THE BIOLOGY OF CHEMICAL SPACE
BY JOHN WATSON, PH.D., TRACY WORZELLA, M.S., AND BRAD LARSON, B.A., PROMEGA CORPORATION
Introduction
enzymes, leading to nearly universal screening of the majormembers of this drug-metabolizing enzyme family.
The sequencing of the human genome was presented to thepublic as a watershed moment in science that would have a
Initial in vitro assays were designed to understand the ADME
significant impact on human health. The challenge now is
(Adsorption, Distribution, Metabolism, Elimination) properties
leveraging our knowledge of the linear DNA sequence to
of compounds, including their impact on cytochrome P450
understand the complex world of human biology. Small-
enzymes (1). More recently, assays for understanding the
molecule drug discovery companies have been at the forefront
in vitro toxicity profile of lead compounds have been
in using molecular methods to tease out relationships between
integrated into most secondary screening programs (2). In
specific biomolecular targets and their roles in disease
addition to assays designed to test a chemical’s effect on a
processes. Those efforts have evolved over the last decade
generic target, in vitro methods are also used to understand
to the point where in vitro bioassays are routinely used to
the selectivity of a compound within a particular target class
prioritize new chemical entities prior to animal testing. The
(3). For example, kinase profiling has found significant utility
intersection between selection and optimization of chemical
for understanding the potential off-target effects for this
leads and in vitro biology is helping pharmaceutical scientists
important group of therapeutics. Importantly, recent results
better understand the biology of chemical space.
suggest that the therapeutic value of some kinase inhibitorsis due to inhibition of multiple kinases (4). These recent
A common misconception is that the biological effect of a
results highlight the importance of screening a large number
particular drug is mediated by interaction with a single
of targets early in the drug discovery process.
molecular target. The reality is that virtually all drugs havecomplex interactions with a large number of biomolecules
Chemical Structure and Biological Targets
within the human body. In aggregate, these interactionsdetermine the efficacy and adverse effects of a particular
Efforts are now underway to quantify the relationship between
compound. However, testing all potential drug leads on
certain chemical classes and their effect on in vitro biology.
human subjects to reveal these effects is not possible.
Most large pharmaceutical companies now have databases
Historically, animal models were the only surrogate systems
on how particular chemical structures affect different biologi-
available prior to human clinical trials. However, animal
cal targets. This information is driving rational drug design
testing is relatively expensive, not feasible for testing large
during lead optimization. Often, the challenge is selecting the
numbers of compounds, and raises ethical concerns. In vitro
best assay to include in the secondary screening program.
tools provide a useful alternative for understanding the bio-
Two major factors that affect the decision-making process
logical properties of potential pharmaceutical compounds.
are assay development costs and the correlation of assayresults with the in vivo biology of the compound. While the
Developing In Vitro Assays for HTS
utility of any individual assay is difficult to quantify, the con-
The use of in vitro assays for large-scale screening of
sensus of most drug discovery biologists is that the pattern
chemical libraries expanded significantly in the 1990s.
seen when comparing large numbers of related assays run
Initial work focused on high-throughput primary screening of
on a number of compounds from a particular chemical
pharmaceutical companies’ proprietary compound libraries.
series shows significant correlation across the series. By
CHEMICAL
The number of leads generated after initial screens had
comparing in vitro data with in vivo data generated during
to be pared down before animal testing. This led to the
clinical studies for compounds in a given series, pharmaceu-
development of secondary screening assays designed to
tical companies can select structures that are more likely to
select compounds with the most drug-like characteristics.
be successful in patients. Table 1 shows a profiling screenusing two steroids and two calcium channel blockers, which
A number of highly publicized adverse drug incidents have
PROFILING
were compared in ten different bioassays. In general, similar
also influenced the use of in vitro assays. In particular, the
compound classes show comparable profiles. The real value
failure of the antihistamines terfenadine and astemizole due
of these kinds of data comes from comparing a large
to inhibition of the hERG channel has driven pharmaceutical
number of structurally related compound groups using a
companies to assay for inhibition of this ion channel earlier
suite of related assays, such as multiple cytotoxicity assays.
in the drug-screening process. Other drugs such as cimeti-
The result is a “bar code” for a particular compound and
dine (H2 blocker) and mibefradil (calcium channel blocker)
compound class. These data are similar to those seen with
have struggled due to their interaction with cytochrome P450
The Biology of Chemical Space Kinase-Glo® Plus Assay PROFILING P450-Glo™ Assay CYP2C19* CHEMICAL Pgp-Glo™ Assay CellTiter-Glo® Assay Caspase-Glo® 3/7 Assay Table 1. Inhibition (Green) correlates with chemical structure. Red = noninhibitory or stimulatory. Kinase-Glo® Plus Assay was run using the Poly E4Y substrate (Sigma Cat.# P0275) for Src kinase or the Kemptide substrate (Cat.# V5601) for PKA. In the Src kinase assay the IC50 was greater than 100µM for nicardipine and nifedipine; the IC50 for dexamethasone was 2.67µM and for progesterone was 6.48µM. For the PKA assay, the IC50 values for nicardipine, nifedipine and dexamethasone were 6.14, 0.18, and 0.29µM, respectively. The IC50 for progesterone was greater than 100µM. CYP450 assays were performed using the P450-Glo™ Assay according to the protocols outlined in Technical Manual #TB340. For the CYP2C19 assay, the IC50 values of nicardipine, nifedipine and progesterone were 0.65, 1.3, and 1.6µM, respectively. Dexamethasone had no effect on CYP2C19 under the conditions assayed. For CYP2D6, nicardipine had an IC50 of 1.8µM, while nifedipine and progesterone had IC50 values greater than 100µM. Dexamethasone did not appear to affect CYP2D6 activity. For CYP1A2, the IC50 values of nicardipine and nifedipine were 6.3 and 0.02µM, respectively. Dexamethasone had an IC50 of 38.66µM, and progesterone had an IC50 of 17.77µM. We used a double- stable HEK293 cell line for the CellTiter-Glo® and Caspase-Glo® 3/7 Assays. Five thousand cells/well were plated, and the drugs were incubated for 4 hours before the assays were run. *P450-Glo™ Assays for CYP2C19 and CYP2D6 are currently in development.
nucleic acid microarrays where any given data point has less
generated by the Molecular Library Screening Center Network
value than the aggregate of overlapping oligo hybridizations.
(MLSCN; see the interview with Jim Inglese, Deputy Director,
The overall profile often provides useful information for
of the NIH Chemical Genomics Center (NCGC), one of the ten
predicting the in vivo behavior of a compound.
centers of the MLSCN, on page 22). Publication of the resultsfrom multiple assays using the MLSCN compound library
One of the frustrations for academic pharmacologists has
should help develop biological profiles useful for under-
been the lack of accessiblity to the proprietary in vitro assay
standing complex biological questions and validation of new
databases held by drug discovery companies. This issue will
drug targets. This public NIH Roadmap Initiative will help to
likely be alleviated by the development of an annotated public
fulfill the promise of the the human genome project. ■
database (PubChem™) containing in vitro assay results
References
Caspase-Glo, CellTiter-Glo and Kinase-Glo are registered trademarks of Promega Corporation.
1. Zlokarnick, G., Grootenhis, P.D.J. and Watson, J.B. (in press)
P450-Glo and Pgp-Glo are trademarks of Promega Corporation.
PubChem is a trademark of The National Library of Medicine.
Products may be covered by pending or issued patents or may have certain limitations. Please
2. Riss, T.L. and Moravec, R.A. (2004) Assay and Drug Dev.
visit our Web site for more information.
Technol. 2, 51–62.
3. Godl, K. et al. (2003) Proc. Nat. Acad. Sci. USA 100, 15434–9.
4. Editorial (2005) Nat. Biotechnol. 23, 267.
Purim Gifts to the Poor By Rabbi David Silverberg Toward the end of Maimonides’ presentation of the laws regarding Purim It is preferable for a person to increase his gifts to the poor than to increase his meal and sending packages to his friends. For there is no greater and more exalted joy than bringing joy to the hearts of the needy, orphans, widows and foreigners – for one who brings jo