-epimerase has an almost equal preference for substrate and
product at equilibrium, an incubation of enzyme and CDP-D-
paratose always gave a mixture of CDP-D-tyvelose and CDP-
D-paratose. These two compounds could be separated by FPLC,
and the identity of purified CDP-D-tyvelose (8) was confirmed
Figure 3. Electronic absorption spectrum of CDP-D-tyvelose 2-epi-
merase before (bottom line) and after (top line) reduction with NaBH4.
The enzyme concentration was 80 µg/mL in 20 mM Tris-HCl buffer,
pH 7.5.
1
by H NMR.
derivatives in equilibrium were isolated and analyzed by high-
resolution FAB MS. Comparison of the results to a control
performed in unlabeled water revealed no significant differences,
suggesting that the hydroxyl substituents of 7 and 8 remained
intact during the catalysis. These results are inconsistent with
Mechanisms B and C.
KIE of C-2 and C-4 Deuterated Substrate. The kinetic
isotope effect (KIE) on V/K and V was calculated according to
eq 2. When deuterium was present at C-2 of CDP-D-paratose,
a KIE of 1.4 ( 0.2 was found for V/K and a KIE of 1.3 ( 0.1
was determined for V. Substrate containing a deuterium at C-4
displayed a KIE of 1.2 ( 0.1 for V/K and 0.90 ( 0.02 for V.
These values are relatively small for a primary KIE, indicating
that C-H bond cleavage is not an important rate-determining
step. Since these two sets of data are similar, especially
considering that the actual error of each value is likely to be
much higher due to the intrinsic inaccuracy of the discontinuous
assay, they cannot be used as evidence to distinguish which
hydrogen is being removed from the sugar substrate during
catalysis.
Reduction of Bound NAD+ with NaBH4. Treatment of
CDP-D-tyvelose 2-epimerase with NaBH4 had a pronounced
effect on its electronic absorption spectrum. As the NAD+ is
reduced to NADH, a decrease in the charge-transfer band
attributed to NAD+ and an increase in absorbance due to NADH
at wavelength longer than 300 nm is expected. As shown in
Figure 3, after addition of NaBH4, a large increase in absorbance
at 360 nm was indeed observed. Interestingly, when substrate
is added to the enzyme before reduction with NaBH4, the large
absorption increase occurs at 350 nm. Such a hypsochromic
shift may be due to a conformational change of the epimerase
in the presence of bound substrate or product.
Incubation of CDP-3,6-dideoxy-D-glycero-D-glycero-4-
hexulose with Reduced Epimerase. All four of the proposed
mechanisms for CDP-D-tyvelose 2-epimerase include a redox
process in which NAD+ and the sugar substrate are converted
to NADH and a keto sugar intermediate (9 or 11), respectively,
before being transformed back to NAD+ and the sugar epimers.
The 4-ketohexose intermediate predicted in Mechanisms C and
D, CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose (11) is also
an intermediate (4) produced by the E1/E3 reaction in the
biosynthetic pathway of CDP-D-tyvelose (see Scheme 1). The
possibility that this compound 4 could be reduced by epimerase
containing NADH was investigated to determine whether 11 is
Kinetic Properties. Due to the highly reversible nature of
the reaction, the kinetic parameters were determined in both
directions. All reactions were performed in buffer at pH 7.5
since an analysis of activity over a range of pH from 6 to 8.5
revealed a pH optimum between 7.0 and 7.5, with a large
decrease near pH 6.5 and a graduate decrease under more
alkaline conditions. In the forward direction, the Km for CDP-
D-paratose was determined to be 6.8 ( 0.4 µM with a kcat of 22
( 1 min-1. In the reverse direction, the Km for CDP-D-tyvelose
and the kcat for the catalysis were determined to be 170 ( 20
µM and 240 ( 10 min-1, respectively. Considering the kinetics
were performed with a discontinuous HPLC assay, the actual
errors for the Km and kcat values are likely to be much higher.
Thus, such a large intrinsic error may have contributed to the
discrepancy in the equilibrium constant (Keq ) 1.22) determined
experimentally from the product and substrate ratio as compared
to that (Keq ) 2.3) calculated from the Haldane relationship of
the kinetic constants.
Incubation with CDP-D-[2-2H]paratose. One possible mech-
anism proposed for C-2 epimerization by CDP-D-tyvelose
2-epimerase involves an intramolecular hydrogen transfer from
C-2 to C-3 as shown in Mechanism B, Scheme 2. To test the
feasibility of this mechanism, CDP-D-[2-2H]paratose was pre-
pared enzymatically from D-[2-2H]glucose and used in the
incubation. This synthetic sample was found to have ap-
proximately 50% deuterium at C-2. After incubation with the
epimerase, the product and substrate were isolated and analyzed
by 1H NMR. Integration of the 2-H and 3-H peaks of the product
and substrate revealed that, in both cases, C-2 was still 50%
deuterated and C-3 contained no deuterium. Hence, Mechanism
B is unlikely a possible mode of action.
Incubations Performed in 2H2O or H218O. The possibility
of hydrogen incorporation from the solvent into CDP-D-tyvelose
and CDP-D-paratose during catalysis was investigated. The
recovered sugar substrate or product from an incubation
performed in deuterated water was analyzed by 1H NMR. The
NMR peak integrations in each case showed that no deuterium
had been incorporated into the substrate or product. An
incubation was also carried out in H218O, and the two sugar
(27) Rizzo, V.; Pande, A.; Luisi, P. L. In Pyridine Nucleotide Coenzymes;
Dolphin, D., Poulson, R., Avramovic, O., Eds.; Wiley and Sons: New York,
1987; Vol. IIA, pp 99-161.
(28) Liu, Y.; Vanhooke, J. L.; Frey, P. A. Biochemistry 1996, 35, 7615-
7620.

Mechanism of CDP-D-TyVelose 2-Epimerase
J. Am. Chem. Soc., Vol. 122, No. 43, 2000 10501
Scheme 4
Scheme 5
the hexose interferes with binding and thus prevents epimer-
ization. However, the presence of a hydroxyl at C-3 of the
substrate is apparently tolerated by the enzyme.
Another compound that proved to be a substrate for CDP-
D-tyvelose 2-epimerase is the 4-fluoro analogue of CDP-D-
paratose (14), exhibiting a Km of 42 ( 2 µM and a kcat of 0.7
( 0.1 min-1. This molecule was originally designed as a probe
for Mechanism D. If the retro-aldol mechanism (Mechanism
D) is operative, incubation of 14 with reduced epimerase may
lead to ring opening by releasing a fluoride ion to give 22
(Scheme 5). Should oxidation takes place at C-2 as in Mech-
anism A, this 4-fluoro analogue is expected to act as a substrate
and be converted to the corresponding C-2 epimer. The latter
was found to be the case.
Inhibitors of CDP-D-tyvelose 2-Epimerase. A comparison
of CMP and CDP as inhibitors of CDP-D-tyvelose 2-epimerase
demonstrated that a 23-fold excess of CMP was necessary to
achieve a 50% decrease in activity, whereas a 12-fold excess
of CDP was enough to completely inhibit this enzyme. The 4-F
analogue of CDP-D-paratose (14) is also an inhibitor of CDP-
D-tyvelose 2-epimerase, albeit less effective than CDP. A 23-
fold excess of 14 displayed a 90% inhibition of activity.
Metal Content of CDP-D-tyvelose 2-Epimerase. Divalent
metal ions were detected in the purified enzyme by ICP analysis.
Interestingly, two nickel ions per tetramer were found in one
case, and two zinc ions per tetramer but no nickel were found
in a different batch. In an effort to determine if metal ions are
necessary for activity, epimerase was treated with metal
chelators. A comparison of epimerase treated with 20 mM
EDTA to a control sample showed that the control lost 40% of
its activity over 5 days at 4 °C, while the epimerase in EDTA
maintained its activity. Interestingly, epimerase dialyzed against
5 mM 1,10-phenathroline displayed a similar rate of lossing
activity to a control. By day 4, both the 1,10-phenanthroline
sample and the control had lost 30% of their activity. ICP
analysis revealed that the zinc had been removed from the
epimerase dialyzed against 1,10-phenanthroline, but not the
control, yet both exhibited the same activity. Thus, neither zinc
nor nickel is necessary for activity.
Stopped-Flow Results. The possible accumulation of tran-
siently formed NADH was investigated by stopped-flow spec-
trophotometry with fluorescence detection. When 75 µM of
epimerase (3 mg/mL) and 50 µM of CDP-D-paratose were
mixed, a very sharp increase in fluorescence (10%) was observed
between 0.001 and 0.01 s. This experiment was repeated with
either C-4 or C-2 deuterated substrate, and the rate of the
increase was compared to that with unlabeled substrate to
determine whether there is any KIE associated with hydride
transfer to NAD+. However, no significant rate change was
found for either of the labeled substrates. These results suggest
that either the increase in fluorescence is not due to NADH
formation or a KIE cannot be detected under these conditions.
Accumulation of NADH Upon Prolonged Incubation with
Excess Substrate. When an excess (10×) of CDP-D-paratose
was incubated with CDP-D-tyvelose 2-epimerase over a long
period of time (4-18 h), an increase in absorbance at 345 nm
was observed. This increase is most likely due to the formation
a real intermediate. The desired epimerase with bound NADH
was prepared by NaBH4 reduction of the native enzyme. After
incubation of 4 with reduced enzyme, the resulting mixture was
analyzed by HPLC. Two new peaks with retention times
corresponding to those of CDP-D-tyvelose (8) and CDP-D-
paratose (7) were observed in the HPLC chromatogram. These
two new peaks were not detected when either enzyme or 4 was
omitted from the incubation. Since the 4-keto sugar 4 could
clearly be processed by the reduced enzyme, it was concluded
that 4 (11) is likely an intermediate in this epimerization
reaction. This result is most consistent with Mechanisms C and
D.
Synthesis of CDP-4-fluoro-4-deoxyparatose (14). Synthesis
of CDP-4-deoxy-4-fluoroparatose 14 was achieved as shown
in Scheme 4. The precursor, methyl 3,6-dideoxy-R-D-ribo-hexo-
pyranoside 16, was prepared from methyl R-D-glucopyranoside
15 in two steps according to literature procedures.26 Regio-
selective protection of the C-2 hydroxyl of 16 was effected by
a tributyltin-activated benzylation procedure.29 Directed fluo-
rination at C-4 of 17 with diethylaminosulfurtrifluoride (DAST)
afforded product 18 with retention of configuration in 35% yield.
The corresponding 4-F epimer also formed, but at a lower yield.
Although fluorination of secondary alcohols using DAST usually
leads to inversion of stereochemistry, examples of retention of
configuration are also known.30 An alternative approach using
the C-4 epimer of 17 as the reactant was also attempted.
Preparation of this 4-epimer was accomplished by standard
Mitsunobu conditions; however, no fluorinated product was
discernible when it was subjected to the DAST reagent. The
final steps of the synthesis shown in Scheme 4 were straight-
forward. Hydrogenation of 18, followed by acetylation, afforded
ester 20. Selective deacetylation at C-1 with tributyltin meth-
oxide, followed by phosphorylation and then coupling with
cytidine 5′-monophosphomorpholidate, gave the desired sugar
dinucleotide 14.
Alternate Substrates for CDP-D-tyvelose 2-Epimerase.
Although incubation of CDP-D-glucose (2) and epimerase failed
to generate any products, incubation with CDP-6-deoxy-D-
glucose did produce a new compound. NMR analysis of this
new product from a large-scale incubation revealed that it is
CDP-6-deoxy-D-mannose. These results clearly show that CDP-
6-deoxy-D-glucose is a substrate and that a hydroxyl at C-6 of
(29) Dasgupta, F.; Garegg, P. J. Synthesis 1994, 1121-1123.
(30) Wolba, D. M.; Razavi, H. A.; Clark, N. A.; Parmar, D. S. J. Am.
Chem. Soc. 1988, 110, 8686-8691.

10502 J. Am. Chem. Soc., Vol. 122, No. 43, 2000
Hallis et al.
of an enzyme-NADH-CDP-D-paratose ternary complex result-
ing from the replacement of the keto sugar intermediate (9 or
11) with the excess substrate. Activity assays of the epimerase
after such an incubation revealed a 30-50% loss of activity,
since the NADH-containing enzyme is catalytically inactive.
A similar accumulation of NADH was also observed with UDP-
galactose 4-epimerase but on a much faster time scale.31
Prolonged Incubation with Deuterated CDP-D-paratose.
The accumulation of NADH after prolonged incubation with
excess substrate provided a convenient means to pinpoint which
hydrogen, 2-H or 4-H, is transferred to the bound NAD+ during
catalysis. If the oxidation occurs at C-2, incubation with C-2
deuterated substrate is expected to give deuterated NADH, and
conversely, if C-4 oxidation is operative, a reaction with C-4
deuterated substrate should produce deuterated NADH. After
the two separate overnight incubations of CDP-D-tyvelose
2-epimerase and either excess C-2 or C-4 deuterated CDP-D-
paratose were performed, the accumulated NADH in each
sample was released by heat denaturation. However, purification
of the resulting NADH proved to be difficult in the presence of
the excess substrate. Therefore, the NADH was converted to
NAD+ by the pro-R stereospecific L-lactate dehydrogenase from
rabbit muscle, which uses NADH to reduce pyruvate to L-lactate.
The NAD+ was then purified by FPLC and analyzed for
deuterium content by positive ion electrospray mass spectrom-
etry. NAD+ from the incubation with C-4 deuterated substrate
displayed a (M + H + 1)+/(M + H)+ of 0% after correction
for natural isotope abundance. Interestingly, NAD+ from the
incubation with C-2-labeled substrate had a (M + H + 1)+/(M
+ H)+ of 10% after the same correction, indicating that the
hydrogen is transferred from C-2 during the epimerization
reaction.
Interestingly, the other two sugar epimerases that act on an
unactivated stereocenter of their substrates are NAD+ inde-
pendent. One of these enzymes is UDP-N-acetylglucosamine
2-epimerase, which catalyzes the interconversion of UDP-N-
acetylglucosamine and UDP-N-acetylmannosamine and is de-
void of any redox cofactor.32,33 On the basis of a recent
positional isotope exchange (PIX) experiment, its catalysis has
been shown to involve a C-O bond cleavage at C-1 to form a
2-acetamidoglucal intermediate.8 The release and the re-ligation
of the UDP moiety at C-1 is a reversible process. What
ultimately leads to the observed epimerization is the deproton-
ation and reprotonation at C-2 taking place at opposite sides of
the glucal intermediate.
The enzyme L-ribulose-5-phosphate 4-epimerase is another
example of a carbohydrate epimerase that is able to invert an
unactivated stereogenic center without using the redox chemistry
of NAD+. The interconversion of L-ribulose-5-phosphate and
D-xylulose-5-phosphate by this epimerase requires a divalent
metal cation per subunit for activity.34 Since the N-terminal
portion of L-ribulose-5-phosphate 4-epimerase is homologous
to a class II L-fuculose-1-phosphate aldolase from E. coli, this
epimerase was thus investigated whether it makes use of an
aldol chemistry.35,7 Results from these and other studies34,36-38
suggest that the epimerase is capable of promoting carbon-
carbon bond cleavage and support a mechanism that follows a
retroaldol cleavage process, generating glycolaldehyde phos-
phate and the metal-bound enolate of dihydroxyacetone. Rota-
tion of the aldehyde carbonyl exposes the opposite face, and
upon aldol addition of the enolate, the epimeric sugar is formed.
As a new member of sugar epimerase that acts on an
unactivated stereocenter, CDP-D-tyvelose 2-epimerase resembles
UDP-D-galactose 4-epimerase more than other epimerases.
Preliminary examination of the translated sequence of CDP-D-
tyvelose 2-epimerase from Y. pseudotuberculosis IVA revealed
a possible NAD+ binding motif at its N terminus. Further
analysis by BLAST search found that this epimerase shows
strong homology (29% identity and 47% similarity) to TDP-
glucose 4,6-dehydratase of Salmonella typhimurium (accession
no. P26391) and UDP-galactose 4-epimerase of E. coli K12
(26% identity and 41% similarity, accession no. P09147). Since
both of these enzymes operate on nucleotide sugars and use
NAD+ as a redox cofactor, the above sequence analysis provided
the initial clue for a catalytic role of NAD+ in the mechanism
of CDP-D-tyvelose 2-epimerase.
Discussion
Inversion of stereocenters is a common transformation found
in sugar biosynthesis and appears to be a convenient strategy
used by nature to generate structural diversity of carbohydrates.
Unlike most epimerases that catalyze stereochemical inversions
adjacent to a carbonyl or a carboxylate, CDP-D-tyvelose
2-epimerase is an unusual enzyme because it epimerizes a
stereocenter that bears no acidic proton and, thus, cannot utilize
a deprotonation-reprotonation mechanism. Only a few enzymes
are known that are capable of inverting such unactivated
stereocenters, among which, the best-studied example is UDP-
galactose 4-epimerase. This enzyme contains a tightly bound
NAD+ and catalyzes the interconversion between UDP-D-
galactose and UDP-D-glucose.6,28 Results from many studies
support a mechanism in which the epimerase binds one of the
substrate epimers and undergoes a conformational change to
activate the cofactor toward hydride transfer. The nucleotide
unit of the substrate serves as an anchor in the active site, but
the 4-keto-hexopyranosyl moiety is allowed to rotate about the
Pâ of UDP and the glycosyl oxygen bond. Hydride transfer from
C-4 of the hexose to the oxidized coenzyme produces NADH
and a UDP-4-ketohexose intermediate. Rotation and exposure
of the opposite face of the 4-ketohexose to the reduced cofactor
permits hydride transfer to either side of the planar C-4, resulting
in the formation of the epimeric products. Other nucleotide sugar
4-epimerases, such as UDP-N-acetylglucosamine 4-epimerase
and UDP-glucuronate 4-epimerase, have been postulated to
follow mechanisms analogous to that of UDP-galactose
4-epimerase.6a
Many experiments designed to differentiate the possible
mechanisms proposed in Schemes 2 and 3 were carried out. As
indicated earlier,9 neither oxygen nor hydrogen from the solvent
was found to be incorporated into the sugar products during
catalysis as expected in Mechanisms B and C. In addition,
labeling studies revealed that the hydrogen at C-2 was retained
in both the product and the recovered substrate, negating a
hydride transfer from C-2 to C-3 as predicted in Mechanism B.
Thus, both Mechanisms B and C are disfavored based on these
observations.
(32) Kawamura, T.; Kimura, M.; Yamamori, S.; Ito, E. J. Biol. Chem.
1978, 253, 3595-3601.
(33) Kawamura, T.; Ishimoto, N.; Ito, E. J. Biol. Chem. 1979, 254, 8457-
8465.
(34) Deupree, J. D.; Wood, W. A. J. Biol. Chem. 1972, 247, 3093-
3097.
(35) Dreyer, M. K.; Schulz, G. E. J. Mol. Biol. 1996, 259, 458-466.
(36) Deupree, J. D.; Wood, W. A. J. Biol. Chem. 1970, 245, 3988-
3995.
(37) Davis, L.; Lee, N.; Glaser, L. J. Biol. Chem. 1972, 247, 5862-
5866.
(31) Nelsestuen, G. L.; Kirkwood, S. J. Biol. Chem. 1971, 246, 7533-
7543.
(38) Salo, W. L.; Fossitt, D. D.; Bevill, R. D.; Kirkwood, S.; Wood, W.
A. J. Biol. Chem. 1972, 247, 3098-3100.

Mechanism of CDP-D-TyVelose 2-Epimerase
J. Am. Chem. Soc., Vol. 122, No. 43, 2000 10503
The two remaining mechanisms, A and D, differ in the
position where oxidation is initiated during the catalysis.
Attempts to distinguish between these two mechanisms by
determining the KIE of substrate labeled with deuterium at C-2
or C-4 led to inconclusive results. Since all of the proposed
mechanisms in Schemes 2 and 3 involve the transient formation
of NADH, stopped-flow spectrophotometry was used to deter-
mine the rate of NADH accumulation with the expectation that
substrate labeled at C-2 or C-4 may display different rates for
the reduction of NAD+ to NADH. Unfortunately, the observed
isotope effects were again ambiguous.
dehydrogenase family, and the stereospecificity of a NAD+-
dependent dehydrogenase can be predicted based on its designa-
tion as either a long-chain or short-chain dehydrogenase as first
defined by Jo¨rnvall et al.39,40
The group I long-chain dehydrogenases are characteristic for
a polypeptide chain of greater than 350 amino acids, having
the NAD(P) binding motif, known as the Rossmann fold, located
at the C-terminus, requiring a zinc ion for the catalytic activity,
and a pro-R stereospecificity with regard to the transfer of
hydride to and from NAD+/NADH. In contrast, the short-chain
dehydrogenases are composed of approximately 250 amino acids
and are zinc ion-independent. The location of their NAD(P)
binding site is at the N terminus, and a YXXXK motif is
conserved in most short-chain dehydrogenases. Most impor-
tantly, the hydride transfer step in short-chain dehydrogenases
is pro-S stereospecific. Although CDP-D-tyvelose epimerase is
composed of 338 amino acids, it does contain a NAD(P) binding
sequence, 7GGCGFL13G, at its N terminus, it does not require
zinc, and it does contain the conserved motif, 164YGCS168K.
According to these characteristics, CDP-D-tyvelose epimerase
is clearly a member of the short-chain dehydrogenase family,
and thus, should also be pro-S stereospecific.
Interestingly, HPLC analysis of an incubation mixture
containing reduced enzyme with bound NADH and the 4-keto-
hexose 11 revealed the conversion of 11 to CDP-D-tyvelose and
CDP-D-paratose. This result is most consistent with the retro-
aldol mechanism (Mechanism D) that involves the formation
of 11 as the intermediate. To further test the implicated retro-
aldol mechanism, a 4-fluoro analogue of CDP-D-paratose (14)
was synthesized as a mechanistic probe. It was anticipated that
14, lacking the 4-OH group required for oxidation to initiate
the retro-aldol reaction, may simply act as a competitive
inhibitor. It was also considered that 14, upon incubation with
reduced enzyme, could undergo ring opening and elimination
of the 4-F to give an acyclic product, 22, as shown in Scheme
5. Since the reactive aldehyde moiety in 22 may trap an active-
site nucleophile, 14 may act as an inactivator for the epimerase.
However, to our surprise, 14 was not processed by the reduced
enzyme (E‚NADH), but instead was recognized as a substrate
for the native enzyme containing NAD+ (E‚NAD+). This
observation is clearly incompatible with Mechanism D in which
4-oxidation is a necessary step. It is possible that processing of
11 by the reduced enzyme, which was cited earlier as evidence
supporting Mechanism D, may simply be a result of an aberrant
reduction by NADH to generate CDP-D-paratose and the NAD+-
containing enzyme. Beyond the initial reduction, the reaction
proceeds just like a normal incubation to give a mixture of CDP-
D-paratose and CDP-D-tyvelose. Structural studies of CDP-D-
tyvelose 2-epimerase will undoubtedly be helpful in revealing
the basis for this anomalous reduction at C-4.
On the basis of this information, the conversion of the NADH
from CDP-D-tyvelose 2-epimerase to NAD+ was performed with
pro-R specific L-lactate dehydrogenase. Upon purification by
FPLC, the NAD+ was analyzed for deuterium incorporation by
mass spectrometry. Only NAD+ from the incubation with C-2-
labeled substrate was determined to contain deuterium. The
deuterium incorporation of 10% is significant considering that
the purified CDP-D-tyvelose 2-epimerase already contained a
large quantity of NADH (∼25%) prior to incubation and less
than 30-50% of the NAD+ is converted to NADH during
turnover. In addition, the C-2-labeled substrate was only 50%
deuterated, and a preference for C-H bond cleavage over C-2H
bond cleavage is expected for the hydride transfer even though
the C-2 deuterated substrate did not display a significant KIE
under steady-state conditions.
In summary, these results provide compelling evidence that
the reaction catalyzed by CDP-D-tyvelose 2-epimerase follows
Mechanism A. Although this epimerization is reminiscent of
that of UDP-galactose 4-epimerase, the motion required in this
case to allow hydride transfer from NADH to either side of the
2-ketohexose intermediate is expected to be different. It is
conceivable that the hexulose and the nicotinamde ring are
arranged in an edge-to-edge configuration and a similar rotation
about the Pâ of CDP and the glycosyl oxygen bond may be
sufficient to expose both faces of the 2-ketohexose to the
pyridine cofactor. However, an understanding of the actual
substrate motion required in the active site to achieve epimer-
ization will have to await further studies of this intriguing
enzyme.
As Mechanism D can be ruled out as a viable pathway,
Mechanism A, the only scenario that has not been eliminated,
must be considered. The key difference between Mechanisms
A and D is the position of the oxidation/reduction. Therefore,
the best evidence to distinguish between these two possibilities
would be a conclusive proof as to the regiochemistry of the
initial oxidation. An opportune discovery led to a solution that
allowed us to address this problem directly. Since prolonged
incubation of CDP-D-tyvelose 2-epimerase with a 10-fold excess
of CDP-D-paratose resulted in NADH accumulation, analysis
of deuterium incorporation into the NADH after incubation with
either C-2 or C-4 deuterated substrate should reveal the origin
of the hydride that is transferred from CDP-D-paratose to NAD+.
A similar phenomenon had also been observed with UDP-
galactose 4-epimerase, where a dead-end complex formed
between the enzyme, NADH, and substrate after prolonged
incubation with excess substrate.31 In this experiment, the
accumulated NADH from CDP-D-tyvelose 2-epimerase was
converted to NAD+ to facilitate purification, since NAD+, not
NADH, showed baseline separation from the substrate. To
ensure that the newly acquired deuterium on NADH was
retained on NAD+, the L-lactate dehydrogenase from rabbit
muscle was used in this transformation. The rationale to use
this pro-R stereospecific dehydrogenase is based on the fact that
CDP-D-tyvelose 2-epimerase is a member of the short-chain
Acknowledgment. This work is supported in part by the
National Institutes of Health Grant GM35906. T.M.H. is a
trainee of a NIGMS Biotechnology Training Grant (2 T32
GM08347). The Yersinia pseudotuberculosis IVA strain was a
generous gift from Professor Robert Brubaker at Michigan State
University. H.-w.L. also thanks the National Institute of General
Medical Sciences for a MERIT Award.
JA0022021
(39) Jo¨rnvall, H.; Persson, B.; Jeffery, J. Eur. J. Biochem. 1987, 167,
195-201.
(40) Reid, M. F.; Fewson, C. A. Crit. ReV. Microbiol. 1994, 20, 13-56.