Inhibition and structure-activity studies of methionine hydroxamic acid derivatives with bacterial peptide deformylase

Article abstract of DOI:10.1006/bioo.2001.1214

The posttranslational deformylation of N-formyl-Met-polypeptides by the metalloenzyme, peptide defomylase, is essential for bacterial growth. Methionine hydroxamic acid derivatives were found to inhibit recombinant Escherichia coli peptide deformylase activity containing either zinc or cobalt. The binding of methionine hydroxamate and hydrazide inhibitors to cobalt-substituted deformylase caused spectral changes consistent with the formation of a pentacoordinate metal complex similar to that of actinonin, a psuedopeptide hydroxamate inhibitor. The spectral and kinetic data support the binding of these N-substituted L-methionine derivatives in a reverse orientation with respect to N-formyl-Met-peptide substrates within the active site. Based on this hypothesis a second generation of N-substituted methionyl hydroxamic acids were evaluated and found to possess greater inhibitory potency. These results may provide the basis for the design of more potent and selective deformylase inhibitors as potential antibacterial agents.

Full text of DOI:10.1006/bioo.2001.1214

Bioorganic Chemistry 29, 211–222 (2001)
doi:10.1006/bioo.2001.1214, available online at http://www.idealibrary.com on
Inhibition and Structure –Activity Studies of Methionine
Hydroxamic Acid Derivatives with Bacterial Peptide
Deformylase
,1
Stephan K. Grant,* Barbara Gordon Green,† and John W. Kozarich†
*
Department of HTS and Automation and †Department of Biochemistry Merck & Co.,
P.O. Box 2000, Rahway, New Jersey 07065
Received February 1, 2001; published online August 9, 2001
The posttranslational deformylation of N-formyl-Met-polypeptides by the metalloenzyme,
peptide defomylase, is essential for bacterial growth. Methionine hydroxamic acid derivatives
were found to inhibit recombinant Escherichia coli peptide deformylase activity containing
either zinc or cobalt. The binding of methionine hydroxamate and hydrazide inhibitors to
cobalt-substituted deformylase caused spectral changes consistent with the formation of a
pentacoordinate metal complex similar to that of actinonin, a psuedopeptide hydroxamate
inhibitor. The spectral and kinetic data support the binding of these N-substituted L-methionine
derivatives in a reverse orientation with respect to N-formyl-Met-peptide substrates within the
active site. Based on this hypothesis a second generation of N-substituted methionyl hydroxamic
acids were evaluated and found to possess greater inhibitory potency. These results may provide
the basis for the design of more potent and selective deformylase inhibitors as potential
antibacterial agents.
᭧ 2001 Academic Press
Key Words: peptide deformylase; metalloenzyme; hydroxamic acid; hydrazide; actinonin;
enzyme inhibition.
INTRODUCTION
Bacteria contain many aminopeptidases that cleave N-terminal amino acids from
peptides. Methionine aminopeptidase (MetAP, EC 3.4.11.18), for example, is an
essential bacterial enzyme that hydrolyzes Met from polypeptides during protein
maturation (1,2). MetAP cannot hydrolyze N-blocked Met-polypeptides such as N-
formyl-Met-peptide formed during protein synthesis in eubacteria (3). Deformylation
of the N-formyl-Met-polypeptide is therefore a prerequisite step for proper posttransla-
tional modification of eubacterial proteins. The deformylation step is catalyzed by
1
To whom correspondence should be addressed. Department of HTS and Automation, Merck Re-
search laboratories, RY80N-1B28, P.O. Box 2000, Rahway, NJ 07065. Fax: (732) 594-0390. E-mail:
stephan grant@merck.com.
2
Abbreviations used: fMV, N-formyl-Met-Val; LMCT, ligand to metal charge transfer; MetAP, methio-
nine aminopeptidase; PDF, peptide deformylase.
2
11
0
045-2068/01 $35.00
Copyright ᭧ 2001 by Academic Press
All rights of reproduction in any form reserved.

212
GRANT, GREEN, AND KOZARICH
the bacterial metalloenzyme, peptide deformylase (PDF, EC 3.5.1.27). Escherichia coli
PDF contains a metal binding motif HEXXH, which is conserved among deformylase
sequences from diverse bacterial sources (4–6). A Cys-thiolate is also conserved in
the active site, forming a tetracoordinate metal center with two imidazole groups
from the HEXXH motif and a catalytic water molecule (4,5). Several metals including
zinc (7,8), cobalt (8), nickel (9,10), and iron (1,12) have been found in preparations
of purified recombinant E. coli PDF. The specific activity of the enzyme changes
greatly depending upon which metal ion is bound within the active site. Cobalt (8)
and nickel (9,10) enzymes have much greater k /K values for N-formyl-Met-peptides
cat
m
3
5
Ϫ1 Ϫ1
such as the chemotactic peptide, N-formyl-Met-Leu-Phe (10 –10 M s ), than
2
3
Ϫ1 Ϫ1
zinc-PDF (10 –10 M s ). The variation of specific activity with different metal
substitution has been attributed to the turnover of substrate, a kcat effect, rather than
substrate binding affinity (8). While the iron-enzyme is more active than other metal-
4
6
Ϫ1 Ϫ1
substituted deformylases (10 –10 M s ), its activity is also much less stable
(8,11–13).
The initial characterization of recombinant zinc E. coli peptide deformylase by
commercially available N-formylated peptides such as N-formyl-Met-Leu-Phe, indi-
cated that the enzyme was specific for N-formylated-Met-peptides or N-formyl-norleu-
cine peptides (7). The solution NMR (16) and crystal structures (17,18) of recombinant
E. coli PDF indicated that the active site is buried within the protein. From these
structural studies a model was proposed where N-formyl-peptide substrates may enter
the active site but not extend beyond the metal binding pocket. This restricted active
site model provides an explanation for why PDF hydrolyzes formamide substrates
and lacks peptidase activity (7).
While the zinc-enzyme appears to be selective for N-formylated-Met-peptides,
studies with nickel-PDF (14) and iron-PDF (15) has extended the substrate specificity
to include N-formylated peptides with hydrophobic aromatic amino acids at the P1Ј
position, such as N-formyl-Phe-Ala-Ser. It has also been reported that N-difluoroacetyl-
Met-Leu-p-nitroanilide and N-trifluoroacetyl-Met-Leu-p-nitroanilide but not N-acetyl-
Met-Leu-p-nitroanilide can be deacetylated by iron-PDF (15,19). Furthermore, our
laboratory has recently reported that peptide aldehydes (8) and biaryl acid analogs
(20) are also inhibitors of recombinant deformylase but are more potent against cobalt-
substituted PDF than the zinc-enzyme. Together the results from these separate studies
suggest that deformylase substrate and inhibitor selectivity may be dependent upon
which metal is bound within the enzyme active site. Metal-dependent inhibitor selectiv-
ity may be an important consideration in the design and development of PDF inhibitors
as potential antibacterial agents.
Recently, the antibiotic activity of actinonin 1, a pseudopeptide hydroxamic acid,
was shown to be attributable to its specific inhibition of peptide deformylase (21).
Actinonin was also found to be a very potent inhibitor of recombinant zinc, nickel,
and iron-substituted E. coli deformylase (21). In this study, we have evaluated the
activity and binding characteristics of a series of methionine hydroxamic acid deriva-
tives against recombinant E. coli deformylase specifically containing either zinc
or cobalt.

INHIBITION OF PEPTIDE DEFORMYLASE
213
EXPERIMENTAL PROCEDURES
Materials. Actinonin 1 and N-formyl-Met-Val were obtained from Sigma (Fig. 1).
N-[4-(Trifluoromethyl)benzoyl]- -methionine hydroxamic acid 2, N-[4-(trifluoro-
methyl)-benzoyl]- -methionylhydrazide 3, and N-[4-(Trifluoromethyl)benzoyl]- -me-
L
L
L
thionine methyl ester 4 were purchased from Maybridge Chemical Company (Corn-
wall, UK). The enantiomeric purity of the commercially obtained compounds 1–4
1
was confirmed by H NMR (J. O’Connell, Merck & Co., pers. commun.). The second
generation methionyl hydroxamic acids were procured from Pharm-Eco Laboratories,
Inc. (Lexington, MA). The compounds were synthesized by coupling various carbox-
ylic acids to DL-methionyl hydroxamic acid with an average 85% purity with none less
than 70% purity (Richard Lombardy, Pharm-Eco Laboratories, Inc., pers. commun.).
Preparation of enzymes. The cloning, expression, and purification of recombinant
truncated and full-length E. coli PDF was previously described (8,16). For production
of recombinant deformylase with homogeneous metal content, bacterial cells were
grown in media with defined metal sources that provided recombinantly expressed
deformylase specifically containing either zinc or cobalt (16). Similar K values for
m
N-formyl-Met-Val were determined for truncated zinc and cobalt deformylase of 8.0 Ϯ
0
.2 and 7.1 Ϯ 0.3 mM, respectively. A K value of 12.4 Ϯ 0.8 mM was determined for
m
the full-length cobalt enzyme with N-formyl-Met-Val. Metal content of homogeneous
metal-substituted enzymes was confirmed by inductively coupled plasma mass spec-
trometry (T. Wang, Analytical Research, Merck & Co., pers. commun.) or atomic
FIG. 1. Structures of Actinonin and N-benzoyl-methionine analog derivatives.

214
GRANT, GREEN, AND KOZARICH
absorption spectrophotometry (Schwarzkopf Microanalytical Laboratory, Woodside,
NY).
Enzyme activity assays. Deformylase activity was monitored by a NAD-formate
dehydrogenase coupled, continuous spectrophotometric assay as described earlier (8).
IC values were determined in duplicate from initial velocity measurements at varying
50
inhibitor concentrations and a single substrate concentration (2 mM N-formyl-Met-
Val). Reversibility was evaluated by preincubation of enzyme and inhibitor followed
by dilution into an assay buffer containing saturating substrate (40 mM N-formyl-
Met-Val). The kinetic mechanism of enzyme inhibition was evaluated by varying
inhibitor and substrate concentrations and fitting initial velocity data to equations for
competitive, noncompetitive, and uncompetitive inhibition. Time-dependent inhibition
was evaluated by fitting nonlinear progress curves to the integrated first-order expres-
Ϫkt
sion, y ϭ v t ϩ (v Ϫ v )(1 Ϫ e )/kobs ϩ A , where v is the steady-state velocity,
s
0
s
0
s
v0 is the initial velocity, and kobs is the observed rate constant. Kinetic parameters for
slow-binding enzyme inhibition were determined from plotting the observed rate
constants versus inhibitor concentration according to kobs ϭ k [I] ϩ K and K ϭ
on
off
I
k /k , where k is the association rate constant, I is the inhibitor concentration, k is
off on
on
off
the dissociation rate constant, and K is the inhibitor equilibrium dissociation constant.
I
Electronic absorption spectra. Optical spectra of cobalt-substituted deformylase
were measured with a Hewlett Packard diode array spectrophotometer (Model
HP8452A) equipped with a thermojacketed cell holder and circulating water bath.
Samples (0.3 mL) were incubated at 25ЊC and the addition of concentrated inhibitors
were made with a Hamilton 10-␮L syringe to minimize sample dilution. Spectra were
recorded before the addition of inhibitor to the protein solution and then after subse-
quent additions of inhibitor. The samples were allowed to equilibrate between addition
of compounds and the spectral signals were stable after 1 min. The difference spectrum
was generated by subtraction of the spectrum of the protein before the addition
of compound.
RESULTS
Inhibition of peptide deformylase by N-benzoyl-methionine derivatives. Three N-
benzoyl-methionine derivatives: N-[4-(Trifluoromethyl)benzoyl]-
droxamic acid 2, N-[4-(trifluoromethyl)benzoyl]- -methionyl hydrazide 3, and N-[4-
trifluoromethyl)benzoyl]- -methionine methyl ester 4, were examined as potential
L-methionine hy-
L
(
L
deformylase inhibitors at 1 mM. Both the hydroxamate 2 and hydrazide 3 inhibited
both zinc and cobalt-substituted PDF activity, but the methyl ester 4 did not. In
support of the proposal that the observed inhibition of PDF by compounds 2 and 3
was specific and not via chelation or inhibition of the coupling enzyme, we demon-
strated that the inhibition of PDF was unaffected by addition of free zinc or cobalt
(0.1 mM) and that these compounds did not inhibit formate dehydrogenase alone.
The hydroxamic acid 2 was a more potent inhibitor of Co-PDF than Zn-PDF with
IC values of 3.2 Ϯ 0.2, 37.0 Ϯ 0.7, and 330 Ϯ 30 ␮M for truncated and full-length
50
Co-PDF and truncated Zn-PDF, respectively. The inhibition of full-length Co-PDF
by hydroxamic acid 2 was competitive with K ϭ 30.0 Ϯ 1.1 ␮M (Fig. 2). In contrast,
i
actinonin 1 gave complete inhibition of E. coli PDF activity when assayed at 1 ␮M.
An accurate IC value for actinonin against zinc or cobalt deformylase could not be
5
0

INHIBITION OF PEPTIDE DEFORMYLASE
215
FIG. 2. Double-reciprocal plot for inhibition of Co-PDF by (4-[trifluoromethyl]benzoyl)-
L-methionine
hydroxamic acid 2 (0, 10, 25, 50, or 100 ␮M). Experiment was performed in duplicate and average values
are shown. Substrate concentration was 4, 5, 6.6, 10, or 20 mM fMV. Velocity data (mOD/min) was fit
to a competitive model of inhibition by a nonlinear least-squares algorithm.
determined because actinonin still inhibited enzyme activity at concentrations equal
to the lowest enzyme concentration that can be accurately detected by the spectrophoto-
metric assay (about 1 nM for Co-PDF and 30 nM for Zn-PDF). This is consistent
with a recent report (21) that estimated a K value of 0.25 nM for actinonin with
i
recombinant Ni-PDF.
Unlike hydroxamate 2, hydrazide 3 displayed time-dependent inhibition of zinc
and cobalt deformylase. Representative plots of the time-dependent inhibition of PDF
by the hydrazide are shown in Fig. 3 and were observed when assayed at substrate
concentrations (40 mM fMV) greater than K . The inhibition curves were psuedo-
m
first-order and were observed when the reaction was initiated with either enzyme or
substrate. When the reaction was initiated with enzyme in the presence of hydrazide
3
, there was a fast initial velocity for product formation followed by a slower steady-
state velocity. This inhibition was fully reversible, since preincubation of enzyme
with inhibitor followed by the addition of excess substrate gave a slow initial velocity
that changed to a faster steady-state velocity. For both experiments the steady-state
velocities were nearly identical. Kinetic parameters for the time-dependent inhibition
were determined by fitting the data to the integrated first-order expression. A plot of
the observed rate constants versus inhibitor concentration is shown in Fig. 4. The
data fit best to a model mechanism for slow-binding enzyme inhibition and gave
slow association rates (k values) for the binding of the inhibitor to the enzyme. The
on
kinetic parameters for the time-dependent inhibition of truncated and full-length PDF
by hydrazide 3 are shown in Table 1. The kinetic parameters for the slow-binding
inhibition of the truncated Zn and Co-enzymes were very similar but the association
rate of the hydrazide was slower with full-length Co-PDF. Spectral studies were

216
GRANT, GREEN, AND KOZARICH
FIG. 3. Slow-binding kinetics for inhibition of PDF by hydrazide 3. (A) Representative time courses
for the enzyme reaction in the presence of inhibitor when initiated by enzyme. Inhibitor concentration
was 0, 0.05, 0.1, 0.2, and 0.3 mM with 40 mM fMV. (B) Representative time courses for the enzyme
reaction after preincubation with inhibitor for 20 min at 25ЊC, followed by initiation with 40 mM fMV.
Experiments were performed in triplicate with average values shown.
initiated with cobalt-substituted PDF to further evaluate the binding of the hydroxamate
and hydrazide inhibitors.
Electronic spectra for inhibitor binding to cobalt-peptide deformylase. Cobalt-
substituted full-length and truncated E. coli deformylase were pink-colored proteins
(8). Distinct spectral bands were observed for both cobalt-substituted enzymes at
about 330, 562, and 660 nm (partially obscured by an instrument spike). These bands
FIG. 4. Plot of observed rate constants versus inhibitor concentration for time-dependent inhibition
of PDF by hydrazide 3. Observed rate constants were calculated by fitting inhibition curves to the integrated
first-order expression. Values were averaged from triplicate experiments with standard deviations shown.

INHIBITION OF PEPTIDE DEFORMYLASE
217
TABLE 1
Kinetic Parameters for Slow-Binding Enzyme Inhibition by Hydrazide 3
Parameter
Truncated Zn-PDF
Truncated Co-PDF
Full-length Co-PDF
on (M sϪ1
Ϫ1
)
120
0.00139
11.6
121
0.00142
11.8
21.6
0.00434
200.5
k
k
off (sϪ1)
K (␮M)
I
correspond to single thiolate-Co(II) ligand to metal charge transfer (LMCT, 330 nm)
transitions and Co(II) d–d transitions (562 and 660 nm), respectively (8,22–25).
Hydroxamic acids typically bind as bidentate ligands to the metal ion of metallopro-
teases causing the formation of a pentacoordinate metal complex (26,27). This change
in the geometry of the metal upon complexation with the hydroxamate causes perturba-
tions in the electronic spectra of cobalt metalloenzymes that are characteristic for the
formation of a pentacoordinate complex (23). Addition of actinonin 1 or hydroxamate
2
to either full-length or truncated Co-PDF caused the loss of the pink color of the
protein. Figure 5 shows representative spectra recorded upon addition of actinonin
to full-length Co-PDF. The inhibitors themselves did not show spectral bands in the
3
3
00- to 700-nm range but hydroxamate 2 did give strong absorbance bands below
00 nm (spectrum not shown). The difference spectra recorded for these two enzyme–
inhibitor complexes were nearly identical to each other in the 300- to 700-nm range.
Similar difference spectra were also observed with these two inhibitors with truncated
Co-PDF (spectra not shown). Overall the following spectral changes were observed
upon addition of actinonin or hydroxamate 2 to Co-PDF: (1) a shift in the ␭max of
the LMCT band from about 330 to 340 nm, (2) the formation of a weak, broad band
at about 490 nm, and (3) a decrease in the d–d transitions at 568 and 660 nm. The
magnitudes for the spectral changes were linear with addition of either actinonin or
hydroxamate 2 until equivalent to the enzyme concentration (about 300 ␮M). Excess
inhibitor did not further perturb the optical spectra. These spectral changes were
similar to those described for the binding of hydroxamic acid inhibitors to other
cobalt-substituted metalloenzymes, resulting from the conversion of a tetracoordinate
cobalt for the free enzyme to a pentacoordinate metal for the enzyme–inhibitor
complex (22,23).
The addition of hydrazide 3 to cobalt deformylase also caused spectral perturbations
that were identical for both full-length or truncated enzymes and changed linearly to
equal concentrations of enzyme and inhibitor. These spectral changes were similar
to those observed for the addition of the hydroxamic acids with the formation of a
weak, broad band at about 490 nm, and decreases in the d–d transitions at 568 and
6
60 nm (Fig. 5C). For the hydrazide–enzyme complex, however, there was a much
greater shift in the ␭max of the LMCT band from about 330 to 358 than observed for
the hydroxamate complex. This result indicates that the binding of the hydrazide
perturbs the cobalt–thiolate bond to a greater extent than by complexation with the
hydroxamic acids.
Structure–activity studies of methionyl hydroxamate inhibitors. In order for the
hydroxamic acid group of the L-methionyl hydroxamate 2 to coordinate directly to

218
GRANT, GREEN, AND KOZARICH

INHIBITION OF PEPTIDE DEFORMYLASE
219
FIG. 6. Model for binding of substrate (A) and hydoxamate inhibitor (B) to Co-PDF.
the metal of deformylase, the inhibitor must bind in a reverse orientation from the
proposed binding of formamide substrates. The enantiomeric purity of the commer-
1
cially obtained hydroxamate 2 was confirmed by H NMR (J. O’Connell, Merck &
Co., pers. commun.). A model for the reverse binding of the hydroxamate 2 is shown
in Fig. 6. In this binding model for the inhibitor, the hydroxamate group coordinates
directly to the metal ion and the N-trifluoromethylbenzoyl group is positioned away
from the metal. Substitution at the amino group of methionyl hydroxamic acid may
therefore provide a means for further probing the active site. Such derivatives also
may allow for the design of more potent and selective inhibitors of deformylase.
Accordingly, second-generation N-substituted DL-methionyl hydroxamic acids with
simple alkyl to large aromatic groups were examined as potential inhibitors of truncated
zinc and cobalt deformylase (Table 2). N-Alkyl ketone substitutions such as acetyl
5
, were found to be poor inhibitors of zinc and cobalt-PDF. Two other N-benzoyl
derivatives 9 and 10 were more potent against the cobalt enzyme but were still poor
inhibitors of zinc-PDF. Substitution of larger aryl groups such as the l-naphthoyl
derivative 11 increased inhibitor potency and the l-naphthylaceto-derivative 12 had the
greatest potency against zinc and cobalt-deformylase with an IC in the submicromolar
50
range against Co-PDF.
DISCUSSION
A detailed structural map of the active site of E. coli deformylase has been defined
by both solution NMR (16) and crystallographic (17,18) structures. Based on the
structures of the native enzyme and enzyme–inhibitor complexes and from structure–
activity studies with N-blocked peptides, a model for the substrate binding site has
emerged. In this model the formyl group of N-formyl-Met-peptide substrates occupies
FIG. 5. Electronic absorption spectra for PDF–inhibitor complex formation with actinonin and
hydrazide 3. (A) Binding of actinonin to full-length Co-PDF (approximately 300 ␮M) in 20 mM Hepes,
pH 7, 25ЊC, with increasing amounts of inhibitor (a–d, 0, 100, 250, 400 ␮M). (B) Difference spectra for
Co-PDF with actinonin. (C) Difference spectra for the binding of hydrazide 3 to full-length Co-PDF
(approximately 200 ␮M) in 20 mM Hepes, pH 7, 25ЊC, with increasing amounts of inhibitor (a–d, 0,
6
0, 120, 180 ␮M). Absorbance of the hydrazide alone accounts for the spectral bands about 300 nm.
Note that an instrument spike partially obscured the bands at 660 nm. These spectral changes represent
a single process without any observable intermediate as evidenced by the isobestic points in the differ-
ence spectra.

220
GRANT, GREEN, AND KOZARICH
TABLE 2
Inhibition of Deformylase by Second-Generation Hydroxamic Acids
IC50 (␮M)
R group
Zn-PDF
Co-PDF
5
6
7
8
9
0
CH
(CH
(CH
Phenyl
1-Naphthylene
2-(Phenoxy)phenyl
(4-CF
1-Naphthylmethylene
3
–
NAa
NA
NA
NA
3
)
2
2
CHCH
CH–
2
–
33.8 Ϯ 2.4
27.6 Ϯ 1.2
5.4 Ϯ 0.2
0.37 Ϯ 0.02
0.24 Ϯ 0.01
1.46 Ϯ 0.02
0.19 Ϯ 0.01
3
)
NA
55.9 Ϯ 5.8
49.7 Ϯ 4.5
NA
1
1
1
1
2
3 6 4
)C H -methylene
15.7 Ϯ 1.0
a
NA, not active as defined by a value less than 50% inhibition at 100 ␮M.
a position close to the metal ion (4,9,14,16,17). Even slightly larger groups such as
acetyl are not acceptable because of unfavorable steric interactions with active site
Gln-50 or Gily-45 residues (16–18). Yet unpredictably, N-benzoyl-
L-Met derivatives
2
and 3 were found to be inhibitors of PDF. One possible explanation for their
inhibitory activity is that these compounds bind within the active site in a reverse
orientation compared to substrate so that it is not the N-benzoyl moiety that binds
near the metal but the respective hydroxamate or hydrazide groups of these compounds.
In this model, the hydroxamate 2 and the hydrazide 3 complex with the active site
metal similar to the pseudopeptide hydroxamic acid, actinonin. Indeed, the binding
of hydroxamate 3 and actinonin to cobalt-substituted PDF were virtually indistinguish-
able and were consistent with the formation of pentacoordinate metal–hydroxamate
complexes.
The spectral changes observed upon complexation of hydrazide 3 with Co-PDF is
also consistent with the binding of the inhibitor in a reverse orientation than substrate,
with the formation of a pentacoordinate complex. The large perturbation of the LMCT
band in the hydrazide–enzyme complex suggests that there is a larger effect upon
the Co-thiolate bond than observed upon complexation with the hydroxamates. Like
aldehydes, hydrazides have a well known affinity for thiols. Previously, we reported
the electronic spectra upon complexation of a peptide aldehyde inhibitor with Co-
PDF (8). However, there is no observable loss of the LMCT band as would be
predicted from covalent labeling of the active site Cys residue. Indeed, the spectral
characteristics of the hydrazide–Co-PDF complex are very different from the alde-
hyde–enzyme complex (8) and more closely resemble the spectra for the hydroxa-
mate–enzyme complexes. While the hydrazide 3 showed slow association kinetics

INHIBITION OF PEPTIDE DEFORMYLASE
221
with Co-PDF and a greater shift in the LMCT band upon binding, the nature for these
differences compared to the hydroxamate 3 is not apparant from these studies alone.
Based on our reverse binding model, we envisioned that the amino-group of a
methionyl hydroxamic acid might be a useful site to append with pharmacophores
of diverse structure and potentially increase inhibitor potency. Indeed, second-genera-
tion N-substituted methionine hydroxamic acids with larger aromatic groups than the
original hydroxamate 2 had greater inhibitory potency for PDF. While these second-
generation hydroxamates were synthesized from the DL-amino acid, these results
demonstrate that potent PDF compounds can be obtained using the Met-hydroxamate
template. The potential antibiotic activity of these N-substituted methionine hydroxa-
mate derivatives is being further investigated.
Another observation that we have made during these studies is that the potency of
N-substituted methionine analog inhibitors can vary considerably depending on the
metal substitution of the enzyme. In general, the IC50 values of the inhibitors were
lower for Co-PDF but the hydrazide 3 was equally potent for zinc and cobalt deformy-
lase. While we hesitate to speculate on the nature for these differences in inhibitor
potency based on this short list of compounds, these results demonstrate that inhibitor
specificity can be dependent on metal substitution. The potential for metal-dependent
inhibitor specificity may be an important consideration since recent reports concerning
the discovery of PDF inhibitors as possible antibacterial agents have utilized recombi-
nant enzyme with either zinc, cobalt, nickel, or iron centers. Moreover, since some
reports have argued that native deformylase is most likely an iron-dependent enzyme
based on turnover number (11,12), it would be important to examine the potency of
these inhibitors with recombinant iron-PDF. Such studies with a set of metal-dependent
PDF inhibitors with biological activity might allow further confirmation of which
metal or metals contribute to native deformylase activity.
ACKNOWLEDGMENTS
A special thanks to Dr. Barbara Leiting and Mr. Frank Marsilio for expression and purification of
recombinant proteins and to Dr. Kenny Wong and Dr. David Pompliano for helpful discussions. We also
thank Dr. John O’Connell for confirmation of the enantiomeric purity of compounds by NMR spectroscopy.
We also are grateful to Richard Lombardy, Pharm-Eco Laboratories, Inc., for providing details about the
synthesis of methionyl hydroxamic acid derivatives.
REFERENCES
1. Meinnel, T., Mechulam, Y., and Blanquet, S. (1993) Biochemie 75, 1061–1075.
2. Chang, S.-Y. P., McGary, E. C., and Chang, S. (1989) J. Bacteriol. 171(7), 4071–4072.
3. Solbiati, J., Chapman-Smith, A., Miller, J. L., Miller, C. G., and Cronan, J. E., Jr. (1999) J. Mol.
Biol. 290, 607–614.
4
5
6
7
8
. Meinnel, T., Lazennec, C., and Blanquet, S. (1995) J. Mol. Biol. 254, 175–183.
. Meinnel, T., Blanquet, S., and Dardel, F. (1996) J. Mol. Biol. 262, 375–386.
. Meinnel, T., Lazennec, C., Villoing, S., and Blanquet, S. (1997) J. Mol. Biol. 267, 749–761.
. Meinnel, T., and Blanquet, S. (1995) J. Bacteriol. 177(7), 1883–1887.
. Durand, D. J., Green, B. G., O’Connell, J. F., and Grant, S. K. (1999) Arch. Biochem. Biophys.
3
67(2), 297–302.
9
. Dardel, F., Ragusa, S., Lazennec, C., Blanquet, S., and Meinnel, T. (1998) J. Mol. Biol. 280, 501–513.
10. Ragusa, S., Blanquet, S., and Meinnel, T. (1998) J. Mol. Biol. 280, 515–523.
11. Rajagopalan, P. T. R., Datta, A., and Pei, D. (1997) Biochemistry 36, 13,910–13,918.
12. Rajagopalan, P. T. R., Yu, X. C., and Pei, D. (1997) J. Am. Chem. Soc. 119, 12,418–12,419.

222
GRANT, GREEN, AND KOZARICH
1
3. Groche, D., Becker, A., Schlichting, I., Kabsch, W., Schultz, S., and Wagner, A. F. V. (1998) Biochem.
Biophys. Res. Commun. 264, 342–346.
14. Ragusa, S., Mouchet, P., Lazennec, C., Dive, V., and Meinnel, T. (1999) J. Mol. Biol. 289, 1445–1457.
15. Hu, Y.-J., Wei, Y., Zhou, Y., Rajagopalan, P. T. R., and Pei, D. (1999) Biochemistry 38, 643–650.
16. O’Connell, J. F., Pryoy, K. D., Grant, S. K., and Leiting, B. (1999) J. Biomol. Nucl. Magn. Reson.
1
3(4), 311–324.
7. Chan, M. K., Gong, W., Rajagopalan, P. T. R., Hao, B., Tsai, C. M., and Pei, D. (1997) Biochemistry.
6, 13,904–13,909.
8. Becker, A., Schlichting, I., Kabasch, W., Schultz, S., and Wagner, A. F. V. (1998) J. Biol. Chem.
73(19), 11,413–11,416.
1
1
3
2
1
9. Wei, Y., and Pei, D. (1997) Anal. Biochem. 250, 29–34.
2
0. Green, B. G., Toney, J. H., Kozarich, J. W., and Grant, S. K. (2000) Arch. Biochem. Biophys.
3
75(2), 355–358.
2
2
2
1. Chen, D. Z., Patel, D. V., Hackbarth, C. J., Wang, W., Dreyer, G., Young, D. C., Margolis, P. S.,
Wu, C., Ni, Z.-J., Trias, J., White, R. J., and Yuan, Z. (2000) Biochemistry 39(6), 1256–1262.
2. Vallee, B. L., and Galdes, A. (1984) in Advances in Enzymology, Vol 56 (Meister, A., Ed), pp.
2
84–413, Wiley, New York.
3. Maret, W., and Vallee, B. L. (1993) in Methods in Enzymology (Riordan, J., and Vallee, B. L., Eds.),
Vol 226, pp. 52–71, Academic Press, San Diego.
24. Rajagopalan, P. T. R., Grimme, S., and Pei, D. (2000) Biochemistry 39, 779–790.
25. Huntington, K. M., Yi, T., and Pei, D. (2000) Biochemistry 39, 4543–4551.
26. Grams, F., Crimmin, M., Hinnes, L., Huxley, P., Pieper, M., Tschesche, H., and Bode, W. (1995)
Biochemistry 34, 14,012–14,020.
2
7. Dhanaraj, V., Ye, Q.-Z., Johnson, L. L., Hupe, D. J., Ortwine, D. F., Dunbar, J. B., Jr., Rubin, J. R.,
Ravovsky, A., Humblet, C., and Blundell, T. L. (1996) Structure 4, 375–386.