Inhibitor Fingerprinting of Metalloproteases
A R T I C L E S
microarray scanner installed with the relevant filters Cy3 - λex/em
)
proteins against immobilized chemical libraries en masse. A
global analysis directly performed with the profiles provided
valuable insights into protein characteristics, revealing activity-
dependent protein fingerprints. For the present study, we selected
different members of the metalloprotease family as our proteins
of interest, as enzymes are among the most difficult (due to
their delicate nature) yet valuable (due to their widespread
involvement in biological processes) classes of proteins to study
in a microarray format.4 Metalloproteases, in particular, are a
group of broad and diverse enzymes critically involved in the
progression of a variety of diseases and bear important roles in
metabolism as well as intra- and extracellular physiology.5
Modulating these proteins using small molecule therapeutics
provides a potential handle for the effective management and
treatment of diseases such as cancer and arthritis as well as
combating infections from pathogens like botulinum and
anthrax.6 This report serves, to our knowledge, as the first large-
scale quantitative application of protein fingerprinting on a small
molecule array that enables the functional discrimination among
closely related protein members, fueling opportunities in drug
design and discovery.
548/595 nm, Cy5 - λex/em ) 633/685 nm. Optimization of procedures
was first performed using thermolysin as a model before general
adoption with the other enzymes.
KD Analysis on SMMs. Using dose-dependent protein application
on the arrays, we were able to extract KD values against all of the
positively binding small molecules simultaneously. A threshold was
set to restrict analysis to only positive binders. A protein concentration
series diluted across a 2-fold concentration range (from 4.8 µM to 150
nM for Thermolysin and from 2.4 µM to 75 nM for Anthrax LF) was
applied on the microarrays in duplicate. The Cy3 channel was selected
to designate the active channel for this experiment. The microarray
data were extracted using the ArrayWoRx software and fitted using
the following relationship, under the assumption that equilibrium is
achieved during the incubation period:
Observed fluorescence of x )
[Maximum fluorescence, x] × [Protein concentration]
KD + Protein concentration
Saturation dynamics observed when plotting ObserVed fluorescence
against the applied Protein concentration were then fitted to the above
equation using the Graphpad Prism software v.4.03 (GraphPad, San
Diego, USA) revealing the binding dissociation constant, KD, of the
various positively identified small molecules. Only fits that correlated
accurately with the experimental results (r > 0.9) were accepted.
Materials and Methods
Materials. All chemicals were purchased at the highest grade
available from commercial vendors and used without further purifica-
tion, unless otherwise noted. All reactions were carried out under an
N2 atmosphere with HPLC grade solvents, unless otherwise stated. ESI
mass spectra were acquired in both the positive and negative mode
using a Finnigan/Mat TSQ7000 spectrometer. Analytical RP-HPLC
separations were performed on a Phenomex C18 column (150 mm ×
3.0 mm), using a Shimadzu Prominence HPLC system equipped with
a Shimadzu SPD-20A detector. Eluents A (0.1% TFA/acetonitrile) and
B (0.1% TFA/water) were used as the mobile phases. Active enzymes
were acquired commercially, specifically Anthrax LF and Thermolysin
from Calbiochem (Merck, Germany) and Collagenase and Carbox-
ypeptidase from Sigma-Aldrich (USA). Fluorogenic substrates were
purchased from List Biological Laboratories (USA), Calbiochem
(Merck, Germany), and Molecular Probes, Invitrogen (California).
High-Throughput Screening on SMMs. Protein samples were
minimally labeled with Cy3 and Cy5 N-hydroxysuccinimide esters
(Amersham, G.E. Healthcare, USA) for 30 min on ice. The unreacted
esters were quenched with a 10-fold molar excess of hydroxylamine
or using 5 mM Tris (pH 8.0), for a further 15 min. Excess dye was
further removed by size exclusion on a Sephadex G-25 column
(Amersham, G.E. Healthcare, USA). The labeled proteins were
confirmed to retain their native enzymatic activity by positive cleavage
of their respective fluorogenic substrates. The designated inactive
channel protein was first heat treated at 95 °C for 10 min and
immediately cooled on ice. Enzymatic activity was confirmed to be
lost by this procedure using activity assays with respective fluorogenic
substrates. The heat-inactivated enzyme was then mixed with protein
from the active channel and reconstituted in a final buffer volume of
100 µL PBS (pH 7.4) containing 1% bovine serum albumin (BSA). A
total of approximately 8 µg of protein (4 µg in each channel) was
applied under a coverslip to the arrays in this manner, and reciprocal
experiments (with the designated dye channels inverted) were performed
in parallel. The samples were incubated with the arrays in a humidified
chamber for 2-4 h at room temperature, before repeated rinses with
water, typically 2 × 2 min washes with gentle shaking. Excessive
washing beyond 10 min was found to reduce correlations with
microplate screening results. Slides were scanned using an ArrayWoRx
IC50 Measurements. Concentration-dependent measurements were
performed to confirm the potency of representative inhibitors within
the library set. Inhibitors exhibiting a range of potencies for selected
metalloproteases were identified from the microarray screens and
evaluated using IC50 measurements. Briefly, dose-dependent reactions
were performed by varying the concentrations of the inhibitor, under
the same enzyme concentration. A 2-fold dilution series from ap-
proximately 25 µM to 390 nM (final reaction concentration) was
prepared for each inhibitor in black 384-well plates. Substrates and
enzymes for Anthrax LF were applied according to the following
conditions: 0.2 pmol of the enzyme was combined with 2 nM
MAPKKide substrate (List Biological Laboratories, USA) and inhibitor
before being read at λex/em ) 490/520 nm, with a cutoff at 515 nM.
The 50 µL reaction was buffered in 20 mM Hepes (pH 7.4)
supplemented with 0.1 mg/mL BSA and 0.01% Tween 20. Bodipy-
FL-Casein (Molecular Probes, Invitrogen, USA) was used as the
substrate for thermolysin and collagenase. Each 50 µL reaction was
buffered in PBS (pH 7.4), with 1 µg of enzyme and 0.5 µg of substrate
used per assay. The plates were allowed to incubate for 1 h at 37 °C
before being interrogated for end-point fluorescence. The IC50 values
were calculated by curve fitting against the concentration-dependent
fluorescent plots using the Graphpad Prism software v.4.03 (GraphPad,
San Diego, USA).
In Silico Docking Experiments. Molecular modeling was performed
on an SGI IRIX 6.5 workstation using the SYBYL suite (version 7.2)
installed with the FlexX docking software. Protein coordinates were
retrieved from the Protein Data Bank; specifically inhibitor complexed
crystal structures with the following accessions were employed:
Anthrax LF: 1YQY. Structures of representative inhibitors were drawn
using the “Sketch Molecule” option, and hydrogens were added. The
biotin linker of the selected inhibitor was excluded to simplify the
docking simulations. The structure was minimized using 100 iterations
at 0.05 kcal/mol Å to relieve any torsional strain, and formal charges
were assigned. The original protein structures were modified through
the removal of water molecules. The docking sphere was set at 10 Å,
centered at the zinc residue in the enzyme active site. Applying these
criteria, the docking was performed for 30 iterations, with the most
optimized configurations displayed. Proteins were displayed as either
MOLCAD Connolly surfaces or ribbon diagrams.
(5) Turk, B. Nat. ReV. Drug DiscoVery 2006, 5, 785-799.
(6) (a) Tyndall, J. D. A.; Nall, T.; Fairlie, D. P. Chem. ReV. 2005, 105, 973-
999. (b) Overall, C. M.; Kleifeld, O. Nat. ReV. Cancer 2006, 6, 227-239.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13111