A basis for reduced chemical library inhibition of firefly luciferase obtained from directed evolution

Article abstract of DOI:10.1021/jm8014525

We measured the "draggability" of the ATP-dependent luciferase derived from the firefly Photuris pennsylvanica that was optimized using directed evolution (Ultra-Glo, Promega). Quantitative high-throughput screening (qHTS) was used to determine IC_50<

Full text of DOI:10.1021/jm8014525

1450
J. Med. Chem. 2009, 52, 1450–1458
A Basis for Reduced Chemical Library Inhibition of Firefly Luciferase Obtained from Directed
Evolution
Douglas S. Auld,* Ya-Qin Zhang, Noel T. Southall, Ganesha Rai, Marc Landsman, Jennifer MacLure, Daniel Langevin,
Craig J. Thomas, Christopher P. Austin, and James Inglese
NIH Chemical Genomics Center, National Institutes of Health, Bethesda, Maryland 20892-3370
ReceiVed NoVember 17, 2008
We measured the “druggability” of the ATP-dependent luciferase derived from the firefly Photuris
pennsylVanica that was optimized using directed evolution (Ultra-Glo, Promega). Quantitative high-throughput
screening (qHTS) was used to determine IC50s of 198899 samples against a formulation of Ultra-Glo luciferase
(Kinase-Glo). We found that only 0.1% of the Kinase-Glo inhibitors showed an IC50 < 10 µM compared
to 0.9% found from a previous qHTS against the firefly luciferase from Photinus pyralis (lucPpy). Further,
the maximum affinity identified in the lucPpy qHTS was 50 nM, while for Kinase-Glo this value increased
to 600 nM. Compounds with interactions stretching outside the luciferin binding pocket were largely lost
with Ultra-Glo luciferase. Therefore, Ultra-Glo luciferase will show less compound interference when used
as an ATP sensor compared to lucPpy. This study demonstrates the power of large-scale quantitative analysis
of structure-activity relationships (>100K compounds) in addressing important questions such as a target’s
druggability.
Firefly luciferase has been used to measure ATP concentra-
tion, the turnover of pro-luciferin substrates,1-3 and as a reporter
of cell-based gene expression; this has enabled the development
of high-throughput screens (HTS) for a wide variety of
biological activities. Numerous assays have been developed for
protein kinases,4,5 including measurement of phosphorylated
peptide product using solid-phase supports as in SPAa6 or
IMAP7 as well as recent generic assays for ATPases that employ
an ADP-specific antibody.8-10 Luciferase-coupled assays do not
require antibodies or fluorescent labels, thus providing a generic
platform for screening that reduces assay development time and
reagent costs.11-13 Therefore, luciferase-based assays remain
one of the most ubiquitous technologies for the measurement
of ATP concentrations in enzyme assays, particularly kinasessone
of the largest, druggable enzyme families that includes sugar,
lipid, and protein kinases.11,12,14 In these assays, the ATP-
dependence of firefly luciferase is used to measure the ATP
concentration where the luminescence signal is inversely
proportional to kinase activity.
A consideration in any assay that uses an enzyme as a reporter
or coupled component is inadvertent inhibition of the reporter
enzyme by compounds from the library.5,15-17 Ideally, enzyme
reporters would be insensitive to compound inhibition but
inhibition of firefly luciferase is commonly observed; both the
ATP and luciferin binding sites can be bound by many scaffolds
commonly found in compound libraries (e.g,. benzimidazoles
and benzthiazoles). The ATP-dependent luciferase enzyme from
the firefly Photinus pyralis (lucPpy) is employed in several
commercial formulations of firefly luciferase used for ATP
detection. In a recent profiling screen of approximately 72000
compounds (see PubChem AID: 411) against lucPpy contained
in a formulation termed PK-Light (Lonza Corp), we found
approximately 3% of this library showed inhibitory concentra-
tion-responses, of which 681 (0.9%) exhibited IC50s < 10 µM,
well within typical compound screening concentration ranges.15
Investigation of the structure-activity relationships (SAR) of
luciferase inhibitors as well as their mode of action revealed
classes of luciferin and adenylate competitive compounds as
well as large classes of compounds that were noncompetitive
with either ATP or luciferin. However, examination of repre-
sentative compounds from the various inhibitor classes identified
in the lucPpy qHTS showed that nearly all these were inactive
in a luciferase formulation termed Kinase-Glo (Promega Corp).
Luciferase assay formulations such as PK-Light and Kinase-
Glo all contain high amounts of luciferin (∼mM) but lack ATP
to allow for sensitive detection of ATP concentration. However,
Kinase-Glo differs from other luciferase-based ATP detection
formulations in that it contains an optimized luciferase derived
from the firefly Photuris pennsylVanica (lucPpe)18 known as
Ultra-Glo (Promega Corp).19 Wild-type lucPpe and lucPpy show
68% similarity.18 Therefore, one possibility to explain the
reduced potency of lucPpy inhibitors in Kinase-Glo is that the
Ultra-Glo luciferase simply possesses a different SAR and
exhaustive profiling of Kinase-Glo would yield a similar number
of inhibitors as obtained for the lucPpy formulation. Indeed, in
our previous study, we observed that small changes in structure
(e.g., the addition of a methyl group) showed a marginal increase
in Kinase-Glo inhibition (IC50 ∼ 10 µM), supporting the idea
that the Ultra-Glo luciferase had an altered SAR. Another
possibility is that Ultra-Glo luciferase is genuinely more resistant
to inhibitors and thus an analogous qHTS profile would show
a reduction in the lucPpy inhibitors without a concomitant
increase in new inhibitor classes. To explore both the amount
and type of compounds associated with inhibition of Ultra-Glo
luciferase, we describe here the qHTS of this luciferase against
198899 samples of the MLSMR that included all the compounds
assayed previously against the lucPpy. The results show a
marked loss of all inhibitor classes associated with lucPpy
without a corresponding increase in new inhibitor chemotypes,
* To whom correspondence should be addressed. Phone: 301-217-5722.
Fax: 301-217-5736. E-mail: dauld@mail.nih.gov.
a Abbreviations: qHTS, quantitative high-throughput screening; lucPpy,
Photinus pyralis firefly luciferase; lucPpe, Photuris pennsylVanica firefly
luciferase; SPA, scintillation proximity assay; Ultra-Glo, optimized variant
of lucPpe used in Promega’s Kinase-Glo formulations; IMAP, immobilized
metal-ion affinity partitioning; MSR, minimum significance ratio; CRC,
concentration-response curve; BEH, bridged ethyl hybrid.
10.1021/jm8014525 This article not subject to U.S. Copyright. Published 2009 by the American Chemical Society
Published on Web 02/12/2009

Reduced Chemical Library Inhibition of Firefly Luciferase
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5 1451
33 control assay plates containing vehicle (DMSO) only that were
inserted uniformly throughout the screen to monitor background
systematic variation in assay signal. Curve fitting was performed
using an in-house developed algorithm (a version of which is
available online).26 Curve-fitting and SAR analysis was performed
as described,15,22 where the noise of the assay was estimated by
calculating the standard deviation (SD) of the activity values
obtained at the lowest tested compound concentration and outliers
were identified and masked by modeling the Hill equation and
determining if the differences exceeded the assay noise. The qHTS
data was depicted using Origin (OriginLab). Data have been
deposited in PubChem (AID 1379).
Synthesis of 2-Phenylbenzothiazole Compounds. To a mixture
of 2-bromobenzo[d]thiazole or 2-bromo-6-methoxybenzo[d]thiazole
(0.25 mmol) and selected arylboronic acids/boronates (0.325 mmol)
in 2 mL of DME in a microwave tube (2-5 mL capacity) was
added Pd(PPh3)4 (5 mol %, 0.012 mmol) and a 2.0 M aqueous
solution of K2CO3 (0.25 mmol). The mixture was irradiated with
MW for 5-45 min at 150 °C, cooled, diluted with 25 mL of ethyl
acetate, and filtered through celite. The solvent was evaporated,
and the crude material was purified on a Biotage silica gel column.
Gradient elution with ethyl acetate in hexanes (proportions changed
based upon the Rf value of the products) gave the coupled products
as solids.
Analytical QC of Compounds. The entire library was subjected
to purity analysis before plating (Galapagos Biofocus DPI, South
San Francisco, CA). Active compounds that were obtained from
commercial sources were reanalyzed for purity. For these resupplied
compounds, the purity analysis was performed via LCMS analysis
on a Waters ACQUITY reverse phase UPLC System and 1.7 µm
BEH column (2.1 mm × 50 mm) using a linear gradient in 0.1%
aqueous formic acid (5% ACN in water increasing to 95% over 3
min). Compound purity was measured based upon peak integration
from both UV/vis absorbance and ELSD, and compound identity
was based upon mass analysis; all compounds passed purity criteria
(>95%). See Supporting Information for proton and high-resolution
mass spectral data of selected compounds.
Follow-up Luciferase Assays. A total of 39 compounds were
obtained and subjected to several luciferase assays using a 1536-
well plate format. Purified firefly luciferases were assayed in a buffer
containing 50 mM Tris-Acetate pH, 10 mM Mg-acetate, 0.01%
Tween, 0.05% BSA, 10 µM D-luciferin, 10 µM ATP, and luciferase
at 10 nM. Alternatively, this same buffer and enzyme concentration
was used to vary either D-luciferin or ATP concentrations to
determine KM or concentration-response curves (CRCs). For
assaying Kinase-Glo, Kinase-GloPlus, or Kinase-GloMax, the same
buffer used in the qHTS was used. Luminescence was collected
on the Perkin-Elmer Viewlux. Each compound was assayed in
duplicate and this experiment repeated on two or three separate
days.
thus providing evidence that Ultra-Glo luciferase is a more
optimal enzyme reporter for ATP detection in HTS. Compari-
sons of the SAR derived from the qHTS of Kinase-Glo to that
previously defined for lucPpy,15 as well as analysis of repre-
sentative inhibitors, illustrate that the reduced inhibitor suscep-
tibility of Ultra-Glo luciferase is due to a loss of inhibitors whose
interactions stretch outside the luciferin pocket including those
that likely interact with the adenylate pocket.
Materials and Methods
Reagents. ATP, BSA, Tween 20, potassium chloride, imidazole,
D-luciferin, and the luciferase control compound 1 (SIB 1757)20
were purchased from Sigma-Aldrich. Magnesium chloride was
acquired from Quality Biological, and dimethylsulfoxide (DMSO),
certified ACS grade, was purchased from Fisher. The Ultra-Glo
luciferase-based detection reagents used were Kinase-Glo, Kinase-
Glo Plus, and Kinase-Glo Max (Promega; Madison, WI). Purified
wild-type lucPpy luciferase was obtained from Sigma (cat. no.
L9506) and purified Ultra-Glo luciferase was obtained from
Promega.
Preparation of Compound Libraries and Control Plates. The
198899 member library was collected from several sources: 185021
compounds from the NIH MLSMR,21 1280 compounds from
Sigma-Aldrich (LOPAC1280), 1120 compounds from Prestwick
Chemical Inc., 361 purified natural products from TimTec (Newark,
DE), three 1000-member combinatorial libraries from Pharmacopeia
(Princeton, NJ), libraries that include pharmacologically active
compounds such as opioids, adrenergics, cholinergics, serotonergics,
dopaminergics, histaminergics, and endocannabinioids from Tocris
(1105 compounds; Ellisville, Missouri), Biomol (256 compounds;
Plymouth Meeting, PA), Spectrum (1952 compounds; Gardena,
CA), as well as 198 compounds synthesized at the NCGC, 1957
compounds from the National Cancer Institute, 48 nucleoside and
nucleotide-based compounds from Biolog Life Science Institute
distributed by Axxora LLC (San Diego, CA) and from various
Centers for Chemical Methodology and Library Development,
including 47 from the University of North Carolina, 81 from Texas
A&M University, 96 from the University of Wisconsin, 252 from
University of Pittsburgh, 989 from the University of Pennsylvania,
and 1136 compounds from Boston University. Interplate dilutions
of the libraries were prepared as described.22,23
Controls were added from a separate 1536-well compound plate
as follows: columns 1 and 2, 16-point titrations in duplicate of ATP
and the control inhibitor 1, respectively (both beginning at 10 mM
in DMSO); column 3, the neutral control (DMSO); column 4, the
control inhibitor (10 mM in DMSO).
Luciferase Assay and qHTS. First, 4 µL/well of substrate/buffer
(10 µM ATP, 50 mM KCl, 7 mM MgCl2, 0.05% BSA, 0.01%
Tween 20, and 50 mM imidazole pH 7.2, final concentration) was
dispensed into Kalypsys solid white 1536-well plates using a bottle-
valve solenoid-based dispenser (Kalypsys). Then 23 nL of com-
pound solution was transferred to the assay plate using a Kalypsys
pin tool equipped with a 1536-pin array24 containing 10 nL slotted
pins (FP1S10, 0.457 mm diameter, 50.8 mm long; V&P Scientific).
Following transfer, 2 µL/well of Kinase-Glo was dispensed for a
final assay volume of 6 µL/well. Following an 8 min incubation at
ambient temperature, luminescence was detected by a ViewLux
(Perkin-Elmer) using a 5 s exposure time and 1× binning. All
screening operations were performed using a fully integrated
Kalypsys robotic system containing one RX-130 and two RX-90
Stau¨bli anthropomorphic robotic arms.25
qHTS Data and SAR Analysis. Screening data was processed
using in-house developed software. Percent activity was computed
from the median values of the uninhibited, or neutral, control (32
wells located in column 3) and the 40 µM of 1, or 100% inhibited,
control (32 wells, column 4), respectively. For assignment of plate
concentrations and sample identifiers, ActivityBase (ID Business
Solutions Ltd., Guildford, UK) was used for compound and plate
registrations. An in-house database was used to track sample
concentrations across plates. Correction factors were generated from
Results
Luciferase qHTS. We used a commercially available detec-
tion system containing Ultra-Glo luciferase, luciferin, and buffer
components (Kinase-Glo) for the screen. This system showed
linear performance of enzyme activity through a range of 2-10
µM of ATP in a manner similar to the lucPpy contained in PK-
Light for which the previous qHTS was performed,15 supporting
similar sensitivity to ATP between the two enzymes. Therefore,
identical buffer systems containing 10 µM ATP and high (∼50
× KM) concentrations of luciferin could be used for both qHTS
studies. The screen was performed in 1536-well plate format
with a final assay volume of 6 µL per well. Compound 1 (Figure
1), identified as an inhibitor of Ultra-Glo luciferase in our
previous study, was used as an inhibitor control.
The Ultra-Glo luciferase activity was screened against ap-
proximately 199K compounds using qHTS, where the activity
is measured across multiple concentrations allowing the genera-
tion of CRCs for all library compounds.22 In the luciferase

1452 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5
Auld et al.
Figure 1. Summary of the qHTS. (a) Results of the Ultra-Glo luciferase qHTS. Concentration-response data are shown for active (red; percentage
of actives noted) and inactive (blue) compounds. (b) Results of the previous qHTS using the lucPpy formulation15 is shown for comparison.
Activities for 352 compounds (green) in the lucPpy qHTS were reduced because of a partially blocked tip. Percentage of actives noted in red. (c)
Control inhibitor (1) titration response curves and structure of 1 are shown. (d) Representation of class 1a, 1b, and 2a CRCs derived from the
Ultra-Glo luciferase qHTS. Data points are shown in red with the fitted CRC curves shown in blue.
Table 1. Final 1536-Well Assay Protocol
qHTS, the library was tested as a series of at least seven 5-fold
stepa
parameter
reagent, µL
value
4
description
10 µM ATP buffer
dilutions at a beginning concentration of approximately 40 µM.
A total of 1088 1536-well microtiter plates were processed.
The assay showed an average signal-to-background ratio ) 8.1
( 2.9 and Z′ ) 0.74 ( 0.13, indicating a performance that was
comparable to the previous qHTS against lucPpy (Figure 1a,b).
The inhibitor control titrations included on every plate were
highly precise as judged by the minimum significance ratio
(MSR)27 for compound 1 that was calculated to be 1.89n ) 1088
(Figure 1c).
The qHTS resulted in titration-response profiles of 198899
samples derived from approximately 1.53 million assayed wells
(Figure 1b). The screen was performed using an optimized 1536-
well protocol (Table 1) so that one CRC was generated per
second on the Kalypsys automated system.25 Automated curve-
fitting was performed to generate CRC fits to the data corre-
sponding to each sample. The CRCs were then separated into
four classes: (1) complete CRCs containing upper and lower
asymptotes, (2) incomplete CRCs having an upper asymptote,
(3) poorly fit CRCs or where the activity was observed only at
the highest tested concentration, and (4) inactive where activity
of all unmasked data points was below 30%. Each CRC class
was further divided based on the efficacy of response (e.g., CRC
classes “1a, 1b”; see Inglese et al.22).
1
2
3
4
5
6
library compounds, nL 23 40 µM to 0.24 nM dilution series
controls, nL
reagent, 2 µL
incubation time, min
assay readouts
23 ATP, compound 1, DMSO
detection buffer
RT incubation
luminescence read
8
5
a Step Notes: (1) Medium-binding white solid Kalypsys plates. 100 µL
pre-dispense, four-tip dispense columns 1-48. Buffer: 50 mM KCl, 7 mM
MgCl2, 10 µM ATP, 0.01% Tween, 0.05% BSA. Mixture kept on ice. (2)
Pin-tool transfer compound library for a (final) range of 40 µM to 0.24
nM. (3) Pin-tool transfer, column 1 16-point titrations in duplicate of ATP
beginning at 10 µM (final), column 2, 16-point titrations in duplicate of
compound 1 beginning at 40 µM (final), column 4, compound 1 at 40 µM
(final). Pin-tool transfer tip wash sequence: DMSO,iPA, MeOH, 3 s vacuum
dry. (4) Kinase-Glo luminescent reagent containing Ultra-Glo luciferase
and high concentrations (∼mM) of luciferin. 100 µL pre-dispense, four-tip
dispense columns 1-48. (5) RT incubation in auxilliary plate hotel. (6) PE
ViewLux, clear filter.
and 232 samples, respectively, represented 0.15% of the
collection (Figure 1d; Table 2). Class 2b was the largest class
of actives, containing 837 samples (0.4%), while class 3 totaled
413 samples (0.2%). Class 1 CRCs spanned a potency range
from 0.6 to 10 µM and included the positive control compound
1 that was present within the Tocris library. In contrast, the
maximum inhibitor potency obtained in the lucPpy qHTS was
50 nM. Potencies for class 2a and 3 should be considered
approximations as they are extrapolated from incomplete CRCs.
The screen identified 0.78% of the library samples as active
(class 1-3). The distribution of potencies within each CRC class
is summarized in Table 2. Compounds associated with the
highest quality CRCs, Class 1a, 1b, and 2a, comprising 43, 11,

Reduced Chemical Library Inhibition of Firefly Luciferase
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5 1453
Table 2. Analysis of qHTS. Numbers in parenthesis represent the % of inhibitors found for this CRC class in the lucPpy qHTS
curve classification
IC50 (µM)
1a
1b
2a
2b
3
total
<0.1
0
3
40
0
43
0
11
0
0
11
0
0
139
93
232
0
0
335
502
837
0
0
159
254
413
0
14
673
849
1,536
0.1-1
1-10
10-100
total per classification
% of Library
0.02, (0.29)
0.01, (0.19)
0.12, (0.34)
0.42, (1.79)
0.21, (0.59)
0.78, (3.19)
Comparison of Activity between lucPpy and Ultra-Glo
Luciferase. For the lucPpy qHTS we observed that 3.1% of
the library samples were active (class 1-3),15 whereas only
0.9% of the samples were active in the Ultra-Glo luciferase
qHTS. Potent inhibitory compounds showing high quality CRCs
were dramatically reduced in Ultra-Glo luciferase compared to
the lucPpy formulation as only 0.15% of compounds in the
Ultra-Glo luciferase qHTS had a CRC class 1a, 1b, or 2a CRC
versus 0.82% for lucPpy qHTS (Table 2). Additionally, we
found only 0.03% Class 1 CRCs in Ultra-Glo luciferase, while
0.48% was found within this CRC class using the lucPpy
formulation (Table 2). Further, of the class 1 CRCs from the
lucPpy luciferase qHTS 50% were found as inactive (class 4)
in the Ultra-Glo luciferase qHTS.
The potency distribution between the qHTS against Ultra-
Glo and lucPpy luciferase is summarized in Figure 2. Only 0.1%
of the Ultra-Glo luciferase inhibitory compounds showed an
IC50 < 10 µM compared to 0.9% using the lucPpy formulation.
Overall, the Ultra-Glo luciferase qHTS showed an average
potency that is 7-fold weaker than what was found in the lucPpy
qHTS.
Characterization of Selectivity between Ultra-Glo and
lucPpy Luciferases. We have identified several prominent
scaffolds that act as luciferase inhibitors, including several
luciferin mimetic compounds such as those containing a
benzothiazole core.15 The same chemical series were again
identified in the Ultra-Glo luciferase screen but with greatly
reduced activity (Figure 3). In general, there was a marked loss
of inhibitors containing benzoxazole (4), benzthiazole (5), or
benzimidazole (6) cores (reduced by 93%, 80%, and 76%,
respectively; Figure 3) that likely bind at luciferyl-adenylate
pocket of luciferase. We also noted that a potent inhibitor class
containing a 3,5-diaryl-oxadiazole core (2, Figure 3), where we
have found that the inhibition cannot be completely relieved
by addition of high substrate concentrations, was also reduced
but to a lesser extent (60% reduction). Similarly, the number
of active compounds in the benzamide class (3, Figure 3) was
reduced by 80% in the Ultra-Glo luciferase qHTS.
To investigate the reason for the reduced inhibitor fraction
in the Ultra-Glo luciferase qHTS experiments, we obtained
represented compounds and characterized these against purified
Ultra-Glo luciferase and lucPpy. We also measured activity of
these compounds in several commercial preparations of Ultra-
Glo luciferase.
To explore the luciferin binding pocket, we examined 22
2-phenylbenzothiazole (7) analogues against the two luciferases
(Table 3). These compounds represented minimal luciferin
mimetic analogues and were generally either unsubstituted or
contained simple halogen or methyl group substitutions. We
determined similar potency values for these compounds using
either purified Ultra-Glo luciferase or lucPpy and KM levels of
substrates against this series of analogues (Table 3). Therefore,
the specificity within the luciferin pocket is similar between
the two luciferases. Consistent with competitive behavior, we
also noted that all these compounds showed greatly weakened
inhibition or became inactive in the commercial Ultra-Glo
luciferase-containing reagents that contain excess luciferin
substrate (Table 3).
However, we noted that a 4-(6-methoxybenzo[d]thiazol-2-
yl)-N,N-dimethylaniline (7p) showed nearly 4-fold weaker
potency against purified Ultra-Glo luciferase than lucPpy and
this compelled us to examine additional analogues (7a-7x;
Table 3) at the 2-position of the benzothiazole that incorporated
larger substitutions. For this purpose, we examined the potency
of several analogues under a range of luciferin and ATP
concentrations (Figure 4) using purified Ultra-Glo and lucPpy
luciferases. Further we compared 7p as well as two analogues
that contained a larger pyrazole-containing substitution to
representative analogues mentioned above that contained simple
halogen substitutions (e.g., 7d, 7e, and 7f). We found that the
halogen substituted analogues behaved as pure luciferin com-
petitive inhibitors with substrate variation, showing similar
effects between the two luciferases (Figure 4a-c). However,
7p and two compounds containing a 1-benzyl-1H-pyrazole at
the 2-position (Figure 4; 7w and 7x) showed different substrate
dependences for the two luciferases. In the case of Ultra-Glo
luciferase, these compounds remained largely luciferin competi-
tive as shown by the general loss of activity at high luciferin
concentrations (Figure 4d-f). However, for lucPpy, these
compounds showed greater potency for lucPpy than Ultra-Glo
luciferase upon either luciferin or ATP variation (compare red
and black fits, Figure 4e,f). In the qHTS, D-luciferin was present
in large excess in both the Ultra-Glo and lucPpy luciferase
formulations and we measured only a 4-fold higher affinity KM
for luciferin with Ultra-Glo luciferase than lucPpy (0.9 ( 0.1
vs 4 ( 0.4 µM, respectively). Therefore, changes in substrate
affinity alone cannot explain the large loss of these inhibitory
Figure 2. Difference in potency distribution between lucPpy and Ultra-
Glo luciferase formulations. Active compounds from the previous
lucPpy qHTS and the Ultra-Glo luciferase qHTS were compared and
the ∆LogAC50 values were calculated. Negative ∆LogAC50 values
represent compounds that were less potent in the Ultra-Glo luciferase
compared to the lucPpy qHTS, while positive ∆LogAC50 values
represent compounds that were more potent in the Ultra-Glo luciferase
compared to the lucPpy qHTS.

1454 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5
Auld et al.
Figure 3. Comparison of inhibitor scaffold representation between lucPpy and Ultra-Glo luciferase formulations. Center pie charts represent the
amount of active (darker gray slices) and related inactive analogues (light-gray area in each pie chart) for prominent luciferase inhibitor scaffolds
(2-6) for the lucPpy (top pie) and Ultra-Glo luciferase (bottom pie) qHTS. The scaffolds and the associated analogues were determined as described
in Auld et al.15 The number of lucPpy active analogues was reduced from 70 to 27 (2, 3,5-diaryl-oxadiazoles), 98 to 23 (3, benzamides), 21 to 5
(4, benzoxazoles), 89 to 6 (5, benzthiazoles), and 55 to 11 (6, benzimidazoles) in the Ultra-Glo luciferase qHTS. The corresponding number of
structurally related inactive analogues increased from 1192 to 1442 in the Ultra-Glo luciferase qHTS.
Table 3. Characterization of Selected 2-Phenylbenzothiazole Analoguesa
analogue no.
R1
R2
R3
R4
IC50UltraGlo
IC50lucPpy
IC50KinGlo
IC50KinGlo Plus IC50KinGlo Max
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
H
H
OMe
OMe
OMe
H
H
H
H
F
H
H
H
F
H
H
H
H
dimethylamine 0.32 ( 0.0
0.2 ( 0.0
4.5 ( 0.0
10.7 ( 3.4
>50
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
Cl
1.4 ( 0.18
0.86 ( 0.08
1.1 ( 0.14
0.5 ( 0.05
4.9 ( 1.1
3.8 ( 0.5
4.3 ( 0.2
30 ( 3
6.1 ( 1.3
4.1 ( 0.4
5.6 ( 1.3
3.8 ( 0.6
1.9 ( 0.4
0.67 ( 0.07
2.7 ( 0.6
7.8 ( 0.3
1.9 ( 0.3
0.36 ( 0.02
7.7 ( 1.4
2.1 ( 0.2
3.1 ( 0.5
F
H
H
F
Cl
H
H
CN
1.3 ( 0.09 >50
1.5 ( 0.1
0.6 ( 0.0
5.4 ( 0.7
3.5 ( 0.0
8.9 ( 0.6
13.4 ( 1
5.7 ( 0.9
2.2 ( 0.0
3.2 ( 0.0
2.8 ( 0.0
2.9 ( 0.2
1.5 ( 0.1
>50
>50
OMe Cl
OMe
OMe O-methylbenzene
OMe
OMe
OMe
>50
H
>50
>50
H
H
H
O-methylbenzene
H
H
H
OMe
acetamide
Me
H
F
OMe
H
H
H
H
inactive
inactive
>50
OMe
H
OMe
H
OMe OMe
>50
OMe
OMe
OMe
OMe
H
H
H
H
H
H
H
H
H
H
H
H
OMe
H
>50
>50
H
>50
dimethylamine
H
0.7 ( 0.04 >50
24.2 ( 1.6
3.2 ( 0.0
1.0 ( 0.0
11.3 ( 1.8
6.3 ( 0.0
3.0 ( 0.2
inactive
OMe
OMe
H
>50
>50
>50
>50
>50
s
t
u
v
CN
acetamide
H
H
a Activity of compounds was determined by measurement of luminescence using either purified luciferase enzyme assays (IC50UltraGlo or IC50lucPpy) or
formulations of UltraGlo (IC50KinGlo, IC50KinGlo Plus, IC50KinGlo Max). Data shown are mean ( SD for at least three replications.
classes in the qHTS, rather a shift to a noncompetitive luciferin
binding mode due to interactions outside the luciferase pocket
appears to result in more prevalent inhibitors for lucPpy
formulations.
To further test possible reasons for this selectivity, we
examined the potencies at Ultra-Glo luciferase for a series of
quinoline compounds identified in the lucPpy qHTS that were
shown to be competitive with ATP and luciferin and modeled
into the adenylate binding pocket of firefly luciferase.15 Con-
sistent with this pocket serving as a mediator of selectivity
between the two luciferases, we found a marked loss of activity
for these quinolines for Ultra-Glo luciferase (Figure 5). Two of
the most potent quinolines for lucPpy were 8 (Figure 5, lucPpy
IC50 ) 0.5 µM) and 9 (Figure 5, lucPpy IC50)4.0 µM). Both 8
and 9 showed IC50s >10 µM, outside the typical screening
concentration range, against Ultra-Glo luciferase (Figure 5).

Reduced Chemical Library Inhibition of Firefly Luciferase
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5 1455
Figure 4. Comparison lucPpy and Ultra-Glo luciferase against 2-phenylbenzothiazole luciferase inhibitors at multiple substrate concentrations.
Graphs of ATP variation or luciferin variation (inset) are shown. Luciferin or ATP was varied at 0.25, 2, 25, and 250 µM, resulting in four sets of
CRCs. In each case, the constant substrate was present at 250 µM. The lucPpy data is shown as solid circles with red CRC fits, and the Ultra-Glo
luciferase is shown as open circles with black fitted lines. The structures of the inhibitors assayed are also shown. The top graphs show compounds
with luciferin competitive behavior against both lucPpy and Ultra-Glo luciferase, while the bottom graphs demonstrate inhibitors that maintain
potency at high ATP or luciferin concentrations for lucPpy but not Ultra-Glo luciferase.
Further, in formulations of Ultra-Glo luciferase, this series is
inactive. This suggests that favorable interactions within the
adenylate pocket are present in lucPpy that are absent in Ultra-
Glo luciferase, suggesting the adenlyate pocket as one of the
regions that determines the large reduction in the inhibitors for
Ultra-Glo luciferase.
formulations of Ultra-Glo luciferase such as Kinase-GloPlus and
Kinase-GloMax (Table 4).
Discussion
We have determined a comprehensive inhibitor profile of
∼199K compounds using an optimized formulation of Ultra-
Glo luciferase and compared this to a previously determined
inhibitor profile that used an analogous lucPpy-containing
formulation.15 The Ultra-Glo luciferase formulation (Kinase-
Glo) showed reductions in both inhibitor abundance and potency
relative to lucPpy with only 0.1% of library showing IC50 <10
µM in the Ultra-Glo luciferase formulation compared to 0.9%
with the lucPpy formulation. Although we noted similar
specificity for compounds purely competitive with luciferin
substrate, we found that one of the principle reasons for the
lower number of inhibitors against Ultra-Glo luciferase was due
to a loss of compounds with interactions stretching outside the
luciferin pocket. These interactions allow for inhibition to persist
even at high D-luciferin for lucPpy, while the absence of these
interactions renders these inhibitors inactive in Ultra-Glo
luciferase formulations. Therefore, formulations such as Kinase-
Glo should be superior ATP sensor reagents for HTS, as these
will have lower compound interference. Indeed, a previous study
Compounds that show a noncompetitive behavior against
reporter enzymes used in HTS, perhaps by exploring inhibitory
sites outside the active site, will be more of a nuisance as the
inhibition will not be completely relieved by increasing substrate
concentrations. Our previous work identified the 3,5-diaryl-
oxadiazoles as potent noncompetitive inhibitors against lucPpy15
and indeed some of the most potent compounds identified here
contained a 3,5-diaryl-oxadiazoles core (2a-l; Table 4). Of
these, the most potent compounds showed IC50s ∼ 50 nM at
low substrate concentrations but showed reduced potencies near
10 µM in the formulated reagents (e.g., 2d,e). Although this
inhibitor class also showed a large reduction of inhibitory
compounds (∼60% less) in the Ultra-Glo luciferase qHTS, we
noted that the 3,5-diaryl-oxadiazoles series was the one com-
pound series that still contained a significant number of
compounds, which demonstrated appreciable inhibition in

1456 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5
Auld et al.
Figure 5. Comparison of lucPpy and Ultra-Glo luciferase inhibition for a series of quinoline analogues. Red fits, solid circles and squares are
lucPpy, and black fits, open circles, and squares are Ultra-Glo luciferase for 8 (squares) or 9 (circles) assayed using KM levels of substrates. Also
shown is the activity of these two compounds using Kinase-Glo for 8 (open triangles) and 9 (upside-down open triangles). The inset shows activity
in the qHTS for a series of related quinolines in the lucPpy qHTS (solid circles, red-fits) and Kinase-Glo (open circles).
Table 4. Characterization of Selected 3,5-diaryl-oxadiazoles from the qHTSa
analogue no.
R1
pyridin-2yl
pyridin-2yl
pyridin-2yl
pyridin-2yl
R2
pyridin-4yl
furan-2yl
IC50UltraGlo
IC50P.Pyralis
IC50KinGlo
IC50KinGlo Plus
IC50KinGlo Max
a
b
c
d
e
f
g
h
i
1.9 ( 0.6
1.3 ( 0.3
14.4 ( 6.9
17.1 ( 7.1
0.08 ( 0.07
1.4 ( 1.2
2.8 ( 1.6
0.2 ( 0.1
1.7 ( 0.5
2.8 ( 1.6
8.3 ( 3.7
24 ( 9.8
8.9 ( 1.9
13.9 ( 0.6
11.3 ( 1.7
10.0 ( 3.4
5 ( 1
14.1 ( 0.0
>50
>50
13.4 ( 2.4
13.5 ( 0.8
>50
inactive
>50
inactive
inactive
inactive
inactive
25.3 ( 4.1
>50
>50
biphenyl-4yl
0.038 ( 0.006
0.040 ( 0.006
0.08 ( 0.01
0.054 ( 0.004
0.58 ( 0.1
0.3 ( 0.05
0.6 ( 0.1
4-phenyl benzoate
2-methoxy phenyl
2,4-dimethoxy phenyl
4-flouro phenyl
2-flouro phenyl
phenyl
pyridin-4yl
furan-2yl
2-methoxy phenyl
13.1 ( 2.5
12.7 ( 1.7
>50
inactive
inactive
inactive
inactive
inactive
inactive
pyridin-2yl
phenyl
13.9 ( 1.1
14 ( 7.7
>50
3-methyl phenyl
4-methoxy phenyl
2-methoxy phenyl
2-methoxy phenyl
pyridin-4yl
>50
j
k
l
4.8 ( 0.7
>50
2.4 ( 0.3
24 ( 10.7
0.5 ( 0.3
>50
4-dimethyaminophenyl
0.32 ( 0.04
inactive
a Activity of compounds was determined by measurement of luminescence using either purified luciferase enzyme assays (IC50UltraGlo or IC50lucPpy) or
formulations of UltraGlo (IC50KinGlo, IC50KinGlo Plus, IC50KinGlo Max). Data shown are mean ( SD for at least three replications.
that used Kinase-Glo for HTS of glycogen synthase kinase-3
noted that no luciferase inhibitors were identified during
confirmation of the primary hits.28
The Ultra-Glo luciferase was derived from lucPpe and was
optimized by directed evolution19,29,30 using a selection strategy
that incorporated several steps and introduced as many as 34
mutations. This included screening for variants of increased
thermostability, signal stability (“glow response”), substrate
usage efficiency through evaluation of KM values, and increased
inhibitor resistance. Inhibitor resistance was determined by
adding L-luciferin into the reaction mix and selecting for variants
that were resistant to this inhibitor29 because studies have shown
that L-luciferin can act as a competitive inhibitor of the
bioluminescent reaction; the enantiomeric position of luciferin
being the site of adenylate attachment.31,32 Some of the
mutations introduced in this process were found to improve the
KM for ATP (from 18 µM to approximately 3 µM, relative to
wild-type lucPpe),29 consistent with changes in the adenylate
pocket and may have improved the resistance to inhibitors.
Although Ultra-Glo luciferase formulations appear to be more
optimal assay reagents for ATP detection in HTS, we note that
Ultra-Glo luciferase reagents intended for uses other than ATP
detection, where luciferin concentrations may be present at lower
concentrations, could contain additional inhibitors than described
here. For example, given the potent binding for many inhibitors
we found using low luciferin concentrations, certain reagents
containing pro-luciferin substrates3 will likely show increased
amounts of luciferase inhibitors compared to the profile
described here. The potent inhibition of firefly luciferases by
luciferin competitive compounds in unformulated reagents
should also caution against the temptation to dilute detection
reagents further than supplier suggestions to save costs. Even
though diluted detection reagents can show adequate signal:
background, increased interference by luciferase inhibitors will
be obtained.
Firefly luciferase is a globular protein composed of N- and
C-terminal domains linked by a hinge region. These two
domains are known to undergo a change upon substrate binding

Reduced Chemical Library Inhibition of Firefly Luciferase
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5 1457
that brings the two domains together, resulting in a more
compact closed conformation.33,34 The final thermostable Ultra-
Glo luciferase variant showed a half-life of more than 100 days
at room temperature, representing an improvement in half-life
at 50 °C of approximately 5000-fold over unoptimized lucPpe.29
Additionally, the increased thermostability of Ultra-Glo lu-
ciferase may have led to increased conformational rigidity,
resulting in a relatively more compact closed conformation, thus
blocking many of the small molecule binding sites present in
unoptimized lucPpy. Indeed, comparison of protein structures
between hyperthermophilic and mesophilic organisms shows
increased hydrophobic packing and decreased flexibility in the
hyperthermophilic proteins.35,36
contain the right structures for high affinity binding to Ultra-
Glo luciferase. A recent study describing a screen of sphingosine
1-phosphate (S1P) receptors against the MLSMR suggested that
chemical library biases may have played a role in the determin-
ing different hit rates between two S1P receptor subtypes.47
Further expansion of the qHTS approach to other target classes
and different types of chemical libraries (e.g., natural products)
will provide experimental testing that can be used to determine
how the nature of the library affects the hit rate of the target
class. The qHTS database already established at our center and
available within PubChem21 could be mined to obtain such
information for relevant target classes.
Given the wide use of enzyme-based reporters in the early
phases of compound discovery efforts, it would seem wise to
critically examine the mechanisms of inhibition of these
common reporter enzymes. This knowledge can be used to
distinguish the SAR related to the targeted biology versus
reporter specific effects that can result in misleading activity.
For example, using our understanding of firefly luciferase
inhibitor SAR, we have recently demonstrated that inhibition
of firefly luciferase in cell-based reporter assays can lead to the
counter-intuitive phenomena of apparent reporter gene activation
due to inhibitor-based stabilization of the reporter enzyme within
the cellular milieu.16 Assay interferences due to inhibition of
reporter enzymes such as firefly luciferase offer the opportunity
to understand the SAR in “medicinal chemistry” terms, and this
knowledge can be used to ensure proper interpretation of the
results and use of this important bioluminescent assay detection
method.
Observing a difference in druggability between related en-
zymes is not without precedence37,38 and a similar mechanism
involving conformational flexibility has been proposed to explain
druggability differences in protein kinases. In the cocrystal of
Raf protein kinase in complex with 4-(4-(3-(4-chloro-3-(trif-
luoromethyl)phenyl)ureido)phenoxy)-N-methylpicolinamide (Bay
43-9006),39 the compound is found to bind to a large open
binding pocket present in the inactive state of the ATP pocket,
thus leading to inhibition through stabilization of the inactive
conformation of the kinase.40 However, in the protein kinase
Tie2, this binding site is closed because it is filled by the
nucleotide binding loop that prevents small molecule inhibi-
tion.41 Therefore, conformational flexibility and relative plastic-
ity have been proposed to explain why certain targets are
associated with many inhibitor scaffolds while others have been
resistant to efforts to develop inhibitors and may represent
another reason for the improved resistance to inhibitors for Ultra-
Glo luciferase.42,43
Acknowledgment. This research was supported by the
Molecular Libraries Initiative of the NIH Roadmap for Medical
Research and the Intramural Research Program of the National
Human Genome Research Institute, National Institutes of Health.
We thank Adam Yasgar, Paul Shinn for compound management
support, Jeremy Smith for analytical chemistry support, and
Carleen Klumpp and Sam Michael for assistance with the
Kalypsys robotic system.
Methods aimed at understanding target druggability are of
great interest as these can lead to better choices of targets geared
for small molecule intervention. The term “druggability” has
been defined as the ability to identify high affinity and selective
small (e.g. ∼500 MW) druglike compounds that show tractable
SAR against a molecular target.38,44,45 Understanding target
druggability has been approached through various methods in
the past, including computational methods employing sequence
alignment information or crystal structures.45,46 A recent study
used an affinity model to predict druggability where an
experimental HTS against 11K compounds was performed to
confirm the results.46 In that study, it was shown that a target
predicted to have low druggability, and for which researchers
at Pfizer were unable to identify suitable leads after considerable
effort, showed a 90% reduction in hits compared to a target
predicted to have high druggability. As well, the maximum
achievable potency was predicted to be 10-fold higher for the
druggable target. These metrics are in line with the experimental
results obtained here as we also observed a 90% reduction in
major inhibitor classes and a 10-fold reduction in the maximum
observed potency when comparing the two luciferase qHTS
experiments, supporting that Ultra-Glo luciferase reagents are
less “druggable”.
Supporting Information Available: General synthetic methods
and procedures as well as characterization of the compounds are
provided. This material is available free of charge via the Internet
at http://pubs.acs.org.
References
(1) Cali, J. J.; Ma, D.; Sobol, M.; Simpson, D. J.; Frackman, S.; Good,
T. D.; Daily, W. J.; Liu, D. Luminogenic cytochrome P450 assays.
Expert Opin. Drug Metab. Toxicol. 2006, 2, 629–645.
(2) Cali, J. J.; Niles, A.; Valley, M. P.; O’Brien, M. A.; Riss, T. L.; Shultz,
J. Bioluminescent assays for ADMET. Expert Opin. Drug Metab.
Toxicol. 2008, 4, 103–120.
(3) Fan, F.; Wood, K. V. Bioluminescent assays for high-throughput
screening. Assay Drug DeV. Technol. 2007, 5, 127–136.
(4) Comley, J. Kinase screening and profiling-spoilt for choice. Drug
DiscoVery World 2006, 7, 27–50.
(5) Inglese, J.; Johnson, R. L.; Simeonov, A.; Xia, M.; Zheng, W.; Austin,
C. P.; Auld, D. S. High-throughput screening assays for the identifica-
tion of chemical probes. Nat. Chem. Biol. 2007, 3, 466–479.
(6) Wu, S.; Liu, B. Application of scintillation proximity assay in drug
discovery. BioDrugs 2005, 19, 383–392.
(7) Sportsman, J. R.; Gaudet, E. A.; Boge, A. Immobilized metal ion
affinity-based fluorescence polarization (IMAP): advances in kinase
screening. Assay Drug DeV. Technol. 2004, 2, 205–214.
(8) Lowery, R. G.; Kleman-Leyer, K. Transcreener: screening enzymes
involved in covalent regulation. Expert Opin. Ther. Targets 2006, 10,
179–190.
(9) Huss, K. L.; Blonigen, P. E.; Campbell, R. M. Development of a
Transcreener Kinase Assay for Protein Kinase A and Demonstration
of Concordance of Data with a Filter-Binding Assay Format. J. Biomol.
Screening 2007, 12, 578–584.
Current technologies allow for rapid determination of potency
values across large chemical libraries in a manner in which
robust SAR is experimentally measured for the entire library.
Such an experiment employed early in the drug discovery
process where both the maximum inhibitor potency and fraction
of active compounds is directly determined could save unneces-
sary time and effort on intractable target classes. However, an
alternative reason for the reduced inhibitor fraction found here
for Ultra-Glo luciferase could be that the MLSMR, which
contains largely small molecule druglike compounds, does not

1458 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 5
Auld et al.
(10) Klink, T. A.; Kleman-Leyer, K. M.; Kopp, A.; Westermeyer, T. A.;
Lowery, R. G. Evaluating PI3 kinase isoforms using Transcreener ADP
assays. J. Biomol. Screening 2008, 13, 476–485.
(11) Schroter, T.; Minond, D.; Weiser, A.; Dao, C.; Habel, J.; Spicer, T.;
Chase, P.; Baillargeon, P.; Scampavia, L.; Schurer, S.; Chung, C.;
Mader, C.; Southern, M.; Tsinoremas, N.; LoGrasso, P.; Hodder, P.
Comparison of miniaturized time-resolved fluorescence resonance
energy transfer and enzyme-coupled luciferase high-throughput screen-
ing assays to discover inhibitors of Rho-kinase II (ROCK-II).
J. Biomol. Screening 2008, 13, 17–28.
(12) Singh, P.; Harden, B. J.; Lillywhite, B. J.; Broad, P. M. Identification
of kinase inhibitors by an ATP depletion method. Assay Drug DeV.
Technol. 2004, 2, 161–169.
(13) Koresawa, M.; Okabe, T. High-throughput screening with quantitation
of ATP consumption: a universal non-radioisotope, homogeneous assay
for protein kinase. Assay Drug DeV. Technol. 2004, 2, 153–160.
(14) Manning, G.; Whyte, D. B.; Martinez, R.; Hunter, T.; Sudarsanam,
S. The protein kinase complement of the human genome. Science 2002,
298, 1912–1934.
(15) Auld, D. S.; Southall, N. T.; Jadhav, A.; Johnson, R. L.; Diller, D. J.;
Simeonov, A.; Austin, C. P.; Inglese, J. Characterization of chemical
libraries for luciferase inhibitory activity. J. Med. Chem. 2008, 51,
2372–2386.
(16) Auld, D. S.; Thorne, N.; Nguyen, D. T.; Inglese, J. A specific
mechanism for nonspecific activation in reporter-gene assays. ACS
Chem. Biol. 2008, 3, 463–470.
(17) Heitman, L. H.; van Veldhoven, J. P.; Zweemer, A. M.; Ye, K.;
Brussee, J.; AP, I. J. False positives in a reporter gene assay:
identification and synthesis of substituted N-pyridin-2-ylbenzamides
as competitive inhibitors of firefly luciferase. J. Med. Chem. 2008,
51, 4724–4729.
(28) Baki, A.; Bielik, A.; Molnar, L.; Szendrei, G.; Keseru, G. M. A high
throughput luminescent assay for glycogen synthase kinase-3ꢀ inhibi-
tors. Assay Drug DeV. Technol. 2007, 5, 75–83.
(29) Wood, K. V.; Hall, M. P.; Gruber, M. G. Thermostable luciferases
from Photuris pennsylVanica and Pyrophorus plagiopthalamus and
methods of production Patent WO 01/20002 A1, 2001.
(30) Zhang, J. H.; Dawes, G.; Stemmer, W. P. Directed evolution of a
fucosidase from a galactosidase by DNA shuffling and screening. Proc.
Natl. Acad. Sci. U.S.A. 1997, 94, 4504–4509.
(31) Lembert, N. Firefly luciferase can use L-luciferin to produce light.
Biochem. J. 1996, 317 (Pt 1), 273–277.
(32) Nakamura, M.; Maki, S.; Amano, Y.; Ohkita, Y.; Niwa, K.; Hirano,
T.; Ohmiya, Y.; Niwa, H. Firefly luciferase exhibits bimodal action
depending on the luciferin chirality. Biochem. Biophys. Res. Commun.
2005, 331, 471–475.
(33) Wang, W. Q.; Xu, Q.; Shan, Y. F.; Xu, G. J. Probing local
conformational changes during equilibrium unfolding of firefly lu-
ciferase: fluorescence and circular dichroism studies of single tryp-
tophan mutants. Biochem. Biophys. Res. Commun. 2001, 282, 28–33.
(34) Rosendahl, M. S.; Leonard, N. J.; DeLuca, M. Dimensional probing
of the ATP binding site on firefly luciferase. Photochem. Photobiol.
1985, 35, 857–861.
(35) Paiardini, A.; Sali, R.; Bossa, F.; Pascarella, S. “Hot cores” in proteins:
comparative analysis of the apolar contact area in structures from
hyper/thermophilic and mesophilic organisms. BMC Struct. Biol. 2008,
8, 14.
(36) Zavodszky, P.; Kardos, J.; Svingor Petsko, G. A. Adjustment of
conformational flexibility is a key event in the thermal adaptation of
proteins. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 7406–7411.
(37) Schafer, M.; Egner, U. Structural Aspects of Drugability and Selectivity
of Protein Kinases in Inflammation. Anti-Inflammatory Anti-Allergy
Agents Med. Chem. 2007, 6, 5–17.
(18) Ye, L.; Buck, L. M.; Schaeffer, H. J.; Leach, F. R. Cloning and
sequencing of a cDNA for firefly luciferase from Photuris pennsyl-
Vanica. Biochim. Biophys. Acta 1997, 1339, 39–52.
(19) Hall, M. P., Gruber, M. G., Hannah, R. R., Jennens-Clough M. L.,
Wood, K. V.; Stabilization of Firefly Luciferase Using Directed
EVolution; John Wiley & Sons: New York, 1998.
(20) Varney, M. A.; Cosford, N. D.; Jachec, C.; Rao, S. P.; Sacaan, A.;
Lin, F. F.; Bleicher, L.; Santori, E. M.; Flor, P. J.; Allgeier, H.;
Gasparini, F.; Kuhn, R.; Hess, S. D.; Velicelebi, G.; Johnson, E. C.
SIB-1757 and SIB-1893: selective, noncompetitive antagonists of
metabotropic glutamate receptor type 5. J. Pharmacol. Exp. Ther. 1999,
290, 170–181.
(38) Schneider, M. A rational approach to maximize success rate in target
discovery. Arch. Pharm. (Weinheim) 2004, 337, 625–633.
(39) Lowinger, T. B.; Riedl, B.; Dumas, J.; Smith, R. A. Design and
discovery of small molecules targeting raf-1 kinase. Curr. Pharm. Des.
2002, 8, 2269–2278.
(40) Wan, P. T.; Garnett, M. J.; Roe, S. M.; Lee, S.; Niculescu-Duvaz, D.;
Good, V. M.; Jones, C. M.; Marshall, C. J.; Springer, C. J.; Barford,
D.; Marais, R. Mechanism of activation of the RAF-ERK signaling
pathway by oncogenic mutations of B-RAF. Cell 2004, 116, 855–
867.
(41) Shewchuk, L. M.; Hassell, A. M.; Ellis, B.; Holmes, W. D.; Davis,
R.; Horne, E. L.; Kadwell, S. H.; McKee, D. D.; Moore, J. T. Structure
of the Tie2 RTK domain: self-inhibition by the nucleotide binding
loop, activation loop, and C-terminal tail. Structure 2000, 8, 1105–
1113.
(21) PubChem. http://pubchem.ncbi.nlm.nih.gov/.
(22) Inglese, J.; Auld, D. S.; Jadhav, A.; Johnson, R. L.; Simeonov, A.;
Yasgar, A.; Zheng, W.; Austin, C. P. Quantitative high-throughput
screening: A titration-based approach that efficiently identifies biologi-
cal activities in large chemical libraries. Proc. Natl. Acad. Sci. U.S.A
2006, 103, 11473–11478.
(23) Yasgar, A.; Shinn, P.; Jadhav, A.; Auld, D. S.; Michael, S.; Zheng,
W.; Austin, C. P.; Inglese, J.; Simeonov, A. Compound Management
for Quantitative High-Throughput Screening. J. Assoc. Lab. Autom.
2008, 13, 79–89.
(24) Cleveland, P. H.; Koutz, P. J. Nanoliter dispensing for uHTS using
pin tools. Assay Drug DeV. Technol. 2005, 3, 213–225.
(25) Michael, S.; Auld, D. S.; Klumpp, C.; Jadhav, A.; Zheng, W.; Thorne,
N.; Austin, C. P.; Inglese, J.; Simeonov, A. A robotic platform for
quantitative high-throughput screening. Assay Drug DeV. Technol.
2008, 6, 637–658.
(42) Fabbro, D. An interview with Doriano Fabbro Ph D., Novartis
Biomedical Research Institute Interviewed by Vicki Glaser. Assay
Drug DeV. Technol. 2004, 2, 109–114.
(43) Egner, U.; Hillig, R. C. A structural biology view of target druggability.
Expert Opin. Drug DiscoVery 2008, 3, 391–401.
(44) Hajduk, P. J.; Greer, J. A decade of fragment-based drug design:
strategic advances and lessons learned. Nat. ReV. Drug DiscoVery 2007,
6, 211–219.
(45) Cheng, A. C.; Coleman, R. G.; Smyth, K. T.; Cao, Q.; Soulard, P.;
Caffrey, D. R.; Salzberg, A. C.; Huang, E. S. Structure-based maximal
affinity model predicts small-molecule druggability. Nat. Biotechnol.
2007, 25, 71–75.
(46) Hopkins, A. L.; Groom, C. R. The druggable genome. Nat. ReV. Drug
DiscoVery 2002, 1, 727–730.
(26) Free Software and Code for Public Use. http://www.ncgc.nih.gov/
resources/software.html.
(47) Schurer, S. C.; Brown, S. J.; Gonzalez-Cabrera, P. J.; Schaeffer, M. T.;
Chapman, J.; Jo, E.; Chase, P.; Spicer, T.; Hodder, P.; Rosen, H.
Ligand-binding pocket shape differences between sphingosine 1-phos-
phate (S1P) receptors S1P1 and S1P3 determine efficiency of chemical
probe identification by ultrahigh-throughput screening. ACS Chem.
Biol. 2008, 3, 486–498.
(27) Eastwood, B. J.; Farmen, M. W.; Iversen, P. W.; Craft, T. J.;
Smallwood, J. K.; Garbison, K. E.; Delapp, N. W.; Smith, G. F. The
minimum significant ratio: a statistical parameter to characterize the
reproducibility of potency estimates from concentration-response
assays and estimation by replicate-experiment studies. J. Biomol.
Screening 2006, 11, 253–261.
JM8014525