Origin of the stereospecificity in binding hydroxamates of α- and β-phenylalanine methylamide to thermolysin revealed by the X-ray crystallographic study

Article abstract of DOI:10.1016/S0968-0896(03)00140-8

Optically active N-formyl-N-hydroxy-α-phenylalanine methylamide (1) and N-formyl-N-hydroxy-β-phenylalanine methylamide (2) were evaluated as inhibitors for thermolysin (TLN) to find that while the D-form is more potent than its enantiomer in the case of t

Full text of DOI:10.1016/S0968-0896(03)00140-8

Bioorganic & Medicinal Chemistry 11 (2003) 2421–2426
Origin of the Stereospecificity in Binding Hydroxamates of
ꢀ- and ꢁ-Phenylalanine Methylamide to Thermolysin
Revealed by the X-ray Crystallographic Study
Seung-Jun Kim,a Dong H. Kim,b,* Jung Dae Park,b
Joo-Rang Woo,a Yonghao Jinb and Seong Eon Ryua,*
aCenter for Cellular Switch Protein Structure, Korea Research Institute of Bioscience and Biotechnology,
52 Euh-eun-dong, Yusong-gu, Deajeon 305-806, South Korea
bCenter for Integrated Molecular Systems, Division of Molecular and Life Sciences,
Pohang University of Science and Technology, San 31Hyoja-dong, Namku, Pohang 790-784, South Korea
Received 20 December 2002; accepted 25 February 2003
Abstract—Optically active N-formyl-N-hydroxy-a-phenylalanine methylamide (1) and N-formyl-N-hydroxy-b-phenylalanine
methylamide (2) were evaluated as inhibitors for thermolysin (TLN) to find that while the d-form is more potent than its enantio-
mer in the case of the hydroxamate of a-Phe-NHMe, in the inhibition with hydroxamate of b-Phe-NHMe, the l-isomer
(Ki=1.66ꢀ0.05 mM) is more effective than its enantiomer. In order to shed light on the stereochemical preference observed in the
.
.
inhibitions, X-ray crystallographic analyses of the crystalline TLN d-1 and TLN l-2 complexes 0were performed to the resolution of
2.1 A. While l-2 binds TLN like substrate does with its benzyl aromatic ring occupying the S1 pocket, the electron density in the
0
.
S1 pocket in the complex of TLN d-1 is weak and could best be accounted for by the methylcarbamoyl moiety. For both inhibi-
tors, the hydroxamate moiety coordinates the active site zinc ion in a bidentate fashion.
# 2003 Elsevier Science Ltd. All rights reserved.
Introduction
Most of small molecule inhibitors for zinc proteases
carry a functional group that ligates the active site
zinc ion, such as carboxylate, thiol, or hydroxamate.4
Of these zinc ligating groups, hydroxamate occupies a
prominent position because of its strong chelating
property toward zinc ion.5 Recently, Jin and Kim
Thermolysin (TLN) is a heat-stable proteolytic enzyme
isolated from Bacillus thermoproteolyticus and carries a
catalytically essential zinc ion at the active site.1 The
enzyme is an extracellular endopeptidase, hydrolyzing the
peptide bond on the imino side of the amino acid residue
having a large hydrophobic side chain, such as Leu, Ile,
and Phe.1 The enzyme has been extensively studied as a
prototypical zinc enzyme and used as a model target
enzyme for developing design strategies of enzyme inhibi-
tors that can be transferred to zinc proteases of medicinal
interest such as angiotensin converting enzyme2 and matrix
metalloproteases.3 Inhibitors of angiotensin converting
enzyme are clinically important as antihypertensive
agents,2 and numerous inhibitors of matrix metallopro-
teases are presently under various stages of clinical investi-
gations as potential antitumor and antiarthritis agents.3
reported that N-formyl-N-hydroxy-LeuOMe is
a
potent inhibitor for TLN but surprisingly its inhibi-
tory activity (Ki=44 mM) was found to be mostly
vested with the d-isomer.6 Subsequently, they noted
that N-formyl-N-hydroxy derivative of b-PheNHMe is
even more potent but the inhibition stereochemistry is
reversed compared with that of the corresponding
hydroxamate of a-amino acid.7 Thus, the Ki value for the
l-enantiomer of N-formyl-N-hydroxy-b-PheNHMe was
shown to be 1.66 mM, but that for its enantiomer that is,
N-formyl-N-hydroxy-b-d-PheNHMe was 68.2 mM. In
order to shed light on the origin of the observed stereo-
chemical preference in the binding of these inhibitors to
TLN, we have synthesized hydroxamates of a- and b-
PheNHMe (1 and 2) in an optically pure form, and
undertaken X-ray crystallographic structure determina-
*Corresponding author. Tel.: +82-54-279-2101; fax: +82-54-279-5877;
e-mail: dhkim@postech.ac.kr (D.H. Kim); ryuse@mail.krib.re.kr (S.E.
Ryu).
0968-0896/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00140-8

2422
S.-J. Kim et al. / Bioorg. Med. Chem. 11 (2003) 2421–2426
tions of the TLN complexes formed with d-1 and l-2,
more potent enantiomers of 1 and 2.
Table 1. Inhibition of thermolysin
Inhibitor
Ki (mM)
l-1
d-1
l-2
d-2
2490ꢀ130
104ꢀ4.8
1.66ꢀ0.05a
68.2ꢀ5.6a
aRef 6.
Results and Discussion
Chemistry
The synthetic route for inhibitor 1 is outlined in Scheme
1. The conversion of a-PheNHMe into 5 was carried
out in one-pot following the general method reported by
Feenstra et al.8 and the N-formylation of 5 to obtain 1
was performed by the method of Jahngen and Rosso-
mando9 using the mixture of formic acid and acetic
anhydride. The synthesis of 2 was reported previously.7
Figure 1. Determination of the Ki value for inhibition of TLN with
d-1. In the plot the intercept of the straight line on [I] would give the
Ki value (see Experimental).
Inhibitory assay
The compounds were assayed as competitive inhibitors
10
for TLN using N-[3-(2-furyl)acryloyl]Gly-Leu-NH2 as
(pdb code:1LNF) as the starting model. The 2Fo-Fc
map produced from the rigid-body refinement showed
the overall feature of the bound inhibitors. In the case
of TLN l-2 complex, subsequent refinement and
rebuilding cycles produced well-defined structure for the
substrate at pH 7.2 and 25 ꢁC. The inhibitory constants
(Kis) were obtained from the respective plot for vo=v
against the concentration of inhibitors and are pre-
sented in Table 1. Figure 1 exemplifies the plots. Incon-
sistent with the previous studies,6,7 the d-isomer was
more potent than its enantiomer by 24-fold in the case
of 1, but in the inhibition with hydroxamate of b-Phe-
NHMe, the l-isomer was more effective than the
d-form by 41-fold. l-2 is found to be more potent than
d-1 by 63-fold.
.
.
bound inhibitor as well as the protein, but for TLN d-1
complex the electron density in the S1 pocket was weak
0
(see below). A summary of the crystallographic statistics
is shown in Table 2.
.
The electron density map for the TLN l-2 complex
(Fig. 2) clearly shows that one molecule of the inhibitor
is bound to the active site of TLN. Both oxygen atoms of
the hydroxmate moiety in l-2 are involved in strong
coordinative bondings to the active site zinc ion having
bond distances of 2.00 and 2.36 A. In addition, the for-
myl oxygen in the hydroxamate is engaged in hydrogen
bonding with one of carboxylate oxygens of Glu-143,
and the hydroxyl in the hydroxamate moiety is hydrogen
Crystal structures of TLN complexes formed with D-1
and L-2
.
Crystals of the TLN inhibitor complexes were obtained
by soaking TLN crystals in a solution containing the
inhibitor under study, and structures of the complexes
were determined by using the structure of native TLN
.
Scheme 1. Reagents, conditions, and yield : (a) p-anisaldehyde, MeOH, rt, 12 h, 91%; (b) mCPBA, CH2Cl2, rt, 12 h, quantitative; (c) HONH2 HCl,
MeOH, rt, 12 h, 71%; (d) (i) Ac2O, HCO2H, 0 ꢁC, 2 h; (ii) 2 N NaOH, rt, 30min, 56%.

S.-J. Kim et al. / Bioorg. Med. Chem. 11 (2003) 2421–2426
2423
.
Table 3. Selected bond distances (A) in TLN inhibitor complexes
Table 2. Crystallographic data
Atoms in interactionsa
TLN d-1
TLN l-2
.
.
TLN: l-2
TLN: d-1
Space group
Unit cell
P6122
P6122
TLN
Inhibitor
Zn2+ - - - - - - - - O2
Zn2+ - - - - - - - - O3
His231-ImÀ - - - - O2
Glu143-CO2H- - - O3
Arg203-GunNH - O1
Asn112-CONH - - N1
Tyr157-OH - - - - - O2
2.11
2.63
2.52
2.98
3.17
5.68
3.44
2.00
2.36
2.55
2.72
2.93
2.88
3.22
a, b, c (A)
a, b, g (ꢁ)
Resolution range (A)
Number of unique reflections
Overall completeness (%)
Rmerge (%)a
93.6, 93.6, 131.9 93.8, 93.8, 131.6
90.0, 90.0, 120.0 90.0, 90.0, 120.0
99–2.1
18,917
91.8
99–2.1
19,725
95.6
8.6
9.1
Number of atoms
Protein atoms
Non-protein atoms
Rcryst (%)b
2432
172
15.7
18.5
2432
151
16.4
19.3
aNumbering of inhibitor atoms is shown in the structure of the inhibitors.
Rfree (%)c
atom of His-231. It may be inferred from the structures
of the both complexes that the carbolxylate of Glu-143
exists as a protonated form but the hydroxyl of the
hydroxamate ligates the zinc ion in a deprotonated
form. Cross et al. reported that upon binding of hydro-
xamate to a coordinated zinc ion such as that in the
active site of zinc proteases, the pKa of the ligand is
attenuated markedly,11 and Holmes and Matthews
described that in the binding of l-LeuNHOH to TLN,
the anionic form of the hydroxamate coordinates to
the zinc ion.12 In the binding of l-2 to TLN, the
nitrogen and oxygen atoms of the methylcarbamoyl
group are engaged in a hydrogen bond with the
peptide carbonyl oxygen of Asn-112 and one of gua-
nidinium nitrogens of Arg-203 of the enzyme,
respectively (Table 3). On the other hand, the
Rms deviations
Bonds(A)
0.006
1.2
23.1
0.72
0.005
1.1
23.1
Angles(ꢁ)
Dihedrals(ꢁ)
Impropers(ꢁ)
0.65
aRmerge ¼ ÆijIiÀ < I >j=Æj< I >j, where I is the intensity for the ith
measurement of an equivalent reflection with the indices h,k,l.
bRcryst ¼ ÆjFo À Fcj=Æ Fo, where Fo and Fc are the observed and cal-
culated structure factor amplitudes, respectively.
cThe Rfree value was calculated from 5% of all data that were not used
in the refinement.
bonded to the imidazole NH 0of His-231. The benzene
ring of l-2 is inserted in the S1 hydrophobic pocket like
that in substrate does. Important binding interactions
between TLN and the bound inhibitors are listed in
Table 3.
0
methylcarbamoyl group in d-1 just occupies the P1
pocket without forming hydrogen bond with the
enzyme. The difference in binding feature between
d-1 and l-2 is illustrated in Figure 4.
0
complex (Fig. 3) is weak and significantly different from
that in the com0plex of TLN l-2. The electron density
.
The electron density in the S1 pocket for the TLN d-1
.
map in the S1 pocket is best accounted for by a
methylcarbamoyl moiety. The benzyl moiety of d-1 was
not found in the refined electron density map, suggest-
ing strongly that the side chain is exposed to the solvent
region in a highly dynamic conformation. Thus, in the
binding of d-1 to TLN, the benzyl and N-methylcarba-
moyl moieties are interchanged compared with the
binding of l-2 to TLN. Other regions of the inhibitor
including the a-carbon and the hydroxamate moiety
were well resolved. Both oxygen atoms of the hydro-
xamate bind to the active site zinc ion in a bidentate
fashion with O–Zn distances of 2.11 and 2.63 A. The
formyl oxygen in d-1 is engaged in a hydrogen bonding
with one of Glu-143 carboxylate oxygens, like that in l-
2, and the aminohydroxyl oxygen is positioned within a
hydrogen bonding distance to an imidazole ring nitrogen
Binding features of d-1 and l-2 to TLN are examined in
an effort to understand the stereo-preference shown by
the inhibitors in their inhibitions of TLN. In addition to
the reversed stereo-preference shown in the binding of
the two inhibitors, the a-carbon of d-1 is moved
towards the zinc ion by about 0.6 A compared with that
of l-2, suggesting that in the bindings of the two inhi-
bitors to TLN, the coordination of hydroxamate to the
zinc ion plays a dominant role as expected from the
strong coordination power5 of hydroxamate group to
zinc ion. The much reduced binding affinity and the
reversed stereospecificity in the complexing of 1 com-
pared with those exhibited by 2 may be envisioned:
because of the strong binding affinity of the hydroxamate
towards the active site zinc ion, the a-carbon of 1 cannot
rest at the locus where the a-carbon of substrate lies
.
Figure 2. The difference electron density map for TLN l-2 was generated with the final model omitting the bound inhibitor. The stereo map con-
toured at a 3.0 s level was presented as superposed with the refined model.

2424
S.-J. Kim et al. / Bioorg. Med. Chem. 11 (2003) 2421–2426
.
Figure 3. The difference electron density map for TLN d-1 was generated with the final model omitting the bound inhibitor. The stereo map con-
toured at a 3.0 s level was presented as superposed with the refined model.
Figure 4. Superimposed stereoview of inhibitors d-1 and l-2 that are bound to TLN at the active site. Thin line represents d-1 and thick line denotes
l-2.
Experimental
upon its binding to TLN, and as a result the side chain
0
phenyl ring in the P1 residue of l-1 would difficultly
Melting points were taken on a Thomas-Hoover capil-
1
0
rest in the S1 pocket with difficulty, and thus l-1 shows
lary melting point apparatus and were uncorrected. H
much weaker binding affinity (Ki=2490ꢀ130 mM). On
the other hand, the relatively small methylcarbamoyl
moiety of d-1 would be accommodated in the pocket,
NMR and 13C NMR spectra were recorded with a
Bruker AM 300 (300 MHz) instrument using tetra-
methylsilane as the internal standard. IR spectra were
recorded on a Bruker Equinox 55 FT-IR spectrometer.
Silica gel 60 (230–400 mesh) was used for flash chroma-
tography and thin layer chromatography (TLC) was
carried out on silica coated glass sheets (Merck silica gel
60F-254). Elemental analyses were performed at
Pohang University of Science and Technology, Pohang,
Korea. Thermolysin (Type X) was purchased from
Sigma Chemical Co. Solutions for kinetic study were
prepared using doubly distilled and deionized water.
and d-1 exhibits
a
moderate binding affinity
(Ki=104ꢀ4.8 mM). In the case of 2 in which a methyl-
ene unit is inserted between the hydroxamate moiety
and the a-carbon of 1, it is not unreasonable to assume
that the hydroxamate hydroxyl is geometrically and
functionally equivalent to the scissile amide carbonyl of
substrate in view of the fact that the distance between
the oxygen atom of the N-hydroxyl and the a-carbon
in 2 is expected to be about the same as that between
the carbonyl oxygen of P1 residue and the a-carbon of
0
P1 residue in substrate. It is, therefore, expected that
N-Hydroxy-L-phenylalanine methylamide (L-5)
the side chain benzyl group of l-2 but not d-2 would
rest at about the same region as the hydrophobic side
To a solution of l-phenylalanine methylamide (1.77 g,
9.8 mmol) in anhydrous methanol (50mL) was added
p-methoxybenzaldehyde (1.3 mL, 10.8 mmol) and anhy-
drous Na2CO3 (1.6 g, 14.7 mmol). The mixture was stir-
red for 12 h under nitrogen atmosphere, then
evaporated, and the residue was dissolved in ethyl ace-
tate. The resulting solution was washed with water
(50mL Â 3), dried over anhydrous MgSO4, and con-
centrated under reduced pressure to give l-3 as a white
solid (2.7 g, 91%). It was used without further purifica-
tion for the preparation of l-4. Compound l-3 (2.7 g,
9.1 mmol) was dissolved in dry CH2Cl2 (100 mL) and
cooled to 0 ꢁC. To this solution was added portionwise
3-chloroperoxybenzoic acid (over 77%, 2.2 g, 10mmol)
and stirred for 12 h at room temperature. The pre-
cipitated 3-chlorobenzoic acid was filtered off and the
filtrate was washed with saturated NaHCO3 solution
(50mL Â 3), brine (50mL Â 3), dried over anhydrous
0
forms complex with TLN. Accordingly, the aromatic
chain of the P1 residue in the substrate would when it
0
ring of l-2 would be readily accommodated in the S1
pocket when it binds TLN, and as a consequence l-2
binds and inhibits TLN much more strongly than its
enantiomer.
Understanding of stereochemical effect of biologically
active chiral compounds at the molecular level is of
paramount importance, especially in connection with
development of chiral drugs. In spite of its importance,
our knowledge on the subject is very limited. The pre-
sent study that unravels the stereochemical preference
shown by the reversed hydroxamates of a- and
b-PheNHMe would be of considerable value for
designing small molecule inhibitors that are effective
towards zinc proteases of etiological importance.

S.-J. Kim et al. / Bioorg. Med. Chem. 11 (2003) 2421–2426
10
2425
MgSO4, and evaporated in vacuo to give oxaziridine
l-4. Compound l-4 (2.8 g, 9.1 mmol) was dissolved in
MeOH (100 mL) and to the resulting solution was
added hydroxylamine hydrochloride (0.7 g, 10 mmol)
was added. The mixture was stirred for 12 h at room
temperature, then evaporated under reduced pressure.
Water was added to the residue, and the mixture was
washed with ether (50mL Â 3). The aqueous phase was
neutralized with 5% NaHCO3 solution and extracted
with CH2Cl2 (50mL Â 3). The organic phase was dried
over anhydrous MgSO4, evaporated in vacuo, and the
residue was recrystallized from methanol to afford l-5
FA-Gly-l-Leu-NH2 (final concentration, 1 mM) and
inhibitor (four concentrations in the range of
0.04–4 mM) in 0.1 M Tris/0.01 M CaCl2, pH 7.2 buffer
(1 mL cuvette, the final content of DMF was controlled
at 5%), and the change in absorbance at 345 nm was
measured immediately. The final concentration of TLN
was 0.45 mM. Initial velocities were then calculated from
the initial linear portion of the change in absorbance
where the amount of substrate consumed was less than
10%. The Ki values were determined from the plot of
vo=v against the concentration of inhibitors based on the
equation, vo=v ¼ 1 þ ½IŠ=Ki in which vo and v represent
the initial velocity in the absence and presence of the
inhibitor, respectively (Fig. 1). In the plot, the intercept
of the straight on [I] would give the Ki value.
as
a
white solid (1.25 g, 71%). Mp 152–154 ꢁC
25
D
25
D
(lit. 151–152.5 ꢁC)7; ½aŠ +7.8ꢁ (c 2, MeOH) (lit. ½aŠ
+6.6ꢁ (c 2, MeOH))7; 1H NMR (CD3OD, 300 MHz)
d 2.67 (s, 3H), 2.74–2.91 (m, 2H), 3.59 (t, 1H), 7.18–7.30
(m, 5H); 13C NMR (CD3OD, 300 MHz) d 25.0, 35.8,
68.4, 126.7, 128.5, 129.2, 137.7, 174.7. Anal. calcd for
C10H14 N2O2: C, 61.84; H, 7.27; N, 14.42. Found: C,
61.56; H, 7.23; N, 14.21.
Crystal growth. TLN (Calbiochem) was crystallized as
described by Holmes and Mathews13 with slight modifi-
cations. Briefly, TLN was dissolved in a 0.05 M Tris
buffer (pH 7.2) solution containing DMSO [45% (v/v)]
and calcium acetate (1.4 M) to have the final protein
concentration of about 100 mg/mL. Crystals were grown
by the hanging-drop vapor diffusion method by using a
reservoir solution containing 0.01 M calcium acetate, 5%
(v/v) DMSO and 0.05 M Tris buffer (pH 7.2). Native
crystals of TLN were equilibrated in the reservoir solu-
tion supplemented with 10mM of the inhibitor for 2–3
days to obtain crystals of TLN-inhibitor complex.
N-Hydroxy-D-phenylalanine methylamide (D-5). Com-
pound d-5 was prepared from d-phenylalanine methyl-
amide in a manner analogous to that used for the
25
preparation of l-5. ½aŠ À6.6ꢁ (c 2, MeOH). Anal. calcd
D
for C10H14N2O2: C, 61.84; H, 7.27; N, 14.42. Found: C,
61.58; H, 7.33; N, 14.24.
N - Hydroxy - N - formyl - L - phenylalanine methylamide
(L-1). To the ice-chilled acetic anhydride (1 mL,
10.3 mmol) was added dropwise formic acid (95–97%,
0.5 mL, 12.8 mmol) and the resulting mixture was stir-
red for 5 min. The solution was heated at 50 ꢁC for
5 min, then cooled to 0 ꢁC. To the resulting solution was
added l-5 (0.203 g, 1.04 mmol) at room temperature.
The solution was stirred until l-5 was no longer detec-
ted when tested by TLC, concentrated, and the residue
was dissolved in MeOH. The methanol solution was
treated with 2 N NaOH, and evaporated under reduced
pressure. The residue was treated with 3 N HCl. The
aqueous solution was extracted with ether (50mL Â 3).
The combined ether extracts were evaporated in vacuo
and the solid residue that obtained was recrystallized
X-ray diffraction data collection and structure refine-
ment. Diffraction data were collected by using a
Rigaku RU300 rotating anode X-ray generator operat-
ing at 40kV Â100 mA and R-axis IV++ imaging plate
detector system. An R-axis data processing program
(Rigaku) and the programs MOSFLM14 and SCALA15
were used for the data processing. After the initial rigid
body refinement, several alternating cycles of model
building and simulated annealing refinements were car-
ried out. The programs O16 and CNS17 were used in the
model building and refinement, respectively. The ran-
domly selected 5% of the data were set aside for the
Rfree calculation. Water molecules were gradually added
to the model with the waterpick routine in the program
CNS. Although extra densities for the added inhibitors
were apparent from the initial rigid body refinement
stage, the inhibitor models were not included until the
last stage of refinement.
from ethyl acetate to give l-1 as a white solid (0.13 g,
25
56%). Mp 169–169.5 ꢁC. ½aŠ +7.6ꢁ (c 0.5, MeOH). IR
D
(CH2Cl2): 1664, 1735, 3324 cmÀ1. 1H NMR (DMSO-d6,
300 MHz) d 2.62 (d, 3H) 2.95–3.17 (m, 2H) 4.42 (dd,
1H) 4.89 (dd, 1H) 7.19–7.30(m, 5H) 7.67 (s, 0.66H) 7.86
(d, 1H) 8.16 (s, 0.34H). 13C NMR (DMSO-d6,
300 MHz) 26.7, 34.2 64.5 127.2 129.1 130.0 138.6 159.0
169.6. Anal. calcd for C11H14N2O3: C, 59.45; H, 6.35;
N, 12.61. Found: C, 59.83; H, 6.23; N, 12.61.
Acknowledgements
The authors express their thanks to Ministry of Science
and Technology and Korea Foundation for Science and
Engineering for their financial supports of this work and
Ministry of Education and Human Resources for the
BK21 fellowship to JDP.
N - Hydroxy - N - formyl - D - phenylalanine methylamide
(D-1). Compound d-1 was prepared from d-5 in a manner
25
D
analogous to that used for the preparation of l-1. ½ꢀŠ
À6.8ꢁ (c 1, MeOH). Anal. calcd for C11H14N2O3: C, 59.45;
H, 6.35; N, 12.61. Found: C, 59.26; H, 6.59; N, 12.76.
References and Notes
Determination of Ki value. Enzyme stock solution was
prepared by dissolving TLN (3 mg) in 2 mL of pH 7.2
buffer 0.1 M Tris/0.01 M CaCl2. Typically, the enzyme
stock solution was added to a solution containing
1. (a) Matthews, B. W. Acc. Chem. Res. 1988, 21, 333. (b)
Fersht, A. Enzyme Structure and Mechanism, 2nd ed.; W.H.
Freeman & Co.: New York, 1977; p 418.

2426
S.-J. Kim et al. / Bioorg. Med. Chem. 11 (2003) 2421–2426
2. (a) Ondetti, M. A.; Rubin, B.; Cushman, D. W. Science
1977, 196, 441. (b) Cushman, D. W.; Cheung, H. S.; Sabo,
E. F.; Ondetti, M. A. Biochemistry 1977, 16, 5484.
3. Whittaker, M.; Floyd, C. D.; Brown, P.; Gearing, A. J. H.
Chem. Rev. 1999, 99, 2735.
4. Powers, J. C.; Harper, J. W. Inhibitors of Metalloproteases;
In Proteinase Inhibitors, Barrett, A. J., Salvesen, G. Ed.;
Elsevier: Amsterdam, 1986; pp 219–298.
10. Feder, J. A. Biochem. Biophys. Res. Common. 1968, 32,
326.
11. Cross, J. B.; Duca, J. S.; Kaminke, J. J.; Madison, V. S.
J. Am. Chem. Soc. 2002, 124, 11004.
12. Holmes, M. A.; Matthews, B. W. Biochemistry 1981, 20,
6912.
13. Holmes, M. A.; Matthews, B. W. J. Mol. Biol. 1982, 160,
623.
5. Browner, M. F.; Smith, W. W.; Castelhano, A. L. Bio-
chemistry 1995, 34, 6602.
14. Leslie, A. G. W. Acta Crystallogr. 1999, D55, 1696.
15. Collaborative Computational Project Number 4 The
CCP4 Suite: Programs for Protein Crystallography, Acta
Crystallogr. 1994, D50, 760.
16. Jones, T. A.; Zou, J. Y.; Cowan, S. W.; Kjeldgaard, M.
Acta Crystallogr. 1991, A47, 101.
17. Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.;
¨
Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Rice, L. M.;
Simonson, T.; Warren, G. L. Acta Crystallogr. 1998, D54, 90 5.
6. Jin, Y.; Kim, D. H. Tetrahedron: Asymmetry 1997, 8, 3699.
7. Jin, Y.; Kim, D. H. Bioorg. Med. Chem. Lett. 1998, 8, 3515.
8. Feenstra, R. W.; Stokkinggreef, E. H. M.; Reichenwein,
A. M.; Lousberg, W. B. H.; Ottenheijm, H. C. J. Tetrahedron
1990, 46, 1745.
9. Jahngen, E. G. E.; Rossomando, E. F. Syn. Commun. 1982,
12, 601.