Mechanistic investigation of UDP-galactopyranose mutase from Escherichia coli using 2- and 3-fluorinated UDP-galactofuranose as probes

Article abstract of DOI:10.1021/ja010473l

The galactofuranose moiety found in many surface constituents of microorganisms is derived from UDP-galactopyranose (UDP-Galp) via a unique ring contraction reaction catalyzed by UDP-Galp mutase. This enzyme, which has been isolated from several bacterial sources, is a flavoprotein. To study this catalysis, the cloned Escherichia coli mutase was purified and two fluorinated analogues, UDP-[2-F]Galf (9) and UDP-[3-F]Galf (10), were chemically synthesized. These two compounds were found to be substrates for the reduced UDP-Galp mutase with the K_m values determined to be 65 and 861μM for 9 and 10, respectively, and the corresponding k_cat values estimated to be 0.033 and 5.7 s^-1. Since the fluorine substituent is redox inert, a mechanism initiated by the oxidation of 2-OH or 3-OH on the galactose moiety can thus be firmly ruled out. Furthermore, both 9 and 10 are poorer substrates than UDP-Galf, and the rate reduction for 9 is especially significant. This finding may be ascribed to the inductive effect of the 2-F substituent that is immediately adjacent to the anomeric center, and is consistent with a mechanism involving formation of oxocarbenium intermediates or transition states during turnover. Interestingly, under nonreducing conditions, compounds 9 and 10 are not substrates, but instead are inhibitors for the mutase. The inactivation by 10 is time-dependent, active-site-directed, and irreversible, with a K₁ of 270 μM and a k_inact of 0.19 min^-1. Since the K₁ value is similar to K_m, the observed inactivation is unlikely a result of tight binding. To our surprise, the inactivated enzyme could be regenerated in the presence of dithionite, and the reduced enzyme is resistant to inactivation by these fluorinated analogues. It is-possible that reduction of the enzyme-bound FAD may induce a conformational change that facilitates the breakdown of the putative covalent enzyme - inhibitor adduct to reactivate the enzyme. It is also conceivable that the reduced flavin bears a higher electron density at N-1, which may play a role in preventing the formation of the covalent adduct or facilitating its breakdown by charge stabilization of the oxocarbenium intermediates/transition states. Clearly, this study has led to the identification of a potent inactivator (10) for this enzyme, and study of its inactivation has also shed light on the possible mechanism of this mutase.

Full text of DOI:10.1021/ja010473l

6756
J. Am. Chem. Soc. 2001, 123, 6756-6766
Mechanistic Investigation of UDP-Galactopyranose Mutase from
Escherichia coli Using 2- and 3-Fluorinated UDP-Galactofuranose as
Probes
Qibo Zhang‡ and Hung-wen Liu*
Contribution from the DiVision of Medicinal Chemistry, College of Pharmacy, and
Department of Chemistry and Biochemistry, UniVersity of Texas, Austin, Texas 78712
ReceiVed February 21, 2001
Abstract: The galactofuranose moiety found in many surface constituents of microorganisms is derived from
UDP-D-galactopyranose (UDP-Galp) via a unique ring contraction reaction catalyzed by UDP-Galp mutase.
This enzyme, which has been isolated from several bacterial sources, is a flavoprotein. To study this catalysis,
the cloned Escherichia coli mutase was purified and two fluorinated analogues, UDP-[2-F]Galf (9) and UDP-
[3-F]Galf (10), were chemically synthesized. These two compounds were found to be substrates for the reduced
UDP-Galp mutase with the Km values determined to be 65 and 861 µM for 9 and 10, respectively, and the
corresponding kcat values estimated to be 0.033 and 5.7 s-1. Since the fluorine substituent is redox inert, a
mechanism initiated by the oxidation of 2-OH or 3-OH on the galactose moiety can thus be firmly ruled out.
Furthermore, both 9 and 10 are poorer substrates than UDP-Galf, and the rate reduction for 9 is especially
significant. This finding may be ascribed to the inductive effect of the 2-F substituent that is immediately
adjacent to the anomeric center, and is consistent with a mechanism involving formation of oxocarbenium
intermediates or transition states during turnover. Interestingly, under nonreducing conditions, compounds 9
and 10 are not substrates, but instead are inhibitors for the mutase. The inactivation by 10 is time-dependent,
active-site-directed, and irreversible with a KI of 270 µM and a kinact of 0.19 min-1. Since the KI value is
similar to Km, the observed inactivation is unlikely a result of tight binding. To our surprise, the inactivated
enzyme could be regenerated in the presence of dithionite, and the reduced enzyme is resistant to inactivation
by these fluorinated analogues. It is possible that reduction of the enzyme-bound FAD may induce a
conformational change that facilitates the breakdown of the putative covalent enzyme-inhibitor adduct to
reactivate the enzyme. It is also conceivable that the reduced flavin bears a higher electron density at N-1,
which may play a role in preventing the formation of the covalent adduct or facilitating its breakdown by
charge stabilization of the oxocarbenium intermediates/transition states. Clearly, this study has led to the
identification of a potent inactivator (10) for this enzyme, and study of its inactivation has also shed light on
the possible mechanism of this mutase.
D-Galactose is a common sugar found ubiquitously in nature.
thyotoxic agent.3 The biosynthetic donor of the D-galactopyra-
nosyl unit in galactoconjugates has been shown to be UDP-
galactopyranose (UDP-Galp, 1), which is derived from UDP-
glucose in a process catalyzed by UDP-galactose 4-epimerase.4
Many bacteria can also directly activate galactose to UDP-Galp
via the combined actions of galactokinase and R-D-galactose-
1-phosphate uridylyltransferase.5 Unlike the formation of UDP-
Galp which is well-studied, the biosynthesis of galactofuranose
has been puzzling for many years. Although early feeding
In free solution, the majority of D-galactose exists as a pyranose
sugar with the thermodynamically less favored D-galactofuranose
as a minor component (<8%).1 Similarly, the pyranose form is
the prevalent form of D-galactose in naturally occurring galac-
toconjugates, where D-galactofuranose is a much rarer constitu-
ent. The occurrence of D-galactofuranose so far has been limited
mainly to surface polymers of microorganisms, such as bacterial
O-antigens, mycobacterial cell walls, fungal glycoconjugates,
and protozoal cell membranes.2 An exception is prymnesin-1
(PRM1), a polycyclic ether toxin from the red tide algae
Prymnesium parVum, in which D-galactofuranose has been
shown as a structural component of this hemolytic and ich-
* To whom correspondence should be addressed. Fax: 512-471-2746.
E-mail: h.w.liu@mail.utexas.edu.
‡ Current address: Albany Molecular Research, Inc., Biocatalysis Divi-
sion, 2660 Crosspark Road, Coralville, IA 52241.
(1) (a) Wertz, P. W.; Garver, J. C.; Anderson, L. J. Am. Chem. Soc.
1981, 103, 3916-3922. (b) Angyal, S. J.; Pickles, V. A. Aust. J. Chem.
1972, 25, 1695-1710.
(2) (a) de Lederkremer, R. M.; Colli, W. Glycobiology 1995, 5, 547-
552. (b) Brennan, P. J.; Nikaido, H. Annu. ReV. Biochem. 1995, 64, 29-
63. (c) Whitfield, C. Trends Microbiol. 1995, 3, 178-185. (d). Previato, J.
O.; Mendonca-Previato, L.; Jones, C.; Wait, R. Glycoconjugates J. 1993,
10, 340.
experiments had revealed that the donor of the galactofuranosyl
unit in microbial surface polymers is UDP-galactofuranose
(3) Igarashi, T.; Satake, M.; Yasumoto, T. J. Am. Chem. Soc. 1999, 121,
8499-8511.
(4) (a) Shibaev, V. N. AdV. Carbohydr. Chem. Biochem. 1986, 44, 277-
339. (b) Frey, P. A. FASEB 1996, 10, 461-470.
(5) (a) Thomas, P.; Bessel, E. M.; Westwood, J. H. Biochem. J. 1974,
139, 661-664. (b) Kalckar, H. M. AdV. Enzymol. 1958, 20, 111.
10.1021/ja010473l CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/22/2001

Mechanistic InVestigation of UDP-Galactopyranose Mutase
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6757
Scheme 1
(UDP-Galf, 2), the precursor of UDP-Galf was only recently
established to be UDP-Galp (1), whose conversion to UDP-
Galf is catalyzed by UDP-galactopyranose mutase.6 The UDP-
Galp mutase has been isolated from several bacterial sources7
and has been shown to be a flavoprotein, with the flavin adenine
dinucleotide (FAD) coenzyme bound noncovalently. The fact
that galactofuranose units are found in many pathogens but not
in human tissues makes UDP-Galp mutase an attractive target
for therapeutic agents.
Scheme 2
The reaction catalyzed by UDP-Galp mutase must involve
ring cleavage and recyclization for the interconversion of 1 and
2. A few possible mechanisms can be envisioned to account
for this unique ring rearrangement. The mechanism depicted in
Scheme 1 is an example in which the redox capability of FAD
is exploited, allowing the oxidation of 2- or 3-OH to produce
the enediols (3/4, or the corresponding enediolates) as possible
intermediates. Subsequent ring opening, recyclization, and
reduction would regenerate the flavin coenzyme and afford the
expected product. A nonoxidative ring cleavage initiated by a
nucleophilic attack at the R-phosphoryl atom to yield an UDP-
enzyme intermediate is another plausible mechanism for the
catalysis of UDP-Galp mutase.8 A reversible 1,2-migration of
the UDP group from C-1 to C-2 to facilitate the requisite ring
opening is also a conceivable alternative.8 However, little
mechanistic investigation of this mutase had been carried out
until a recent study9 in which a reversible cleavage of the
anomeric C-O bond of 1 was elegantly demonstrated to be
part of the catalysis. Clearly, mechanisms involving PR-O bond
cleavage could be ruled out on the basis of this finding.9 As
illustrated in Scheme 2, a new mechanism initiated by distortion
of the ring to allow attack of O4 on C-1 to release UDP, or by
elimination of UDP first to form an oxocarbenium ion 5
followed by O4 attack on C-1, was proposed by Blanchard and
his coworkers.9 Subsequent ring opening between C-1 and O5
of the bicyclo-acetal intermediate 6, via either 7 or 8, followed
by the rebound of UDP at C-1 should lead to the formation of
2. Since oxidation/reduction is not a requisite step of this
mechanism and there is no net change in the redox states of
the substrate or the product, whether the enzyme-bound FAD
plays an active role in the reaction mechanism remains obscure.
To study this transformation, we have recently amplified the
glf gene of Escherichia coli6 by polymerase chain reaction
(PCR) and expressed it in E. coli BL-21(DE3) host cells.10 The
encoding UDP-Galp mutase was purified to near homogeneity.
Using UDP-Galf as the substrate, a Km of 194 µM and a kcat of
1.5 s-1 for the catalysis in the reverse direction (2 f 1) were
obtained. Interestingly, when the enzyme was reduced by
sodium dithionite, its catalytic efficiency was increased by more
than 2 orders of magnitude (Km ) 22 µM, kcat ) 27 s-1). While
both the FAD coenzyme and the cysteine residues of this mutase
were reduced upon treatment with dithionite, reoxidation of the
flavin coenzyme alone was enough to abolish the increment of
the activity. A comparable rate enhancement was also noted
when the flavin coenzyme was selectively reduced by photore-
duction in the presence of 5-deazariboflavin under anaerobic
(6) (a) Stevenson, G.; Neal, B.; Liu, D.; Hobbs, M.; Packer, N. H.; Batley,
M.; Redmond, J. W.; Lindquist, L.; Reeves, P. J. Bacteriol. 1994, 176,
4144-4156. (b) Nassau, P. M.; Martin, S. L.; Brown, R. E.; Weston, A.;
Monsey, D.; McNeil, M. R.; Duncan, K. J. Bacteriol. 1996, 178, 1047-
1052.
(7) (a) Ko¨plin, R.; Brisson, J.-R.; Whitfield, C. J. Biol. Chem. 1997,
272, 4121-4128. (b) Weston, A.; Stern, R. J.; Lee, R. E.; Nassau, P. M.;
Monsey, D.; Martin, S. L.; Scherman, M. S.; Besra, G. S.; Duncan, K.;
McNeil, M. R. Tuberc. Lung Dis. 1998, 78, 123-131. (c) Chen, P.; Bishai,
W. Infect. Immun. 1998, 66, 5099-5106. (d) McMahon, S. A.; Leonard,
G. A.; Buchanan, L. V.; Giraud, M.-F.; Naismith, J. H. Acta Crystallogr.
1999, D55, 399-402.
(8) Barlow, J. N.; Blanchard, J. S. Carbohydr. Res. 2000, 328, 473-
480.
(9) Barlow, J. N.; Girvin, M. E.; Blanchard, J. S. J. Am. Chem. Soc.
1999, 121, 6968-6969.
(10) (a) Zhang, Q.; Liu, H.-w. J. Am. Chem. Soc. 2000, 122, 9065-
9070. (b) Zhang, Q.; Liu, H.-w. Bioorg. Med. Chem. Lett. 2001, 11, 145-
149.

6758 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Zhang and Liu
found m/z 351.1103. Spectral data for the â isomer (13â): 1H NMR
(CDCl3) δ 1.99 (3H, s), 2.04 (3H, s), 2.06 (3H, s), 2.07 (3H, s), 4.16
(1H, dd, J ) 12.0, 6.9, 6-H), 4.25 (1H, dd, J ) 12.0, 4.5, 6-H), 4.32
(1H, dd, J ) 4.5, 3.3, 4-H), 4.91 (1H, dd, J ) 48.9, 0.9, 2-H), 5.07
(1H, ddt, J ) 21.6, 4.5, 0.9, 3-H), 5.34 (1H, ddd, J ) 6.9, 4.5, 3.3,
5-H), 6.29 (1H, d, J ) 10.8, 1-H); 13C NMR (CDCl3) δ 20.6, 20.7
(2C), 20.9, 62.4, 69.1, 76.2 (d, J ) 31), 83.2, 97.3 (d, J ) 185), 98.8
(d, J ) 38), 169.0, 169.7, 170.2, 170.5; 19F NMR (CDCl3) δ -191.3
(ddd, J ) 49.3, 21.1, 11.1); high-resolution FABMS calcd for C14H20-
FO9 [M + H]+ 351.1091, found m/z 351.1099.
conditions. These findings clearly demonstrated the involvement
of FAD in regulating the catalytic efficiency of this enzyme.10a
To gain further insight into the mechanism of this interesting
transformation, we have synthesized UDP-2-deoxy-2-fluoro-
galactofuranose (UDP-[2-F]Galf, 9) and UDP-3-deoxy-3-fluoro-
galactofuranose (UDP-[3-F]Galf, 10) and examined the com-
petence of these two compounds as substrates and/or inhibitors
for UDP-Galp mutase. Reported in this paper are the results of
these experiments that not only confirmed an active role played
by the enzyme-bound flavin in regulating the catalysis, but also
provided preliminary evidence suggesting the possible involve-
ment of a covalent intermediate in the mechanism of this mutase.
Dibenzyl (3,5,6-Tri-O-acetyl-2-deoxy-2-fluoro-r-D-galactofuran-
osyl)phosphate (14). A solution of 13 (3.7 g, 10 mmol, R:â ) 1:2) in
anhydrous dichloromethane (30 mL) was cooled with an ice-water
bath. To this solution was added bromotrimethylsilane (9 mL), and
the resulting mixture was stirred overnight at room temperature. The
solution was then evaporated to dryness under reduced pressure. The
residual oil was redissolved in anhydrous toluene (5 mL), and the
mixture was evaporated to dryness to remove more volatile silane
derivatives. The residue and dibenzyl phosphate (4.3 g, 15 mmol) were
then mixed with anhydrous toluene (10 mL). To this solution was added
triethylamine (2.1 mL, 15 mmol), and the mixture was stirred overnight
at room temperature. After removal of solvent, the residue was
chromatographed on silica gel and eluted with EtOAc/hexanes (1:1).
The fast-moving unreacted starting material (2.8 g, 65%, R:â ) 1:3.6)
was recovered. The desired product was eluted next from the column
and was purified again on silica gel to give pure 14 in 8% yield (540
mg). Only a small amount of â isomer was detected in the crude mixture
by 19F NMR. Spectral data for 14: 1H NMR (CDCl3) δ 1.95 (3H, s),
2.02 (3H, s), 2.11 (3H, s), 4.11 (1H, dd, J ) 11.7, 6.5, 6-H), 4.17 (1H,
dd, J ) 6.5, 4.8, 4-H), 4.28 (1H, dd, J ) 11.7, 4.8, 6-H), 5.06 (1H,
dddd, J ) 52.2, 6.5, 4.2, 1.8, 2-H), 5.06-5.12 (4H, m, benzylic CH2’s),
5.21 (1H, dt, J ) 6.5, 4.8, 5-H), 5.58 (1H, dt, J ) 15.6, 6.5, 3-H), 5.94
(1H, dd, J ) 6.0, 4.2, 1-H), 7.32-7.36 (10H, m, Ar-H’s); 13C NMR
(CDCl3) δ 20.55, 20.61, 20.7, 61.9, 69.36 (d, J ) 6), 69.37, 69.6 (d, J
) 5), 72.6 (d, J ) 23), 79.0 (d, J ) 8), 91.9 (dd, J ) 198, 7), 96.4 (dd,
J ) 18, 4), 127.8 (2C), 128.0 (2C), 128.6 (4C), 128.7 (2C), 135.5 (d,
J ) 7), 135.6 (d, J ) 8), 169.6, 169.9, 170.4; 19F NMR (CDCl3) δ
-205.2 (dd, J ) 52.8, 15.0); 31P NMR (CDCl3) δ -2.4; high-resolution
FABMS calcd for C26H31FO11P [M + H]+ 569.1588, found m/z
569.1555.
Experimental Section
General. Unless otherwise specified, all chemicals were purchased
from Aldrich or Sigma and used without further purification. NMR
spectra were recorded on Varian 200, 300, or 500 MHz spectrometers.
13C NMR spectra were recorded with proton broad-band decoupling,
and the reported spin couplings are from fluorine and/or phosphorus.
31P NMR spectra were recorded with proton decoupling and externally
referenced with 85% phosphoric acid. 19F NMR spectra were recorded
without proton decoupling, and the external reference was fluoro-
trichloromethane. The J-values are given in hertz (Hz). For NMR
spectra recorded in D2O, t-BuOH was added as an internal reference;
the chemical shifts are 1.27 and 31.2 ppm for 1H NMR and 13C NMR,
respectively.11 All UV-visible spectra were taken on a Beckman
DU650 spectrophotometer. Fast-atom bombardment (FAB), chemical
ionization (CI), and electrospray ionization (ESI) mass spectra were
recorded by the MS facility at the Department of Chemistry of the
University of Minnesota. High-performance liquid chromatography
(HPLC) analysis and/or purification were conducted with a Hewlett-
Packard 1090A system equipped with a photodiode array detector. To
detect nucleotide sugar, the detector wavelength was set at 262 nm.
Analytical thin-layer chromatography was carried out on Merck silica
gel 60 G-254 plates, and the spots were visualized either under UV
light or by heating plates previously stained with solutions of vanillin/
ethanol/H2SO4 (1:98:1) or phosphomolybdic acid (7% in EtOH).
Synthesis of UDP-[2-F]Galf (9). 1,3,5,6-Tetra-O-acetyl-2-deoxy-
2-fluoro-r-D-galactofuranose (13r) and 1,3,5,6-Tetra-O-acetyl-2-
deoxy-2-fluoro-â-D-galactofuranose (13â). The starting material,
2-deoxy-2-fluoro-D-galactose (12, 7.0 g, 38 mmol), was prepared from
11 in two steps. The fluorination step was accomplished by using
SELECTFLUOR as the reagent,12 and the product was deacetylated
using NaOMe/MeOH. The product 12 was dissolved in dry pyridine
(100 mL) and heated at 110 °C for 2 h. To this solution was slowly
added acetic anhydride (60 mL), and the resulting mixture was stirred
for an additional 1.5 h at 110 °C. The reaction was cooled to room
temperature and evaporated to dryness under reduced pressure. The
residual oil was chromatographed on silica gel and eluted with ethyl
acetate/hexanes (1:2). The pyranose products (6.4 g, 48%) were eluted
first from the column, followed by the elution of the furanose products.
While the furanose products in the earlier fractions were a mixture of
both R and â isomers (2.37 g, R:â ) 1:2), that in the latter fractions
was mainly the â isomer (1.4 g). A small amount of R isomer was also
purified to homogeneity by repeated chromatography on silica gel for
analytical purposes. The combined yield for the furanose products was
Triethylammonium (3,5,6-Tri-O-acetyl-2-deoxy-2-fluoro-r-D-ga-
lactofuranosyl)phosphate (15). A solution of 14 (420 mg, 0.74 mmol)
in ethyl acetate (10 mL) and triethylamine (0.8 mL) was stirred in the
presence of Pd/C (10%) at room temperature overnight under a
hydrogen atmosphere. The mixture was filtered through a Celite pad,
and the filtrate was evaporated to dryness under reduced pressure to
give 15 in 97% yield (350 mg): 1H NMR (CDCl3) δ 1.28 (9H, t, J )
7.5), 2.01 (3H, s), 2.07 (3H, s), 2.09 (3H, s), 3.03 (6H, q, J ) 7.5),
4.03 (1H, t, J ) 5.7, 4-H), 4.19 (1H, dd, J ) 12.0, 6.3, 6-H), 4.37 (1H,
dd, J ) 12.0, 4.2, 6-H), 4.92 (1H, dt, J ) 51.6, 4.8, 2-H), 5.23 (1H, q,
br, J ) 6.0, 5-H), 5.51 (1H, dt, J ) 16.2, 5.4, 3-H), 5.82 (1H, dt, J )
7.5, 3.9, 1-H); 13C NMR (CDCl3) δ 8.5, 20.76, 20.77, 20.9, 45.4, 62.2,
70.4, 74.1 (d, J ) 25), 77.6 (d, J ) 7), 92.4 (dd, J ) 200, 6), 95.6 (dd,
J ) 17, 4), 169.9, 170.5, 170.8; 19F NMR (CDCl3) δ -206.1 (ddd, J
) 52.8, 15.8, 4.5); 31P NMR (CDCl3) δ -0.1; high-resolution FABMS
calcd for C12H18FO11PNa [M + Na]+ 411.0468, found m/z 411.0450.
Bis(triethylammonium) (2-Deoxy-2-fluoro-r-D-galactofuranosyl)-
phosphate (16). A solution of 15 (328 mg, 0.67 mmol) in a mixture
(20 mL) of methanol/water/triethylamine (5:2:1) was allowed to stand
at room temperature for 2 days. The reaction was monitored by TLC
(ethanol:NH4OH:H2O ) 5:3:1) and was stopped when the conversion
was complete. The mixture was evaporated to dryness under reduced
pressure to afford the desired product 16. The yield was quantitative
(345 mg): 1H NMR (D2O) δ 1.30 (18H, t, J ) 7.5), 2.95 (12H, q, J )
7.5), 3.66 (1H, dd, J ) 11.7, 7.2, 6-H), 3.73 (1H, dd, J ) 11.7, 4.2,
6-H), 3.77-3.88 (2H, m, 4-H, 5-H), 4.57 (1H, dt, J ) 17.7, 7.2, 3-H),
5.00 (1H, dddd, J ) 53.1, 7.2, 4.5, 1.2, 2-H), 5.65 (1H, dd, J ) 5.4,
4.5, 1-H); 13C NMR (D2O) δ 9.7, 48.1, 63.7, 73.2, 73.6 (d, J ) 21),
82.2 (d, J ) 10), 96.0 (dd, J ) 197, 7), 96.1 (dd, J ) 18, 5); 19F NMR
(D2O) δ -206.1 (dd, J ) 53.3, 17.5); 31P NMR (D2O) δ 1.4; high-
1
28%. Spectral data for the R isomer (13r): H NMR (CDCl3) δ 2.03
(3H, s), 2.10 (3H, s), 2.11 (3H, s), 2.14 (3H, s), 4.13 (1H, dd, J )
12.0, 6.0, 6-H), 4.13 (1H, t, J ) 6.0, 4-H), 4.28 (1H, dd, J ) 12.0, 4.2,
6-H), 5.10 (1H, ddd, J ) 52.2, 6.0, 4.2, 2-H), 5.26 (1H, td, J ) 6.0,
4.2, 5-H), 5.57 (1H, dt, J ) 16.2, 6.0, 3-H), 6.27 (1H, d, J ) 4.2,
1-H); 13C NMR (CDCl3) δ 20.66, 20.68, 20.8, 21.0, 62.0, 70.0, 73.7
(d, J ) 24), 79.1 (d, J ) 8), 92.1 (d, J ) 203), 92.4 (d, J ) 18), 169.4,
169.7, 169.8, 170.4; 19F NMR (CDCl3) δ -206.5 (dd, J ) 52.3, 15.8);
high-resolution FABMS calcd for C14H20FO9 [M + H]+ 351.1091,
(11) Zhang, Q.; Cone, M. C.; Gould, S. J.; Zabriskie, T. M. Tetrahedron
2000, 56, 693-701.
(12) Burkart, M. D.; Zhang, Z.; Hung, S.-C.; Wong, C.-H. J. Am. Chem.
Soc. 1997, 119, 11743-11746.

Mechanistic InVestigation of UDP-Galactopyranose Mutase
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6759
resolution FABMS calcd for C6H12FO8PNa [M + Na]+ 285.0152, found
m/z 285.0152.
was kept at 75 °C for 6 h. At the end of the incubation, solvent was
evaporated under reduced pressure. The residue was chromatographed
on a silica gel column which was eluted with EtOAc/hexanes (1:1).
The desired product was isolated as a mixture of R and â anomers in
71% total yield (3.1 g). Spectral data for this mixture: 1H NMR (CDCl3)
δ 2.05 (3H, s), 2.06 (3H, s), 2.10 (3H, s), 2.11 (3H, s), 4.08-4.20
(3H, m), 4.25-4.35 (4H, m), 4.48 (1H, ddd, J ) 24.0, 4.5, 3.5, 4-H),
4.79 (1H, ddd, J ) 52.5, 3.5, 1.5, 3-H), 4.95 (1H, dt, J ) 54, 4.5,
3-H), 5.26-5.37 (4H, m); 13C NMR (CDCl3) δ 20.70, 20.72, 20.9,
21.0, 62.4, 62.6, 70.2 (d, J ) 6), 71.6 (d, J ) 7), 75.3 (d, J ) 23), 78.5
(d, J ) 27), 79.3 (d, J ) 24), 80.9 (d, J ) 28), 96.5 (d, J ) 7), 96.8
(d, J ) 184), 97.0 (d, J ) 185), 102.4 (d, J ) 4), 171.0, 171.2, 171.31,
171.33; 19F NMR (CDCl3) δ -194.4 (ddd, J ) 53.6, 23.4, 18.9), -186.2
(ddd, J ) 52.8, 24.0, 15.2); high-resolution CIMS calcd for C10H19-
FNO7 [M + NH4]+ 284.1146, found m/z 284.1169.
1,2,5,6-Tetra-O-acetyl-3-deoxy-3-fluoro-r,â-D-galactofuranose (22).
A solution of 21 (3.1 g, 12 mmol, a mixture of R and â anomers) in
pyridine (10 mL) and acetic anhydride (10 mL) was stirred at room
temperature for 2 h. Removal of solvent to dryness under reduced
pressure gave the desired product 22 as a mixture of R and â anomers
in quantitative yield (3.78 g). This mixture was pure enough to be used
in the next step of synthesis without purification: 1H NMR (CDCl3) δ
2.08 (3H, s), 2.09 (3H, s), 2.10 (3H, s), 2.12 (3H, s), 2.13 (9H, s), 2.15
(3H, s), 4.13 (1H, dd, J ) 12.5, 6.0), 4.19 (1H, dd, J ) 11.5, 7.0),
4.29 (1H, ddd, J ) 20.5, 7.5, 5.5), 4.34 (1H, dd, J ) 12.0, 4.0), 4,35
(1H, dd, J ) 12.0, 3.5), 4.52 (1H, dt, J ) 23.5, 4.0), 4.93 (1H, ddt, J
) 51.3, 4.2, 0.9), 5.21 (1H, ddd, J ) 56.0, 6.5, 5.5), 5.25-5.40 (4H,
m), 6.21 (1H, s), 6.36 (1H, d, J ) 4.5); 13C NMR (CDCl3) δ 20.4,
20.60, 20.66, 20.69, 20.79, 20.81, 21.0 (2C), 62.0, 62.2, 69.1 (d, J )
8), 70.5 (d, J ) 5), 75.6 (d, J ) 23), 78.6 (d, J ) 27), 80.0 (d, J ) 29),
82.6 (d, J ) 28), 92.9 (d, J ) 10), 93.3 (d, J ) 190), 94.8 (d, J )
187), 98.8 (d, J ) 3), 169.0, 169.1, 169.2, 169.6, 170.0, 170.1, 170.5;
19F NMR (CDCl3) δ -199.4 (dt, J ) 55.9, 21.2), -188.6 (ddd, J )
51.9, 24.0, 18.2); high-resolution FABMS calcd for C14H19FO9Na [M
+ Na]+ 373.0911, found m/z 373.0924.
Dibenzyl (2,5,6-Tri-O-acetyl-3-deoxy-3-fluoro-r-D-galactofura-
nosyl)phosphate (23r) and Dibenzyl (2,5,6-Tri-O-acetyl-3-deoxy-
3-fluoro-â-D-galactofuranosyl)phosphate (23â). A solution of 22 (2.0
g, 5.7 mmol) in anhydrous dichloromethane (10 mL) was cooled with
an ice-water bath. To this solution was slowly added bromotrimeth-
ylsilane (5 mL), and the mixture was stirred at room temperature for
18 h. The solution was evaporated to dryness under reduced pressure.
Anhydrous toluene (5 mL) was then added and dried in vacuo to
facilitate the evaporation of more volatile silane derivatives. The
resulting dry residue and dibenzyl phosphate (2.4 g, 8.4 mmol) was
redissolved in anhydrous toluene (10 mL). To this solution was added
triethylamine (1.2 mL, 8.4 mmol), and the mixture was stirred over-
night at room temperature. After removal of solvent in vacuo, the
residue was chromatographed on a silica gel column which was eluted
with EtOAc/hexanes (1:1). The fast-moving unreacted starting material
(724 mg, 36%) was recovered. The R isomer of the desired product
was eluted next from the column and was further purified on silica gel
with CHCl3/methanol (99:1) to give pure 23r in 8% yield (252 mg).
The next-eluted component was the â isomer of the desired product
(23â) that was isolated in 17% yield (539 mg). Spectral data for the R
isomer 23r: 1H NMR (CDCl3, 500 MHz) δ 1.96 (3H, s), 2.01 (3H, s),
2.05 (3H, s), 4.13 (1H, dd, J ) 12.0, 6.5, 6-H), 4.32 (1H, dd, J ) 12.0,
4.5, 6-H), 4.34 (1H, dt, J ) 20.0, 7.0, 4-H), 5.04 (4H, d, J ) 7.5,
benzylic CH2’s), 5.15 (1H, dt, J ) 56.5, 7.0, 3-H), 5.26 (1H, dddd, J
) 21.0, 7.0, 4.5, 2.5, 2-H), 5.29 (1H, m, 5-H), 5.99 (1H, dd, J ) 7.0,
4.5, 1-H), 7.34 (10H, m, Ar-H’s); 13C NMR (CDCl3, 125 MHz) δ 20.2,
20.6 (2C), 61.9, 69.4 (2C, d, J ) 6), 69.8 (d, J ) 3), 75.8 (dd, J ) 23,
7), 78.6 (d, J ) 26), 91.9 (d, J ) 188), 97.0 (dd, J ) 10, 5), 127.9
(2C), 128.0 (2C), 128.61 (2C), 128.63 (3C), 128.7, 135.5 (d, J ) 7),
135.6 (d, J ) 7), 169.8, 170.1, 170.4; 31P NMR (CDCl3) δ -2.5; 19F
NMR (CDCl3) δ -203.7 (dt, J ) 56.4, 20.3); high-resolution FABMS
calcd for C26H31FO11P [M + H]+ 569.1588, found m/z 569.1633.
Spectral data for the â isomer 23â: 1H NMR (CDCl3) δ 2.02 (3H, s),
2.07 (3H, s), 2.11 (3H, s), 4.10 (1H, dd, J ) 12.0, 6.9, 6-H), 4.24 (1H,
dd, J ) 12.0, 4.0, 6-H), 4.49 (1H, dt, J ) 24.2, 4.0, 4-H), 4.89 (1H,
dd, J ) 51.6, 4.0, 3-H), 5.07 (2H, d, J ) 8.0, benzylic CH2), 5.08 (2H,
Bis(triethylammonium) UDP-[2-F]Galf (9). Compound 16 (310
mg, 0.67 mmol) was dried by repetitive coevaporation with anhydrous
pyridine (3 mL). To this residue was added UMP-morpholidate (1.2 g,
1.7 mmol) in anhydrous pyridine (3 mL), and the solution was
evaporated again to dryness under vacuum. This was followed by the
addition of 1H-tetrazole (155 mg, 2.0 mmol) in anhydrous pyridine (3
mL), and the resulting solution was stirred at room temperature for 34
h. Removal of the solvent under vacuum gave a solid residue which
was dissolved in methanol and chromatographed on a Sephadex LH-
20 column (2.5 × 120 cm) using methanol as the eluent. Fractions
containing the desired product, judging by TLC analysis (EtOH:
NH4OH:H2O ) 5:3:1), were pooled and evaporated to dryness under
reduced pressure. Compound 9 was further purified by HPLC using a
C18 column (10 × 250 mm). The eluent was 1% aqueous acetonitrile
in 50 mM triethylammonium acetate, pH 6.8, and the flow rate was 5
mL/min. The retention time of 9 under these conditions was 14.4 min.
The yield was 43% (70 mg) after lyophilization: 1H NMR (D2O, 500
MHz) δ 1.30 (18H, t, J ) 7.2, Et3N-Me), 3.22 (12H, q, J ) 7.2, Et3N-
CH2), 3.66 (1H, dd, J ) 12.0, 7.0, 6-H), 3.73 (1H, dd, J ) 12.0, 5.0,
6-H), 3.82 (1H, dt, J ) 7.0, 5.0, 5-H), 3.89 (1H, dd, J ) 7.0, 5.0,
4-H), 4.20-4.26 (2H, m, 5′-H2), 4.28-4.31 (1H, m, 4′-H), 4.36-4.41
(2H, m, 2′-H, 3′-H), 4.57 (1H, dt, J ) 17.5, 7.0, 3-H), 5.05 (1H, dddd,
J ) 53.0, 7.0, 5.3, 1.5, 2-H), 5.79 (1H, t, 5.3, 1-H), 5.97 (1H, d, J )
8.0, 5′′-H), 6.03 (1H, d, J ) 4.5, 1′-H), 7.94 (1H, d, J ) 8.0, 6′′-H);
13C NMR (D2O) δ 9.8 (Et3N-Me), 48.1 (Et3N-CH2), 63.5 (C-6), 66.5
(d, J ) 5, C-5′), 71.2 (C-2′), 73.2 (d, J ) 21, C-3), 73.3 (C-5), 75.2
(C-3′), 82.6 (d, J ) 10, C-4), 84.9 (d, J ) 9, C-4′), 89.8 (C-1′), 95.7
(dd, J ) 198, 7, C-2), 96.8 (dd, J ) 18, 5, C-1), 104.2 (C-5′′), 143.3
(C-6′′), 153.3 (C-2′′), 167.6 (C-4′′); 31P NMR (D2O) δ -11.1 (d, J )
20), -12.8 (d, J ) 20); 19F NMR (D2O) δ -204.6 (dt, J ) 52.8, 18.1);
high-resolution FABMS calcd for C15H23FN2O16P2Na [M + Na]+
591.0405, found m/z 591.0441.
Synthesis of UDP-[3-F]Galf (10). 1,2-O-Isopropylidene-3-deoxy-
3-fluoro-r-D-galactofuranose (19). 1,2:5,6-Di-O-isopropylidene-3-
deoxy-3-fluoro-R-D-galactofuranose (18, 4.7 g, 18 mmol), which was
prepared from 17 by a literature procedure,13 was dissolved in 80%
aqueous acetic acid (40 mL) and kept at 70 °C for 25 min. The mixture
was evaporated to dryness under reduced pressure, and the residual oil
was subjected to chromatography on silica gel and eluted with EtOAc/
hexanes (1:1). The desired product 19 was isolated in 93% yield (3.7
g): 1H NMR (CDCl3) δ 1.33 (3H, s), 1.52 (3H, s), 2.24 (1H, s, exch.),
2.89 (1H, s, exch.), 3.66 (1H, dd, J ) 11.7, 4.5, 6-H), 3.78 (1H, dd, J
) 11.7, 3.9, 6-H), 3.86 (1H, dddd, J ) 7.5, 4.5, 3.9, 0.9, 5-H), 4.31
(1H, dddd, J ) 24.0, 7.5, 1.8, 0.9, 4-H), 4.78 (1H, dd, J ) 15.3, 4.2,
2-H), 5.00 (1H, dd, J ) 51.0, 1.8, 3-H), 5.98 (1H, d, J ) 4.2, 1-H);
13C NMR (CDCl3) δ 25.9, 26.5, 63.4, 69.9 (d, J ) 9), 84.4 (d, J )
32), 85.9 (d, J ) 25), 95.3 (d, J ) 182), 105.6, 113.5; 19F NMR (CDCl3)
δ -187.3 (ddd, J ) 50.1, 24.1, 15.0); high-resolution CIMS calcd for
C9H19FNO5 [M + NH4]+ 240.1247, found m/z 240.1263.
5,6-Di-O-acetyl-1,2-O-isopropylidene-3-deoxy-3-fluoro-r-D-ga-
lactofuranose (20). A solution of 19 (3.7 g, 17 mmol) in pyridine (10
mL) and acetic anhydride (10 mL) was stirred at room temperature for
2 h. Removal of the solvent under reduced pressure afforded the
acetylated product 20 in quantitative yield (5.1 g). The product was
pure enough to be used in the next step without purification: 1H NMR
(CDCl3) δ 1.34 (3H, s), 1.57 (3H, s), 2.06 (3H, s), 2.10 (3H, s), 4.17
(1H, dd, J ) 12.2, 5.8, 6-H), 4.30 (1H, ddd, J ) 24.8, 7.0, 3.0, 4-H),
4.36 (1H, dd, J ) 12.2, 4.0, 6-H), 4.73 (1H, dd, J ) 14.2, 3.8, 2-H),
4.96 (1H, dd, J ) 50.8, 3.0, 3-H), 5.34 (1H, ddd, J ) 7.0, 5.8, 4.0,
5-H), 5.90 (1H, d, J ) 3.8, 1-H); 13C NMR (CDCl3) δ 20.7, 20.9, 26.3,
26.7, 62.7, 69.7 (d, J ) 9), 82.6 (d, J ) 28), 84.2 (d, J ) 32), 94.1 (d,
J ) 182), 105.1, 114.0, 170.1, 170.5; 19F NMR (CDCl3) δ -187.7
(ddd, J ) 50.8, 25.0, 14.5); high-resolution FABMS calcd for C13H20-
FO7 [M + H]+ 307.1193, found m/z 307.1181.
5,6-Di-O-acetyl-3-deoxy-3-fluoro-r,â-D-galactofuranose (21). A
solution of 20 (5.0 g, 16 mmol) in 80% aqueous acetic acid (100 mL)
(13) (a) Brimacombe, J. S.; Foster, A. B.; Hem, R.; Westwood, J. H.;
Hall, L. D. Can. J. Chem. 1970, 48, 3946-3952. (b) Kovac, P.; Glaudemans,
C. P. J. Carbohydr. Res. 1983, 123, 326-331.

6760 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Zhang and Liu
d, J ) 8.2, benzylic CH2), 5.22 (1H, d, J ) 15.0, 2-H), 5.33 (1H, dt,
J ) 6.9, 4.0, 5-H), 5.84 (1H, d, J ) 5.2, 1-H), 7.35 (10H, m, Ar-H’s);
13C NMR (CDCl3) δ 20.6, 20.7, 20.8, 62.1, 69.0 (d, J ) 6), 69.5 (d, J
) 6), 69.7 (d, J ) 5), 80.7 (dd, J ) 29, 11), 82.8 (d, J ) 29), 94.4 (d,
J ) 187), 102.6 (dd, J ) 5, 3), 127.9 (2C), 128.1 (2C), 128.6 (4C),
gene (glf)6 for the mutase derived from E. coli C600 (ATCC 23724).
The E. coli BL21 (DE3)/pQZ-1 was grown in Luria-Bertani (LB) broth
supplemented with kanamycin (50 µg/mL) at 37 °C for 18 h with
vigorous agitation. The cells were harvested, and the C-terminal His6-
tagged mutase was purified to near homogeneity using Ni-NTA resin
(Qiagen) according to the protocols recommended by the manufacturer.
Details of the cloning, expression, and purification of the enzyme had
been described in an earlier report.10a The purified mutase (Glf) was
found to be stable in phosphate buffer containing 15% glycerol. The
yield was about 100 mg of the pure protein from 6 L of culture. The
enzyme exhibits a typical unresolved flavin spectrum with absorbance
maxima at 450 and 376 nm. A stoichiometry of 0.52 bound FAD per
protein monomer was estimated by quantitation of FAD released from
an enzyme sample of known concentration.
Protein Assay. Protein concentrations were routinely determined
by the Bradford method14 using bovine serum albumin as the standard.
This assay was calibrated by comparing the results from quantitative
amino acid analysis performed on aliquots of the same sample by the
Microchemical Facility at the Institute of Human Genetics of the
University of Minnesota.
Activity Assay. Whether the mutase could catalyze the turnover of
a substrate or be inhibited by an inhibitor was determined by incubating
the substrate/inhibitor sample (1 mM) and an appropriate amount of
the mutase in 30 µL of 100 mM potassium phosphate buffer (pH 7.6).
For assays conducted under reducing conditions, 20 mM freshly
prepared sodium dithionite was included in the mixture. The reaction
was usually carried out at 37 °C for 2 min, and the resulting mixture
was immediately frozen with liquid nitrogen to terminate the incubation.
The reaction mixture was analyzed by HPLC using a C18 column
(Microsorb-MV, Varian, 4.6 × 250 mm) which was eluted with 1.5%
acetonitrile in 50 mM triethylammonium acetate buffer, pH 6.8. The
detector was set at 262 nm, and the flow rate was 1.0 mL/min. The
extent of conversion was determined by comparing the integration of
the substrate and product peaks. The conversion was routinely controlled
to be within 15% by properly adjusting the enzyme concentration.
Determination of Kinetic Parameters. The kinetic parameters of
those compounds that could serve as substrates were determined by
preparing a series of samples containing the purified mutase and the
specific compound being tested in a total volume of 30 µL of 100 mM
potassium phosphate (pH 7.5). The incubation mixture also contained
20 mM freshly prepared sodium dithionite. Each sample was incubated
at 37 °C for 5 min and frozen with liquid nitrogen to terminate the
reaction. The contents of these samples were analyzed by HPLC as
described above, and the ratio of product and substrate peaks was used
to determine the activity. The kinetic parameters were deduced by fitting
these data to the Michaelis-Menten equation. For UDP-[2-F]Galf (9),
the incubation mixture contained 217 nM enzyme and one of the
following concentrations of 9: 0.083, 0.10, 0.125, 0.167, 0.25, 0.50,
1.0, and 2.0 mM. For UDP-[3-F]Galf (10), the incubation mixtures
contained 7.2 µM enzyme and one of the following concentrations of
10: 0.20, 0.25, 0.30, 0.50, 0.70, 1.0, 1.5, 2.0, 3.0, 4.0, and 8.0 mM.
Production and Characterization of UDP-[2-F]Galp (26). A
mixture of UDP-[2-F]Galf (9, 15 mg) and the purified mutase (6.3 mg)
in 1 mL of 100 mM potassium phosphate buffer containing sodium
dithionite (10 mg) was kept at room temperature for 16 h. HPLC
analysis as described above indicated that the conversion was greater
than 95%. The turnover product 26 was purified by an HPLC C18
semipreparative column eluted with 50 mM triethylammonium acetate
buffer (pH 6.8) containing 1.5% acetonitrile. Lyophilization of the
pooled HPLC fractions yielded 14 mg of 26: 1H NMR (D2O, 500 MHz)
δ 1.30 (18H, t, J ) 7.5, Et3N-Me), 3.22 (12H, q, J ) 7.5, Et3N-CH2),
3.74 (1H, dd, J ) 12.0, 4.5, 6-H), 3.78 (1H, dd, J ) 12.0, 7.5, 6-H),
4.12 (1H, t, J ) 3.0, 4-H), 4.17-4.34 (5H, m, 3-H, 5-H, 4′-H, 5′-H2),
4.40 (2H, m, 2′-H, 3′-H), 4.72 (1H, ddt, J ) 49.0, 9.5, 3.5, 2-H), 5.84
(1H, dd, J ) 7.5, 3.5, 1-H), 6.01 (1H, d, J ) 8.0, 5′′-H), 6.02 (1H, d,
J ) 4.5, 1′-H), 8.00 (1H, d, J ) 8.0, 6′′-H); 13C NMR (D2O, 125 MHz)
δ 9.8 (Et3N-Me), 48.2 (Et3N-CH2), 62.3 (C-6), 66.5 (d, J ) 4, C-5′),
69.2 (d, J ) 17, C-3), 71.2 (d, J ) 9, C-4), 71.3 (C-2′), 73.4 (C-5),
75.4 (C-3′), 85.0 (d, J ) 9, C-4′), 89.7 (dd, J ) 184, 8, C-2), 89.9
128.7 (2C), 135.4 (d, J ) 7), 135.5 (d, J ) 7), 169.1, 170.0, 170.4; 31
P
NMR (CDCl3) δ -3.0; 19F NMR (CDCl3) δ -180.3 (ddd, J ) 51.4,
24.3, 15.8); high-resolution FABMS calcd for C26H31FO11P [M + H]+
569.1588, found m/z 569.1613.
Triethylammonium (2,5,6-Tri-O-acetyl-3-deoxy-3-fluoro-r-D-ga-
lactofuranosyl)phosphate (24). A solution of 23r (240 mg, 0.42 mmol)
in ethyl acetate (5 mL) and triethylamine (0.4 mL) was stirred overnight
at room temperature under a hydrogen atmosphere in the presence of
Pd/C (10%). The mixture was filtered through a Celite pad, and the
filtrate was evaporated to dryness under reduced pressure to give 24 in
96% yield (198 mg): 1H NMR (CDCl3) δ 1.20 (9H, t, J ) 7.2), 1.94
(3H, s), 2.02 (3H, s), 2.03 (3H, s), 2.95 (6H, q, J ) 7.2), 4.09 (1H, dd,
J ) 12.0, 6.0, H-6), 4.13 (1H, dt, J ) 20.7, 6.0, 4-H), 4.32 (1H, dd, J
) 12.0, 3.9, 6-H), 5.02-5.30 (3H, m, 2-, 3-, 5-H’s), 5.72 (1H, dd, J )
7.2, 3.8, 1-H); 13C NMR (CDCl3) δ 8.4, 20.7 (2C), 20.9, 45.3, 62.0,
70.9 (d, J ) 3), 75.8 (dd, J ) 21, 6), 77.3 (d, J ) 26), 93.0 (d, J )
187), 95.2 (dd, J ) 10, 4), 170.3, 170.47, 170.49; 19F NMR (CDCl3)
δ -201.9 (dt, J ) 57.0, 19.2); 31P NMR (CDCl3) δ -0.33; high-
resolution FABMS calcd for C12H18FO11PNa [M + Na]+ 411.0468,
found m/z 411.0472.
Bis(triethylammonium) (3-Deoxy-3-fluoro-r-D-galactofuranosyl)-
phosphate (25). A solution of 24 (310 mg, 0.63 mmol) in 30 mL of a
mixture of methanol/water/triethylamine (5:2:1) was stirred at room
temperature for 3 days until TLC (ethanol:NH4OH:H2O ) 5:3:1)
indicated that the reaction was complete. The mixture was evaporated
to dryness under reduced pressure, and the desired product was obtained
in quantitative yield (290 mg): 1H NMR (D2O) δ 1.30 (18H, t, J )
7.2), 3.22 (12H, q, J ) 7.2), 3.67 (1H, dd, J ) 12.0, 6.6, 6-H), 3.75
(1H, dd, J ) 12.0, 4.5, 6-H), 3.86 (1H, td, J ) 6.6, 4.5, 5-H), 4.14
(1H, dt, J ) 21.0, 6.6, 4-H), 4.51 (1H, dddd, J ) 22.2, 6.6, 4.7, 1.2,
2-H), 5.14 (1H, dt, J ) 57.3, 6.6, 3-H), 5.59 (1H, t, J ) 4.7, 1-H); 13
C
NMR (D2O) δ 9.7, 47.8, 63.3, 73.2 (d, J ) 3), 77.0 (dd, J ) 21, 7),
81.3 (d, J ) 25), 97.9 (d, J ) 182), 98.1 (dd, J ) 12, 6); 19F NMR
(D2O) δ -202.8 (dt, J ) 57.3, 22.6); 31P NMR (D2O) δ 0.87; high-
resolution FABMS calcd for C12H28FNO8P [M + Et3N + H]+ 364.1537,
found m/z 364.1533.
Bis(triethylammonium) UDP-[3-F]Galf (10). Preparation of 10
followed a procedure analogous to that described for the synthesis of
9. The reaction mixture contained 25 (270 mg, 0.58 mmol), UMP-
morpholidate (1.0 g, 1.45 mmol), and 1H-tetrazole (135 mg, 1.89 mmol)
in 9 mL of anhydrous pyridine. The resulting solution was stirred at
room temperature for 30 h. Removal of the solvent under vacuum gave
a solid residue which was chromatographed on a Sephadex LH-20
column (2.5 5 120 cm) using methanol as the eluent. The crude 10
was further purified by HPLC (C18 column, eluted with 1% acetonitrile
in 50 mM triethylammonium acetate buffer, pH 6.8, flow rate 5.0 mL/
min). The retention time of 10 under these conditions was 19.5 min,
and the yield was 15% (32 mg): 1H NMR (D2O) δ 1.31 (18H, t, J )
7.2, Et3N-Me), 3.24 (12H, q, J ) 7.2, Et3N-CH2), 3.68 (1H, dd, J )
12.0, 6.5, 6-H), 3.76 (1H, dd, J ) 12.0, 4.0, 6-H), 3.87 (1H, td, J )
6.5, 4.0, 5-H), 4.17 (1H, dt, J ) 20.5, 6.5, 4-H), 4.22-4.29 (2H, m,
5′-H2), 4.31 (1H, s, br, 4′-H), 4.38-4.42 (2H, m, 2′-H, 3′-H), 4.56
(1H, dm, J ) 22.0, 2-H), 5.14 (1H, dt, J ) 57, 6.5, 3-H), 5.72 (1H, t,
5.0, 1-H), 5.99 (1H, d, J ) 8.0, 5′′-H), 6.01 (1H, d, J ) 5.0, 1′-H),
7.99 (1H, d, J ) 8.0, 6′′-H); 13C NMR (D2O) δ 9.8 (Et3N-Me), 48.2
(Et3N-CH2), 63.3 (C-6), 66.5 (d, J ) 5, C-5′), 71.2 (C-2′), 73.7 (d, J )
4, C-5), 75.3 (C-3′), 76.9 (dd, J ) 21, 7, C-2), 81.9 (d, J ) 25, C-4),
84.8 (d, J ) 9, C-4′), 89.9 (C-1′), 97.4 (d, J ) 182, C-3), 99.4 (dd, J
) 12, 6, C-1), 104.2 (C-5′′), 143.2 (C-6′′), 153.4 (C-2′′), 167.8
(C-4′′); 31P NMR (D2O) δ -10.9 (d, J ) 20), -12.4 (d, J ) 20); 19F
NMR (D2O) δ -201.5 (dt, J ) 57.4, 20.5); high-resolution negative
ion FABMS calcd for C15H22FN2O16P2 [M - H]- 567.0429, found m/z
567.0465.
Enzyme. UDP-galactopyranose mutase used in this study was
purified from a recombinant strain of Escherichia coli BL21 (DE3)10a
which contained an expression plasmid (pQZ1) harboring the encoding
(14) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.

Mechanistic InVestigation of UDP-Galactopyranose Mutase
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6761
(C-1′), 94.4 (dd, J ) 22, 6, C-1), 104.3 (C-5′′), 143.3 (C-6′′), 153.4
(C-2′′), 167.8 (C-4′′); 31P NMR (D2O) δ -11.1 (d, J ) 20), -12.8 (d,
J ) 20); 19F NMR (D2O) δ -206.2 (ddd, J ) 49.4, 12.1, 3.7); negative
ion high-resolution ESIMS calcd for C15H22FN2O16P2Na [M - H]-
567.0429, found m/z 567.0416.
individually incubated with native UDP-Galp mutase (2 µM) on ice
for 1 h in 24 µL of 100 mM potassium phosphate buffer (pH 7.6). The
residual enzyme activity was determined as described in the section,
Activity Assay. For the protection experiments, compounds 10 and 27
(1 mM each) were individually included in the assay mixture containing
UDP-Galf (2, 1 mM) as the substrate. Either native UDP-Galp mutase
(2.1 µM) or reduced enzyme (53 nM) was added to the mixture, and
the activity was determined as described.
Kinetic Analysis of Inactivation of UDP-Galactopyranose Mutase
by UDP-[3-F]Galf. In a typical inactivation experiment, an appropriate
amount of UDP-[3-F]Galf (10) was incubated with mutase (15 µM) in
30 µL of 50 mM potassium phosphate buffer, pH 7.6, at 25 °C. At
various time intervals, 3 µL aliquots of the incubation mixture were
added to the standard assay cocktail (27 µL) to initiate the reaction.
The residual enzyme activity was determined as described in the section,
Activity Assay.
Test for Reversibility of the Inactivation by UDP-[3-F]Galf. A
sample containing UDP-[3-F]Galf (10, 3 mM) and mutase (6 µM) was
incubated for 1 h on ice. The inactivated enzyme was dialyzed against
300 mL of 50 mM potassium phosphate buffer containing 15% glycerol
(pH 7.6) over 9 h with two changes of buffer. The residual enzyme
activity was determined prior to and after the dialysis. A control experi-
ment was also performed under identical conditions without the inac-
tivator. It was noted that the activity of the control sample decreased
significantly (>50%) when the dialysis was allowed to run for longer
than 9 h.
Production and Characterization of UDP-[3-F]Galp (27). A
mixture of UDP-[3-F]Galf (10, 10 mg) and the purified mutase (7.1
mg) in 1.1 mL of 100 mM potassium phosphate buffer containing
sodium dithionite (10 mg) was kept at room temperature overnight.
The turnover product 27 was isolated by HPLC as described above for
the purification of 26. Lyophilization of the pooled HPLC fractions
yielded 8 mg of the product: 1H NMR (D2O, 500 MHz) δ 1.30 (18H,
t, J ) 7.5, Et3N-Me), 3.22 (12H, q, J ) 7.5, Et3N-CH2), 3.77 (1H, dd,
J ) 12.0, 5.5, 6-H), 3.81 (1H, dd, J ) 12.0, 7.0, 6-H), 4.12 (1H, tt, J
) 8.5, 3.5, 2-H), 4.19-4.36 (5H, m, 4-H, 5-H, 4′-H, 5′-H2), 4.40 (2H,
m, 2′-H, 3′-H), 4.84 (1H, ddd, J ) 49.5, 8.5, 3.5, 3-H), 5.72 (1H, dt,
J ) 7.0, 3.5, 1-H), 5.99 (1H, d, J ) 8.5, 5′′-H), 6.01 (1H, d, J ) 3.5,
1′-H), 7.98 (1H, d, J ) 8.5, 6′′-H); 13C NMR (D2O, 125 MHz) δ 9.8
(Et3N-Me), 48.2 (Et3N-CH2), 62.2 (C-6), 66.5 (d, J ) 6, C-5′), 68.6
(dd, J ) 19, 9, C-2), 68.9 (d, J ) 17, C-4), 71.2 (C-2′), 73.9 (d, J )
6, C-5), 75.4 (C-3′), 84.8 (d, J ) 9, C-4′), 91.7 (C-1′), 92.5 (d, J )
181, C-3), 97.4 (dd, J ) 11, 7, C-1), 104.3 (C-5′′), 143.2 (C-6′′), 153.7
(C-2′′), 168.2 (C-4′′); 31P NMR (D2O) δ -10.5 (d, J ) 21), -12.2 (d,
J ) 21); 19F NMR (D2O) δ -201.6 (doublet of quintet, J ) 49.4, 6.2);
negative ion high-resolution ESIMS calcd for C15H22FN2O16P2Na [M
- H]- 567.0429, found m/z 567.0456.
Equilibrium Constants. The equilibrium constant, Keq, was calcu-
lated from the ratio of product to substrate at equilibrium. The ratios
were determined for both the forward and reverse directions by
integration of the corresponding peaks from HPLC chromatograms.
Specifically, a 200 µL incubation mixture containing 65 µM mutase,
20 mM sodium dithionite, and 1 mM UDP-[2-F]Galf (9) or UDP-[3-
F]Galf (10) in 100 mM potassium phosphate buffer (pH 7.6) was
allowed to reach equilibrium, as determined by the constant product/
substrate ratio by HPLC. The same procedure was also repeated with
a 200 µL incubation of 65 µM enzyme and 1 mM UDP-[2-F]Galp
(26) or UDP-[3-F]Galp (27) in 100 mM potassium phosphate buffer
(pH 7.6) containing 20 mM sodium dithionite.
Positional Isotope Exchange (PIX) Experiments. The labeled
UDP-[1-13C,18O]Galp used in these experiments was prepared according
to a known procedure.9,15 The desired product was purified by reversed-
phase HPLC as described in the preparation of 9 and 10 and was verified
by 13C NMR and mass spectroscopy. The m/z 568.0594 determined by
the high-resolution negative ion ESIMS matches well with [M - H]-
568.0554 calculated for C1413CH23N2O1618OP2. For the experiments
performed with native enzyme, 10 µL of UDP-Galp mutase (3 µM
final concentration) was added to a solution of UDP-[1-13C,18O]Galp
(10 mM) in 100 mM potassium phosphate buffer (pH 7.6) prepared
with D2O. The reaction was conducted in an NMR tube, and the total
volume of the incubation mixture was 610 µL. The 13C NMR spectra
of the resulting sample were recorded before the addition of mutase
and at 10, 40, 80, 180, and 300 min after the addition of mutase. For
the experiments performed with reduced enzyme, the same conditions
were employed except that the substrate concentration was 16 mM and
freshly prepared sodium dithionite (20 mM) was included in the buffer.
Again, 13C NMR spectra were recorded before the addition of enzyme
and at 5, 10, 20, and 40 min after the addition of mutase.
Results and Discussion
The fluorine-containing sugars, while none are naturally
occurring, have attracted much attention due to their high clinical
potential as enzyme inhibitors/chemotherapeutic agents.16 Two
fluorinated galactofuranose derivatives, UDP-[2-F]Galf (9) and
UDP-[3-F]Galf (10), were synthesized in this work and used as
probes to investigate the catalytic mechanism of UDP-Galp
mutase.17 Since the equilibrium of this enzymatic transformation
greatly favors the pyranose over the furanose form (Keq
)
0.057),10a,18 both probes were prepared as furanose sugars. The
availability of these furanose derivatives had allowed the
enzyme-catalyzed interconversion to be monitored in the reverse
reaction (2 f 1), thus facilitating the kinetic and mechanistic
analyses.
A number of factors were considered in the design rationales
of 9 and 10. First, UDP-Galp mutases isolated from different
sources are all flavoproteins. In light of the redox capability of
this coenzyme, a mechanism relying on redox recycling of the
flavin coenzyme as the driving force for catalysis is certainly
feasible. Although failure to recognize the cleavage of the
anomeric C-O bond as a requisite step in the interconversion
of 1 and 2 has made the mechanism depicted in Scheme 1
invalid, a modified version, incorporating the anomeric C-O
bond scission as part of the pathway (see Scheme 5), is still
worth considering. Since oxidation of 1 at C-2 or C-3 of the
substrate is a key step of this revised mechanism,19 one of the
fluorinated probes is not expected to be processed by the mutase.
Instead, it may act as an inhibitor. Second, replacement of a
hydroxy group with an isosteric and isoelectronic fluorine at
loci adjacent to the anomeric center on the sugar substrate has
been shown to slow the catalysis of glycosidases by destabilizing
the corresponding carbocation transition states.20 In our case,
(16) (a) In Fluorinated Carbohydrates, Chemical and Biochemical
Aspects; Taylor, N. F., Ed.; American Chemical Society: Washington, DC,
1988. (b) Tsuchiya, T. AdV. Carbohydr. Chem. Biochem. 1990, 48, 91-277.
(17) Burton, A.; Wyatt, P.; Boons, G.-J. J. Chem. Soc., Perkin Trans. 1
1997, 2375-2382.
(18) Lee. R.; Monsey, D.; Weston, A.; Duncan, K.; Rithner, C.; McNeil,
M. Anal. Biochem. 1996, 242, 1-7.
(19) It has been suggested that the reaction may proceed via a 2-keto
intermediate; however, no detailed scheme has been proposed.6a
Inhibition of UDP-Galactopyranose Mutase by Fluorinated
Analogues. To test whether these fluorinated analogues are inhibitors
for the mutase, compounds 9, 10, 26, and 27 (1 mM each) were
(15) (a) Thiem, J.; Wiemann, T. Angew. Chem., Int. Ed. Engl. 1991, 30,
1163-1164. (b) Hayashi, T.; Murray, B. T.; Wang, R.; Wong, C.-H. Bioorg.
Med. Chem. 1997, 5, 497-500. (c) Risley, J. M.; Van Etten, R. L.
Biochemistry 1982, 21, 6360-6365.

6762 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Zhang and Liu
Scheme 3
Scheme 4
the fluorine group in 9 and 10 is expected to destabilize the
putative oxocarbenium ion intermediates/transition states (5 and
7), if the mechanism shown in Scheme 2 is operative. The
reduction of turnover rate may also allow the reaction inter-
mediates (5 and/or 7) to react with enzyme active-site residues,
resulting in enzyme inactivation. This strategy has been suc-
cessfully applied to elucidate the mechanism of the glycosidase-
catalyzed reaction which also involves cleavage of the anomeric
C-O bond.20 Thus, these fluorinated probes hold promise to
provide insight into the mechanism of this enzyme.
the presence of 1H-tetrazole. The reaction mixture was purified
by size exclusion chromatography and reversed-phase HPLC
under conditions similar to those used for the purification of
UDP-Galf.10 The structure of the final product was confirmed
by NMR (1H, 13C, 31P, 19F) and high-resolution FABMS. The
1
large electronegativity of fluorine helps spread the H NMR
signals associated with the galactofuranose moiety over a wide
range, thus facilitating the signal assignments by COSY. The
purified product was found to be stable in aqueous solution at
room temperature and neutral pH, since no decomposition was
Synthesis of UDP-[2-F]Galf (9). The first key step for the
synthesis of 9 was to convert the readily available 2-deoxy-2-
fluoro-D-galactose (12)12 to the thermodynamically less favored
furanose derivative. Among various methods tested, the acety-
lation of 12 at elevated temperature gave the best results and
led to tetraacetyl galactofuranose 13 (R:â ) 1:2) in 28% yield.
The prevalence of the trans configuration between the anomeric
OH and the fluorine substituent at C-2 is characteristic for
2-deoxy-2-fluorofuranose.21 As shown in Scheme 3, conversion
of 13 to 14 was achieved using bromotrimethylsilane and then
with triethylammonium dibenzyl phosphate. A large amount of
13 (65%) was recovered, and the â anomer in the recovered 13
was significantly enriched (R:â ) 1:3.6). The lack of spin
coupling between H-1 (δ 5.94, J1,2 ) 4.2 and J1H-P ) 6.0 Hz)
1
detected by H NMR within 24 h.
Synthesis of UDP-[3-F]Galf (10). Diisopropylidene-3-deoxy-
3-fluoro-R-D-galactofuranose (18)13 was converted, through a
stepwise deprotection and protection, to tetraacetyl furanose 22
as an ∼1:1 mixture of R,â anomeric isomers (Scheme 4). Upon
treatment with bromotrimethylsilane, the mixture was partially
converted to the corresponding tri-O-acetyl-3-deoxy-3-fluoro-
D-galactofuranosyl bromide, which was sensitive to silica gel.
Without purification, this relatively unstable compound was
subjected to excess triethylammonium dibenzyl phosphate in
toluene,22 and the resulting mixture was separated on silica gel
to give three main fractions. The fastest migrating material was
found to be unreacted 22. The second fraction contained
dibenzyl (2,5,6-tri-O-acetyl-3-deoxy-3-fluoro-R-D-galactofur-
anosyl)phosphate (23r). Its H-1 signal appears as a doublet of
doublets (J1,2 ) 4.5, JH-1,P ) 7.0), with the J1,2 value indicative
of a cis relationship between 1-H and 2-H.23 The major product
of this reaction was found in the most polar fraction. On the
basis of various spectral data, this product was assigned as
dibenzyl (2,5,6-tri-O-acetyl-3-deoxy-3-fluoro-â-D-galactofur-
anosyl)phosphate (23â). The anomeric proton resonates at δ
5.84 as a doublet (JH-1,P ) 5.2). Since no coupling exists be-
tween 1-H and 2-H, these protons must be trans to each other.23
The downfield shift of the C-1 resonance (δ 102.6) as compared
to that of a furanose with an R anomeric configuration is also
typical for a â isomer. Having the key intermediate 23r in hand,
the synthesis of the target compound was completed via inter-
mediates 24 and 25, by a sequence similar to that used for the
preparation of UDP-[2-F]Galf (9). Pure UDP-[3-F]Galf (10) was
1
and 2-F of 14 revealed in both H and 19F NMR spectra is
consistent with an R-anomeric configuration. The appearance
of C-1 at 96.4 ppm in 13C NMR is also in good agreement with
the assigned R configuration.22
Compound 14 was then debenzylated by catalytic hydrogena-
tion in the presence of triethylamine to afford 15 as a
monotriethylammonium salt in good yield. Deacetylation of 15
was readily achieved by treatment with methanol/water and
triethylamine, providing 2-deoxy-2-fluoro-R-D-galactofuranosyl
phosphate (16) as a bistriethylammonium salt in quantitative
yield. The final product UDP-[2-F]Galf (9) was prepared from
16 and uridine phosphomorpholidate in anhydrous pyridine in
(20) (a) Withers, S. G. Can. J. Chem. 1999, 77, 1-11. (b) Zechel, D.
L.; Withers, S. G. In ComprehensiVe Chemistry of Natural Products
Chemistry; Barton, D., Nakanishi, K., Meth-Cohn, O.; Eds.; Pergamon: New
York, 1999; Vol. 5, pp 279-314 and references therein.
(21) (a) McAtee, J. J.; Schinazi, R. F.; Liotta, D. C. J. Org. Chem. 1998,
63, 2161-2167. (b) Niihata, S.; Ebata, T.; Kawakami, H.; Matsushida, H.
Bull. Chem. Soc. Jpn. 1995, 65, 1509-1512.
(22) (a) de Lederkremer, R. M.; Nahmad, V. B.; Varela, O. J. Org. Chem.
1994, 59, 690-692. (b) Tsvetkov, Y. E.; Nikolaev, A. V. J. Chem. Soc.,
Perkin Trans. 1 2000, 889-891.
(23) Bundle, D. R.; Lemieux, R. U. Methods Carbohydr. Chem. 1976,
7, 79-86.

Mechanistic InVestigation of UDP-Galactopyranose Mutase
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6763
Scheme 5
obtained as a bistriethylammonium salt and was found to be
stable at room temperature and neutral pH in aqueous solution.
Incubation of UDP-[2-F]Galf and UDP-[3-F]Galf with
Reduced UDP-Mutase. When UDP-[2-F]Galf (9, 3 mM) was
incubated with UDP-Galp mutase (70 µM) in the presence of
20 mM sodium dithionite in a closed vial at 37 °C, over 50%
of the substrate was converted to a new product within 5 min,
as indicated by HPLC. The retention time for 9 on the C18
column was 15.3 min, and that for the new product was 9.3
min. Similar results were also found when UDP-[3-F]Galf (10,
3 mM) was incubated with the reduced mutase. The retention
times for 10 and its turnover product were 25.2 and 8.5 min,
respectively. These products were purified and identified as
UDP-[2-F]Galp (26) from 9 and UDP-[3-F]Galp (27) from 10
by NMR (1H, 13C, 31P, 19F, and COSY) and high-resolution MS.
Since the fluorine substituent is chemically inert, the fact that
both 9 and 10 can be recognized and processed by the E. coli
mutase clearly indicates that mechanisms (such as Scheme 5)
initiated by the oxidation of 2- or 3-OH during the intercon-
version of UDP-Galp and UDP-Galf can now be ruled out. This
is also consistent with our early conclusion that FAD is not
directly involved in the catalysis of UDP-Galp mutase.10a It is
important to note that a similar approach has been used in a
recent report to study the catalysis of mutase from Klebsiella
pneumoniae.8 On the basis of the observation that both chemi-
cally synthesized 26 and 27 are substrates for the native Kleb-
siella enzyme, a conclusion debunking the involvement of redox
chemistry in the mechanism of this enzyme was also reached.
Equilibrium Constants. The reaction catalyzed by UDP-
Galp mutase is a reversible process with a Keq value of 0.057
for the conversion of 1 to 2 at 37 °C.10a To determine the
equilibrium constants for the interconversion of the fluorinated
analogues, separate reactions containing each of the furanose
and the pyranose analogue were allowed to equilibrate at 37
°C. The reaction was monitored by HPLC until a constant
product/substrate ratio was reached. Under these conditions, the
equilibrium constants for the forward reaction (26 to 9, and 27
to 10) were determined to be 0.020 and 0.011 for the 2- and
3-fluorinated analogues, respectively. The replacement of a
hydroxy group at C-2 or C-3 of the galactose moiety with a
fluorine substituent apparently enhances the thermodynamic
preference for the pyranose over the furanose form by 0.65 and
1.0 kcal/mol, respectively. The enhanced preference for the
pyranose form has also been observed in D-ribose when its
2-hydroxy group is replaced with a flourine.24
Table 1. Kinetic Parameters for Reduced UDP-Galactopyranose
Mutase
substrates
Km (µM) kcat (s-1
)
kcat/Km
ratio
UDP-Galf (2)
UDP-[2-F]Galf (9)
UDP-[3-F]Galf (10)
22
65
861
27
0.033
5.7
1.2273
1
0.00051 4.1 × 10-4
0.0066
5.4 × 10-3
Kinetic Properties. The kinetic parameters for turnover of
9 and 10 in the reverse direction under reducing conditions were
determined by fitting the initial reaction rates to the Michaelis-
Menten equation based on nonlinear regression. The Km values
were determined to be 65 and 861 µM for UDP-[2-F]Galf (9)
and UDP-[3-F]Galf (10), respectively. The corresponding kcat
values were estimated to be 0.033 and 5.7 s-1 (Table 1). In
comparison to that of UDP-Galf (2), the catalytic efficiencies
(kcat/Km) for UDP-[2-F]Galf and UDP-[3-F]Galf have decreased
by approximately 4 and 3 orders of magnitude, respectively.
Since the Km values for UDP-[2-F]Galf (9) and UDP-Galf (2)
are comparable, the substitution of 2-OH with a fluorine appears
to have little effect on the interactions in the Michaelis complex.
In contrast, the large Km value for UDP-[3-F]Galf (10) implicates
the significance of the 3-hydroxy group of the galactose in
binding to the reduced enzyme to form the Michaelis complex.
Replacing the 3-OH with a fluorine atom may have altered the
hydrogen-bonding network essential for the interactions in the
Michaelis complex.
As indicated by the kcat values, both 9 and 10 are poorer
substrates than UDP-Galf (2). The rate reduction is especially
profound for 9, whose fluorine substituent is immediately
adjacent to the anomeric center. Such rate retardation is likely
due to destabilization of the oxocarbenium ion intermediates
or transition states by the electron-withdrawing fluorine group.
This inductive effect is expected to be distance-dependent and
should result in suppression of the anomeric C-O bond
cleavage. Precedents of similar findings are well documented
in the glycosidase-catalyzed hydrolysis of pyranosides contain-
ing fluorine substituent at C-2 or C-5.20 Thus, the above results
support a mechanism for the catalysis of UDP-Galp mutase
involving oxocarbenium ions, such as 5 and/or 7, as intermedi-
ates and/or transition states (Scheme 2). The same argument
has also been made for the Klebsiella enzyme.8
Positional Isotope Exchange (PIX) Catalyzed by Native
and Reduced UDP-Galp Mutase. As mentioned earlier, the
(24) Sanderson, P. N.; Sweatman, B. C.; Farrant, R. D.; Lindon, J. C.
Carbohydr. Res. 1996, 284, 51-60.

6764 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Zhang and Liu
Table 2. Inhibition of UDP-Galactopyranose Mutase by
Fluorinated Analoguesa
inhibitors
9
10
26
27
residual activity of native enzyme
preincubated with inhibitor (%)b
39
1.4
46
29
residual activity of native enzyme (%)c
residual activity of reduced enzyme (%)d
nde
nd
13
90
nd
nd
81
94
a See Experimental Section for details. b Native enzyme (2 µM) was
preincubated with inhibitor (1 mM) for 1 h on ice, and the residual
activity was determined using 2 as the substrate (1 mM) as described
in the section, Activity Assay. c Native enzyme (2.1 µM) was added
to a mixture of substrate (2, 1 mM) and inhibitor (1 mM), and the
residual activity was determined as described in the section, Activity
Assay. d Native enzyme (53 nM) was added to a mixture of substrate
(2, 1 mM) and inhibitor (1 mM) containing dithionite (20 mM), and
the residual activity was determined as described in the section, Activity
Assay. e Not determined.
Galp mutase (70 µM) at 37 °C for 20 min. Since the equilibrium
for this catalysis heavily favors the conversion of furanose to
pyranose, the pyranose derivatives, 26 and 27, are not expected
to be substrates under the same conditions. To test whether these
fluorinated analogues are inhibitors for the mutase, compounds
9, 10, 26, and 27 (1 mM each) were individually incubated with
native UDP-Galp mutase (2 µM) on ice for 1 h, and the resulting
mixtures were assayed for their capabilities to catalyze the
turnover of UDP-Galf (2). As shown in Table 2, all four ana-
logues demonstrated varied inhibitory effects against the native
enzyme. It is clear that the 3-F compounds are better inhibitors
than the corresponding 2-F analogues, with UDP-[3-F]Galf (10)
being the most potent inhibitor. Greater than 98% of the enzyme
activity was lost when the mutase was pretreated with 10 prior
to activity assay. However, the inhibition was less effective when
UDP-[3-F]Galf (10) or UDP-[3-F]Galp (27) (1 mM each) was
included in the assay mixture containing UDP-Galf (2, 1 mM)
as the substrate. Our data showed that the mutase was 81%
active in the presence of equimolar amounts of 27 and 2 and
retained 13% of its activity upon incubation with equimolar
amounts of 10 and 2. Since the normal substrate 2 can protect
the enzyme against 10 and 27, as demonstrated by the com-
petition assays, the inactivation of UDP-Galp mutase by these
flurorinated analogues is most likely active-site-directed.
Kinetic Study of Inactivation of UDP-Galp Mutase by
UDP-[3-F]Galf. Since UDP-[3-F]-Galf (10) is the most potent
inhibitor in this series, it was selected for further analysis. As
illustrated in Figure 2, the activity of this mutase decreased, in
a time-dependent fashion, by nearly 80% after 20 min of
incubation with 10. Specifically, the residual enzyme activities
were determined to be 26%, 18%, 3.4%, and 1.5% after preincu-
bation at room temperature for 30 min with 0.05, 0.1, 0.5, and
1.0 mM 10, respectively. The kinetic parameters of the inhibition
of 10 on the activity of UDP-Galp mutase were estimated by
analyzing the plot of kobs versus inhibitor concentration as
depicted in Figure 3. Figure 3A shows a plot of the natural log
of the fraction of remaining enzyme activity versus time at
various concentrations of 10. The values of kobs as deter-
mined from the slopes of the individual lines were double-
reciprocally plotted against the concentrations of 10, as shown
in Figure 3B. The KI and kinact of the inactivation were deter-
mined from Figure 3B and have values of 270 µM and 0.19
min-1. Although the rate of inactivation is slow, the KI is close
to the Km value (194 µM) found for UDP-Galf (2) under the
same conditions.10a Thus, UDP-[3-F]Galf (10) appears to fit well
in the active site of native mutase.
Figure 1. Positional isotope exchange catalyzed by (A) native UDP-
galactopyranose mutase and (B) reduced UDP-galactopyranose mutase.
Darken atoms indicate 18O labels. See Experimental Section for detailed
conditions.
cleavage of the anomeric C-O bond of 1 during catalysis had
been demonstrated in a PIX experiment using mutase isolated
from K. pneumoniae.9 To confirm that the same pathway is also
followed by the E. coli enzyme and to investigate if the
mechanism remains unaltered for the native and reduced
enzymes, the same PIX experiments were carried out with both
the native and reduced enzymes. The required doubly labeled
UDP-Galp (99% 13C at C-1 and >90% 18O at O1) was prepared
enzymatically as described in the literature.9,15 This labeled
compound and native mutase were incubated at room temper-
ature and analyzed by 13C NMR. The C-1 signal of the labeled
UDP-Galp appeared as a doublet (due to its coupling to 31P) at
time 0 in the proton-decoupled 13C NMR spectra. As illustrated
in Figure 1, a new set of doublets at downfield became visible
as incubation continued. The diminishing of the intensity of the
original doublet occurred concurrently with the increase of the
downfield doublet and was a result of scrambling of the 18O
from O1 to the two nonbridging diphosphate oxygen positions.
Therefore, similar to the mechanism of the K. pneumoniae
enzyme, cleavage of the anomeric C-O bond must also take
place during the catalysis of the E. coli mutase. Analogous
results were also noted when dithionite was included in the
reaction mixture, but with greatly enhanced efficiency (Figure
1B). These findings firmly established that the reaction catalyzed
by the native as well as the reduced enzymes proceeds through
the same mechanism which involves the cleavage of the
anomeric C-O bond.
Inhibition of Mutase by Fluorinated Analogues. While
compounds 9 and 10 are substrates for the reduced E. coli
mutase, surprisingly, no turnover products could be detected
when 9 and 10 (3 mM each) were incubated with native UDP-
Irreversibility of the Inactivation of Mutase by UDP-
[3-F]Galf. It is important to note that the inactivation of the

Mechanistic InVestigation of UDP-Galactopyranose Mutase
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6765
Scheme 6
Figure 2. Time-dependent inactivation of UDP-galactopyranose mutase
by UDP-[3-F]Galf (10) (O, control; 9, with UDP-[3-F]Galf). UDP-
galactopyranose mutase (15 µM) was incubated with UDP-[3-F]Galf
(10, 200 µM) in a reaction volume of 30 µL at room temperature. At
the indicated times, aliquots of the reaction were taken out, diluted
10-fold, and assayed as described in the Experimental Section. The
control reaction had an equivalent volume of buffer instead of UDP-
[3-F]Galf.
was incubated with UDP-[3-F]Galf (10) or UDP-[3-F]Galp (27)
and an equivalent of 2 under reducing conditions (20 mM
dithionite) for 2 min at 37 °C, little inhibition was observed.
As shown in Table 2, the mutase retained greater than 90% of
its activity in both cases. It is worth mentioning that compounds
10 and 27 are stable toward dithionite, and no turnover of 10
to 27 was discernible within the detection limit under the above
experimental conditions. Thus, the very little inhibitory effect
of 10 and 27 under the reducing state is likely a result of
depleting the effective concentration of enzyme for the conver-
sion of the normal substrate 2, and consequently lowering the
observed catalytic activity. Evidently, the reduced enzyme is
immune to inactivation by 10, and more significantly, the
enzyme being inactivated by 10 can be reductively reactivated.
Mechanistic Insights Derived from the Inactivation of
Mutase by UDP-[3-F]Galf. As mentioned earlier, the catalytic
mechanism of UDP-Galp mutase has been proposed to be
initiated either by ring distortion to allow attack of O4 on C-1
to release UDP, or by elimination of UDP first followed by O4
attack on C-1 (Scheme 2).9 In both cases, formation of
oxocarbenium ions, such as 5 and/or 7, as intermediates and/or
transition states is postulated. The fact that the electron-
withdrawing fluorine substituent reduces the kcat of UDP-[2-F]-
Galf (9) more significantly than that of UDP-[3-F]Galf (10)
provides strong support for the proposed mechanism. While 9
and 10 are substrates for the reduced mutase, both are inhibitors
for the native mutase. Since the inhibition likely involves
covalent adduct formation, trapping an active-site nucleophile
by the oxocarbenium type intermediates/transition states (5/7)
to form stable adducts (Scheme 6, 33/34) is a conceivable
inactivation mechanism. Although the corresponding adduct (33/
34) derived from 9 will be preferentially stabilized by the
adjacent electron-withdrawing 2-F group, its formation via 5/7
should be considerably suppressed by the inductive effect of
2-F. Thus, it is not too surprising that the mutase was only
partially inactivated after treatment with 9 for 60 min. In
contrast, the suppression of the formation of oxocarbenium-
type intermediates/transition states should be less significant in
the case of 10. Hence, complete inactivation was observed upon
incubation with 10 under identical conditions.
Figure 3. (A) Time- and concentration-dependent inactivation of UDP-
galactopyranose mutase by UDP-[3-F]Galf (10) (O, 0 µM UDP-[3-F]-
Galf; 9, 100 µM; 0, 133 µM; ×, 200 µM; b, 400 µM). UDP-
galactopyranose mutase (15 µM) was incubated with the indicated
amounts of UDP-[3-F]Galf in a reaction volume of 30 µL at room
temperature. At the indicated times, aliquots of the reaction were directly
used to assay the activity of the enzyme. The values of kobs were
determined from the slopes of the linear fit for the inactivation data
obtained with the respective concentration of UDP-[3-F]Galf. (B) Plot
of kobs as a function of UDP-[3-F]Galf concentration ([I]). The inset
shows the double-reciprocal plot of kobs versus UDP-[3-F]Galf con-
centration. The data from plot B were used to calculate kinact and KI,
which are reported in the text.
mutase by 10 is practically irreversible. An inactivated enzyme
sample with 2% residual activity was found to remain inactive
with little recovery of activity (5% residual activity) after
extensive dialysis. Since the inactivation is time-dependent,
irreversible, and active-site-directed, it likely results from the
formation of a covalent adduct between the enzyme and the
inactivator. While the above observations may also be ascribed
to 10 being a tight binding inhibitor, such a scenario, however,
is incompatible with the fact that KI of 10 is greater than the
Km of the substrate 2. Unfortunately, attempts to further
characterize the inactivated enzyme by MS analysis led to
ambiguous results. The inhibitor-enzyme adduct, if it indeed
exists, may be labile under the sample preparation conditions.
A strategy requiring less destructive treatment will have to be
developed to fully analyze the inactivated enzyme in the future.
Reductive Reactivation of Inactivated Mutase. To our great
surprise, upon treatment with dithionite, this inactivated enzyme
could be fully reactivated, having activity restored to the same
level as the reduced enzyme. As expected, when the mutase
The disparity on the response of the native and reduced
enzyme toward the inactivation by 10 is intriguing. Early studies
had shown that when the enzyme was reduced by dithionite,
its catalytic efficiency was increased by more than 2 orders of
magnitude.10a While both the FAD coenzyme and the cysteine

6766 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Zhang and Liu
residues of this mutase were fully reduced under these condi-
tions, the activity enhancement was demonstrated to be solely
associated with the reduction of the flavin coenzyme. Reduction
of FAD, which involves transformation of the coenzyme from
a highly conjugated planar frame to a bent butterfly structure,
may induce a conformational change within the enzyme that
may become more conducive to catalysis.25 Since the reduced
flavin bears a higher electron density at N-1, this anionic species
may be used to stabilize the transiently formed oxocarbenium
intermediates/transition states to facilitate catalysis.25 It is
conceivable that the proposed conformational change resulting
from reduction of the enzyme-bound FAD may also facilitate
the collapse of the putative covalent enzyme-inhibitor adducts
(33/34) to reactivate the enzyme. Likewise, the reduced FAD
may play a role to prevent the formation of the covalent adduct
or facilitate its breakdown by charge stabilization of the
oxocarbenium intermediates/transition states (5/7). Clearly, more
experiments are needed to address whether conformational
effects, electronic effects, or a combination of both dictate the
ability of reduced FAD to enhance the rate of the mutase
reaction and to protect as well as to salvage it from the
inactivation by 9 and 10.
these enzymes are not well defined, reduction of the flavin
coenzyme in several cases have been shown to have no effect26,29
or to be detrimental.28 In contrast, the unexpected increment of
activity of UDP-Galp mutase by the reduction of its FAD
coenzyme makes this catalyst unique among members of this
family.30 It should be mentioned that the catalysis of chorismate
synthase is an exception, since its flavin coenzyme in the
reduced form has been demonstrated to be involved in the
catalytic cycle, mediating a radical mechanism.25 However, a
similar pathway for UDP-Galp mutase invoking radicals is
unlikely as both the oxidized and reduced enzymes are active.
Using two fluorinated UDP-galactofuranose analogues, UDP-
[2-F]Galf (9) and UDP-[3-F]Galf (10), as mechanistic probes,
we have now confirmed an active role played by the enzyme-
bound flavin in the catalysis of UDP-Galp mutase. Reduction
of the FAD coenzyme in this mutase can not only enhance the
catalytic efficiency of the mutase, but also protect the enzyme
from being inactivated by 9 and 10. Compound 10 is an effective
inhibitor for the mutase, and the inactivation is reversible only
under reduction conditions. Study of this inactivation has led
to the discovery of possible involvement of a covalent inter-
mediate in the mechanism of this mutase. It remains to be
established whether the covalent adduct formation is catalytically
relevant or is simply an incident of derail from the normal
turnover.
In summary, several enzymes that require flavin and yet
catalyze reactions involving no net redox chemistry are known.
Members of this group include chorismate synthase,25 aceto-
lactate synthase,26 mandelonitrile lyase,27 tartronate-semialde-
hyde synthase,28 and YerE.29 While the roles of the flavin in
Acknowledgment. This work was supported in part by a
grant from the National Institutes of Health (GM54346).
H.-w.L. also thanks the National Institute of General Medical
Sciences for a MERIT Award.
(25) (a) Kerwin, J. F., Jr.; Lancaster, J. R., Jr.; Feldman, P. L. J. Med.
Chem. 1995, 38, 4343-4362. (b) Macheroux, P.; Schmid, J.; Amrhein, N.;
Schaller, A. Planta 1999, 207, 325-334 and references therein.
(26) Chipman, D.; Barak, Z.; Schloss, J. V. Biochim. Biophys. Acta 1998,
1385, 401-419.
(27) Xu, L.-L.; Singh, B. K.; Conn, E. E. Arch. Biochim. Biophys. 1986,
250, 322-328.
(28) Cromartie, T. H.; Walsh, C. T. J. Biol. Chem. 1976, 251, 329-
333.
JA010473L
(30) A relevant example is NifL of Azotobacter Vinlandii, a flavoprotein
that modulates transcriptional activation of nitrogen-fixation genes via the
conformational change of flavin oxidation reduction (Hill, S.; Austin, S.;
Eydmann, T.; Jones, T.; Dixon, R. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
2143-2148).
(29) Chen. H.; Guo, Z.; Liu, H.-w. J. Am. Chem. Soc. 1998, 120, 11796-
11797.