4718 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 22
Loll et al.
repeatedly with diethyl ether. The organic phase was dried
over CaSO4 and filtered; its volume at this point was 150 mL.
Pyridine was added to the solution (4.9 g, 62 mmol), after
which acetyl chloride (3.9 g, 50 mmol) was added dropwise
with stirring. A white precipitate appeared immediately. After
40 min excess pyridine was precipitated by the addition of 1
mL of concentrated HCl; the solution was then filtered and
concentrated in a rotary evaporator. The remaining material
was dissolved in water, frozen, and lyophilized; 1.35 g of a clear
colorless oil remained after lyophilization (21%): 1H NMR (300
MHz, D2O) δ 5.04 (q, 1H, J ) 7.0 Hz), 2.16 (s, 3H), 1.50 (d,
3H, J ) 7.1 Hz); 13C NMR (75 MHz, D2O) δ 178.2, 176.2, 72.5,
22.8, 18.8. Anal. (C5H8O4) C, H, O.
Mea su r em en t of Liga n d Bin d in g Con sta n ts via 13C
NMR Sp ectr oscop y. Stock solutions of 400 mM ligand and
24 mM vancomycin were prepared in 50 mM potassium
phosphate buffer, pH 6.0, containing 10% D2O. The stock
solutions were mixed with buffer to yield a final solution of
12 mM vancomycin and 0-200 mM ligand. A solution of 400
mM AcDLac, 12 mM vancomycin was prepared by the addition
of solid vancomycin to the 400 mM AcDLac stock. NMR spectra
were acquired immediately after sample preparation. 13C NMR
spectra were acquired on a Varian XL-300 spectrometer at a
13C frequency of 75 MHz. Spectra were acquired with com-
posite pulse proton decoupling, using a 90° (15-ms) excitation
pulse and a 2-s interpulse delay. Spectra were acquired into
32K data points with a spectral width of 233 ppm. Ap-
proximately 12 000 acquisitions were averaged per spectrum.
Spectra were acquired at room temperature (297 ( 2 K).
Limiting chemical shift changes were estimated from plots of
13C chemical shift versus concentration of AcDLac. Binding
constants were calculated from plots of ∆δ/∆δlim versus (∆δ/
∆δlim) × (concentration).
Presumably, the ligands are so weakly bound that under
crystallization conditions they are unable to displace the
symmetry-related molecule. However, the addition of
AcDA to any of the tetragonal crystals immediately
causes them to crack and dissolve, indicating that the
more tightly bound ligand can displace a symmetry-
related vancomycin molecule from this binding pocket.
From the structures currently available, we can
identify a minimum set of drug-ligand interactions
necessary to trigger this structural change in vancomy-
cin. The set includes: (1) three hydrogen bonds formed
between the carboxylic acid of the ligand and amide
protons of the antibiotic, (2) interactions between the
R-methyl group of the ligand and the aromatic ring of
residue 4, (3) the hydrogen bond donated by a nitrogen
or oxygen of the ligand to the carbonyl oxygen of residue
4 of the antibiotic, and (4) various contacts between an
acyl group attached to the C-terminal residue and both
monomers in a face-face dimer. This acyl group is
acetyl in AcG and AcDA but would be D-alanyl in the
natural ligand. Each of the relatively low-affinity struc-
tures described herein exhibits only three of these four
interactions, implying that all four interactions must
be formed in order for vancomycin to adopt the high-
affinity conformation.
Con clu sion s
Binding of high-affinity ligands by vancomycin is
associated with a structural change in the drug and the
formation of ligand-mediated face-face dimers. The
structural change and dimerization may be coupled,
since while low-affinity ligands bind in a manner
compatible with face-face dimer formation, they elicit
neither the structural change nor face-face dimers. We
have identified a minimal set of interactions, all of
which must occur between the drug and its ligand in
order for high-affinity binding to be observed. These
results show that the loss of an intermolecular hydrogen
bond is only one of the factors needed to explain the
failure of vancomycin to recognize depsipeptide ligands
found in resistant bacteria, and it will be important to
compensate for these other factors when designing
agents targeted against vancomycin-resistant organ-
isms. A question that remains is whether the high-
affinity conformation of vancomycin seen with AcDA
reflects the full extent of the structural change that is
induced in vancomycin by ligands, or if tighter binding
by longer and more physiologically relevant ligands will
elicit further structural changes.
Cr ysta lliza tion . Crystals were prepared by the hanging
drop vapor diffusion method; 5.0-µL drops containing 30 mg/
mL vancomycin in water were mixed with equal volumes of
reservoir buffer and suspended from a siliconized cover slip
over 1.0 mL of reservoir buffer. The plates were sealed with
plastic tape and maintained at 291 K. The reservoir buffers
contained 100 mM ligand, which was used as a buffering agent
at pH 4.6, and 2.0-2.4 M NaCl. Racemic lactic acid was used
in the reservoir buffer for the DLac complex. Bipyramidal
crystals formed within 1 week, growing to maximum dimen-
sions of 0.2 × 0.3 × 0.5 mm. Shortly prior to data collection,
the crystals were transferred to a solution of 30% v/v glycerol
in reservoir buffer. After 1-2 min, the crystals were mounted
in nylon loops and flash-cooled by plunging into liquid N2.
Da ta Collection a n d P r ocessin g. Diffraction data were
collected at beamline A1 at the Cornell High Energy Synchro-
tron Source and beamline X-12B at the National Synchrotron
Light Source. Beamline A1 at CHESS was fitted with a binned
2K×2K CCD detector,10 while beamline X-12B utilized a 30-
cm MAR image plate detector. At CHESS, several data
collection passes were made, using the unattenuated beam for
high-resolution data and attenuation for the more intense low-
resolution data. A similar strategy was used at NSLS, except
that instead of attenuating the beam the exposure time was
drastically reduced. In addition, the crystal used at X-12B was
returned still frozen to the home laboratory, where additional
low-resolution data were collected using a 30-cm MAR image
plate detector mounted on a rotating anode source equipped
with focusing mirrors and a Ni foil. Crystals were maintained
at ca. 100 K in a stream of N2 gas at all times during data
collection. Images were processed using the programs DENZO
and SCALEPACK.11 Observations containing overloaded pixels
were not included in the data set.
Exp er im en ta l Section
Ma ter ia ls a n d Meth od s. Pharmaceutical grade vancomy-
cin hydrochloride (Vancocin, Eli Lilly) was used without
further purification. 2-Acetoxy-D-propanoic acid was synthe-
sized as described below; N-acetylglycine was obtained from
Sigma and lithium D-lactate from Aldrich. Other materials
were obtained commercially and were of the highest purity
available. TLC was carried out using silica-60 plates (EM
Science) and a mobile phase of 1% acetic acid in ethyl acetate.
HPLC analyses were conducted with an analytical Vydac C18
column, using 50 mM K2HPO4 as the mobile phase and
identifying analytes by absorbance at 220 nm. NMR spectra
were acquired on a Varian XL-300 spectrometer.
Refin em en t. Refinement was carried out against F2 using
the program SHELXL-93.12 All three of the complexes are
isomorphous with the vancomycin:Ac complex structure de-
termined previously.3,4 Thus, starting structures were obtained
by removing the carbohydrate groups, ligand, and residue 1
and 3 side chains from the vancomycin:Ac structure and
subjecting this fragment to rigid body refinement. R values
after rigid body refinement were typically about 0.45. Missing
Syn th esis of AcLa c. The sodium salt of D-lactic acid (5.5
g, 49 mmol) was dissolved in 25 mL of a 65:35 (v/v) mixture of
water and concentrated HCl. This solution was extracted