A new, simple, high-affinity glycosidase inhibitor: Analysis of binding through X-ray crystallography, mutagenesis, and kinetic analysis [9]

Full text of DOI:10.1021/ja0002870

J. Am. Chem. Soc. 2000, 122, 4229-4230
A New, Simple, High-Affinity Glycosidase Inhibitor:
4229
Analysis of Binding through X-ray Crystallography,
Mutagenesis, and Kinetic Analysis
Spencer J. Williams,§ Valerie Notenboom,† Jacqueline Wicki,§
David R. Rose,† and Stephen G. Withers*,§
Protein Engineering Network of Centres of Excellence
Department of Chemistry, UniVersity of British Columbia
2036 Main Mall, VancouVer, Canada, V6T 1Z1
Ontario Cancer Institute and
Figure 1. (a) Hypothetical transition-state of a retaining â-xylanase
proceeding through an intermediate in a chair conformation. (b) Proposed
mode of inhibition of an isofagomine lactam.
Department of Medical Biophysics
UniVersity of Toronto, Toronto, Canada, M5G 2M9
ReceiVed January 26, 2000
Retaining glycosidases catalyze the hydrolysis of glycosides
with overall retention of anomeric stereochemistry, normally with
the assistance of two carboxyl groups, one functioning as an acid/
base and the other as a nucleophile.1 The reaction usually proceeds
through a covalent glycosyl enzyme, and the transition-states
leading to and from this intermediate have been shown to possess
considerable oxacarbenium-ion character (Figure 1a).2 Many
glycosidase inhibitors have been developed over the last 30 years,
including a number of sugar-shaped heterocycles containing a
basic nitrogen at positions that correspond to C1, O5, and the
glycosidic oxygen, sites that develop partial charge at the transition
state.3,4 Of these, 1-N-iminosugars are particularly powerful
inhibitors of retaining â-glycosidases. However, unlike the
substrates for most glycosidases and most other iminosugar
inhibitors, 1-N-iminosugars lack a hydroxyl at C2.5,6 This is
unfortunate since the interaction of active-site residues, particularly
of the catalytic nucleophile, with OH2 has been suggested to
provide upward of 30-40 kJ mol-1 to the stabilization of the
transition state of enzyme-catalyzed glycoside hydrolysis.7-10 The
possibility therefore exists to develop much stronger inhibitors
of this type if this interaction can be included. However, 1-N-
iminosugars with a hydroxyl at C2 are not expected to be stable
compounds, since they will presumably dehydrate in common
with many hemiaminals.11 An attractive way of incorporating this
interaction into a 1-N-iminosugar is provided in compound 1. By
binding as the protonated iminol tautomer,12 1 could thereby
provide a hydroxyl group at C2 and a positive charge at the
anomeric position and in part reflect the planarity of the sugar
ring at the transition state (Figure 2). Investigation of the mode
of interaction of such a compound may cast light on the role of
OH2 in the transition state of the enzyme-catalyzed reaction.
Xylobiose-derived inhibitors with sp2-hybridized nitrogen atoms
in place of the glycosidic oxygen, namely the imidazole 2 and
the lactam oxime 3, and two compounds with sp3-hybridized
Figure 2. Potential tautomerisation of compound 1.
Table 1. Kinetic Parameters for Xylobiose-Derived Iminosugars
and p-Nitrophenyl â-Cellobioside with Wild-Type Cex and the
N126A Mutanta
p-NP
Ki (µM)
wild-type
1
2
3
4
5
â-cellobioside
0.34 0.15a 0.37a 5.8a 0.13a kcat/Km
26.3 s-1 mM-1
)
N126A
2000 200 1300 4800 1.6 kcat/Km
)
0.0044 s-1 mM-1
∆∆G°
-5.3 -4.4 -5.0 -4.1 -1.5 ∆∆Gq )
(kcal mol-1
)
5.4 kcal/mol-1
a Data taken from ref 13.
nitrogen atoms in place of C1 or O5, the deoxynojirimycin 4 and
the isofagomine 5, were recently shown to be effective inhibitors
of Cex, a retaining family 10 xylanase from Cellulomonas fimi
(Table 1).13 It was hoped that by installing a carbonyl into 5,
affording 1, we could provide a simple, new class of nonbasic
inhibitors of glycosidases, and of Cex in particular. Notably, there
exist relatively few effective, nonbasic glycosidase inhibitors, the
best of these being kifunensine,14 the glyconotetrazoles15 and
various glyconolactones and their isosteric lactams.16
(1) Sinnott, M. L. Chem. ReV. 1990, 90, 1171-1202.
(2) Zechel, D. L.; Withers, S. G. Glycosyl transferase mechanisms in
ComprehensiVe Natural Products Chemistry; Poulter, C. D., Ed.; Pergamon:
Amsterdam, 1999; Vol. 5, pp 279-314.
Compound 1 was prepared in a straightforward fashion from
the known lactam 6.13 Benzoylation of 6 afforded the monobenz-
oate 7 that was subsequently xylosylated with tri-O-acetyl-R-D-
xylopyranosyl trichloroacetimidate to afford the disaccharide 8.
Deacetylation was achieved under standard conditions to give 1
which was purified by flash chromatography and assayed as an
inhibitor of Cex (Scheme 1). 1 was shown to act as a competitive
inhibitor with a Ki value of 340 nM.17,18 The Ki value determined
is impressive considering that glyconolactams ordinarily bind
rather poorly.19 To probe whether this high affinity is due to its
binding in the iminol tautomer proposed earlier, the X-ray crystal
structure of the complex of 1 with Cex was determined (Figure
3).20 This structure shows the inhibitor binding in the -1 and
(3) Stutz, A. E. Iminosugars as glycosidase inhibitors: nojirimycin and
beyond; Wiley-VCH: Weinheim, 1999.
(4) Heightman, T. D.; Vasella, A. T. Angew. Chem., Int. Ed. 1999, 38,
750-770.
(5) Bols, M. Acc. Chem. Res. 1998, 31, 1-8.
(6) Ichikawa, Y.; Igarashi, Y.; Ichikawa, M.; Suhara, Y. J. Am. Chem. Soc.
1998, 120, 3007-3018.
(7) White, A.; Tull, D.; Johns, K.; Withers, S. G.; Rose, D. R. Nat. Struct.
Biol. 1996, 3, 149-154.
(8) Notenboom, V.; Birsan, C.; Nitz, M.; Rose, D. R.; Warren, R. A. J.;
Withers, S. G. Nat. Struct. Biol. 1998, 5, 812-818.
(9) Mega, T.; Matsushima, Y. J. Biochem. 1983, 94, 1637-1647.
(10) Roeser, K. R.; Legler, G. Biochim. Biophys. Acta 1981, 657, 321-333.
(11) Paulsen, H.; Todt, K. AdV. Carbohydr. Chem. 1968, 23, 115-232.
(12) O-Alkyl imidates are weakly basic, for example, the pKa of phenyl
N-methylacetimidate is 6.2. See Ha¨felinger, G. Aspects of amidines and imidic
acid deriVatiVes in The chemistry of amidines and imidates; Patai, S., Ed.;
Wiley & Sons: London, 1975; Vol. 1, pp 1-84.
(14) Kayakiri, H.; Takase, S.; Shibata, T.; Okamoto, M.; Terano, H.;
Hashimoto, M.; Tada, T.; Koda, S. J. Org. Chem. 1989, 54, 4015-4016.
(15) Ermert, P.; Vasella, A. HelV. Chim. Acta 1991, 74, 2043-2053.
(16) Legler, G. AdV. Carbohydr. Chem. Biochem. 1990, 48, 319-385.
(17) Compound 1 was a poorer inhibitor of Bcx (Ki ) 9 mM), a retaining
family 11 xylanase from Bacillus circulans.
(13) Williams, S. J.; Hoos, R.; Withers, S. G. J. Am. Chem. Soc. 2000,
122, 2223-2235.
10.1021/ja0002870 CCC: $19.00 © 2000 American Chemical Society
Published on Web 04/14/2000

4230 J. Am. Chem. Soc., Vol. 122, No. 17, 2000
Communications to the Editor
Scheme 1
iminol, the proton on nitrogen is directed in the plane of the ring
and is not able to form a strong hydrogen bond with the
nucleophile. Nonetheless, a strong electrostatic interaction between
the nucleophile and the iminium ion is still possible. Also
observed in the crystal structure is a relatively short contact of
2.59 Å between OE2 of the nucleophile and the lactam oxygen,
reminiscent of that (2.37 Å) observed in the X-ray crystal structure
of the cellobiosyl enzyme formed with the H205N/E127A Cex
double mutant.8 This strong interaction is suggestive of a hydrogen
bond between OE2 of the nucleophile and O2 of the inhibitor. A
significant H-bonding interaction (2.97 Å) is also seen between
the amide nitrogen of N126 and O2, again consistent with a
hydrogen bond from a 2-hydroxyl. N126 is a highly conserved
residue in family 10 and in clan GH-A and, on the basis of crystal
structures of the 2-deoxy-2-fluoro-cellobiosyl7 and 2-deoxy-2-
fluoro-xylobiosyl enzyme,23 has been suggested as a residue that
hydrogen bonds directly with OH2.
a (a) BzCl, Pyr, -40°, 68%; (b) tri-O-acetyl-R-D-xylopyranosyl trichlo-
roacetimidate, BF3.Et2O, (CH2Cl)2, 71%; (c) NaOMe, MeOH, 76%.
To determine the importance of interactions of O2 of 1 with
the N126 residue, we measured the Ki values of the inhibitor with
the Cex N126A mutant and compared this with Ki values for the
inhibitors 2-5. Table 1 shows Ki values for compounds 1-5 with
wild-type Cex and the N126A mutant, as well as values for the
contributions of the interactions with N126 to inhibitor binding.
As can be seen for compounds 2-4 (all of which possess a
2-hydroxyl) binding to the N126A mutant is considerably weaker
than to wild-type enzyme, as expected due to the loss of important
hydrogen bonding in the N126A mutant.24 By contrast, binding
of 5 to the mutant is compromised to a much lesser extent, consis-
tent with the absence of an interaction at that position. Signifi-
cantly, the consequence of the mutation of N126 upon binding
of 1 is very similar to that seen for inhibitors 2-4, which contain
a 2-hydroxyl group. This strongly implies that there are similar
interactions in the two cases and thus that 1 binds in its iminol
form. It is interesting to compare these results with the effect
upon catalysis of mutating N126, as reflected in kcat/Km values.
The loss in transition state stabilization observed (∆∆Gq ) 5.4
kcal mol-1) is very similar to that seen for binding of these inhib-
itors, implying that these inhibitors are, at least in part, mimicking
the reaction transition state. However, a more detailed analysis
will be required to properly probe this behavior. The tautomer-
ization energy for the amide-iminol conversion is likely to be of
Figure 3. Stereodiagram of the 2|Fo| - |Fo| electron density from the
protein model, contoured at 1σ. The stick model of the isofagomine is
superimposed on the density in an E5 conformation but was not used in
the phase calculation.
-2 subsites in an E5 conformation, consistent with its binding as
the iminol tautomer, with C6, N1, C2, C3 and O2 lying very
nearly in a plane.21,22 While the amide tautomer of 1 necessarily
requires atoms N1, C2, C3 and O2 to lie in a plane, upon
tautomerisation to the iminol there is an additional requirement
that C6 lies in this same plane, as observed. The catalytic
nucleophile, Glu233 lies directly below N1 at a distance of 2.98
Å. The observation of significant interactions between the
nucleophile and both N1 and O2 is most consistent with 1 binding
in the protonated iminol form. Indeed, if the amide tautomer was
bound to Cex, these interactions with the nucleophile would be
strongly destabilizing. If 1 is in fact bound as the protonated
(18) Inhibition constants were determined at 37 °C using a 0.05 M NaH2-
PO4/Na2HPO4 buffer (pH 7.0) and 2,4-dinitrophenyl â-cellobioside as a
substrate. Measurements were started by addition of Cex. Measurements of
the increase of absorption at 400 nm per min in a continuous assay yielded
reaction rates. Michaelis parameters (Vmax and Km) were extracted from these
data by best fit to the Michaelis-Menten equation. Estimates of Ki values
(for N126A Cex) were obtained by measuring rates at a fixed substrate
concentration with a range of inhibitor concentrations (6-10) which encom-
passed the Ki value ultimately determined, generally from 0.3 to 3 Ki. The
observed rates were plotted in the form of a Dixon plot and the Ki value was
determined by an intersection of this line with a horizontal line drawn through
1/Vmax. Full Ki determinations were performed by measurement of rates at a
series of seven substrate concentrations (generally from 0.3 to 3 Km) in the
presence of a range of inhibitor concentrations (typically 5 concentrations)
which bracket the Ki value.
the order of 11 kcal mol-1 25
, indicating that the concentration of
the iminol form in solution is very low. Thus, if the Ki value ob-
served results from the small amount of the iminol present, then
the true Ki value for this tautomer must be several orders of mag-
nitude lower. An important approach to tighter binding inhibitors
may therefore involve inclusion of structural elements that
stabilize this tautomer. Compound 1 therefore represents an
example of a possible new class of potent glycosidase inhibitors
and work is continuing to explore their generality as glycosidase
inhibitors.
(19) Although low Ki values for glyconolactams have been observed in
many cases, the values observed have been similar to those of the corre-
sponding deoxynojirimycin. In this case the lactam 1 binds more than 10-
fold more tightly than the xylobiose-derived deoxynojirimycin 4. See ref 16.
(20) Cex crystals were grown in a 0.1 M NaOAc buffer (pH 4.6) containing
16% PEG 4000. Cex crystals were soaked in artificial mother liquor containing
inhibitor for several hours prior to data collection. Diffraction data were
recorded in-house to 2.0 Å resolution at 100 K on a Mar345 image plate
using Osmic mirror focused Cu KR X-rays, generated from a rotating anode
operating at 100 mA and 50 kV. Oscillations of 1° were collected in 120 s
exposures. Recorded reflections were indexed, integrated, and scaled in the
Denzo/Scalepack MarHKL suite (a) Otwinowski, Z.; Minor, W. Processing
of X-ray diffraction data collected in oscillation mode. In Methods in
Enzymology; Carter, C. W., Jr., Sweet, R. M., Eds.; Academic Press: New
York, 1997; Vol 276, pp 307-326. The data set was 99% complete with Rsym
of 5.7% and an average I/s of 13.3, using 258 208 observations for 21 259
unique reflections. The crystals belonged to tetragonal space group P41212
with cell dimensions a ) b ) 86.99 Å and c ) 80.36 Å and R ) â ) γ )
90°. The structure was refined using wild-type Cex as a starting model (PDB
code 2EXO) with CNS. (b) Bru¨nger, A. T; Adams, P. D.; Clore, G. M.;
DeLano, W. L.; Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski,
J.; Nilges, M.; Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren,
G. L. Acta Crystallogr. 1998, D54, 905-921; to an R-factor of 20.9% (R-
free 25.0%). In order not to bias ligand geometry, dihedral angles were left
unimposed. The rmsd for the CR backbone atoms between this complex and
that of native Cex was 0.315 Å.
Acknowledgment. We thank the Natural Sciences and Engineering
Research Council of Canada and the Protein Engineering Network of
Centres of Excellence of Canada for financial support. S.J.W. is a Killam
Postdoctoral Fellow. Valerie Notenboom is a Connaught scholar. Karen
Rupitz is thanked for technical assistance.
Supporting Information Available: Experimental details (PDF). This
material is available free of charge via the Internet at http://pubs.acs.org.
JA0002870
(23) Notenboom, V.; Birsan, C.; Warren, R. A. J.; Withers, S. G.; Rose,
D. R. Biochemistry 1998, 37, 4751-4758.
(24) The X-ray structure of the N126A Cex mutant is essentially identical
to that of the wild-type other than a slight movement of Trp84. This residue,
which lies behind N126 in wild-type Cex, is observed to have moved towards
the void left by the Asp to Ala mutation (a 30° rotation about ø-1) and may
slightly obstruct the active site.
(25) The tautomerization energy for the formamide to formimidic acid
conversion has been estimated to be 11 ( 4 kcal mol-1. Sygula, A. J. Chem.
Res., S 1989, 56-57.
(21) The torsional angles measured about the C6-N1-C2-C3 and C6-
N1-C2-O2 systems of 1 were 21° and 175°, respectively.
(22) The single-crystal X-ray structure of the lactam 6 showed it to be in
5
a H4 conformation, Williams, S. J. and Withers, S. G. Unpublished data.