The role of sugar substituents in glycoside hydrolysis

Article abstract of DOI:10.1021/ja992044h

A series of monosubstituted deoxy and deoxyfluoro 2,4-dinitrophenyl (DNP) β-D-glycopyranosides was synthesized and used to probe the mechanism of spontaneous β-glycoside hydrolysis. Their relative rates of hydrolysis followed the order 2-deoxy > 4-deoxy > 3-deoxy ? 6-deoxy > parent > 6-deoxy- 6-fluoro > 3-deoxy-3-fluoro > 4-deoxy-4-fluoro > 2-deoxy-2-fluoro. Hammett correlations of the pH-independent hydrolysis rates of each of the 6-, 4-, 3- , and 2-position substituted glycosides with the σ₁ value for the sugar ring substituent were linear (r = 0.95 to 0.999, π(I) = -2.2 to -10.7), consistent with hydrolysis rates being largely dictated by field effects on an electron-deficient transition state. The relative rates of hydrolysis of the DNP glucosides can be rationalized on the basis of the stabilities of the oxocarbenium ion-like transition states, as predicted by the Kirkwood- Westheimer model. The primary determinant of the rate of hydrolysis within a series appears to be the field effect of the ring substituent on O5, the principal center of charge development at the transition state. Differences in the rates of hydrolysis between different series of hexopyranosides may not arise solely from field effects and likely also reflect differences in steric factors or solvation.

Full text of DOI:10.1021/ja992044h

1270
J. Am. Chem. Soc. 2000, 122, 1270-1277
The Role of Sugar Substituents in Glycoside Hydrolysis
Mark N. Namchuk, John D. McCarter, Adam Becalski, Trevor Andrews, and
Stephen G. Withers*
Contribution from the Department of Chemistry, UniVersity of British Columbia,
VancouVer, British Columbia, Canada V6T 1Z1
ReceiVed June 16, 1999
Abstract: A series of monosubstituted deoxy and deoxyfluoro 2,4-dinitrophenyl (DNP) â-D-glycopyranosides
was synthesized and used to probe the mechanism of spontaneous â-glycoside hydrolysis. Their relative rates
of hydrolysis followed the order 2-deoxy > 4-deoxy > 3-deoxy ≈ 6-deoxy > parent > 6-deoxy-6-fluoro >
3-deoxy-3-fluoro > 4-deoxy-4-fluoro > 2-deoxy-2-fluoro. Hammett correlations of the pH-independent
hydrolysis rates of each of the 6-, 4-, 3-, and 2-position substituted glycosides with the σI value for the sugar
ring substituent were linear (r ) 0.95 to 0.999, FI ) -2.2 to -10.7), consistent with hydrolysis rates being
largely dictated by field effects on an electron-deficient transition state. The relative rates of hydrolysis of the
DNP glucosides can be rationalized on the basis of the stabilities of the oxocarbenium ion-like transition
states, as predicted by the Kirkwood-Westheimer model. The primary determinant of the rate of hydrolysis
within a series appears to be the field effect of the ring substituent on O5, the principal center of charge
development at the transition state. Differences in the rates of hydrolysis between different series of
hexopyranosides may not arise solely from field effects and likely also reflect differences in steric factors or
solvation.
Introduction
Scheme 1
A
B
Carbohydrate oligomers are ubiquitous in nature and, in
addition to having structural and storage roles, are involved in
a number of biologically important processes such as cell-cell
recognition and energy metabolism. Thus the mechanism of their
hydrolysis has been the focus of a great deal of research.1-9
Hydrolysis of glycopyranosides is believed to occur by an
essentially DN + AN mechanism, leading to the generation of a
cyclic oxocarbenium ion as shown in Scheme 1A. Modeling
and experimental studies suggest that the stability of the resultant
cation in solution is variable and dependent on the identities
and locations of the ring substituents.1,2,5-7,9-13 Studies carried
out with fully hydroxylated glycosides indicate that the transition
state for glycoside hydrolysis in these systems is highly
preassociative, consistent with the relative instability of these
species in solution.1,2,6,10,11
Within this framework, little is known about the role of
substituents on the glycopyranose ring in determining the relative
rates of glycoside hydrolysis. In general, it has been observed
* To whom correspondence should be addressed. Tel: (604) 822-3402.
Fax: (604) 822-2847. E-mail: withers@chem.ubc.ca.
(1) Amyes, T. L.; Jencks, W. P. J. Am. Chem. Soc. 1989, 111, 7900-
7909.
(2) Banait, N. S.; Jencks, W. P. J. Am. Chem. Soc. 1991, 113, 7951-
7958.
(3) Bennet, A.; Sinnott, M. L. J. Am. Chem. Soc. 1986, 108, 7287-
7294.
(4) Capon, B. Chem. ReV. 1969, 69, 407-498.
(5) Huang, X.; Surry, C.; Hiebert, T.; Bennet, A. J. J. Am. Chem. Soc.
1995, 117, 10614-10621.
(6) Sinnott, M. L.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102, 2026-
that the rate of hydrolysis of glycosides increases as the number
of axial hydroxyl groups on the glycone increases. Edward14
rationalized these observations on steric grounds, wherein steric
2032.
(7) Zhu, J.; Bennet, A. J. J. Am. Chem. Soc. 1998, 120, 3887-3893.
(8) Cocker, D.; Sinnott, M. L. J. Chem. Soc., Perkin Trans. 2 1975,
1391-1393.
(9) Horenstein, B. A.; Bruner, M. J. Am. Chem. Soc. 1996, 118, 10371-
10379.
(10) Knier, B. L.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102, 6789-
6798.
10.1021/ja992044h CCC: $19.00 © 2000 American Chemical Society
Published on Web 02/02/2000

Mechanism of Glycoside Hydrolysis
J. Am. Chem. Soc., Vol. 122, No. 7, 2000 1271
compression due to 1,3 diaxial interactions in the ground state
4C1 conformation is released as the transition state half-chair is
formed. Such ground-state interactions will be larger with axial
substituents, thus according to this rationale the hydrolysis rates
of galactosides, for example, are greater than those of glucosides.
Several studies have indicated that polar effects15 contribute
strongly to the relative rates of glycoside hydrolysis. Generally,
the rate of hydrolysis of a glycoside increases as a consequence
of deoxygenation,16-19 presumably because replacement of an
electron-withdrawing hydroxyl group with a hydrogen stabilizes
the electron-deficient transition state. However, the increase in
the rate of hydrolysis of the deoxy glycosides is not simply a
function of the distance of the substitution from the anomeric
center. In the studies carried out to date, the relative rates of
hydrolysis of monodeoxygenated glycosides and R-glucosyl
phosphates is 2-deoxy > 4-deoxy > 3-deoxy > 6-deoxy >
parent.17,19 Conversely, when a hydroxyl is replaced by a more
electronegative fluorine atom, the rate of glycoside hydrolysis
decreases. Indeed, a study of the rates of hydrolysis of a series
of deoxyfluoro R-glucosyl phosphates showed the exact inverse
order, parent > 6-deoxyfluoro > 3-deoxyfluoro > 4-deoxyfluoro
> 2-deoxyfluoro.20 Unfortunately, detailed interpretation of these
rates was not possible because the observed rate constants
contain contributions from at least two hydrolytic pathways,
that via the neutral species and that via the conjugate acid.
Further, the substitution can affect not only the rate constant
for bond cleavage via both pathways but also, through effects
on the basicity of the glycosidic oxygen, the concentration of
the conjugate acid species.
The intent of this study is to further investigate the contribu-
tion of the ring substituents to the relative rates of glycoside
hydrolysis. To this end, a series of monosubstituted deoxy and
deoxyfluoro 2,4-dinitrophenyl â-D-glycopyranosides was syn-
thesized. (Abbreviations: DNP, 2,4-dinitrophenyl; R-DKIE,
R-deuterium kinetic isotope effect; all, allopyranoside; gal,
galactopyranoside; glc, glucopyranoside; man, mannopyrano-
side; d, deoxy; F, deoxyfluoro) Both the equatorial and the axial
epimers of the parent and the fluorinated glycosides were
synthesized to explore the dependence of hydrolysis rate on the
configuration at each of the glycone ring positions. Deoxy and
deoxyfluoro substitutions are excellent probes of field effects
since both are smaller than a hydroxyl group,19 thereby
minimizing steric contributions to the observed rates, while the
electronic effects differ substantially. DNP glycosides were
chosen for these studies because the rate of glycoside hydrolysis
with this aglycone has been shown8 to be pH-independent from
pH 2 to 8. Thus rate constants determined in the pH-independent
region will be unaffected by protonation equilibria, providing
true first-order rate constants for the heterolysis of the glycosidic
linkage, free from any complications associated with parallel
pathways through neutral and conjugate acid species.
Results
Synthesis of Glycosides. Synthesis of fluorinated sugars was
achieved either by deoxyfluorination with diethylaminosulfuryl
trifluoride or by electrophilic fluorination using acetyl hypof-
luorite according to literature procedures, as described in the
Supplementary Information. Deoxygenation was generally
achieved by radical reduction of the bromo sugar substituted at
the site of interest. Per-O-acylated sugars were converted to their
hemiacetals by selective anomeric deprotection and then con-
verted to the dinitrophenyl glycoside by reaction with fluoro
2,4-dinitrobenzene and deprotected using methanolic HCl.
Kinetic Studies. General. Rates of hydrolysis of the DNP
glycosides were determined under the conditions used by Cocker
and Sinnott.8 However, reactions were followed for long periods
(>3 half-lives), and rate constants were extracted by direct fit
to a first-order expression rather than via initial rates analysis.
To minimize possible solute-solute interactions, hydrolyses
were carried out with dilute solutions of the glycosides (∼10-4
M) in buffered 0.4 M KCl. Hydrolyses were followed at three
different glycoside concentrations, and rate constants were
found, within experimental error, to be independent of the
starting concentration of substrate.
Eliminating Alternative Mechanisms. Although it is likely
that hydrolysis of the DNP glycosides proceeds via the
heterolytic mechanism shown in Scheme 1A, control experi-
ments were performed to demonstrate that other modes of
hydrolysis did not contribute significantly to the observed release
of DNP. One possibility was that glycoside cleavage occurred
by attack of water on C1′ of the aryl ring21 rather than at the
anomeric center of the sugar. However, electron impact mass
spectral analysis of the products of hydrolysis of DNPglc at 60
°C in [18O]H2O, under conditions otherwise identical to those
used in the kinetic studies, showed no significant incorporation
of 18O into the 2,4-dinitrophenol released on hydrolysis of the
substrate. This is consistent with hydrolysis of the DNP
glycosides proceeding only via attack of water on the sugar ring.
Horton22 has demonstrated that under basic conditions (0.25
M KOH) the hydrolysis of p-nitrophenyl R-D-glucoside proceeds
by a series of migrations initiated by the attack of the sugar C2
hydroxyl on the ipso position of the aryl ring to produce 2-O-
p-nitrophenyl-D-glucose (Scheme 1B), which rearranges further
and subsequently eliminates the phenolate. This pathway seems
unlikely for hydrolysis of DNP â-D-glucosides since these are
1,2-trans glycosides, and because these experiments were carried
out at near neutral pH (6.5). Further, it has been demonstrated8
that this migration/elimination process did not contribute
significantly to the release of phenol in the first 10% of the
reaction for the hydrolysis of DNPgal. Additional evidence
against the migration/elimination pathway was obtained by
HPLC analysis of aliquots taken at several times during the
hydrolysis of DNPglc. No new UV-absorbing, sugar-containing
peaks, which would be produced by migration of the DNP group
to the C2 or C3 hydroxyl, were observed.
(11) Young, P. R.; Jencks, W. P. J. Am. Chem. Soc. 1977, 99, 8238-
8247.
(12) Buckley, N.; Oppenheimer, N. J. J. Org. Chem. 1996, 61, 8039-
8047.
(13) Buckley, N.; Oppenheimer, N. J. J. Org. Chem. 1996, 61, 8048-
8062.
(14) Edward, J. T. Chem. Ind. (London) 1955, 1102-1104.
(15) The term polar effect refers to the observed influence of an
unconjugated, sterically remote substituent on the rate of a reaction.
Depending on the mechanism of transmission, polar effects in aliphatic
systems may be described as inductive or field effects. Inductive effects
are caused by through-bond polarization of electron density, whereas field
effects are due to through-space interactions.54 Several studies have
convincingly demonstrated that polar effects in aliphatic systems in solution
are best described as field effects31,55-58 and that the inductive component
is relatively minor.
(16) Franks, F.; Ravenhill, J. R.; Reid, D. S. J. Solution Chem. 1972, 1,
3-9.
(17) Mega, T.; Matsushima, Y. J. Biochem. 1983, 94, 1637-1647.
(18) Overend, W. G.; Rees, C. W.; Sequeria, J. S. J. Am. Chem. Soc.
1962, 84, 3429-3440.
(19) Withers, S. G.; Percival, M. D.; Street, I. P. Carbohydr. Res. 1989,
187, 43-66.
(21) Hengge, A. C. J. Am. Chem. Soc. 1992, 114, 2747-2748.
(22) Horton, D.; Luetzow, A. E. J. Chem. Soc. Chem. Commun. 1971,
79-81.
(20) Withers, S. G.; MacLennan, D. J.; Street, I. P. Carbohydr. Res. 1986,
154, 127-144.

1272 J. Am. Chem. Soc., Vol. 122, No. 7, 2000
Namchuk et al.
Table 1. Kinetic Data for the Spontaneous Hydrolysis of the DNP
Glycosidesa
no. of ∆H‡ (37 °C)
∆S‡
(J/mol/K)
rate constantc
DNP glycosideb temps
(kJ/mol)
at 37 °C (sec-1
)
DNPglc (2)
13
4
8
3
4
3
3
4
4
4
9
8
4
4
4
4
4
4
117.8 (2.1) 33.8 (0.9)
120.8 (1.5) 13.3 (0.3)
121.4 (6.2) 29.6 (2.4)
5.58 × 10-6
1.45 × 10-7
8.23 × 10-7
3.74 × 10-7
1.99 × 10-6
2.23 × 10-5
1.25 × 10-4
2.60 × 10-5
2.61 × 10-5
1.25 × 10-6
1.04 × 10-5
3.36 × 10-6
9.40 × 10-6
1.61 × 10-4
1.18 × 10-5
9.51 × 10-7
1.09 × 10-6
3.05 × 10-8
2FDNPglc (10)
3FDNPglc (8)
4FDNPglc (6)
6FDNPglc (4)
3dDNPglc (15)
4dDNPglc (13)
6dDNPglc (11)
DNPgal (17)
2FDNPgal (25)
3FDNPgal (23)
4FDNPgal (21)
6FDNPgal (19)
6dDNPgal (27)
DNPall (33)
123 (8.2)
126 (16)
27.9 (3.0)
53 (11)
116.7 (1.1) 42.0 (0.6)
109.5 (0.3) 33.0 (0.2)
108.3 (0.7) 16.3 (0.1)
107.0 (1.8) 12.0 (0.3)
104.5 (1.7) -21.2 (0.6)
90.7 (4.9)
-48.1 (4.8)
107.1 (0.8) -4.74 (0.06)
101.5 (2.3) -14.1 (0.5)
104.3 (3.9) 18.4 (1.1)
111.3 (0.2) 19.4 (0.06)
119.1 (1.4) 23.4 (0.5)
105.9 (0.5) -18.0 (0.2)
126.3 (0.7) 18.1 (0.2)
3FDNPall (35)
DNPman (37)
2FDNPman (39)
a Standard errors for the activation parameters are given in paren-
theses. These, however, do not accurately reflect the precision of these
parameters, as noted in the Results section. Consequently, the activation
parameters are not interpreted to any significant extent. b Numbers in
brackets are compound numbers used in the Supporting Information.
c Rates were calculated by extrapolation or interpolation of Eyring plots.
Figure 1. Eyring plots for the hydrolysis of DNPglc from three separate
experiments. Line correlations for each plot are excellent (r ) 0.997-
1.00). Slight variations in the slope and intercepts of such plots result
in much larger variation in the activation entropies and enthalpies than
that observed for the rate constants (see text).
Determination of Rate Constants. Rate constants and
activation parameters determined for the hydrolyses of the DNP
glycosides are presented in Table 1.23 Reassuringly, the rate
constants (4.6 × 10-6 s-1 for DNPgal and 1.2 × 10-6 s-1 for
DNPglc) determined previously by initial rates analysis8 were
in close agreement with the rate constants (4.7 × 10-6 s-1 for
DNPgal and 8.7 × 10-7 s-1 for DNPglc) determined in this
study by following the reaction for several half-lives and fitting
the time course of phenol release to a first-order equation. The
wide range of hydrolysis rates observed with these glycosides
made it impossible to obtain reliable data for all compounds at
a single temperature. Thus the rate constants presented in Table
1 at 37 °C were determined by extrapolation or interpolation
of Eyring plots derived from hydrolyses at various temperatures.
As a check of the precision of the rate constants determined in
this way, this complete analysis at a series of temperatures (total
temperature range of 55 °C) was repeated three times for
DNPglc, and calculated rate constants of 5.4 × 10-6, 5.6 ×
-6
10-6, and 5.4 × 10
sec-1 were obtained (Figure 1). Even
though these calculated rate constants agree well and the plots
are as good as most published plots of this type, values of
activation entropy and enthalpy determined from these same
plots were not as reproducible, the activation enthalpies ranging
from 115 to 122 kJ/mol and the activation entropies from 25.8
to 46.6 J/mol‚K. This indicates that the precision of the
activation parameters is not well represented by the statistically
calculated errors, as noted previously.24 For this reason, only
rate constants (hence, ∆G‡) will be interpreted in any detail.
The 2dDNPglc proved too labile for purification and charac-
terization. Rate constants for its hydrolysis were therefore
estimated with the aid of the excellent correlation (r ) 0.997,
F ) 1.1) discovered between the rate constants for hydrolysis
of the DNP glucosides and those previously determined for the
analogous series of R-glucosyl phosphates (Figure 2).
Figure 2. Comparison of rate constants for hydrolysis of DNP
glucosides (37 °C) and the analogous R-glucosyl phosphates (25 °C).
Discussion
Field Effects Dominate. Data obtained in this study are
consistent with previous results from other systems. A plot of
the logarithm of the rate constants for hydrolysis of the DNP
glucosides versus the logarithm of the rate constants for the
analogously substituted R-glucosyl phosphates showed a cor-
relation of r ) 0.997 (Figure 2). The presence of such a
correlation indicates that the two reactions proceed through
similar oxocarbenium ion-like transition states. The slope of
the plot was positive (F ) 1.1) indicating a slightly greater
sensitivity to ring substitutions for the R-glucosyl phosphates.
Assuming field effects are important, this result indicates a
somewhat greater degree of charge development at the transition
state for hydrolysis of the glycosyl phosphates. While this
difference may be due to the different leaving groups, it is also
consistent with differences in the stereochemistry at the anomeric
(23) Some compounds are referred to by trivial designations for clarity,
e.g., 3dDNPglc for 2,4-dinitrophenyl 3-deoxy glucopyranoside rather than
2,4-dinitrophenyl 3-deoxy-D-xylo-hexopyranoside. Note that 2dDNPglc,
3dDNPglc, and 4dDNPglc may with equal justification be referred to,
respectively, as 2dDNPman, 3dDNPall, and 4dDNPgal.
(24) Linert, W.; Jameson, R. F. Chem. Soc. ReV. 1989, 18, 477-505.

Mechanism of Glycoside Hydrolysis
J. Am. Chem. Soc., Vol. 122, No. 7, 2000 1273
Table 2. Correlation Coefficients (r) and Reaction Constants (FI)
for Hammett Plots from Each Set of DNP Glycosides
Oppenheimer on the 2-substituted ribosides and that by Timell
on the 6-substituted glucosides.25,26
Transition State Structure. While the solvolysis of DNP
glycosides clearly proceeds via an oxocarbenium ion-like
transition state,6,8 a species conventionally represented with
partial double bond character and partial positive charge between
C1 and O5, the distribution and extent of charge development
between these centers is unclear and likely variable. Chemical
intuition regarding the predominance of the formal charge at
O-5 to maintain a stable octet of electrons on each atom is
supported by ab initio calculations (6-31G*) ,which were used27
to assess the extent of oxocarbenium ion formation in the acid-
catalyzed hydrolysis of 2-methoxy tetrahydropyran. These
studies indicated that lone pair donation from O5 into the nσ*
orbital at C1 was important to attainment of the transition state
and that cleavage of the C1-O1 bond takes place concurrently
with formation of a C1-O5 double bond. Thus the transition
state for this system more closely resembles an oxocarbenium
than a C1 carbonium ion. A semiempirical modeling study
(PM3)28 calculated the partial charge distributions for the ring
carbons and oxygen in R-D-glucose and in a glucopyranosyl
oxocarbenium ion. This study, along with AMPAC modeling
of the mannopyranosyl oxocarbenium ion,29 indicated that the
greatest increases in partial positive charge between the ground
state and the transition state were at O5 and C1, with the greatest
increase in positive charge development at O5. Though these
model studies are suggestive, none considers the contribution
of the solvent, the nucleophile, or the leaving group to the
stability of the oxocarbenium ion at the transition state.
However, these model studies, together with experimental results
substitution position
r
FI
6-glucosides
6-galactosides
4-glucosides
4-galactosides
3-glucosides
3-galactosides
3-allosides
2-glucosides
2-galactosides
2-mannosides
0.99
0.99
0.99
0.99
0.99
0.95
0.95
0.96
0.98
0.97
-2.2
-2.5
-5.1
-3.1
-2.9
-3.5
-2.7
-8.3
-8.7
-10.7
center. The heterolysis of R-glycosides proceeds via a transition
state more dissociative than that of the analogous â-linked
compounds, presumably for stereoelectronic reasons. This is
seen in comparison of secondary deuterium kinetic isotope
effects for the acid-catalyzed hydrolysis of R-methyl glucoside
(kH/kD ) 1.188) and â-methyl glucoside (kH/kD ) 1.089).3 It is
also seen in the greater percentage of product of inverted
anomeric configuration formed upon ethanolysis of â-glycosides
compared to that of R-glycosides.6
The relative rates of hydrolysis determined in this study for
the deoxy and deoxyfluoro DNP glucosides were 2-deoxy >
4-deoxy > 6-deoxy ≈ 3-deoxy > parent > 6-deoxyfluoro >
3-deoxyfluoro > 4-deoxyfluoro > 2-deoxyfluoro (Table 1),
which is essentially the same order seen for the substituted
R-glucosyl phosphates. This order does not reflect the distance
of the modified substituent from C-1 and thus cannot be
rationalized simply by the field effect felt at C-1. However, field
effects likely play a major role in determining the rates of
glycoside hydrolysis, because in all cases deoxygenation of the
glucoside increases the rate of hydrolysis with respect to the
parent glycoside, while fluorination has the inverse effect.
Indeed, studies by Oppenheimer of the hydrolysis rates of a
series of 2-position substituted nicotinamide â-D-ribofurano-
sides25 revealed a strong correlation of rate constant with
substituent σI (r ) 0.99, FI ) -6.7) consistent with the
importance of field effects at the 2-position. Further, such effects
are also seen at the 6-position of glucosides, as is seen if the
previously published rate constants for the acid-catalyzed
hydrolysis of a series of 6-substituted methyl glucosides26 are
plotted against σI according to the equation log k ) FIσI + log
ko, when a correlation coefficient of r ) 0.97 (FI ) -3.0) is
observed.
2
of H, 13C, and ring 18O kinetic isotope effects of the acid-
catalyzed hydrolyses of methyl glucosides,3 do confirm that
charge development at the transition state of glycoside hydrolysis
is delocalized between C1 and O5.
Kirkwood-Westheimer Analysis. Glucosides. Extending
the previous σI analysis to examine the relative rates of
hydrolysis of the DNP glucosides substituted at different
positions, the observed relative order of rates appears to be
determined by field effects according to the Kirkwood-
Westheimer model.30 Our analysis was similar to that applied
to examination of the effect of ring substituents on the pKa of
carboxylic acids in 2,2,2-bicyclo systems.31 This model predicts
that the relative values of k (the equilibrium constant in this
case) for two structures differing by a single substituent (x and
y) can be calculated using eq 1 , where e is the charge generated
To gain insight into the consequences of substitution at each
position and the sources of the rate differences, Hammett plots
of log k versus σI were constructed for each group of glycosides
substituted at a particular position (Figure 3). Although these
plots contain only three data points in most cases, such plots
are very informative, yielding the correlation coefficients (r)
and sensitivities (FI) shown in Table 2. These plots provide
reasonable evidence that the major factor determining the rates
of hydrolysis of the substituted DNP glycosides at a given ring
position is the field effect of the substituent on the oxocarbenium
ion-like transition state. The sensitivities of these systems to
the substituent, as reflected in FI values, are consistent with their
general proximity to the reaction center, with the greatest values
being at the 2-position (FI ) -8.3 to -10.7) and the smallest
being at the 6-position (FI ) -2.2 to -2.5). These sensitivities
are quite consistent with those noted earlier from work by
log(kx/ky) ) (e/2.3kT)[(µ cos θ/R2DE)x - (µ cos θ/R2DE)y]
(1)
and θ is the angle between a line of length R joining the point
where charge is generated and the midpoint of the dipole
associated with the substitution. DE is the dielectric constant of
the space through which the field effect is transmitted, and µ is
the magnitude of the dipole associated with the substitution.
Extension of this analysis to comparison of the relative rates
of reaction examined in the current study is afforded by the
use of transition state theory. Both modeling studies28,29 and
(27) Andrews, C. W.; Fraser-Reid, B.; Bowen, J. P. J. Am. Chem. Soc.
1991, 113, 8293-8298.
(28) Kajimoto, T.; Liu, K. K. C.; Pederson, R. L.; Zhong, Z.; Ichikawa,
Y.; Porco, J. A., Jr.; Wong, C. H. J. Am. Chem. Soc. 1991, 113, 6187-
6196.
(29) Winkler, D. A.; Holan, G. J. Med. Chem. 1989, 32, 2084-2089.
(30) Westheimer, F. H.; Kirkwood, J. G. J. Chem. Phys. 1938, 6, 513-
517.
(25) Handlon, A. L.; Oppenheimer, N. J. J. Org. Chem. 1991, 56, 5009-
5010.
(26) Timell, T. E.; Enterman, W.; Spencer, F.; Soltes, E. J. Can. J. Chem.
1965, 43, 2296-2305.
(31) Baker, F. W.; Parish, R. C.; Stock, L. M. J. Am. Chem. Soc. 1967,
89, 5677-5685.

1274 J. Am. Chem. Soc., Vol. 122, No. 7, 2000
Namchuk et al.
Figure 3. Hammett plots correlating rate constants for hydrolysis of the DNP glycosides at 37 °C with σI. Each plot represents the effects of
changing the substituent at a single position with a single stereochemistry. The plot of 2-position substituted glucosides includes an additional point
measured for 2,4-dinitrophenyl 2-chloro-2-deoxy â-D-glucopyranoside (k ) 1.61 × 10-6 sec-1). Details of parameters extracted from these plots
are provided in Table 2.
experiment3 indicate that the transition state of a glucosyl
oxocarbenium ion is flattened toward a 4H half-chair with partial
charge development at both C1 and O5. The majority of the
charge generated resides at O5.3,28,29 Through space distances
between O5 or C1 and the dipole midpoints of the substituted
centers, along with values for θ, were calculated for such a
glucosyl oxocarbenium ion using InsightII (Molecular Simula-
tions Inc., San Diego, CA.). Minimization using either steepest
descents or conjugate gradient algorithms converged on
essentially identical conformations. Because substitutions are
made on the same system at one temperature, e, k, and T are
constant. Further, given the conservative substitutions examined
in the current study, and the relative insensitivity of field effects
to the ring system examined31 it is reasonable to assume that
DE is constant. Values of σI were substituted for µ as measures
of the substituent dipole.32
Values of relative rates for each of the substituted DNPgly-
cosides {log(calcdkrel)} can therefore be calculated by the
summing of all of the relative field effects due to the substituents
on the ring felt at both C1 and O5 using different charge

Mechanism of Glycoside Hydrolysis
J. Am. Chem. Soc., Vol. 122, No. 7, 2000 1275
distributions between these two centers. Field effects, when
uncomplicated by steric factors as in this case, have been shown
previously to be strictly additive in a closely analogous system
involving comparison of rates of hydrolysis for a series of
deoxy- and deoxyfluoro R-glucosyl phosphates.19 Therefore,
determination of log(calcdkrel) for DNPglc requires calculation
of relative field effects for hydroxyl groups at C2, C3, C4, and
C6, while that for 2FDNPglc requires calculation of the relative
field effects for a fluorine at C2 and hydroxyl groups at the 3-,
4- and 6-positions, and so on. Since the Kirkwood-Westheimer
model best describes rigid ring systems,31 it is difficult to
estimate the field effect for the 6-position substituents because
they are able to rotate around the C5-C6 bond. The weighted
populations for the three staggered rotamers of methyl R-D-
glucoside in D2O were therefore used to estimate values of R
and θ and thus the relative field effect for this position.33
Thus the logarithm of the calculated relatiVe rate of a
substituted DNP glucoside, log(calcdkrel), is given by eq 2, where
C6
log(calcdk ) )
e [σ cos θ /(RC1i)2] +
C1 Ii i
∑
rel
i)C2
eO5[σIi cos θi/(RO5i)2] (2)
eC1 is the fractional charge on C1 and eO5 is the fractional charge
on O5.
Values of log(calcdkrel) were calculated for a range of possible
charge distributions, where the total charge at the transition state
was shared between O5 and C1. Figure 4 shows a series of
plots of calcdkrel values (calculated for various charge distributions
between O5 and C1) versus the observed rate constants obsk, of
the DNPglycosides. The calculated relative rates based on the
magnitude of the field effect at O5 and C1 as predicted by the
Kirkwood-Westheimer model correlate with the observed rates
of hydrolysis (r ) 0.92-0.98, slopes ) 1.21-1.33). The best
agreement of the calculated and empirical data was obtained
when the majority of the charge generated at the transition state
was localized at O5, a finding in agreement with previous studies
of the transition state structure for glycoside hydrolysis (vide
supra). The optimal fit of the data was obtained when 97% of
the charge was localized on O5 (r2 ) 0.967). Further, the
observed relative rates of hydrolysis for compounds substituted
at C3 and C4 (i.e., 4-deoxy > 3-deoxy, 4-fluoro < 3-fluoro)
are predicted by our model at all points when >60% of the
charge generated was localized at O5. These data, taken together
with the Hammett plots shown in Figure 3, strongly suggest
that the relative rates of hydrolysis of these glucosides are
dictated largely by the stabilities of the oxocarbenium ion-like
transition states and that these stabilities are heavily dependent
on field effects associated with the various substituents.
Galactosides and Mannosides. The calculations of relative
rates based upon Kirkwood-Westheimer theory used in this
analysis thus far have only dealt with cases in which the
substituent has the same configuration as in the parent glucoside.
Because the theory takes into account the orientation of the
dipole associated with the substituent with respect to the center
Figure 4. Comparison of the calculated and experimentally determined
rate constants for hydrolysis of DNP glycosides. Relative rate constants
for substituted glycosides were calculated using eq 2 and the measured
rate constant for DNPglc assuming the charge distributions (O5 versus
C1) shown and plotted versus observed rate constants. The quality of
the fit to the data improved as the amount of charge on O5 increased
(r ) 0.92-0.98 using σI for the calculations, r ) 0.91-0.97 using µ;
best fit obtained at 97% of charge at O5, r2 ) 0.967 using σI, r2 )
0.942 using µ). Glycosides substituted at the 3-position (filled circles)
and 4-position (open squares) are highlighted to demonstrate that the
observed relative rates of hydrolysis are best predicted by the model
when the majority of the positive charge obtained at the transition state
resides on O5. Data points on the graph represent, from the left in
order of increasing obsk, 2-fluoro-, 4-fluoro-, 3-fluoro-, 6-fluoro-, parent,
3-deoxy-, 6-deoxy, and 4-deoxy.
(32) To validate the use of σI as a substitute for µ, the calculations used
to obtain Figure 4 were carried out using both σI (H ) 0.0, OH ) 0.25,
F ) 0.5) (Charton, M. J. Org. Chem. 1964, 29, 1222-1227) and bond
dipole moment µ (H ) 0.4 D, OH ) 0.7 D, F ) 1.39 D) (Minkin, V. I.;
Osipov, O. A.; Zhdanov, Y. A. Dipole Moments in Organic Chemistry;
Plenum Press: New York, 1970; p 88). The data obtained with the two
parameters were very similar (see Figure 4 and its legend) though slightly
better correlations were obtained with σI.
(33) Abraham, R. J.; Chambers, E. C.; Thomas, W. A. Carbohydr. Res.
1992, 226, C1.

1276 J. Am. Chem. Soc., Vol. 122, No. 7, 2000
Namchuk et al.
ion.35 In our case, with sterically conserVatiVe substituents, rate
reductions are observed with electronegative substituents. Pos-
sibly release of ground-state strain was more important in their
case than they had realized.
Solvent Involvement. Although the analysis of activation
entropies and enthalpies has generally been avoided in this study,
it is noteworthy that activation enthalpies for the galactosides
are significantly lower than those for the corresponding gluco-
sides, balanced somewhat by more negative ∆S‡ values, with
the net result of a slight decrease in ∆G‡. Such entropy/enthalpy
compensations have been noted in a variety of systems such as
the ionization of carboxylic acids36 and binding of carbohydrates
to lectins37 and have been commented on recently in relation
to solvation and ligand binding.38 A rationale proposed for these
observations (see, for example, ref 39) invokes a reorganization
of the solvent structure as the reaction proceeds, the extent of
this being different for different substrates, thereby influencing
the reaction rate in different ways. In other systems, more
negative ∆S‡ values have been interpreted as reflecting increased
solvent organization at the transition state. Indeed, differences
in the hydration structure of glucose and galactose and their
related pentopyranose congeners16,40-45 have been observed in
a number of studies, galactose apparently being more strongly
solvated than glucose in aqueous solution.46,47 Thus if the
transition states are not solvated to exactly equivalent extents,
effects upon rates, which would show up in the activation
enthalpies, would be expected. Such effects have been noted
by others.46,47
Figure 5. Comparison of rate constants for hydrolysis of DNP
glucosides and DNP galactosides. A correlation coefficient of r ) 0.98
was observed if the point for the 4-deoxyglucoside (filled circle) was
omitted, as represented by the line shown.
of charge it should be possible to extend this analysis to the
epimeric galactosides, allosides, and mannosides, as long as the
transition state structures are similar in all cases. Considerable
similarity of transition states for the hydrolysis of glucosides
and galactosides is indeed indicated by the excellent linear free
energy relationship (r ) 0.98 if the 4-position is excluded)
observed between the rates of hydrolysis of the two series of
compounds (Figure 5), as well as by the nearly identical R-DKIE
values obtained for the DNP galactoside (kH/kD ) 1.11 ( 0.01)
and the DNP glucoside (kH/kD ) 1.09 ( 0.02). Predictions of
rates for allosides and galactosides should therefore be feasible.
Indeed, calculations predict rates 2- to 28-fold greater for
galactosides than for the corresponding glucosides, factors quite
consistent with the 5- to 13-fold greater rate constants actually
observed. The observed rate constant for hydrolysis of the
4-deoxy “galactoside” appears to be lower than that predicted
by the correlation shown in Figure 5. This may be due to the
fact that, unlike any other galactosides (including 4FDNPgal),
the 4-deoxy “galactoside” lacks an axial C4 substituent and thus
lacks the release of steric compression proposed by Edward14
as important to lowering of activation barriers when axial
substituents are present.
Conclusions
Hammett correlations of the pH-independent hydrolysis rates
of the 6-, 4-, 3-, and 2-position substituted glycosides with the
σI value for the substituent were linear (r ) 0.95 to 0.999, FI )
-2.2 to -10.7), consistent with hydrolysis rates being dictated
largely by field effects. The slopes of the plots were similar in
magnitude and sign to those seen in other systems involving
electron-deficient transition states. The relative rates of hy-
drolysis of the DNP glucosides, i.e., 2-deoxy > 4-deoxy >
3-deoxy ≈ 6-deoxy > parent > 6-deoxy-6-fluoro > 3-deoxy-
3-fluoro > 4-deoxy-4-fluoro > 2-deoxy-2-fluoro, can be
rationalized on the basis of the relative stabilities of the oxocar-
benium ion-like transition states (predicted by the Kirkwood-
Westheimer model), which appear to be principally a function
of field effects exerted by the ring substituents on O5, the
principal center of charge development at the transition state.
In addition the results indicated that differences in the rates of
There is reason to believe that this steric compression
argument may not apply to mannosides, other factors possibly
taking precedence given the proximity of the 2-hydroxyl to the
incoming nucleophile and the departing aglycone and especially
given previous suggestions of a role for the 2-hydroxyl in
orienting the incoming nucleophile.3,6 In addition, modeling
studies29 have suggested that the mannopyranosyl oxocarbenium
(35) Miljkovic, M.; Yeagley, D.; Deslongchamps, P.; Dory, Y. L. J. Org.
Chem. 1997, 62, 7597-7604.
(36) Stewart, R. The Proton, Applications to Organic Chemistry;
Academic Press: Orlando, 1985.
(37) Hindsgaul, O.; Khare, D. P.; Bach, M.; Lemieux, R. U. Can. J.
Chem. 1985, 63, 2653-2663.
3
ion adopts one of two H half-chair conformations rather than
the 4H conformation. Indeed, the data in this study indicate that
the relief of steric strain at the transition state is not a key factor
for sugars with an axial 2-substituent, since rate constants for
hydrolysis of DNPman and 2FDNPman are 5-fold lower than
those for their corresponding glucosides. Interestingly, calcula-
tions by Woods and co-workers34 had indeed predicted that
lyxopyranosides (axial 2-hydroxyl) would generate a less stable
oxocarbenium ion than xylopyranosides (equatorial 2-hydroxyl).
Our results do, however, contradict the recent claim that axial
electronegative substituents at C4 accelerate glycopyranoside
solvolysis via electrostatic stabilization of the oxocarbenium
(38) Gallicchio, E.; Kubo, M. M.; Levy, R. M. J. Am. Chem. Soc. 1998,
120, 4526-4527.
(39) Jencks, W. P. Catalysis in Chemistry and Enzymology; Dover
Publications Inc.: New York, 1987.
(40) Barone, G.; Castronuovo, G.; Doucas, D.; Ella, V.; Mattia, C. A. J.
Phys. Chem. 1983, 87, 1931-1937.
(41) Franks, F.; Reid, D. S.; Suggett, A. J. Solution Chem. 1973, 2, 99-
103.
(42) Galema, S. A.; Blandamer, M. J.; Engberts, J. B. F. N. J. Am. Chem.
Soc. 1990, 112, 9665-9666.
(43) Hoiland, H.; Holvik, H. J. Solution Chem. 1978, 7, 587-592.
(44) Miyajima, K.; Machida, K.; Nakagaki, M. Bull. Chem. Soc. Jpn.
1985, 58, 2595-2599.
(45) Liu, Q.; Brady, J. W. J. Am. Chem. Soc. 1996, 118, 12276-12286.
(46) Galema, S. A.; Hoiland, H. J. Phys. Chem. 1991, 95, 5321-5326.
(47) Galema, S. A.; Blandamer, M. J.; Engberts, J. B. F. N. J. Org. Chem.
1992, 57, 1995-2001.
(34) Woods, R. J.; Andrews, C. W.; Bowen, J. P. J. Am. Chem. Soc.
1992, 114, 859-864.

Mechanism of Glycoside Hydrolysis
J. Am. Chem. Soc., Vol. 122, No. 7, 2000 1277
calculated from the slope of the line and values for ∆S‡ from the y
intercept. Rate constants shown were calculated from the equations
defining the lines obtained for Eyring plots of the data.
hydrolysis between different series of hexopyranosides may not
arise solely from field effects but may also reflect differences
in steric factors or solvation.
High-Temperature Hydrolysis Rates (>50 °C). Approximately 1
mL aliquots of buffered DNP glycoside solutions (approx 0.2, 0.1, and
0.05 mM) were pipetted into 1 mL Wheaton vials, flushed with nitrogen,
sealed using a Bunsen burner, and then immersed in a water bath at
the required temperature. After equilibrating for 5 min, samples were
removed at time intervals and immediately frozen. These solutions were
later thawed, and the A390 for each of the vials was measured at room
temperature using a Pye-Unicam PU 8000 UV-visible spectropho-
tometer equipped with a sipper cell. Data were analyzed as described
previously.
Low-Temperature Hydrolysis Rates. Samples were prepared as
described above, placed in quartz cuvettes fitted with Teflon plugs,
and sealed with Parafilm. These cuvettes were then placed in a Perkin-
Elmer λ II spectrophotometer equipped with a thermostated cell block
and temperature probe, and absorbance readings at 390 nm were
recorded as a function of time. These absorbances were automatically
recorded and then fit to a first-order equation as previously described.
Secondary rDKIE Measurements. The isotope effects for spon-
taneous hydrolysis of glycosides were determined at 45 ( 0.1 °C using
the procedure employed for the low-temperature hydrolysis rates. Rate
constants were determined at least six times for each of the deuterated
and protiated compounds. Isotope effects were determined from the
ratios of consecutive measurements of the rates for the protiated and
deuterated compounds, and errors are provided as the standard deviation
in the isotope effects determined.
Experimental Section
Syntheses of DNP Glycosides. Full experimental details for the
synthesis of the compounds described and full characterization data
are provided as Supplementary Information. The acetylated deoxy- and
deoxyfluoro-glycopyranoses were selectively deacetylated at the 1-posi-
tion, either by treatment with ethanolamine48 or hydrazine acetate49 or
by conversion of the per-O-acetylated carbohydrate to the R-glycosyl
bromide, followed by silver carbonate-catalyzed hydrolysis.50 Protected
DNP glycosides were synthesized by reaction of the protected hemi-
acetal with fluoro-2,4-dinitrobenzene.51 These compounds were then
deacetylated by use of methanolic HCl.52
Kinetic Studies. General. Conditions used for measuring the rates
of spontaneous hydrolysis of the 2,4-dinitrophenyl glycosides are
essentially identical to those used by Sinnott for the determination of
the rates of hydrolysis of a series of DNP galactosides; however, rate
constants were determined by fits to full hydrolysis profiles rather than
via initial rates.8 All rates were determined at pH 6.50 in 25 mM sodium
phosphate buffer and 0.40 M KCl. Glycoside concentrations were
approximately 0.15 mM. All solutions were prepared with Nano-pure
quality water. Hydrolysis rates were determined at three or more
temperatures with each glycoside. The temperature was regulated by a
Neslab RTE-210 circulating bath. Temperatures were recorded from
the available digital readout and were checked using an external
thermometer and found to be accurate within (0.2 °C. The temperatures
during the experiment did not fluctuate more than (0.2 °C. Each rate
was determined by fitting the A390 of the solution at specific times to
a first-order equation using the work station on an Applied Photophysics
MV 17 stopped flow spectrophotometer and employing the available
fitting routine. Typically seven points were used to determine the rate,
and the reactions were allowed to proceed to at least 3 half-lives. The
rates at each temperature were measured three times and averaged. As
a check of the first-order fits obtained, several rate constants were also
determined by plotting ln((A∞ - At)/A∞) vs time. The rates obtained
were essentially identical to those found using the fitting routine.
Activation parameters for the substrates were obtained from Eyring
plots of the logarithm of the averaged value for the rate constant
determined at each temperature divided by temperature versus the
reciprocal of the absolute temperature. These data were analyzed by
linear regression using GraFit,53 which provided the slope and intercept
for the line obtained and errors from these values. Values for ∆H‡ were
Acknowledgment. We thank Dr. Ross Stewart for his help,
encouragement, and advice throughout this project. Financial
support from the Natural Sciences and Engineering Research
Council of Canada and a GREAT Scholarship from the Science
Council of British Columbia (for M.N.N.) are gratefully
acknowledged. We also wish to thank Drs. M. L. Sinnott, N. J.
Oppenheimer, S. Brown, and P. Charifson for reading early
versions of this manuscript and providing helpful comments.
Supporting Information Available: Full experimental
details for the synthesis of all compounds described, including
their characterizations. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA992044H
(53) Leatherbarrow, R. J. Gra-Fit, Version 2.0; Erithacus Software Ltd.:
Staines, U.K., 1990.
(54) March, J. AdVanced Organic Chemistry; John Wiley & Sons: New
York, 1985.
(55) Emsly, J. W.; Feeney, J.; Sutcliffe, C. H. High-Resolution Nuclear
Magnetic Resonance Spectroscopy; Pergamon Press: Oxford, 1966.
(56) Taft, R. W. Prog. Phys. Org. Chem. 1983, 14, 247-269.
(57) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic
Chemistry; Harper & Row: New York, 1987.
(58) Cole, T. W. J.; Mayers, C. J.; Stock, L. M. J. Am. Chem. Soc. 1974,
96, 4555-4557.
(48) Grynkiewicz, G.; Fokt, I.; Szeja, W.; Fitak, H. J. Chem. Res. 1989,
152-156.
(49) Excoffier, G.; Gagnaire, D.; Utille, J. P. Carbohydr. Res. 1975, 39,
368-373.
(50) Haynes, L. J.; Neweth, F. H. AdV. Carbohydr. Chem. 1955, 10, 207-
219.
(51) Koeners, H. J.; de Kok, A. J.; Romers, C.; van Boom, J. H. Rec.
TraV. Chim. Pays-Bas 1980, 99, 355-362.
(52) Ballardie, F.; Capon, B.; Derek, J.; Sutherland, G.; Cocker, D.;
Sinnott, M. J. Chem. Soc., Perkin Trans. 1 1973, 2418-2419.