Targeting Metalloproteins by Fragment-Based Lead Discovery

Article abstract of DOI:10.1111/j.1747-0285.2011.01136.x

It has been estimated that nearly one-third of functional proteins contain a metal ion. These constitute a wide variety of possible drug targets including metalloproteinases, dehydrogenases, oxidoreductases, hydrolases, deacetylases, or many others in which the metal ion is either of catalytic or of structural nature. Despite the predominant role of a metal ion in so many classes of drug targets, current high-throughput screening techniques do not usually produce viable hits against these proteins, likely due to the lack of proper metal-binding pharmacophores in the current screening libraries. Herein, we describe a novel fragment-based drug discovery approach using a metal-targeting fragment library that is based on a variety of distinct classes of metal-binding groups designed to reliably anchor the fragments at the target's metal ions. We show that the approach can effectively identify novel, potent and selective agents that can be readily developed into metalloprotein-targeted therapeutics. We describe a novel fragment based drug discovery approach using a metal targeting fragment library. The library is based on a variety of distinct classes of metal-binding groups designed to reliably anchor the fragments at the target's metal ions. We demonstrate that the approach can effectively identify novel, potent and selective agents that can be readily natured into lead molecules for metalloprotein-targeted therapeutics.

Full text of DOI:10.1111/j.1747-0285.2011.01136.x

ª 2011 John Wiley & Sons A/S
doi: 10.1111/j.1747-0285.2011.01136.x
Chem Biol Drug Des 2011; 78: 211–223
Research Article
Targeting Metalloproteins by Fragment-Based
Lead Discovery
Sherida Johnson1, Elisa Barile1,
pursued: (i) a single fragment can be elaborated in order to obtain
a tight binder or (ii) multiple fragments binding at distinct sites can
be connected by an appropriate linker to obtain a potent inhibitor
(Figure 1A,B) (2,4,5). In most cases, fragments are structurally char-
Biancamaria Farina1, Angela Purves1, Jun
Wei1, Li-Hsing Chen1, Sergey Shiryaev1,
Ziming Zhang1, Irina Rodionova1, Arpita
Agrawal2, Seth M. Cohen2, Andrei
acterized in the bound form either by X-ray crystallography (6) or by
high-field NMR techniques (7). Compared with high-throughput
Osterman1, Alex Strongin1 and Maurizio
screening (HTS), FBLD is purported to have several advantages,
Pellecchia1,*
which include a more efficient exploration of chemical diversity
space and higher ligand efficiencies (LE) (8–14). LE is the binding
free energy for a ligand per number of heavy atoms, which repre-
sent the goodness of the fit between the ligand and the receptor
(8,13). Monitoring the ligand efficiency during the fragment or hit
maturation process helps in assessing the additional contribution
that each given functionality brings to the overall activity of a com-
pound series (3,8).
1Sanford-Burnham Medical Research Institute, 10901 N. Torrey
Pines Rd., La Jolla, CA 92037, USA
2Department of Chemistry and Biochemistry, University of California,
San Diego, 9500 Gilman Drive, La Jolla, CA 92093, USA
*Corresponding author: Maurizio Pellecchia,
mpellecchia@sanfordburnham.org
It has been estimated that nearly one-third of
functional proteins contain a metal ion. These con-
stitute a wide variety of possible drug targets
including metalloproteinases, dehydrogenases,
oxidoreductases, hydrolases, deacetylases, or
binding group (ZBG) was used to bind to the catalytic zinc(II) ion
many others in which the metal ion is either of
(partially occupying one binding site), while a series of hydrophobic,
catalytic or of structural nature. Despite the pre-
One of the earliest applications of FBLD was against the zinc(II)-
dependent matrix metalloproteinase-3 (MMP-3, also known as
stromelysin) (5). In this early study, an acetohydroxamic acid's zinc-
aromatic fragments were examined for their ability to occupy the
dominant role of a metal ion in so many classes of
S1' selectivity pocket (schematically illustrated in Figure 1C) (5). The
weakly interacting second-site-binding hydrophobic fragments (Kds
in the high lM range) were then linked to the acetohydroxamic acid
moiety (Kd in the mM range) to generate ligands with nanomolar
affinity for MMP-3 (Figure 1C) (5,15). Despite this early, important
application of FBLD to metalloproteinases, only a limited number of
subsequent studies have actively applied FBLD to such targets. Fur-
thermore, there are only a few studies that describe the develop-
ment of libraries that are specifically designed to target
metalloproteins (16–19).
drug targets, current high-throughput screening
techniques do not usually produce viable hits
against these proteins, likely due to the lack of
proper metal-binding pharmacophores in the cur-
rent screening libraries. Herein, we describe
a
novel fragment-based drug discovery approach
using a metal-targeting fragment library that is
based on a variety of distinct classes of metal-
binding groups designed to reliably anchor the
fragments at the target’s metal ions. We show that
the approach can effectively identify novel, potent
and selective agents that can be readily developed
into metalloprotein-targeted therapeutics.
In order to demonstrate the utility and advantages of applying FBLD
to metallo-targets, a focused library of about 500 fragments was
synthesized. The compounds were selected based on their ability to
bind metal ions, by using various different ZBG anchors (20,21). By
incorporating a ZBG into each member of the library, it is anticipated
that initial binding fragments are found that more efficiently occupy
a first site around the metal ion (Figure 1D). For each metal anchor-
ing motif, a diverse set of small substituents were appended in
order to generate a diverse fragment library. In this way, a set of
molecular metal-coordinating fragments is obtained that are highly
amenable to elaboration into potent and selective inhibitors (Fig-
ure 1D). Hence, a library was assembled using a variety of frag-
ments, each derivatized with a potential metal-chelating group
(Figure 2), including not only the well-known hydroxamic acid but
Key words: FBLD, metal-chelating fragments, metalloproteins, ZBG
Received 28 January 2011, revised 21 April 2011 and accepted for pub-
lication 24 April 2011
Fragment-based lead design (FBLD), also called fragment-based drug
discovery, is becoming a well-established approach for the discovery
of biologically active compounds. FBLD involves the rapid screening
of libraries of relatively few (hundreds to thousands) fragment com-
pounds against a target of interest (1–5). Once fragments are iden-
tified that bind to a target, generally one of two approaches are
211

Johnson et al.
A
B
C
Figure 1: Schematic represen-
tation of various Fragment-based
lead design approaches. The evolu-
tion of fragments into optimized hit
compounds generally follows one
of two routes involving either (A)
maturation of the initial fragment
or (B) linking two fragments occu-
pying adjacent sites (58). The latter
approach was successfully utilized
(C) for the metalloproteinase matrix
metalloproteinase-3 (15). The pres-
ent approach provides initial zinc-
binding group-containing fragment
hits that can likewise be matured
using either (D) the evolution or (E)
merging two fragments occupying
non-overlapping sites.
D
E
other less-investigated ZBGs (20–22). Specifically, the library con-
tained about 200 hydroxamates [one of them is a known lethal
factor (LF) inhibitor synthesized in-house (23)], 48 were commer-
cially available of which nine are known histone deacetylase (HDAC)
inhibitors, 23 reverse hydroxamates (limited owing to the availability
of starting materials), 96 sulfonylhydroxamates, 64 maltol deriva-
tives, and 23 thiomaltol derivatives (22). Although it has been
reported that hydroxamic acids have potentially limiting pharmacoki-
netic properties possibly due to the hydrolysis of the hydroxamic
acid to the corresponding shorter and less-active carboxylic acid,
simple strategies can be employed to eliminate the metabolic liabil-
ity of hydroxamate-containing inhibitors by modifying the substitu-
ents around the hydroxamate moiety (15,24–27). As a consequence,
there has been a successful HDAC inhibitor, known as Vorinostat
(Zolinzaꢀ, formerly known as SAHA), that has been recently
approved by the FDA for the treatment of the cutaneous T-cell lym-
phoma (28), while other hydroxamate inhibitors for other metallopro-
teins are at various stages in clinical development (14,17,29).
Nonetheless, we believe it is advantageous to explore additional
ZBGs as illustrated in Figure 2.
In order to evaluate the effectiveness of our approach, we tested
our library against a variety of Zn(II)-metalloproteins, including
anthrax LF, botulinum neurotoxin A (BoNT ⁄A), gelatinase-2 (MMP-9),
stromelysin (MMP-3), histone deacetylases 1 and 8 (HDAC 1,
A
B
Figure 2: Depiction of the metal-chelating groups chosen for
the synthesis of the library.
212
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
HDAC8), and matrix metalloproteinases MT1-MMP and MT6-MMP.
Enzymatic assays identified 'hits' for all targets, including unique
fragment hits for the majority of them. In addition, the metallopro-
teins showed different distributions of hits for each ZBG, indicating
that the choice of ZBG contributes substantially to the binding spec-
ificity. Some fragments proved to be scaffold precursors for known
clinical candidates as metalloproteinase inhibitors, while others
could be easily optimized into potent and selective compounds. We
further proved that the approach could also be applicable to other
metal ions such as the Mn2+ present in the active site of the bac-
terial essential enzyme 1-deoxy-D-xylulose-5-phosphate reducto-
isomerase (DXR; IspC) chosen as an example (30). Hence, the
findings reported here show that our approach and the derived me-
talloprotein-focused library for FBLD will be a significant source of
lead compounds for the development of new inhibitors for a large
variety of drug targets.
N-hydroxyformanilides were obtained using commercially available
nitroso compounds and converting them to the reverse hydroxamate
using glyoxylic acid as described elsewhere (37). Similar to the pro-
cedure of the hydroxamic acids, the sulfonylhydroxamates were
obtained starting from commercially available sulfonyl chloride in
the presence of hydroxylamine hydrochloride and sodium bicarbon-
ate (38). However, several of the sulfonylhydroxamates are easily
hydrolyzed and are not stable in DMSO stock solutions, but stable
in powder form. Full experimental details on the synthesis of com-
pounds 16–21 were carried out as described by Kim et al. (39).
Compounds 10–12 were synthesized as described by Levin et al.
(40). Characterization of the compounds is reported in the Appendix
S1, Figure S1. Molecular docking studies were performed as
described by Johnson et al. (41).
Metalloenzyme assays
Methods
MAPKKideꢁ assay for LF
The fluorescence peptide cleavage assay was performed in a 96-
well plate (100 lL ⁄ well) in which each reaction mixture contained
MAPKKideꢀ (4 lM) and LF (50 nM) (List Biological Laboratories,
Campbell, CA, USA) in 20 mM HEPES, pH 7.4, and the screening
compounds (mixture of 20; each compound at 10 lM). Kinetics of
the peptide cleavage was examined for 30 min using a fluorescent
plate reader (Victor V; Perkin Elmer, Waltham, MA, USA) at excita-
tion and emission wavelengths of 485 and 535 nm, respectively,
and IC50 values were obtained by dose–response measurements.
For selected compounds, Lineweaver–Burk analysis was also carried
out to verify that the compounds are competitive against the sub-
strate. The Km and Vmax values of the MAPKKideꢀ cleavage by LF
were determined at 25 ꢂC using the same experimental condition
described earlier for the fluorescence screening assay but with
increasing MAPKKideꢀ concentrations (10, 5, 2.5 lM). The Ki and
Km(app) were calculated at 5 and ⁄ or 10 lM inhibitor concentration.
Preparation of fragment library
Starting materials and solvents were purchased and used without
further purification from commercial suppliers (Sigma-Aldrich,
St. Louis, MO, USA; Alfa Aesar, Ward Hill, MA, USA; Fisher Scienti-
fic, Pittsburgh, PA, USA; Maybridge, Pittsburgh, PA, USA; Acros, Geel,
Belgium; ICI, Portland, OR, USA; Cambridge Isotope Laboratories Inc.,
Andover, MA, USA and others). Compound 5 was purchased from
AKOS Consulting and Solutions GmbH (Steinen, Germany).
The library was constrained to be in line with the general 'rule of
three' (31) for fragments: (i) the MW of all fragments were gener-
ally <300 Da; (ii) excluding the ZBG, the number of H-bond donors
and H-bond acceptors was <3; and (iii) high aqueous solubility was
sought for ease of screening. As such, the library consisted of
small-molecule derivatives containing one of several basic ZBGs: hy-
droxamic acids, hydroxysulfonic acids, reverse hydroxamates, 3,4-hy-
droxypyridinones, and 3,4-hydroxypyridinethiones (Figure 2). These
groups were selected because of their established ability to bind
metal ions and prior identification as effective ZBGs for use in me-
talloproteinase inhibitors (20–22).
MMP-9 assay
This assay was performed as outlined in the Anaspec (Fremont,
CA, USA) MMP Assay kit (Cat. No. 71151 ⁄ 71155). The fluores-
cence peptide cleavage assay was performed in a 96-well plate
(50 mL ⁄ well) in which each reaction mixture contained 5-
FAM⁄ QXLTM520 (60 lL; diluted 1:100 in assay buffer) and MMP-
9 (10 mg ⁄ mL; pro-MMP-9 was first activated with 1 mM APMA
for 2 h) in EnzolyteTM 520 MMP-9 assay buffer and the screen-
ing compounds at 20 lM. Kinetics of the peptide cleavage was
examined every 5 min for 30 min using a fluorescent plate
reader (Victor V; Perkin Elmer) at excitation and emission wave-
lengths of 485 and 535 nm, respectively, and % inhibition values
were obtained. Similarly, MMP-3 assay was performed as
described previously (22).
The syntheses of both hydroxypyridinones and hydroxypyridinethiones
have been well established by several reported procedures (22).
Similarly, the hydroxamates, reverse hydroxamates, and sul-
fonylhydroxamates were all synthesized according to previous litera-
ture reports (32–38). In the synthesis of the hydroxamic acids,
different protocols were utilized, including conversion of the acid to
the acyl chloride using oxalyl chloride or thionyl chloride (32,33), or
coupling the acid to the hydroxamic acid via ethyl-(N¢,N¢-dimethyla-
mino)propylcarbodiimide-hydrochloride (EDC) coupling reagent along
with cyanuric chloride (34), or using derivatized esters that were
converted to the hydroxamic acid using 50% hydroxylamine in dime-
thylacetamide ⁄ methanol (35). Two types of reverse hydroxamates
were incorporated into this library, N-arylacetohydroxamates and N-
hydroxyformanilides. The N-acetoarylhydroxamates were synthesized
in two steps starting from the nitro compound to the hydroxylamine
using Raney nickel followed by acetylation with acetyl chloride (36).
SNAPtideꢀ assay for BoNT ⁄ A
The fluorescence peptide cleavage assay (50 lL) was performed in
a 96-well plate in which each reaction mixture contained SNAP-
tideꢀ (8 lM) and botulinum neurotoxin type A (5 nM) (List Biological
Laboratories) in 20 mM HEPES, 0.1% Tween 20, pH 8.2, and the
Chem Biol Drug Des 2011; 78: 211–223
213

Johnson et al.
screening compounds. Kinetics of the peptide cleavage was exam-
ined for 30 min using a fluorescence plate reader (VictorTM V; Per-
1 h at 37 ꢂC with MMPs (0.02 lM each) in 50 mM HEPES, 10 mM
CaCl2, and 1 lM ZnCl2, pH 6.8. The digests were analyzed using
SDS–PAGE. Where indicated, GM6001 (0.5 lM) and the novel com-
pounds (at 5 and 25 mM each) were added to the reactions.
2
kin Elmer) at excitation and emission wavelengths of 485 and
535 nm, respectively, and IC50 values were obtained by dose–
response measurements. The Km and Vmax values of the SNAPtideꢀ
cleavage by BoTN ⁄ A were determined at 25 ꢂC using the same
experimental condition described previously for the fluorescence
screening assay, but with increasing SNAPTideꢀ concentrations
(100, 60, 30, 10, 1 lM).
Cell-based assays
MEK-1 cleavage
The N-terminal MEK1 antibody (anti-MEK1, N-terminal) was
obtained from Calbiochem (Rockland, MA, USA) and the b-actin
antibody (monoclonal-anti-b-actin) was purchased from Sigma
(St. Louis, MO, USA). Anti-rabbit and anti-mouse horseradish perox-
idases were purchased from GE Healthcare (Piscataway, NJ, USA).
HeLa (human cervical carcinoma) cells were obtained from the
American Type Culture Collection (Manassas, VA, USA) and were
cultured in Dulbecco's modified Eagle's medium (DMEM, Cellgro)
supplemented with 10% fetal bovine serum (FBS; Omega Scientific,
Tarzana, CA, USA) and 1% penicillin ⁄ streptomycin (Omega Scien-
tific) at 37 ꢂC in a humidified incubator containing 5% CO2. Tripli-
cate samples of approximately 5.0 · 104 HeLa cells were plated in
individual wells of a 48-well plate and incubated for 48 h. The cells
were exposed for 90 min at 37 ꢂC to a preincubated solution of
2.5 · 10)8 M PA83, 5 · 10)9 M LF, and test compounds in serum-
free DMEM medium. The analyzed inhibitors were dissolved in
DMSO reaching a final DMSO concentration of 1%. Controls
included untreated cells and LF ⁄ PA-only-treated cells. Cells were
later lysed and directly collected with reducing gel sample buffer
and analyzed for MEK1 cleavage by immunoblotting as previously
described (44).
HDAC1 and HDAC8 assays
HDAC1 and HDAC8 inhibitors were identified using the Enzo Life
Sciences' (Farmingdale, NY, USA) HDAC1 and HDAC8 Fluorimetric
Drug Discovery kit (Cat. No. BML-AK511 ⁄ AK518). The assay was
carried out in two steps in a 96-well plate. First, a tetrapeptide
substrate containing an acetylated lysine side chain, Fluor de LysTM
SIRT-1 (5 mM) for HDAC1 and Fluor de LysTM HDAC8 (25 mM) for
HDAC8, was incubated with the enzyme for 45 min. (0.1 mg
HDAC1, 37 ꢂC, and 0.5 U HDAC8, 30 ꢂC). Second, the deacetylated
substrate was allowed to react with the Fluor de LysTM Developer
II in order to produce a fluorophore. The fluorescence is measured
using a fluorescent plate reader (Victor V; Perkin Elmer) at excita-
tion and emission wavelengths of 355 and 460 nm, respectively. Ini-
tially, the test samples were screened at 50 lM, and IC50 values
were obtained by dose–response measurements.
DXR assay
Enzymatic activity was measured at 37 ꢂC in a 200-lL assay mix-
ture: MnCl2 0.5 mM, NADPH 200 mM, Tris pH 8.5 100 mM, deoxy-
xylulose 5-phosphate 250 mM, BSA 10 mg ⁄ mL. Concentration of
substrate was chosen at Km value (42). Specific activity of enzyme
was approximately 1 mmol ⁄ mg ⁄ min. Inhibition was measured in
comparison with the activity of the enzyme in the presence of 1%
DMSO. The oxidation of NADPH was monitored at 340 nm using a
Beckman DTX-880 multimode microplate reader. Consumption of
substrate for activity measurements was <10%. Fosmidomycin, a
known DXR inhibitor, used as a positive control showed the value
of IC50 expected from published IC50 8.2 nm (43).
Macrophage protection assay
RAW 264.7 Murine monocyte macrophages (4.5 · 104 cells ⁄ well)
were plated into 96-well tissue culture plates in Hyclone DMEM
(4500 mg ⁄ L glucose, 110 g ⁄ L sodium pyruvate) supplemented with
5% FBS, 2 mM Glutamax (Invitrogen, Carlsbad, CA, USA), and 1%
penicillin ⁄ streptomycin (Omega Scientific). Cells were cultured over-
night at 37 ꢂC in a humidified incubator containing 5% CO2. They
were replenished with fresh serum-free medium (0.1 mL ⁄ well) and
exposed to a preincubated solution of test compounds (increasing
concentrations from 0.015 to 33.3 mM), protective antigen (PA)83
(500 ng ⁄ mL), and LF (37.5 ng ⁄ mL). The analyzed inhibitors were dis-
solved in DMSO reaching a final DMSO concentration of 1%. Con-
trols included untreated cells, LF ⁄ PA-only-treated cells and cells
treated with the known LF inhibitor 15. After incubation for 3.5 h
at 37 ꢂC, cell viability was assessed using ATPlite assay from Per-
kin Elmer. Each data point represents triplicates of each concentra-
tion in one experiment. Viability was normalized to control cells
that were treated with the vehicle, DMSO.
DXR NMR screen
NMR spectra were acquired on a 700 MHz Bruker Avance spec-
trometer equipped with a TCI-cryoprobe. All 1D 1H experiments
were carried out with 100 lM of testing compounds in Tris buffer
(20 mM Tris buffer pH 7.5, 20 mM Mn2+) in the presence or the
absence of 10 lM DXR protein. 1D T1q experiments were measured
with a variable spin-lock time of 10, 200, or 1000 ms, a recycle
delay of 1.5 s, and water suppression based on the watergate
sequence. Data were analyzed with aid of 1D proton spectrum of
the individual, and hits were further confirmed by testing the indi-
vidual compound.
Protein expression of DXR
Escherichia coli strain JW0168 from the ASKA collection (45) was
used for recombinant DXR protein overexpression. Preculture was
grown in Luria Broth (LB) overnight with chloramphenicol 50 lg ⁄ mL,
and 40 lL of the preculture was used to inoculate 50 mL LB in a
250-mL flask. Induction was achieved overnight at an OD600 = 0.8.
Alpha-1-antitrypsin SDS gel assay
Inhibitors block the cleavage of Alpha-1-antitrypsin (AAT) by MT1-
MMP and MT6-MMP. A1-antitrypsin (2 lM) was coincubated for
214
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
Chem Biol Drug Des 2011; 78: 211–223
215

Johnson et al.
216
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
Chem Biol Drug Des 2011; 78: 211–223
217

Johnson et al.
After shaking for 4 h at 37 ꢂC, cells were collected, lysed, and puri-
fied using Ni-NTA column (46). Protein was mixed with 1 mM EDTA
and buffer changed by Bio-Spin chromatography column. Protein
SDS-polyacrylamide gel is shown in Figure 1, and the concentration
of protein after buffer exchange was 2.8 mg ⁄ mL. Concentration of
protein was measured by Bradford assay.
inhibition assay, and hits were defined as those that exhibited
>50% inhibition at 100 lM; IC50 values were subsequently obtained
for the most potent compounds (Table 1) with the exception of DXR
hits, which were initially found via NMR screening using T1q NMR
experiments and further validated using a colorimetric assay. Over-
all, the screening of our fragment library gave 14 hits against LF
(4% hit rate), 19 hits against BoNT ⁄A (4.8% hit rate), nine hits
against MMP2 (2% hit rate), 16 hits against MMP3 (4% hit rate),
81 hits against HDAC1 (17% hit rate), 189 hits against HDAC8
(34% hit rate), and 47 hits (11%) against DXR (see below for our
criteria for definition of hits against this target).
Results and Discussion
In order to evaluate the potential of the designed library for the
identification of viable fragment hits, we profiled it against a vari-
ety of Zn-metalloproteins including two bacterial toxins (LF and
BoNT ⁄A), human metalloproteinases (MMP-3, MMP-9, MT1-MMP,
and MT6-MMP), two hydrolases (HDAC1 and HDAC8) as well as a
Mn-metalloprotein, DXR. Each compound was tested in an enzyme
Several striking observations can be made: first, compounds show
rather marked inhibition constants (in the low micromolar range)
and very good LE; second, both cross-reactive and rather selective
fragments could be identified; third, fragment selectivity is a
A
C
D
B
E
Figure 3: Application of the Fragment-based lead design (FBLD) approach to the discovery and optimization of lethal factor (LF) inhibitors.
Inhibition constants and LE, expressed in kcal ⁄ mol, are reported. (A) Chemical structures of fragment 'hits' 4 and 5 against LF; the panel also
reports the chemical structures of LF inhibitors 13, 14, and 15 (23,53). (B) Chemical structures of the merged compounds 18 and 19 and of
the fragments from which these were derived. (C). Dose–response curve for the inhibition of LF by compound 18. (D). Dose–response protec-
tive effect of compound 18 (EC50 = 7.2 lM) from LF ⁄ PA-induced cell death of macrophage RAW cells. The inhibition of LF-induced cleavage
of MEK1 in HeLa cells is also reported. Known LF inhibitor 15 and the FBLD-derived compound 18 are equally effective in inhibiting LF-
induced MEK1 cleavage in HeLa cells. (E) Molecular model of compound 18 docked into the catalytic pocket of LF (PDB ID 1YQY). The protein
surface was generated with MOLCAD and color coded according to cavity depth (brown, deep; blue, shallow) with the Zn2+ ion shown in
magenta.
218
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
function of both the ZBG and the scaffold that carries it. Hence,
unlike in typical FBLD approaches, the library produces hits that are
nearly comparable in activity and selectivity with what one could
obtain in the successful HTS campaigns against these targets
(41,47,48). Moreover, the LE are significantly higher than both typi-
cal FBLD and HTS hits (8). Furthermore, as different metal-chelating
groups have varying affinities for different targets, the fragments
are expected to have an intrinsically higher selectivity than typical
FBLD hits (49–51). At the same time, we expect that our approach
would lead to fragments that could be rapidly optimized into poten-
tial lead candidates, either by fragment maturation (Figure 1D) or
by fragment merging (Figure 1E). While these methods work best
when supported by the experimentally derived structure of the bin-
ary complexes [either by X-ray (6) or by NMR spectroscopy (7)],
recently Chen and Shoichet (52) reported that virtual docking could
also effectively supplement these endeavors. Likewise, in our
approach, these tasks could be enormously supported by structure-
based computational modeling, given that positioning the hit frag-
ments in the binding site would be facilitated by the geometry of
the metal coordination.
Screening the fragment library against the other metalloprotease
targets also resulted in significant hits (50% inhibition at 50 lM),
which could ultimately be optimized. In particular, screening the
library against BoNT ⁄A gave 19 hits that were predominately the
thiomaltol derivatives, with the exception of three compounds (one
hydroxamate, compound 9, Table 1; two maltol derivatives,
IC50 > 10 lM), the top hits were compounds 11–13 (Table 1). This
same trend was also observed for MMP-9 in which we identified
22 total hits comprising 17 thiomaltol derivatives, two maltol deriv-
atives (IC50 30–40 lM), and three hydroxamate derivatives (com-
pound 5, Table 1, and two other hydroxamates with
IC50's > 30 lM). Hits identified against MMP-3 were mainly the hy-
droxamate analogs in which compounds 7, 10, and 11 represent
the most potent compounds with IC50 values of 1.0, 3.0, and
4.0 lM, respectively. In the case of HDAC1 and HDAC8, the hits
were mainly the hydroxamate derivatives such as compound 8
(Table 1) and analogs thereof (not shown).
The results from screening the library via T1q NMR against DXR
led to 47 hits by monitoring the broadening or decrease in intensity
of the peaks of the compounds (200 lM) in the presence of a
substoichiometric (5 lM) amount of protein, compared with the
spectra of the ligands without protein. Of the 47 hits, 35 were
For example, in devising inhibitors against anthrax LF, fragments 4
and 5 were found as hits in our kinetic assay screen (Figure 3).
These two fragments 4 and 5 could have been used together
with the X-ray structure of the enzyme for further optimizations
(Figure 3A) (23). The evolution of these fragments would then fol-
low simple medicinal chemistry guided by the structure of the
docked hits as schematically illustrated in Figure 3. Proper evolu-
tion of these fragments would indeed lead to compounds 18 and
19 with IC50 values in the low nanomolar range and with high
selectivity for LF (23,53). In particular, compound 19 (Figure 3A)
has been reported to be a viable clinical candidate against
Anthrax (23).
A
In addition to the fragment maturation strategy, we mentioned that
optimized compounds could be designed by merging two weak
inhibitory fragments (Figure 1E). In the current example, compounds
20 (IC50 = 100 lM) and 3 (IC50 = 47 lM), found as hits in the
kinetic screen against LF (Figure 3B), were merged leading to the
low micromolar compound 22 (Figure 3B,C). Docking studies per-
formed using the X-ray coordinates of LF (PDB ID 1YQY) suggest
that compound 22 fits nicely into the binding pocket of the enzyme,
placing the hydroxamate moiety coordinating the metal ion, extend-
ing into the adjacent S1¢ pocket (Figure 3E). Other fragments with
the same scaffold were in the library, but were not chosen because
of lack of activity in the kinetic assay. To demonstrate this, we
merged compound 3 with an inactive compound (compound 21,
IC50 > 100 lM), which did not result in a molecule with significantly
improved affinity (compound 23, Figure 3B).
B
To further analyze the efficacy of compound 22 in cell, we con-
firmed its ability of protecting macrophages from LF ⁄ PA-induced cell
death (41,54), and in preventing MEK1 cleavage in Hela cells (Fig-
ure 3D) (44). These data further corroborate the underlying principle
of FBLD that the combination of two albeit weak binder fragments
can result in high-affinity ligands. Overall, the Structure-Activity
Relationships (SAR) of this fragment indeed led to potent LF inhibi-
tors, exhibiting selectivity, cell permeability, and efficacy in vivo.
Figure 4: Screening results for compound 24 against DXR. (A)
T1q NMR spectra of compound 24 (100 lM) in buffer without pro-
tein (red) and with protein (10 lM) (black). Peak intensity reductions
are highlighted. (B) Molecular surface of DXR in complex with com-
pound 24. The surface is generated with MOLCAD and color coded
based on cavity depth: blue (shallow), brown (deep). The metal ion
is depicted as a purple sphere. (PDB ID 2JCZ).
Chem Biol Drug Des 2011; 78: 211–223
219

Johnson et al.
Table 2: Table of hits targeting DXR
#
Structure
LE (kcal ⁄ mol)
IC50 (lM)
50 € 20a
24
0.37
25
26
27
28
29
0.26
0.45
0.50
0.40
0.48
70 € 11
2.31 € 0.22
0.716 € 0.048
5.21 € 0.79
1.19 € 0.13
LE, ligand efficiencies.
aComplete inhibition could not be obtained; hence, reported IC50 is an approximate value.
hydroxamates, one sulfonyl hydroxamate, four reverse hydroxa-
mates, five maltol derivatives, and two thiomaltol derivatives. As an
example, compound 24 shown in Figure 4A represents the outcome
of the T1q screen against DXR as described in the material and
methods. These 47 compounds were then tested using a colorimet-
ric enzymatic assay in which six of the 47 compounds were con-
firmed to inhibit DXR in the low lM range (Table 2). Docking
studies were performed on all hits in which compound 24 gives a
representation of how well it fits inside the pocket, showing the
hydroxamate moiety chelating the Mn2+ (Figure 4B).
applied to any other hits reported in Table 1, leading to a rapid
generation of lead candidates for these targets. For example, we
recently reported on the discovery and maturation of hydroxamic
acid-based inhibitors of MT6-MMP using this strategy (55). More-
over, further potent, non-hydroxamic acid–based compounds 15–17
were obtained that effectively inhibit MT6-MMP and MT1-MMP
cleavage of AAT (Figure 5). The maturation of these compounds into
nanomolar inhibition was achieved with very limited SAR guided by
the docked structure of initial hits. Of note is that the docking was
performed with a homology-modeled structure, given that the X-ray
coordinates of MT6-MMP are not yet available (Figure 5). Given
their role in tumor progression and neurodegeneration (55), inhibi-
tors of MT1-MMP and MT6-MMP hold great potential for the
development of novel therapeutic agents.
Hence, with several examples, we demonstrated that this fragment
library can rapidly and efficiently lead to potent inhibitors of metal-
loproteins. In fact, similar fragment maturation strategies could be
220
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
2. Shuker S.B., Hajduk P.J., Meadows R.P., Fesik S.W. (1996) Dis-
covering high-affinity ligands for proteins: SAR by NMR. Sci-
ence;274:1531–1534.
A
3. Carr R.A., Congreve M., Murray C.W., Rees D.C. (2005) Frag-
ment-based lead discovery: leads by design. Drug Discov
Today;10:987–992.
4. Siegal G., Ab E., Schultz J. (2007) Integration of fragment
screening and library design. Drug Discov Today;12:1032–1039.
5. Hajduk P.J., Greer J. (2007) A decade of fragment-based drug
design: strategic advances and lessons learned. Nat Rev Drug
Discov;6:211–219.
6. Jhoti H., Cleasby A., Verdonk M., Williams G. (2007) Fragment-
based screening using X-ray crystallography and NMR spectros-
copy. Curr Opin Chem Biol;11:485–493.
B
7. Pellecchia M., Bertini I., Cowburn D., Dalvit C., Giralt E., Jahnke
W., James T.L., Homans S.W., Kessler H., Luchinat C., Meyer B.,
Oschkinat H., Peng J., Schwalbe H., Siegal G. (2008) Perspec-
tives on NMR in drug discovery: a technique comes of age. Nat
Rev Drug Discov;7:738–745.
8. Bembenek S.D., Tounge B.A., Reynolds C.H. (2009) Ligand effi-
ciency and fragment-based drug discovery. Drug Discov
Today;14:278–283.
9. Hann M.M., Leach A.R., Harper G. (2001) Molecular complexity
and its impact on the probability of finding leads for drug dis-
covery. J Chem Inf Comput Sci;41:856–864.
10. Schuffenhauer A., Ruedisser S., Marzinzik A.L., Jahnke W.,
Blommers M., Selzer P., Jacoby E. (2005) Library design for frag-
ment based screening. Curr Top Med Chem;5:751–762.
11. Mestres J., Veeneman G.H. (2003) Identification of ''latent
hits'' in compound screening collections. J Med Chem;46:
3441–3444.
12. Gribbon P., Sewing A. (2005) High-throughput drug discovery:
what can we expect from HTS? Drug Discov Today;10:17–
22.
13. Kuntz I.D., Chen K., Sharp K.A., Kollman P.A. (1999) The maximal
affinity of ligands. Proc Natl Acad Sci U S A;96:9997–10002.
14. Fattori D., Squarcia A., Bartoli S. (2008) Fragment-based
approach to drug lead discovery: overview and advances in vari-
ous techniques. Drugs R D;9:217–227.
15. Hajduk P.J., Shuker S.B., Nettesheim D.G., Craig R., Augeri D.J.,
Betebenner D., Albert D.H., Guo Y., Meadows R.P., Xu L., Mi-
chaelides M., Davidsen S.K., Fesik S.W. (2002) NMR-based mod-
ification of matrix metalloproteinase inhibitors with improved
bioavailability. J Med Chem;45:5628–5639.
Figure 5: Application of the Fragment-based lead design
approach to the discovery and optimization of MT6-MMP and MT1-
MMP inhibitors. (A) Inhibition of alpha-1-antitrypsin cleavage by
catalytic domains of MT1-MMP and MT6-MMP at 5 and 25 lM. The
non-specific peptide inhibitor GM6001 was also tested (0.5 mM). (B)
Molecular surface of MT6-MMP in complex with compound 12. The
surface is generated with MOLCAD and color coded based on cavity
depth: blue (shallow), brown (deep). The metal ion is depicted as a
purple sphere. The model was obtained using the 3D structure of
MMP-2 (26% identities) and 3D Swiss Protein Modeler (PDB ID 1YQY).
In summary, screening this library composed of hydroxamates,
reverse hydroxamates, sulfonyl hydroxamates, hydroxypyranones, hy-
droxypyridinethiones, and sulfonamide acids resulted in an array of
hits with the exception of the reverse hydroxamates. In conclusion,
we have devised a novel metal-targeted fragment library and dem-
onstrated that its application in FBLD projects leads to novel, selec-
tive, and highly efficient ligands that could be matured into lead
candidates. Finally, because it has been estimated that nearly one-
third of drug targets contain a metal ion (56), we envision that our
approach will find wide utility in targeting a variety of additional
drug targets for which HTS has notoriously failed in producing via-
ble hits such as dehydrogenases, carboxylases, oxidoreductases,
transcription factors, and potentially many others as recently dem-
onstrated by a similar approach (57).
16. Wang J., Uttamchandani M., Li J., Hu M., Yao S.Q. (2006) Rapid
assembly of matrix metalloprotease inhibitors using click chem-
istry. Org Lett;8:3821–3824.
Acknowledgments
We thank the National Institutes of Health, grants AI070494,
AI059572 to MP, and CA083017 to AS and MP, for generous support.
We also thank Dr. Spyroulias and his laboratory at the Department of
Pharmacy, University of Patras, Greece, for the kind gift of the expres-
sion plasmid for the catalytic domain of LF used for the Appendix S1.
17. Flipo M., Beghyn T., Charton J., Leroux V.A., Deprez B.P., Deprez-
Poulain R.F. (2007) A library of novel hydroxamic acids targeting
the metallo-protease family: design, parallel synthesis and
screening. Bioorg Med Chem;15:63–76.
18. Sternson S.M., Wong J.C., Grozinger C.M., Schreiber S.L. (2001)
Synthesis of 7200 small molecules based on a substructural
analysis of the histone deacetylase inhibitors trichostatin and
trapoxin. Org Lett;3:4239–4242.
References
19. Price S., Osbourn S.E. (2005) Solid-phase synthesis of N-form-
ylhydroxylamines (reverse ⁄ retro-hydroxamates). Org Lett;7:3761–
3763.
1. Erlanson D.A., McDowell R.S., O'Brien T. (2004) Fragment-based
drug discovery. J Med Chem;47:3463–3482.
Chem Biol Drug Des 2011; 78: 211–223
221

Johnson et al.
20. Jacobsen F.E., Lewis J.A., Cohen S.M. (2007) The design of
inhibitors for medicinally relevant metalloporteins. ChemMed-
Chem;2:152–171.
21. Jacobsen F.E., Lewis J.A., Cohen S.M. (2006) A new role for old
ligands: discerning chelators for zinc metalloproteinases. J Am
Chem Soc;128:3156–3157.
36. Boge N., Kruger S., Schroder M., Meier C. (2007) A new short
and effcieint synthetic route to C8-N-acetylarylamine 2¢¢-deoxy-
guanosine phosphoramidites. Synthesis;24:3907–3914.
37. Corbett M.D., Corbett B.R. (1980) Reaction of nitroso aromatics
with glyoxylic acid. A new path to hydroxamic acids. J Org
Chem;45:2834–2839.
22. Agrawal A., Johnson S.L., Jacobsen J.A., Miller M.T., Chen L.H.,
Pellecchia M., Cohen S.M. (2010) Chelator fragment libraries for
targeting metalloproteinases. ChemMedChem;5:195–199.
23. Shoop W.L., Xiong Y., Wiltsie J., Woods A., Guo J., Pivnichny
J.V., Felcetto T. et al. (2005) Anthrax lethal factor inhibition. Proc
Natl Acad Sci U S A;102:7958–7963.
24. Bouchain G., Delorme D. (2003) Novel hydroxamate and
anilide derivatives as potent histone deacetylase inhibitors: syn-
thesis and antiproliferative evaluation. Curr Med Chem;10:2359–
2372.
25. Michaelides M.R., Dellaria J.F., Gong J., Holms J.H., Bouska
J.J., Stacey J., Wada C.K. et al. (2001) Biaryl ether retrohydrox-
amates as potent, long-lived, orally bioavailable MMP inhibitors.
Bioorg Med Chem Lett;11:1553–1556.
26. Curtin M., Glaser K. (2003) Histone deacetylase inhibitors: the
abbott experience. Curr Med Chem;10:2373–2392.
27. Pikul S., Ohler N.E., Ciszewski G., Laufersweiler M.C., Almstead
N.G., De B., Natchus M.G., Hsieh L.C., Janusz M.J., Peng S.X.,
Branch T.M., King S.L., Taiwo Y.O., Mieling G.E. (2001) Potent
and selective carboxylic acid-based inhibitors of matrix metallo-
proteinases. J Med Chem;44:2499–2502.
28. Attenni B., Ontoria J.M., Cruz J.C., Rowley M., Schultz-Fademr-
echt C., Steinkuhler C., Jones P. (2009) Histone deacetylase
inhibitors with a primary amide zinc binding group display anti-
tumor activity in xenograft model. Bioorg Med Chem
Lett;19:3081–3084.
38. Scozzafava A., Supuran C.T. (2000) Carbonic anhydrase and
matrix metalloproteinase inhibitors: sulfonylated amino acid
hydroxamates with MMP inhibitory properties act as efficient
inhibitors of CA isozymes I, II, and IV, and N-hydroxysulfonamides
inhibit both these zinc enzymes. J Med Chem;43:3677–3687.
39. Kim M., Kim J.Y., Song K., Kim J., Lee J. (2007) Convenient
total synthesis of taranabant (MK-0364), a novel cannabinoid-1
receptor inverse agonist as an anti-obesity agent. Tetrahe-
dron;63:12845–12852.
40. Levin J.I., Rush T.S.I., Lovering F., Hu Y., Li J., Li W., Wu J.J.,
Hotchandani R., Xiang J.S., Du X., Cole D.C., Tam S.Y. (2005)
Preparation of Amino acid Biarylsulfonamides as Metalloprotein-
ase Inhibitors. New York, USA: Wyeth; p. 119.
41. Johnson S.L., Chen L.H., Harbach R., Sabet M., Savinov A., Cot-
ton N.J., Strongin A., Guiney D., Pellecchia M. (2008) Rhodanine
derivatives as selective protease inhibitors against bacterial tox-
ins. Chem Biol Drug Des;71:131–139.
42. Kuzuyama T., Takahashi S., Takagi M., Seto H. (2000) Character-
ization of 1-deoxy-D-xylulose 5-phosphate reductoisomerase, an
enzyme involved in isopentenyl diphosphate biosynthesis, and
identification of its catalytic amino acid residues. J Biol
Chem;275:19928–19932.
43. Kuzuyama T., Shimizu T., Takahashi S., Seto H. (1998) Fosmido-
mycin, a specific inhibitor of 1-dexoy-D-xylulose 5-phosphate re-
ductoisomerase in teh nonmevaolnate pathway for terpenoid
biosynthesis. Tetrahedron Lett;39:7913–7916.
29. Aranapakam V., Davis J.M., Grosu G.T., Baker J., Ellingboe J.,
Zask A., Levin J.I. et al. (2003) Synthesis and structure-activity
relationship of N-substituted 4-arylsulfonylpiperidine-4-hydroxa-
mic acids as novel, orally active matrix metalloproteinase inhibi-
tors for the treatment of osteoarthritis. J Med Chem;46:2376–
2396.
30. Dhiman R.K., Schaeffer M.L., Bailey A.M., Testa C.A., Scherman
H., Crick D.C. (2005) 1-Deoxy-D-xylulose 5-phosphate reducto-
isomerase (IspC) from Mycobacterium tuberculosis: towards
understanding mycobacterial resistance to fosmidomycin. J Bac-
teriol;187:8395–8402.
44. Rainey G.J., Wigelsworth D.J., Ryan P.L., Scobie H.M., Collier
R.J., Young J.A. (2005) Receptor-specific requirements for
anthrax toxin delivery into cells. Proc Natl Acad Sci
USA;102:13278–13283.
45. Kitagawa M., Ara T., Arifuzzaman M., Ioka-Nakamichi T., Inamot-
o E., Toyonaga H., Mori H. (2005) Complelte set of ORF clones
of Escherichia coli ASOA library (a complete set of E. coli K-12
ORF archive): unique resources for biological reserach. DNA
Res;12:291–299.
46. Yang C., Rodionov D.A., Rodionova I.A., Li X., Osterman A.L.
(2008) Glycerate 2-kinase of Thermotoga maritima and genomic
31. Congreve M., Carr R., Murray C., Jhoti H. (2003) A 'rule of three'
for fragment-based lead discovery? Drug Discov Today;8:876–
877.
32. Booher R.N., Kornfeld E.C., Smalstig E.B., Clemens J.A. (1987)
Synthesis of D-oxa tricyclic partial ergolines as dopamine agon-
ists. J Med Chem;30:580–583.
33. Summers J.B., Mazdiyasni H., Holms J.H., Ratajczyk J.D., Dyer
R.D., Carter G.W. (1987) Hydroxamic acid inhibitors of 5-lipoxy-
genase. J Med Chem;30:574–580.
reconstruction of related metabolic pathways.
riol;190:1173–1182.
47. Johnson S.L., Chen L.H., Pellecchia M. (2007) A high-throughput
screening approach to anthrax lethal factor inhibition. Bioorg
Chem;35:306–312.
48. Schepetkin I.A., Khlebnikov A.I., Kirpotina L.N., Quinn M.T.
(2006) Novel small molecule inhibitors of anthrax lethal factor
identified by high-througput screening. J Med Chem;49:5232–
5244.
J Bacte-
34. Giacomelli G., Porcheddu A., Salaris M. (2003) Simple one-flask
method for the preparation of hydroxamic acids. Org
Lett;5:2715–2717.
35. Miller T.A., Witter D.J., Belvedere S. (2005) Thiophene and Ben-
zothiophene Hydroxamic Acid Derivatives, A61K ed, Boston, MA:
p 1–186.
49. Yan Y.L., Miller M.T., Cao Y., Cohen S.M. (2009) Synthesis of hy-
droxypyrone- and hydroxythiopyrone-based matrix metalloprotein-
ase inhibitors: developing a structure-activity relationship. Bioorg
Med Chem Lett;19:1970–1976.
50. Agrawal A., de Oliveira C.A., Cheng Y., Jacobsen J.A., McCam-
mon J.A., Cohen S.M. (2009) Thioamide hydroxypyrothiones
222
Chem Biol Drug Des 2011; 78: 211–223

Targeting Metalloproteins by FBLD
supersede amide hydroxypyrothiones in potency against anthrax
lethal factor. J Med Chem;52:1063–1074.
51. Agrawal A., Romero-Perez D., Jacobsen J.A., Villarreal F.J., Co-
hen S.M. (2008) Zinc-binding groups modulate selective inhibi-
tion of MMPs. ChemMedChem;3:812–820.
52. Chen Y., Shoichet B.K. (2009) Molecular docking and ligand
specificity in fragment-based inhibitor discovery. Nat Chem
Biol;5:358–364.
a M. (2006) Structure-activity relationships by interligand NOE-
based design and synthesis of antiapoptotic compounds target-
ing Bid. Proc Natl Acad Sci U S A;103:12602–12606.
59. Siliphaivanh P., Harrington P., Witter D.J., Otte K., Tempest P.,
Kattar S., Kral A.M., Fleming J.C., Deshmukh S.V., Harsch A.,
Secrist P.J., Miller T.A. (2007) Design of novel histone deacety-
lase inhibitors. Bioorg Med Chem Lett;17:4619–4624.
Supporting Information
53. Johnson A.T., Jiao G.S., Kim S. (2008) Hydroxamic Acid Derivatives
of Phenoxy-Acetic Acids an Analogs Useful as Therapuetic Agents
for Treating Anthrax Poisoning. Alea, HI, USA: PanThera Biophar-
ma, LLC.
Additional Supporting Information may be found in the online ver-
sion of this article:
54. Johnson S.L., Chen L.H., Barile E., Emdadi A., Sabet M., Yuan
H., Wei J., Guiney D., Pellecchia M. (2009) Structure-activity
relationship studies of a novel series of anthrax lethal factor
inhibitors. Bioorg Med Chem;17:3352–3368.
55. Radichev I.A., Remacle A.G., Shiryaev S.A., Purves A.N., Johnson
S.L., Pellecchia M., Strongin A.Y. (2010) Biochemical character-
ization of the cellular glycosylphosphatidylinositol-linked mem-
brane type-6 matrix metalloproteinase. J Biol Chem;285:16076–
16086.
Appendix S1. NMR methods and Expression procedure of the
core protease domain of anthrax lethal factor (LF672–776).
Figure S1. [1H–15N] Heteronuclear Single Quantum Coherence
(HSQC) overlaid spectra of the representative residues of 100 lM
15N labeled apoLF672–776 alone (red), in presence of 200 lM ZnCl2
(blue) and in presence of 200 lM ZnCl2 and 200 lM of a generic LF
scaffold from the synthesized library (green).
56. Butler A. (1998) Acquisition and utilization of transition metal
ions by marine organisms. Science;281:207–210.
57. Jacobsen J.A., Fullagar J.L., Miller M.T., Cohen S.M. (2011)
Identifying chelators for metalloprotein inhibitors using a frag-
ment-based approach. J Med Chem;54:591–602.
Please note: Wiley-Blackwell is not responsible for the content or
functionality of any supporting materials supplied by the authors.
Any queries (other than missing material) should be directed to the
corresponding author for the article.
58. Becattini B., Culmsee C., Leone M., Zhai D., Zhang X., Crowell
K.J., Rega M.F., Landshamer S., Reed J.C., Plesnila N., Pellecchi-
Chem Biol Drug Des 2011; 78: 211–223
223